[Journal of Astrophysics and Astronomy] AstroSat - Five Years in Orbit (Volume 42, August 2021)

Citation preview

J. Astrophys. Astr. (2021) https://doi.org/10.1007/s12036-021-09745-z

© Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:15 https://doi.org/10.1007/s12036-021-09753-z

Ó Indian Academy of Sciences Sadhana(0123456789().,-volV)FT3 ](0123456789().,-volV)

Editorial

It gives us great pleasure to present this Special Issue on ‘‘AstroSat: Five Years in Orbit’’. The first Indian multi-wavelength space observatory, AstroSat, has been the realisation of the dream of many scientists and engineers. This issue is a compendium of articles on the recent science results based on the observations made with this observatory, while also providing a glimpse of the methods adopted to operate the satellite. AstroSat, the first dedicated astronomy mission of the Indian Space Research Organisation (ISRO), was launched from the Satish Dhawan Space Centre, Sriharikota, India on 28 September, 2015, by the PSLVC30 launch vehicle into a 650-km circular orbit with an inclination of 6 degree. It is the first satellite to combine both NUV and FUV capabilities along with broad spectral coverage in X-rays. It has many more firsts in the Indian context—the first time several major Indian scientific institutes and international partners have contributed to the design, development, testing and qualification of the payloads in addition to conducting scientific research with the observations, the first Indian satellite to have a payload mass fraction greater than 50%, the first time the PSLV was used to launch a 1500 kg class satellite in a nearequatorial orbit, and the first Indian spacecraft operated as a proposal-driven space observatory. AstroSat was in performance verification mode for the first six months in orbit, followed by one year of observations for Guaranteed Time observers, primarily from the instrument developing teams. After this the observatory time was gradually opened up for researchers from India and abroad, and with the completion of five years in orbit, there have been more than 150 refereed publications using observations from AstroSat. On the completion of five years of AstroSat, the Journal of Astrophysics and Astronomy circulated an invitation to contribute articles to mark the occasion.

It was wonderful to receive a very enthusiastic response to this call, which has enabled us to assemble this bouquet of over sixty peer-reviewed articles addressing various aspects of the mission and its science outcome. This issue begins with an article tracing the history of initiation of the idea of AstroSat, and a brief overview of some of the results from the first five years of operation. This is followed by four review articles providing an overview of the unique aspects and the major science achievements of the LAXPC, UVIT and SXT instruments. A series of nearly twenty articles that follow provide a broad perspective of various aspects of the project implementation and new developments that were undertaken to operate this spacecraft as an observatory, present several new calibration results including that of the UV grating and describe a variety of data analysis pipelines. Wide-ranging original science results are presented in the subsequent forty articles or so. The largest group of papers is devoted to UV studies of galaxies in different environments, their structure and star formation—including the Milky Way, its satellites and the neighbouring Andromeda galaxy. Several papers also report multi-wavelength studies of Active Galaxies, providing new constraints on their central engines. Stellar populations in different open and globular clusters are studied in a number of contributions, revealing their evolutionary history. Fascinating new details of Planetary Nebulae, and interesting results from the studies of individual stars are also presented. Fast, broadband X-ray timing capability is a key strength of AstroSat and several articles in this issue employ this to study diverse aspects of stellar mass compact objects—white dwarfs, neutron stars and black holes—in binary systems. Some of the articles also discuss the detection and detailed study of fast high energy transient sources with AstroSat.

15

Page 2 of 2

The success of AstroSat has contributed to a major growth in the Indian Space Astronomy community. The resulting maturity and confidence has generated a slew of ideas for future Indian astronomy missions. This volume concludes with a discussion of the exciting prospects of future growth in this area. Finally, words are insufficient to acknowledge the help we have received from a large number of contributors, without whom this issue would not have been possible. We are extremely thankful to Chairman ISRO, Dr. K. Sivan, for providing the Foreword. We also express our sincere thanks to the Chief Editor of Journal of Astrophysics and Astronomy, Prof. Annapurni Subramaniam, who has been instrumental in not only guiding us through, but also contributing significantly to the editorial activities of this issue. We would also like to take this opportunity to express our

J. Astrophys. Astr.

(2021) 42:15

heartfelt gratitude to all the contributors for enthusiastically agreeing to write for this issue despite their other commitments. We are also grateful to all the team members of ISRO, and the various institutions, who have made AstroSat what it is today. Special thanks are due to the contributors for the cover page, and to Ms. Shylaja, Ms. Cicilia and Ms Srimathi and the entire editorial team at the Indian Academy of Sciences for their invaluable help in bringing out this issue. We believe this is only the beginning. We look forward to many more interesting results from AstroSat in the years to come. S. Seetha D. Bhattacharya Guest Editors

 Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:19 https://doi.org/10.1007/s12036-021-09692-9

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

REVIEW

AstroSat: Concept to achievements S. SEETHA1,* and K. KASTURIRANGAN2 1

Raman Research Institute, C.V. Raman Avenue, Bengaluru 560 080, India. ISRO Headquarters, Antariksh Bhavan, New BEL Road, Bengaluru 560 231, India. *Corresponding Author. E-mail: [email protected] 2

MS received 16 November 2020; accepted 1 January 2021 Abstract. AstroSat has completed 5 years of successful in-orbit operation on 28 September 2020. AstroSat is ISRO’s first Indian multi-wavelength satellite operating as a space Observatory. It is the only satellite which can simultaneously observe in the Far UV and a wide X-ray band from 0.3 to 80 keV using different instruments. This astronomy mission was conceived, following the success of several piggy back astronomy experiments flown earlier on Indian satellites. AstroSat is the result of collaboration between ISRO and several astronomy institutions within India and abroad. There are over 150 refereed publications resulting from data from AstroSat, in addition to Astronomy Telegrams, Circulars and Conference proceedings. This paper provides a brief summary of the evolution of the concept of AstroSat, how it was realized and scientific outcome from this mission. Keywords. AstroSat—space mission—accretion—X-ray binaries—stars—stellar clusters—galaxies.

1. Introduction AstroSat is ISRO’s first Indian multi-wavelength astronomy satellite being operated as a space observatory. It has completed the design life of five years in orbit on 28 September 2020 and holds promise for further successful operations leading to significant science for at least 5 more years, if not more. It is a satellite having multiple instruments covering both the UV and the X-ray wavebands. This paper provides a brief summary of the environs under which AstroSat was conceptualized, the mode in which it was developed, tested, and is being operated. This paper also gives a glimpse of some of the significant science results from AstroSat.

1.1 Historical perspective India’s space program has completed over 50 years since its inception. Space Science experiments have This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

been intertwined with this program, starting with sounding rockets for upper atmospheric studies and X-ray astronomy, moving on to Indian satellites for X-ray and Gamma ray piggyback experiments, finally leading to dedicated missions. In parallel, Tata Institute of Fundamental Research (TIFR) Mumbai has developed a balloon facility at Hyderabad dedicated for conducting science experiments with balloons, which can now be launched to heights of 42 km. The developments of satellites and launch vehicles have largely been driven by the national priority of space applications for development. However the recognition that these could in turn provide opportunities for new space science experiments was inherent in the vision of the leaders of the space program. This unique approach of developing scientific capabilities without making science as the prime reason for resource investment, gave the Indian Space program a distinct character. The scientific heritage and the instrumentation capabilities in the country ensured that the space science program matured into an indispensable component of the Indian Space Program.

19

Page 2 of 19

1.2 Heritage Starting from the 1940s, the foundation for ‘space science’ began with the experiments to study cosmic rays. This led to development of Balloon fabrication and launch facility by TIFR at Hyderabad that has the capability of carrying a payload of up to several hundred kg to an altitude of 30 to 40 km. This enabled Indian scientists of TIFR and Physical Research Laboratory (PRL) to develop instruments for studies of hard X-rays ([20 keV) from bright X-ray sources. A large number of Balloon experiments for X-ray astronomy studies were conducted from Hyderabad during 1967-1980 resulting in several new and interesting results. In parallel the PRL group initially and later the TIFR group developed instruments for the rocket borne X-ray experiments launched from ISRO’s launch facilities at Thumba and Shriharikota in late sixties and up to early eighties for studies of cosmic sources in *1–10 keV energy range. These provided invaluable training and experience to the Indian scientists for the design and fabrication of the satellite-borne instruments. Meanwhile, scientists at the PRL, Ahmedabad also conducted research on similar areas of study with emphasis on the solar activity influence on cosmic rays. Further X-ray astronomy experiments from PRL were flown on sounding rockets. Both the teams at TIFR and PRL, also placed a lot of stress on development of instrumentation for these experiments. It is worth pointing out that both the rocket facility at Thumba and the balloon facility at Hyderabad had specific locational advantage for conducting X-ray astronomy studies owing to the low background resulting from the low latitude cut off of the cosmic ray intensity and therefore the reduction of secondary background. The ground was therefore well prepared when the opportunity arose to fly instruments on board the first Indian satellite ‘Aryabhata’, and both these groups at PRL and TIFR contributed to two of the astronomy instruments on this satellite. As the newly created Indian Space Research Organisation (ISRO) gathered momentum, indigenous experimental rockets and satellites provided opportunities for several ‘‘piggy back experiments’’. The first of these was the Gamma Ray Burst (GRB) experiment developed at ISRO Satellite Centre (ISAC, now U.R. Rao Satellite Centre (URSC)) flown on the SROSS series of satellites starting in 1987, with a capability of carrying a payload of 5–10 kg. The GRB experiment onboard the SROSS-C2 satellite (Kasturirangan et al. 1997) recorded over 50 GRBs, of high

J. Astrophys. Astr.

(2021) 42:19

signal to noise ratio (Sinha et al. 2001). Gamma ray bursts being transient sources with little prior knowledge of the direction, required large field of view instruments to detect the bursts wherever they occurred in the sky. Some of the bursts detected by this experiment were useful for triangulation with other international satellites for localization of the sources. Although it turned out that this effort was superseded by the launch of the large Compton Gamma Ray Observatory (CGRO) by NASA, it must be said our experiment was an unqualified success. Encouraged by the success of the GRB instrument, the TIFR and ISAC groups submitted a joint proposal to ISRO for an X-ray Astronomy experiment weighing about 50 kg for launch on an Indian satellite. It was aimed at studies of timing and spectral characteristics of X-ray binaries. This also required inertial pointing of the satellite to specific sources in the sky. This demanded provision for higher mass and power on the satellite. The development of the Polar Satellite Launch Vehicle (PSLV) by ISRO primarily for remote sensing satellites, enabled the launch of satellites with mass of about 1000 kg into a polar sun synchronous orbit. When the PSLV became operational, ISRO agreed to accommodate the proposed Indian X-ray Astronomy Experiment (IXAE) to be flown on the Remote sensing satellite IRS-P3. Four argon-filled proportional counters (area *1600 sq cm) and the front end electronics was designed and provided by the TIFR group and the ISAC group designed and supplied the signal processing electronics (Agrawal et al. 1997). Developed in a record time of 18 months, this instrument named as Indian X-ray Astronomy Experiment (IXAE), was launched on March 21, 1996 in a polar orbit by the PSLV. Using a star sensor the IXAE detectors were pointed towards the sky for about 2 months every year to study specific sources. The IXAE performed well for about 5 years and studied about 20 X-ray sources. This experiment produced many new results especially on a new transient black hole binary GRS 1915?105 (Paul et al. 1998). The success of the IXAE was a major milestone in the evolution of the idea of AstroSat as it gave confidence that Indian scientists can make the gas filled detectors that can work in space (Agrawal 2017). One of the scientists associated with the IXAE proposed to ISRO a dedicated Indian X-ray astronomy satellite with a more complex and heavier X-ray experiment which will lead to internationally competitive science. ISRO suggested expanding the scope by convening a meeting of Indian astronomers where this and other proposals were further deliberated, for a dedicated astronomy mission.

J. Astrophys. Astr.

(2021) 42:19

2. AstroSat This is the golden age for multi-wavelength astronomy. Ground-based optical telescopes have extended to tens of meters in size, radio astronomy telescopes extending their capabilities to long wavelengths, huge arrays and improving the spatial resolution with interferometry, new facilities like mm wave astronomy making a mark and a large number of space astronomy satellites covering almost the entire wavelength range from optical to high energy gamma rays. These have been made possible by the aspirations of the scientists world over to continuously improve on the capabilities and broaden and sharpen the knowledge base already achieved. The success of the GRB and the IXAE propelled a series of meetings of the Indian astronomical community to explore the possibility of a major Indian astronomical observatory in space. By mid 1990s, NASA’s Hubble Space Telescope and CGRO were already operational in orbit and the Chandra X-ray mission and ESA’s XMM-Newton were well into final stages of development. Even more ambitious missions were being funded or under planning world over. Given this international scenario, and the proven capabilities and resources within India, the critical question was to ask ‘‘What should be the nature and scope of the Indian mission?’’ Traditionally, the quest for higher sensitivity and higher angular resolution has been the motivating factor for newer missions. It became clear from the discussions that this is not the only goal we could aim for. Indian astronomers worked towards defining a mission that could, while supplementing other missions, have its own niche objectives which would encompass the research interests of the scientific community and technical capabilities of the engineering teams within the country. The final consensus was a multiwavelength astronomical observatory, based on the following reasoning. To understand the nature of cosmic sources, their radiation processes and environment, it is necessary to measure their emission over a wide range of the electromagnetic spectrum. Since intensity of several classes of cosmic sources varies with time, it is necessary to make simultaneous observations in different wavebands. Most of the space observatories are dedicated to a particular waveband, e.g. X-ray, UV etc. Consequently, multiwavelength studies usually have to be made from coordinated observations with different satellites.

Page 3 of 19

19

Often there are logistic problems in making simultaneous and coordinated studies of a specific object from different satellites and ground based telescopes. The most efficient and effective way to pursue multiwavelength studies is to have a dedicated satellite mission which will carry several instruments covering the desired spectral bands so that simultaneous observations in all the desired wavebands can be made from the same satellite. There are however, different observational constraints for instruments operating in different wavebands and therefore it was also decided that the scientific objectives could include some specific aims which could also be realized using individual instruments/wavebands. 2.1 Objectives The AstroSat was therefore proposed as a multiwavelength astronomy mission with wide spectral coverage extending over Near and Far Ultraviolet (NUV, FUV), soft and hard X-ray bands (Agrawal 2006; Rao et al. 2009). It has provided an opportunity to astronomers to carry out cutting edge research in the frontier areas of X-ray astronomy and Ultraviolet astronomy and allow them to address some of the outstanding problems in high energy astrophysics. The scientific objectives of AstroSat are the following: (1) Understand high-energy emission processes in various astrophysical systems: Multiwavelength studies of various cosmic sources over a wide spectral band extending over visible, UV and X-ray bands. (2) Correlated time variations of intensity in UV, soft and hard X-ray bands to investigate the origin and mechanism of the emission of radiation in different wave bands. (3) Understand variability timescale in various astrophysical systems through broad-band and longduration observations: studies of periodic (pulsations, binary light curves, QPOs etc.) and aperiodic (flaring activity, bursts, flickering and other chaotic variations) variability. (4) Search for black hole sources by limited surveys in galactic plane; detection and detailed studies of stellar-mass black holes. (5) Study of non-thermal emission in supernova remnants and galaxy clusters.

19

Page 4 of 19

(6) Study magnetic fields in strongly magnetized systems: Measuring magnetic fields of neutron stars by detection and studies of cyclotron lines in the X-ray spectra of X-ray pulsars. (7) Detection of X-ray transients and long duration temporal studies. (8) High-resolution UV studies of stars, emission nebulae and galaxies: study of UV emission from hot stars (WD, CV, WR, LBV, b-Cephei, etc.) in galaxies, morphological studies of nebulae and supernova remnants and nearby galaxies. (9) Limited sky survey in UV: multi-band limited sky survey in ultra-violet in 130–300 nm band. To accomplish the above scientific objectives, the instrument configuration was flown as a combination of four types of X-ray detectors and twin telescopes covering visible, NUV and FUV bands (Singh et al. 2014; Agrawal 2017). The instrumentation flown had the following features: • Large Area X-ray Proportional Counters (LAXPC) (3 nos) to cover the energy range 3–80 keV, and to have an effective area of *6000 cm2 at 15 keV and a time resolution of 10 microsecond (Agrawal et al. 2017; Yadav et al. 2017). • Twin Ritchey Chretian UV Imaging Telescope (UVIT) covering the wavelength band of 130–180 nm and 200–300 nm (using several filters) and having an angular resolution better than 1.8 arcsec (Tandon et al. 2017). • Soft X-ray Telescope (SXT) covering the energy range of 0.3–8 keV with an angular resolution of 3–4 arcmin (Singh et al. 2017). • Pixellated Cadmium Zinc Telluride Imager (CZTI) operating in the energy range 20–100 keV with an area of 1000 sq cm, with a collimator and coded mask. Above 100 keV, it operates as a detector without collimation (Bhalerao et al. 2017a, b). • Scanning Sky Monitor (SSM) operating in the X-ray range of 2.5–10 keV, with a one-dimensional coded mask (Ramadevi et al. 2017). • A Charge Particle Monitor (CPM) capable of detecting protons [1 MeV (Rao et al. 2017). Design and fabrication of three of the X-ray astronomy instruments namely LAXPC, SXT and CZTI and the CPM were the responsibility of TIFR, and the responsibility for UVIT was with the Indian Institute of Astrophysics, (IIA) and the SSM instrument by ISAC. Figure 1 is a collage of the flight

J. Astrophys. Astr.

(2021) 42:19

models of the main five payloads and in the center is the picture of assembled satellite on ground. The realization of the above instruments was made possible with the involvement of many institutions in addition to various Centres of Indian Space Research Organisation (ISRO). They are Tata Institute of Fundamental Research (TIFR), Mumbai, Indian Institute of Astrophysics, (IIA) Bengaluru, Inter-University Centre for Astronomy and Astrophysics, (IUCAA) Pune, Raman Research Institute, (RRI) Bengaluru, Physical Research Laboratory, (PRL) Ahmedabad along with a collaboration with Canadian Space Agency for the detectors and electronics of UVIT and with University of Leicester, UK for the detectors of SXT. There were several challenges in realizing the payloads and the satellite, a few of them being: • Achieving an overall angular angular resolution of \1.8 arc-sec for the UVIT telescope with a field-of-view of 28 arcmin, with indigenous development of UV mirrors for the first time. • Development and testing of indigenous goldcoated foil optics for soft X-ray telescope, and ensure an aligned assembly of these mirrors. • Development of large high pressure gas filled counters with effective area *6000 cm2 with energy measurement up to 80 keV, and a temporal resolution of 10 microsecond. A major challenge in the LAXPC instrument was also to develop a light weight multi-layer collimator yet opaque to X-rays up to 80 keV. • Qualification of position sensitive gas-filled counters and rotation mechanism for SSM. • Qualifying of commercial Cadmium Zinc Telluride detectors for astronomy purposes, and demonstrating the polarisation capabilities of CZTI. • Contamination control of UVIT optics. • Development of doors for the UVIT and SXT, and deployment mechanism. • Capability for large onboard data storage with read/write capability. • Inter-payload alignment and measurement on ground and measurement after launch using celestial sources. • S/C maneuvering with avoidance of the Sun along two axes, and S/C pointing and stability. • Enabling data storage in photon counting mode for the payloads. • Background simulations, estimate and modeling for the various detectors.

J. Astrophys. Astr.

(2021) 42:19

Page 5 of 19

19

Figure 1. The five main payloads of AstroSat, and the assembled satellite in the centre, before launch. Image Courtesy: Payload and project teams and S. Megala.

It may be noted here that the Neil Gehrels Swift Observatory and XMM-Newton do have capability to observe in NUV bands in addition to Xray band. However they do not have FUV capabilities. On the other hand, the Hubble Space Telescope has extremely good resolution in Far UV, but over a narrow field-of-view. The advantage of UVIT is it can observe faint objects with ~1.5 arcsec resolution with an FoV of nearly 0.5 degree, ideal for imaging galaxies. UVIT also has about 3 times better angular resolution than the earlier GALEX mission. LAXPC with its large area of 3 counters enabled an effective area of LAXPC is ~6000 sq. cm at 15 keV (Antia et al. 2017). Above 30 keV, the effective area of the 3 LAXPCs together was four to five times greater than that of the Proportional Counter Array (PCA) on Rossi X-ray timing Explorer (RXTE).

2.2 AstroSat mission management AstroSat with a mass of 1513 kg and a payload mass of 855 kg (Navalgund et al. 2017) was launched on 28 September 2015 by the Polar Satellite Launch Vehicle, PSLV-XL C30, into an orbit with inclination of 6 degree and an altitude of 650 km. This was

the first time the payload mass was [50% of the satellite mass on an Indian satellite launched by PSLV. The inclination of the orbit was chosen to avoid the satellite traversing through the inner portions of the South Atlantic Anomaly region with high count rates of charged particles. The altitude was chosen to minimise the effect of atomic oxygen on the UVIT optics. This was also the first time PSLV launched a satellite in a low Earth orbit at a near equatorial inclination. Four of the payloads namely the LAXPC, SXT, UVIT and CZTI have their view direction aligned to the ?ve roll axis of the satellite (see Fig. 2). These payloads therefore are pointed towards the same source for observation. The sky monitor is pointed in a perpendicular direction and is rotated in a step and stare mode to scan as much of the sky in a rotation. For further details of pointing and maintenance of orbit, see Seetha and Megala (2017) and other papers in this volume. The satellite was operated for the first six months for performance verification of the payloads, and the next six months was guaranteed time (GT) for the instrument teams (Pandiyan et al. 2017). In ensuing years, the GT gradually reduced and the observing time was gradually made open to both Indian and later international astronomers too.

19

Page 6 of 19

J. Astrophys. Astr.

(2021) 42:19

Figure 2. A drawing showing the post launch configuration of the payloads, with satellite axes reference. Image Courtesy: Project team.

2.3 AstroSat as an Observatory AstroSat is operated as the first Indian Space Observatory. Observations are proposal driven. The AstroSat Proposal Processing System (APPS) is a web based software facility, developed through IUCAA, Pune, and hosted and administered by ISRO Space Science Data Centre (ISSDC), Bangalore. All proposal related activities like proposal submission, review, approval etc. are done through APPS. The call for proposals for observations are made through an Announcement of Opportunity (AO) typically issued in March of each year for an year’s observation cycle starting from October. Announcement for various observation cycles are made by ISRO HQ in coordination with ISSDC and the Science Working Group (SWG). Proposals can be submitted by any scientist capable of utilising the data, with 55% of the available time being allotted for proposals from India. In addition to proposals for AO, proposals can be submitted anytime for Targets of Opportunity (TOO), (in case of any sudden outburst or state change to be observed in specific sources) and Calibration proposals (Cal). Submitted proposals are reviewed and approved through the APPS utility, by the AstroSat Time Allocation Committee (ATAC) for AO proposals, TOO committee for TOO proposals and Calibration Scientist for the Cal. Proposals. These committees are supported by recommendations by the AstroSat Technical Committee (ATC), which considers payload operation constraints like sun angle, avoidance of

bright object, and sensitivity in the various instruments etc. Based on the approvals, the mission and operations team at U.R. Rao Satellite Centre, and ISRO Telemetry, Tracking and Command Network (ISTRAC) create the necessary list of targets to be observed, and generate command files, taking into consideration Sun avoidance, RAM angle constraints, data readout capability and thermal constraints. The operations of the payload and spacecraft are commanded at ISTRAC. Data is received and data products are again generated at ISSDC, and verified through Payload Operation Centres (POCs) and further made available for dissemination to the respective proposal Principal Investigators (PI). The data of observation under AO proposals has a lock-in period of twelve months with proposal PIs, after which it is made available to the public as archival data. The TOO data do not have any lock-in period and is released to public, along with proposer. The observations of TOOs have to be scheduled by sliding/ shifting the pre-planned observations. Both proprietary data (during lock-in) and archival data are hosted on https://astrobrowse.issdc.gov.in maintained by ISSDC at ISTRAC. Anybody can download data through this website after registering in APPS. The AstroSat calibration database (CALDB) and other tools/software required for analysing the AstroSat data are hosted/linked at the AstroSat support cell (ASSC, http://astrosat-ssc.iucaa.in/) hosted by IUCAA Pune. ASSC team also provides assistance to proposers and conducts workshops for training users for

J. Astrophys. Astr.

(2021) 42:19

proposal preparation and data analysis. ASSC is supported by ISRO. All developments of necessary tools for operating the AstroSat as an observatory has been done for the first time in India for a space mission, and modified versions could be used for future astronomy satellites. In the last five years, over 150 refereed publications, have resulted based on observations from AstroSat.

3. Scientific outcome In this section, we provide a brief overview of the scientific results from AstroSat. This is to highlight the uniqueness of AstroSat and is by no means exhaustive. The references provided are mainly those resulting from AstroSat observations, and therefore we request the reader to consider pursuing the paper with the references therein for a better understanding of the scientific contributions on the topic.

3.1 Multiwavelength observations 3.1.1 Her X-1. Her X-1/HZ Her an accretion powered pulsar is an ideal target for multiwavelength observations, with optical radiation from the companion star, X-ray radiation from the inner parts of the disc and the accretion onto a magnetised neutron star and UV/EUV radiation from the disc and the X-ray irradiated portion of the optical companion viewable due to the very conducive edge on view of this binary system with an inclination of about 85 degree. In addition to the spin and orbital variations, the X-ray data of this binary exhibits a 35 day cycle with two bright states referred to as main ON, and short ON with OFF or low states in between. This variation is explained in terms of a precessing accretion disc. Spectral Observations using SXT on AstroSat at Low and High state of Her X-1 is consistent with earlier observations and a precessing disc model (Leahy & Chen 2019). FUV emission from this source indicates a variability starting near orbital phase of 0.3. FUV emission is estimated from two components after ruling out other contributions (Leahy et al. 2020): (a) due to the disc emission and (b) due to emission from heated portion of the companion star HZ Her. The geometry of the system is modeled by including FUV data. The current FUV observations

Page 7 of 19

19

are commensurate with a thin disc which is tilted and twisted and does not require a thick inner disc, which was used to explain X-ray data. However since the emission mechanisms have to match both the orbital phase and the X-ray emissions over the complex 35 day cycle, more detailed observations of both UV and X-rays over the different phases of the 35 day cycle may be required to further refine the model. Closer to the neutron star, the X-ray observations of Her X-1 exhibit a cyclotron resonance scattering feature (CRSF, or referred as cyclotron line) produced by resonant scattering of photons off electrons moving perpendicular to the magnetic field, as accreted material flows along the magnetic field lines onto the polar cap of the neutron star. In the case of Her X-1, the cyclotron feature has been observed around 35 keV. The cyclotron peak energy is positively correlated with X-ray luminosity of the source which is explained in terms of reduction of emitting region with increasing luminosity/accretion rate (Becker et al. 2012). The time variation of the cyclotron feature has been studied for over three decades using data from various satellites and the peak energy of this feature is found to vary from 35 to 44 keV. It was found to be *35 keV before 1991, and a high value of 44 keV was observed in 1993 and 1996. Subsequently, a gradual decrease with time of this value, with episodes of constancy have been observed. The variation with time has been explained in terms of formation of a mound on the polar cap and changes in magnetic field due to this formation (Staubert et al. 2016 and references therein). Detailed modeling using previous and AstroSat data and considering various scenarios has been done by Bala et al. (2020). The authors fit the data with variation in both luminosity and time. They find that the best-fit corresponds to a case for which there is no time dependence beyond MJD 54487. The variation due to both luminosity and time, has been explained with the modeling of the dynamical variation of a mound over the polar cap of the neutron star. There is continuous inflow of matter leading to formation of a mound on the polar cap due to accretion along the magnetic field lines, and outflow of matter leaking through the edges (Mukherjee & Bhattacharya 2012; Mukherjee et al. 2013a, b). While the mound is gradually growing in size to few tens of meter, the magnetic field strength shifts from the centre to the edges of the mound, hence exhibiting a decrease in field strength with height of the mound. This explains decrease in cyclotron energy with increasing mound height. When the inflow and outflow balance each other there can be a constancy of the energy of the

19

Page 8 of 19

cyclotron feature. However if the mound reaches a critical height, which is estimated by the authors to be about 100 m for Her X-1, it is very likely that the instabilities in the mound can cause sudden increase in outflow, leading to a sudden collapse of the mound and increase in cyclotron energy like that observed between 1991–1993. 3.1.2 Blazar RGB J0710?591. One of the main objectives of AstroSat was to conduct simultaneous observations over a broad spectral band, to estimate the spectral energy distributions (SEDs) for different types of objects, thus enabling the identification of physical processes in these objects. Blazars are Active Galactic Nuclei (AGNs) which have jets beamed towards the observer. Blazars exhibit a double peaked SED, with the lower energy synchrotron peak in the radio to X-ray regime, and the high energy peak in the X-ray to gamma ray energy regime. The synchrotron peak in high-energy-peaked blazars (HBLs) typically lies in UV–X-ray energies. RGB J0710?591 is one such blazar which was observed with AstroSat in 2016. These observations indicate that the X-ray part of the SED is unusually curved and can be explained in terms of synchrotron emission from a non-thermal distribution of high energy electrons with a declining energy density around the peak (Goswami et al. 2020). Such electrons can be produced from a region where particles are accelerated under Fermi process and the radiative losses in acceleration site decide the maximum attainable Lorentz factor. The UV part of the spectrum appears to be of a different emission component than X-ray. When this SED is compared with the SED of the same source obtained in 2015, using data from SWIFT and NuSTAR satellites, it is found that while the UV part is only marginally different, the X-ray part of the SED is significantly different.

3.2 Stars and stellar systems UV observations are extremely sensitive to blue objects which include early type stars, blue stragglers, planetary nebulae, white dwarfs etc. 3.2.1 Globular clusters and open clusters. For long, Globular Clusters (GC) were considered to be consisting of co-evolving stars with single age and composition. In the last decade, observations have shown globular clusters could contain stars of more than one population. Figure 3 shows an image of the

J. Astrophys. Astr.

(2021) 42:19

globular cluster 5466. UVIT on AstroSat was used to explore if the blue objects in GCs showed any evidence of multiple populations. Stars corresponding to different branches have been identified in both NGC 1851 (Subramaniam et al. 2017) and NGC 2808 (Jain et al. 2019), and both show there are more than one population among the horizontal branch stars. Observations of NGC 1261 on the other hand, exhibits blue horizontal branch (BHB) stars in UV but the red horizontal branch is too faint (Rani et al. 2020). Analysis of four other globular clusters NGC 4147, NGC 4590, NGC 5053 and NGC 7492, yield *150 blue horizontal branch stars (BHBs), and *40 blue straggler stars (Kumar et al. 2019). The temperature distribution of the BHBs is bi-modal, indicative of two groups. Blue straggler stars (BSS) are stars which are bluer than the other stars and have evolved off the main sequence in a stellar cluster. They have often been referred to as rejuvenated stars having gained mass through some process. NGC 188 is an old (7 Gyr) open cluster with 20 identified BSS. One of the other candidates WOCS5886, also a member of NGC 188 was observed in both NUV and FUV using multiple filters of UVIT on AstroSat. The UV observations indicate the presence of excess flux in the UV region as compared to a spectral fit for a single star. Based on a spectral energy distribution fitting with these observations combined with those from other satellites/observatories, WOCS5886 a member of NGC 188 is now identified (Subramaniam et al. 2016) to be a binary consisting of BSS and a hotter, post AGB/HB star with temperature of the BSS being 6000 K and the temperature of the companion being 17000 K, clearly indicating mass transfer as the likely mode for formation of BSS. This system is the first of its kind to be found in an open cluster. The next step was whether BSS-WD systems could be identified, with the expected luminosity of WD being lower. Detection of white dwarfs with masses \0.4M indicates evolution in binary systems, as single star evolution would take longer than the age of the Universe. Detection of WD-BSS binaries would therefore provide evidence for mass transfer from the pre-WD to the BSS, thus providing a clue to the formation of both the individual stars. UV observations with AstroSat have resulted in the detection of WD-BSS binaries in open cluster M67 (Sindhu et al. 2019; Jadhav et al. 2019) and globular cluster 5446 (Sahu et al. 2019).

J. Astrophys. Astr.

(2021) 42:19

Page 9 of 19

19

Figure 3. Image of the field containing globular cluster NGC 5466 observed with UVIT. Here, blue corresponds to ˚ , while green represents the observation with the F148W filter in the far ultra-violet (FUV) at a mean wavelength of 1481 A ˚ . Also shown are observation with the N263M filter in the Near Ultra-Violet (NUV) at a mean wavelength of 2632 A (i) expanded view of the cluster (top) and (ii) expanded view of a galaxy (right) that lies within the UVIT field. The exposure times in FUV and NUV are 5990 seconds and 8243 seconds respectively. Image Courtesy: UVIT, POC, IIA. For other AstroSat images and details, see outreach page of the Astronomical Society of India, https://astron-soc.in/outreach/apom/.

In the globular cluster 5446, NH84 has been identified to be a WD-BSS binary, the second such system in globular clusters, the first being in 47 Tuc. Thirty stars have been studied in open cluster M67 (Jadhav et al. 2019), of which two have been identified as WD?BSS systems (WOCS1007, WOCS 2007), one as WD?yellow Giant and two WD?Main sequence binaries (see also Subramaniam et al. 2020). The BSS and the YG systems have extremely low mass WD (with mass\0.3M) as companion, clearly indicating their donor status. Figure 4 is a sketch of likely scenario of how BSS-WD binaries might have formed and evolve. The other systems studied include close binaries which could be progenitors of WD?BSS systems, and single WDs with masses [0.5M, which could in turn have formed from BSS as progenitors.

3.2.2 Planetary nebulae. NGC 40 or bow-tie nebula is a planetary nebula with a Wolf-Rayet star at its centre. For the first time a Far UV halo has been detected in the F169M sapphire filter of UVIT (Rao et al. 2018a). The FUV halo is most likely due to UV fluorescence emission from the Lyman bands of H2 molecules, and trails the optical and IR halos, in the direction opposite to the central star’s proper motion. NGC 6302 is another planetary nebula in which lobes are observed in the same FUV band (Rao et al. 2018b). The lobes are seen extending to 5 arcmin from the central hot and luminous bright object. In addition jets extending perpendicular to the line joining the lobes have also been observed. 3.2.3 Flare stars. The habitability of a planet not only depends on its location from the central star but

19

Page 10 of 19

J. Astrophys. Astr.

(2021) 42:19

Figure 4. A sketch of the likely scenarios of formation of blue Stragglers, by the mass transfer pathway or the merger pathway due to collisions or binary mergers. The AstroSat findings of several white dwarf binaries including the extremely low mass white dwarfs in a binary system with blue stragglers suggest the mass transfer pathway for low density environments. Image courtesy: Snehalata Sahu.

also the environment it is exposed to. Activity on the central star can lead to mass loss from the planet’s atmosphere. Proxima Centauri, the closest star to the solar system and also host to an Earth-like planet was observed by AstroSat, Chandra and HST to study the activity phase of this star. Flares have been observed in both X-ray and FUV observations from this star during 2017 (Lalitha et al. 2020). It is found that UV emission peak precedes the X-ray peak by 300–400 s. This could be due to the Neupert effect similar to that observed during some of the solar flares.

3.3 Cataclysmic variables Cataclysmic Variables (CVs) are binaries with a white dwarf (WD) accreting matter from its companion, usually via the formation of an accretion disc, unless the magnetic field of the WD is sufficient to disrupt it. Novae form a subset of CVs. The accreted matter in novae increases in mass leading to explosive thermonuclear burning resulting in a nova outburst. Repeated outbursts are observed in Recurrent Novae (RNe), which could be progenitors for Type Ia supernovae. Of the confirmed RNe in our galaxy, four have red giants as companions similar to symbiotic stars and are referred to as Symbiotic Recurrent Novae. V3890 Sgr is one such symbiotic RNe which

has so far exhibited two outbursts with a recurrence time of 28 years and hence it is a relatively rare event. The recent third outburst was during September 2019, observations of which were made in two slots with AstroSat (Singh et al. 2020). In the first slot between 8.1–9.9 days after outburst, the source initially exhibited a hard X-ray spectrum due to shock between nova ejecta and companion star, and on day 8.57, a super soft source (SSS) was evident in the spectra, the first view of the WD surface. The SSS was variable initially, but became brighter and stable after 8.9 day. The second slot of observation was from 15.9–19.6 d, with the source being brighter and variable, before fading away, with the SSS phase ending on day 26.18. The mass of the WD determines the duration of the SSS phase, with low mass WD having a longer phase, and indicates a high mass WD in V3890 Sgr. The source also exhibits temporary vanishing of SSS phase only re-appearing after a day, which could provide important clues regarding ejecta properties.

3.4 X-ray binaries X-ray binaries are close binaries in which a companion star transfers matter to a compact star which could be a neutron star or a black hole. The primary goal for observations with the X-ray instruments on AstroSat

J. Astrophys. Astr.

(2021) 42:19

were to obtain high time resolution data on variability and broad band spectrum in X-rays, in order to understand the physical processes underlying these phenomena. 3.4.1 Neutron star binaries. The results on neutron star binaries can be divided into three parts: (a) broad band spectrum of Atoll and Z type sources which are Low Mass X-ray Binaries (LMXBs), (b) bursting sources which are also usually from LMXBs with neutron stars with relatively low magnetic field, and (c) accretionpowered pulsars and cyclotron lines usually from high mass X-ray binaries (HMXBs) and the neutron star has a magnetic field of ~1012 Gauss. 3.4.1.1. Broad band spectrum of Z type and Atoll sources. GX5-1 a Z type source was observed with data on two branches in its hardness intensity diagram (HID). This was divided into 5 portions and analysed (Bhulla et al. 2019) The spectra could be fitted with a disc component rather than a black body component as was done in earlier studies and a thermal Comptonised component. This fit did not improve with inclusion of additional black body component. The low energy coverage with SXT, provides a better constraint on the column density. The disc flux to total flux changes monotonically from horizontal branch to normal branch, and disc flux itself is correlated to the disc flux ratio, indicating that disc flux ratio may be the component which places the source in a particular branch of the HID. Temporal analysis indicate a QPO ~50 Hz, whose frequency increases from 30 Hz to 50 Hz while its rms decreases, with disc flux ratio. On the other hand, LMC X-2 another Z type source was observed in 2016, and the source exhibited the complete Z pattern in the HID. The spectra were fitted with a Comptonization model alone and addition of a disk or a black body component, did not improve the fit (Agrawal and Nandi 2020). The authors therefore conclude that the disc if present is obscured by the Comptonisation region. Data from atoll source 4U1705-44 (only LAXPC) in the banana state shows a significant flux component >20 keV and has been fitted with three different models. The authors find that a model with thermal Comptonisation + Gaussian for a broad iron line + non-thermal power-law fits the data best (Agrawal et al. 2018). The temporal analysis of various states of the banana state, also shows peaked noise features in the 1–13 Hz range, with most observations indicating a broad feature ~10 Hz.

Page 11 of 19

19

3.4.1.2. Bursting sources. A few neutron star low mass X-ray binaries also exhibit bursts. The bursts are considered to occur due to the thermonuclear burning of the accumulated accreted matter on the neutron star. 4U 1728-34 exhibited a burst during the observations in March 2016. Burst oscillations have also been observed during the rise portion of the burst, which is consistent with earlier burst oscillations (Verdhan Chauhan et al. 2017). This source also exhibited kiloHertz QPOs at 815 Hz, with an evolution to 850 Hz over a span of 3 ks of data. These QPOs are also observed [10 keV, for the first time with AstroSat. One of the outstanding issues in modeling bursts has been whether the pre-burst persistent emission spectrum can be considered non-changing and can be subtracted as background to estimate the burst spectrum. In order to assess this the burst data of 4U 1728-34 has been divided into smaller parts to study the evolution of the burst spectrum. This was made possible due to the large area of LAXPC, providing at least 5 bins with sufficient count rates. Based on this analysis it is found that the persistent emission does get enhanced and it is due to the burst emission possibly increasing the disc emission by increasing the accretion rate (Bhattacharyya et al. 2018). Another bursting source 4U 1636–536 revealed the presence of a rare triplet of X-ray bursts, having gaps of few minutes between them. Though bursts with small gaps have been reported in few other sources, these LAXPC observations is shortest triplet with gap between first and second burst being 7 min, and with gap between second and third burst being 5.5 min (Beri et al. 2019). Since bursts occur due to thermonuclear burning, this time gap along with the state in color–color diagram and absence of mHz oscillations have to be taken together to constrain models for short recurrence time bursts. The source 4U 1323-62 has been observed with LAXPC during February 2017, and during two days of observation six thermonuclear bursts were detected (Bhulla et al. 2020). The recurrence time between three consecutive bursts were found to be *2.66 hr. Rise time of the bursts were typically 10 s and total duration being 75 to 100 s. The bursts are detected only up to 20 keV and energy-dependent profiles of the bursts indicate they have maximum peak counts and last longest for the 3–6 keV range. A QPO is detected *1 Hz. Based on spectral modelling, a photospheric radius of 94–40 km is estimated assuming isotropic emission of burst.

19

Page 12 of 19

3.4.1.3. Cyclotron lines and accretion powered pulsars. The detection of cyclotron resonant scattering feature (CRSF) has been observed at 5 keV in SXP 15.3 (Maitra et al. 2018) indicative of a magnetic field of 6 9 1011 G and is the second pulsar in small Magellanic cloud after SMC X-2 with a cyclotron line detection, and hence a confirmed magnetic field strength of the neutron star. The CRSF centroid energy varies with pulse phase, with an increase in energy and spectral hardening during an intensity dip. The characteristics mentioned above and the doublepeaked pulse profile of SXP 15.3 indicate a fan-beam like geometry dominating the emitting region as is expected for super-critically accreting sources. Two other wind fed pulsars sources in which cyclotron features have been detected are 4U1907?09 at 18.5 keV (Varun et al. 2019a), and 4U 1538–522 at *22 keV (Varun et al. 2019b). An energy-resolved pulse variation has been observed up to 60 keV in both these pulsars. Energy-dependent pulse profiles are also found in 4U1909?07, but with no cyclotron feature (Jaisawal et al. 2020). Be/Xray-binaries are binaries with a highly elliptical orbit and X-ray emission occurs when the neutron star is close to the periastron, and it accretes matter from the circumstellar disc of its companion Be star. This sudden increase in X-ray intensity is often observed as an outburst. The Be/X-ray binary GRO J2058?42 exhibited an outburst in April 2019, a part of which was observed by AstroSat (Mukerjee et al. 2020). Spectral analysis led to detection of three absorption features around *10 keV, 20 keV and 38 keV. Magnetic field strength of 1.21 9 1012 Gauss has been estimated for the cyclotron feature at 10.81 keV. A QPO at 90 mHz and its harmonics, were also observed for the first time, indicative of formation of a transient disc during the outburst. The spin period of 194.22 s and the spin derivative are similar to that observed earlier using BATSE data. Another Be/X-ray binary 3A 0726–260 (4U 0728–25) has been observed with AstroSat during 2016 (Roy et al. 2020). X-ray pulsations are observed from 0.3 keV to 40 keV at a period of 103.144 ± 0.001 s using both SXT and LAXPC. The period is detected above 20 keV for the first time. Pulse profile changes from single-peaked at lower energies to double-peaked at higher energies. Spectral-fit components include a broad iron line in addition to a power law with high energy cutoff.

J. Astrophys. Astr.

(2021) 42:19

3.4.1.4. Ultraluminous X-ray sources. Ultra-luminous X-ray sources (ULXs) are X-ray binaries with luminosities C1039 erg/s beyond the Eddington limit for a typical neutron star/black hole. Many of these are extragalactic and considered to be candidates for intermediate mass black holes. With AstroSat two ULXs have been observed, both being Be/X-ray binaries with neutron star companions. RX J0209.6-7427 a source discovered by ROSAT went into type II outburst in 2019, after about 26 years. Type II outbursts are much brighter and rarer than the type I bursts (caused during periastron passage of neutron star). The exact cause of these bursts are not fully understood and are probably due to sudden mass eruption from the Be star or due to accretion from the warped disc of the Be star which does not lie in the orbital plane of the binary system. AstroSat observations of this 2019 outburst extends for the first time pulse profile observations beyond x-ray energies of 12 keV (Chandra et al. 2020). The pulse period is found to be 9.28204 s. and the profiles are similar in the 3–80 keV range, with pulse fraction being maximum in the 12–20 keV range. Spin up rate of 1.75 9 10–8 s/s estimated using LAXPC data, is much higher than the usual 10–11 s/s observed in pulsars. The rapid spin-up is indicative of accretion disc which is likely rotating in the same direction as the neutron star. Spectra of both SXT and LAXPC data in the energy range 0.5 to 50 keV, is fitted with a highly absorbed power law with a high energy cutoff and an Fe-L Gaussian feature at 0.88keV. An X-ray flux of 3.7 ± 0.1 9 10–9 erg cm–2 s–1 which implies Xray luminosity to be 1.6 ± 0.1 9 10 39 erg/s for a distance of 60 kpc for the source. This pulsar may be an ULX located in the Magellanic Bridge in the vicinity of SMC. The second ULX is Swift J0243.6?6124 a new transient discovered in October 2017. It was observed with AstroSat on 7th October (Obs-1) and 26th October 2017 (Obs-2) (Beri et al. 2020). Both the observations are during the rise of the outburst, with Obs-2 being brighter than obs-1. Pulsations are detected at a period of 9.850 s in both the observations, in the energy range 0.3 to 150 keV (including CZTI observations up to 150 keV in Obs-2). The pulse profile in this source is double-peaked and shows a complex structure \7 keV in Obs-1. Pulse profiles of both observations are energy-dependent with pulse fractions increasing with energy. The luminosities estimated based on a spectral fit were found to be sub-

J. Astrophys. Astr.

(2021) 42:19

Eddington (LX * 7910 37 erg/s ) for Obs-1 and superEddington (LX * 6910 38 erg/s ) for Obs-2. No cyclotron feature has been detected in both these sources, although strong magnetic fields are estimated by other methods. Observations of these two pulsars are important for future studies.

Page 13 of 19

19

with hardness. During these observations the source is predominantly in a disc dominated state, and the hard photons [10 keV lag with respect to soft photons. For the first time such small variations (*7%) in centroid frequency and time lags have been correlated with other parameters like hardness, though they are not sufficient to draw any new conclusions.

3.4.2 Black hole binaries 3.4.2.1. Mass and spin parameter. Black hole X-ray binaries are often observable only during an outburst, as the number of persistent black hole binaries is very small. In addition, during an outburst, the source may exhibit several spectral states. The mass and spin are fundamental parameters of a black hole. One method of estimating the spin, is by fitting the redshifted iron line in the spectra of black holes. The other method is by fitting the continuum spectrum, when the X-ray emission from the source is in a disc dominated high soft state, and the disc is expected to extend to the innermost stable circular orbit (ISCO). 4U1630-47 was observed to be in outburst mode in 2016. Observations in October 2016 indicated the source was in soft state during which the spectra could be fitted with primarily a disc spectrum. Chandra data and AstroSat data were obtained and analysed (Pahari et al. 2018a, b). This is the first time spectral fit in high soft state of 4U1630-47 has been used to estimate the spin to be 0.92 with over 99% confidence. In addition evidence of lines due to wind has also been found using Chandra data. The same source was observed during both the (2016) and (2018) outbursts and spectral data using both SXT and LAXPC was analysed over the energy range 0.7–20 keV (Baby et al. 2020). The authors estimate the mass limits of the compact object using three methods to be 3–9 solar mass. In GRS 1915?105, QPOs are detected in the range 3-5 Hz (Misra et al. 2020). Joint SXT and LAXPC spectral data are fitted with Kerrd model providing inner radius and accretion rate. Results show that the combined effect of QPO frequency and accretion rate have to be considered for finding the correlation with the gravitational radius. The QPOs are identified to be due to the general relativistically modified sound speed from the inner disk. The spin parameter is estimated to be 0.973?/–0.002. During observations in 2017, the source exhibits high-frequency QPOs which varies from 67.4 and 72.3 Hz (Belloni et al. 2019). The centroid frequency of the QPO is also found to show a variation along the hardness intensity diagram, the frequency increasing

3.4.2.2. Black hole binaries with high mass compan-ion. Cyg X-1 was studied in its hard state during January 2016 and found to have a predominantly thermal Comptonised fit (Misra et al. 2017). In the hard state, the overall understanding is, the disc is truncated at a radius far out from the ISCO. The seed photons from the disc are Comptonised in a region which lies inward of the truncated radius, and occasionally can also lead to a reflected component in the form of an iron line. The temporal spectrum exhibits two broad peaks at 0.4 and 3 Hz. It is found that there is a lag of the high energy photons, in the range 20–40 keV with respect to the low energy photons of 5–10 keV, which varies as a function of frequency and decreases from 50 ms at lower frequencies to 8 ms at 2–5 Hz frequency. Further analysis of more data sets of the same source in hard state are done, and it is found that the power density spectrum (PDS) exhibits two broad peaks for January 2016 data set and three broad peaks in the other five data sets (Bari et al. 2019). The large energy coverage of LAXPC and the time resolution allows for studying the frequency and energy dependence of the rms and the time lag of the hard photons based on a stochastic propagation model. The power density spectrum (PDS) of black hole binaries exhibit broad continuum noise-like features, which could be due to perturbations occurring in the disc. The energydependent variability in these features of the continuum observed in many black hole binaries, is explained in terms of propagation of perturbations from the outer regions of the disc, all the way to the inner regions. Bari et al. (2019) explain this and the frequency dependence of the time lags, by considering the inner Comptonisation region to be a single temperature, optically thin, geometrically thick disc, instead of multiple Comptonisation regions. The frequencydependent time lag between different energy bands may be due to an underlying time lag between seed photon fluctuations and subsequent variation of heating rate of the hot inner flow. Cyg X-3 is an X-ray binary with an estimated mass 2.4M for the compact object, which could either be a neutron star or a black hole. However, broadband

19

Page 14 of 19

spectral properties favour a black hole. Cyg X-3 exhibits an asymmetric orbital modulation of *4.8 hr, which is attributed to periodic covering of the compact object by a plasma cloud. Radio jets have been observed in several X-ray binaries and the connection between the X-ray emission and radio flaring has been an interesting problem in recent years. This phenomenon can be well studied during an outburst and also in bright persistent sources. Cyg X-3 had an outburst in 2017, and it was observed using SXT and LAXPC on April 1-2, during which the X-ray source was caught displaying spectral state transition, coincident with a rise of radio flux (Pahari et al. 2018a, b). The initial 10–12 hours of this observation was modelled with a disc spectrum and a Bremsstrahlung to account for the wind fed by the companion star. This state has no significant photons [17 keV and is termed as hypersoft state (HPS). The source then becomes more luminous with the spectrum having an additional hard component modelled by a power law with a photon index of *1.49, termed as Very High State (VHS). The authors find that during this intensity and spectral transition the radio flux intensity at 11.2 GHz using RATAN-600 radio telescope increases from 100 to 476 mJy. In addition it is found that while the 3–6 keV count rate is strongly correlated during the HPS with the 15–60 keV count rate with the Spearman Rank Correlation Coefficient (SRCC) of 0.82, it reduces to 0.17 during VHS. The hard component is therefore attributed to a different origin, and when the power law is extrapolated to radio frequencies, it matches the radio flux intensity levels. Hence the power law emission is attributed to synchrotron emission from the slowly moving radio jet base. This is supported by the fact that the temperature of the inner disc reduces and the radius increases indicative of receding of the disc with formation of a central blob of plasma. The data obtained with AstroSat (SXT and CZTI) in 2015, when combined with *45 years of archival data exhibits a simple secular variation of the binary period, without any second derivative (Bhargava et al. 2017). Small variations are however indicated, most likely associated with jet emission. Data has been collected on this source intermittently for over a year from October 2015, and exhibits different spectral states. In addition to orbital variations, a likely period around 35.8 days is detected (Pahari et al. 2017). During the flaring hard X-ray states and only rising phases of orbital modulation, a QPO of *20 mHz is observed. The fractional RMS of the QPO tends to show an

J. Astrophys. Astr.

(2021) 42:19

increasing trend with energy, and a soft lag (soft photons lagging hard photons) of few seconds is observed. The rising portion of the orbital modulation is when the accreted material is high in this wind fed binary. The origin of the QPO and soft lag is therefore attributed to accretion of oscillating clumpy material, with the amplitude of the oscillations being boosted by in-phase oscillation of the material of the corona, with Compton down scattering of the hard photons. 3.4.2.3. Transients. MAXI J1535-571 is an X-ray transient discovered in 2017. Observations using AstroSat were done when the source was in hard intermediate state. The source exhibits *2Hz QPOs with change in QPO frequency from 1.85 to 2.88 Hz (Sreehari et al. 2019). The fundamental frequency at 2.21 Hz is observed in the energy range 3–50 keV while its harmonics are only detected at energies below 35 keV. Using the two-component accretion flow, the mass of the compact object is estimated to be in the range 5–7.8 solar mass. With a more detailed analysis of the QPO, it was found that the change in frequency is found to be correlated with power law index of the spectral fit but not with luminosity (Bhargava et al. 2019). Further, the wide band spectrum appears to be fitted best with a model of spinning black hole (Sridhar et al. 2019) with estimates of the mass of the black hole to be 10.39±0.61M and the dimensionless spin parameter to be 0.67. Another new transient MAXI J1820?070 went into outburst in March 2018. AstroSat observations were made during its hard state and the spectra are fitted with a disc, thermal Comptonisation and reflection component. The hard X-ray component is believed to be from the regions inner to that of a truncated disc due to a Comptonisation region, with the seed photons being from the disc. A QPO is observed at 47.7 mHz and a likely one *109 mHz (Mudambi et al. 2020). In addition the continuum in the temporal Power Density Spectrum (PDS) can be fitted with 3 broad humps. The fractional RMS at a given frequency of the continuum decreases with energy and the time lag of hard photons with respect to the soft photons increases with energy. This energy and frequency dependence is explained in terms of the stochastic propagation model (Bari et al. 2019) in terms of propagation of perturbations from the outer regions of the disc, all the way to the inner regions. The perturbation time delay is of the order of 100 ms and is frequency-dependent, with the frequency dependence being similar to that of Cyg X-1.

J. Astrophys. Astr.

(2021) 42:19

3.5 Galaxies and active galactic nuclei Dwarf galaxies are the most abundant type of galaxy in the universe. Many of them having been formed in the early Universe, they can provide important clues to the formation and evolution of galaxies. Star formation in these galaxies continue throughout cosmic time. Star formation in spite of their low metallicity and mass may indicate different pathways due to internal triggers like feedback from massive stars. Wolf–Lundmark–Melotte, or WLM is an isolated irregular dwarf galaxy in the local group at a distance of 995 kpc. The UV emission extend to 1.7 kpc from the galactic centre, with FUV contour being smaller than NUV, which may be indicative of NUV emission from older stars compared to FUV emission (Mondal et al. 2018). The flux is mostly from the stars within a radius of 1 kpc. Multiple star-forming regions of temperatures [17500 K are observed with tens of regions with sizes 20–50 pc, and hundreds of regions with size \10 pc. Several hot regions have temperatures [35000 K indicative of O star population, and since their life time is 10 Myr, these regions may have formed recently. Star formation rate is estimated to be 0.008M/year. It is found that the hot regions are surrounded by cooler regions, indicating a clumpy nature, and possibly a feedback from the hotter regions triggering star formation in surrounding regions. NGC 2336 on the other hand is a spiral galaxy viewed nearly face on, and hence ideal for studies of star forming regions (Rahna et al. 2018). 78 starforming knots have been identified in the NUV band with 3 of them being in the inter-arm region and 6 of them being in co-rotation ring around the bar. Of these 57 knots are also identified in FUV. The knots have a mean size of 485 pc in FUV and 408 pc in NUV. The star formation rates vary from 6.9 910 -4 to 2.2 910-2M/yr in NUV and from 4.5910-4 to 1.8 910-2 M/yr in FUV depending on the mass of the star-forming knots, with average being 4.7 910 -3 M/yr in NUV and 4.1910 -3 M/yr in FUV. Using the FUV–NUV color, it is found that the blue knots are towards the centre. It is suggested that star forming is driven by spiral density wave. The number of early type galaxies is found to grow due to merging of galaxies to form an elliptical galaxy. Tidal forces can expel large amounts of molecular gas into the intergalactic medium, which later condenses and triggers star formation. Study of star formation in the tidal tails of merging galaxies often resembles dwarf galaxies and are referred to as

Page 15 of 19

19

tidal dwarf galaxies and forms an interesting topic especially in low density environment. NGC 7252 (atoms for peace galaxy) is one such galaxy with the merger having started 600–700 Myr ago and is now observed with a single nucleus remnant, with two tidal tails. The central part also has young clusters of stars. UVIT observations of this galaxy is used to identify 7 star-forming regions in the tails (including two regions already known (George et al. 2018a). Star formation rates in the already known regions is found to be 0.02 and 0.03 M/yr, comparable to other dwarf galaxies. A weak dependence of star formation rate density is observed with distance from the centre, which could be indicative of initial conditions in the merger. The central region of the galaxy has been studied in detail. Based on the FUV-NUV color a blue circumnuclear ring of radius 1000 with a radius core of 400 is detected (George et al. 2018b). The blue ring has stellar formation regions, and is comprised of stellar populations with ages B300 Myr, with embedded starforming clumps of younger age (B150 Myr and 250 Myr). The red core region is devoid of star formation and extinction being the cause of the color is ruled out based on ground-based observations. The authors suggest that this could be the result of a central black hole which could launch a jet and a blast wave from the jet can create a bow shock in the vertical direction, pushing the gas outwards creating a cavity at the centre thus suppressing star formation in the core. If a critical density is achieved in the outer regions, star formation can be triggered in the form of a ring around the core. Jelly fish galaxies are another set of galaxies which exist in clusters and in which gas is stripped off due to ram pressure. The intra-cluster medium consists of hot X-ray emitting plasma. The interstellar medium of spiral galaxies, which usually exist in the outer regions of the cluster, start experiencing a force opposite to their orbital velocity as they fall into the intra-cluster plasma. This leads to ram stripping of gas which modifies the morphology and converts rich star forming galaxies to partial or fully gas stripped galaxies, and making them as passively evolving systems. The gas gets stripped and also undergoes shock compression, and therefore can be sites of star formation. Jelly fish galaxies, so named because of their appearance, can therefore have tentacles and still retain part of their spiral form, and are detected in Ha images. JO201 is a jelly fish galaxy located in the Abell 85 cluster with a redshift of 0.056 with a corresponding luminosity distance of 250 Mpc. UV images of JO201

19

Page 16 of 19

in Abell 85 cluster have been obtained using UVIT on AstroSat (George et al. 2018c). 89 star-forming knots are detected in NUV, of which 85 are confirmed with Hα images using the Multi-Unit Spectroscopic Explorer (MUSE) on the very large telescope (VLT) at ESO (Bellhouse et al. 2017). Of the 89 knots, 80 are confirmed based on redshift and extinction measurements of Hα. 24 knots are present on the disk and the remaining 56 are outside the disk within the inter-galactic medium. Star formation rates (SFR) for the knots of the disc range from ~0.05 to 2.07M/yr. The SFR of knots outside the disk has a median value of ~0.05 M/yr and range from ~0.01 to 0.6 M/ yr. Further results indicate quenching of star formation in the central region close to the AGN and a region outside it with enhanced star formation providing evidence of feedback mechanism (George et al. 2019). 3.6 Polarization The Cadmium Zinc Telluride Imager (CZTI) on AstroSat becomes a wide angle detector at high energies above 100 keV. This enables CZTI to detect GRBs, which can occur anywhere in the sky. This property of CZTI combined with the capability of the detectors to be able to detect Compton scattered X-rays in adjacent pixels of the CZTI has led to the detection of over 200 GRBs, and more important detect polarisation in some of these bursts. Polarisa-tion studies began with detection and study of spin phase dependence of polarisation in Crab (Vadawale et al. 2018). Polarisation has been detected in many GRBs observed using CZTI (Chattopadhyay et al. 2019). Of these 5 of them have detection above 3 sigma. Polarisation studies can help in understanding the processes leading to GRBs. Recent spectropolarimetric studies of the prompt emission using AstroSat, Swift and Fermi data of GRB160325A, and afterglow measurements using Swift XRT/UVOT measurements have demonstrated this capability (Sharma et al. 2020). The afterglow observations indicate that the jet emission is pointed towards the observer. The spectra of prompt emission consisting of two episodes separated by 9s, has been modeled using the fireball model. The first of the two episodes is due to quasithermal Comptonisation in the photosphere, and the second episode, due to synchrotron emission prod-uced in internal shocks in the optically thin region above the photosphere, and having high polarisation. CZTI observations have also been used to differentiate

J. Astrophys. Astr.

(2021) 42:19

between the emission due to gravitational wave source GW 170104 and gamma ray burst source GRB170105A (Bhalerao et al. 2017a, b).

3.7 Deep survey using UVIT 3.7.1 Andromeda galaxy. Our neighbouring spiral galaxy M31 or Andromeda galaxy located at 783 kpc has been studied in detail using the UVIT. The central bulge covered in 7 pointings of the galaxy is found to have 31 bright sources clustered around a central 5 arcmin radius around the nucleus (Leahy et al. 2018). SEDs are constructed for 26 of these sources along with the HST Panchromatic Hubble Andromeda Treasury (PHAT) covering the wavelength from 275 to 1600 nm in six bands. Most of the SEDs are double peaked, their fits indicative of a hot and cool star. The fit for hot stars correspond to main sequence (5–20M) and the cooler stars to giant branch. Further observations of the galaxy with 18 pointings has resulted in a UVIT catalog of total of 74,907 sources identified in 4 FUV filters or 2 NUV filters, with largest number (31000) being in the CaF2 filter of FUV having a bandwidth of 125–175 nm. A matching is done with the UVIT source catalog and the Chandra catalog (Leahy & Chen 2020). Totally 67 sources are common in various bands. The UV and X-ray data are analysed using power law and black body models. 3.7.2 Lyman continuum from z = 1:42 galaxy. Observational evidence of star formation in distant galaxies, supporting the concept of the epoch of re-ionisation in the early Universe, is one of the interesting problems in astronomy. Detection of Lyman Continuum (LyC\900 Å) is an important marker indicating the escape of ionising radiation. Inter-galactic matter prevents direct observation of the LyC from very distant galaxies. Hence direct LyC observations can be made in UV for nearby galaxies z < 0.4. Observations of galaxies with 2.5 < z < 3.5 is made possible as the LyC is redshifted to optical/IR bands. For this purpose, there has been a systematic study of deep fields both from ground- and space-based observatories. One of the goals of UVIT was also to conduct deep survey of specific areas in the sky. For this purpose, deep fields which had been surveyed by Hubble Space Telescope were chosen in order to maximise the scientific outcome from the observed celestial objects. One such field was the GOODS-South field observed using the UVIT instru-

J. Astrophys. Astr.

(2021) 42:19

Page 17 of 19

19

Figure 5. The LyC flux detected from different galaxies at different z. The red star at the centre is from Saha et al. (2020), indicating detection of rest frame EUV photons at z * 1.42. Image Courtesy: Kanak Saha.

ment onboard AstroSat. From this field, the object AUDFs01 (AstroSat ultra deep field source 01) was chosen for in depth study (Saha et al. 2020) from the data of Hubble Extreme Deep field of HST. Based on HST data, it was identified to be a clumpy galaxy exhibiting broad and strong Ha and [OIII] lines and found to be at z * 1.42. The Lyman break wavelength corresponds to the NUV band of the UVIT for this redshift of z = 1.42, and a rest frame EUV ˚ corresponded to the FUV band of wavelength *600 A UVIT. This object AUDFs01 was found to have a good S/N in the UV bands of UVIT on AstroSat, and therefore enabled the detection of LymanC leakage from this source. This observation of the UV Imaging telescope of AstroSat, is made possible due to the reduced background of the UVIT detectors combined with the long exposure time. As to why only UVIT detected this galaxy in FUV, we quote the authors, ‘one can also think that UVIT is more sensitive to low-surface-brightness objects because it has a lower spatial resolution and thereby samples more angular area per unit detector element.’ This source AUDFs01 is classified as a clumpy galaxy with 4 clumps. The escape fraction of Lyman Continuum is estimated to be at least 20%, and is thus a good candidate of small faint galaxies providing ionising radiation, which could be an analogue for the Epoch of re-ionisation. This result is therefore the first detection of Lyman Continuum from a galaxy at z * 1.42, in the hitherto unexplored range of 0.4 \ z \ 2.5, often referred to as the ‘redshift desert’ for LyC.

Figure 5 shows the observed LyC fluxes of the detected galaxies from literature (Izotov et al. 2016a, b, 2018a, b; Steidel et al. 2018). It may also be mentioned that the observation of Saha et al.(2020) for the first time is from rest frame EUV photons (hence depicted as red star symbol), and this could provide crucial input towards questions such as: (a) at what redshift does the peak of star formation density occur? (b) does escape fraction of LyC evolve? Further detection of many more LyC galaxies are expected to provide answers to these outstanding queries. 4. Beyond AstroSat Taking cognizance of the spectacular success of the AstroSat mission in terms of the basic concept, the configuration, a number of state-of-art scientific instruments backed up by a sophisticated highly maneuverable spacecraft with precision pointing capabilities and operating the spacecraft truly as an observatory in space, have all together made India develop the full confidence in realizing an end-to-end astronomical observational space objective. The mission management including multiplicity of observers from different parts of the world placing their own demands on the observational time and the ability to realize the same by the mission management and ground system operations puts India among a select group of countries carrying out such a complex operation successfully.

19

Page 18 of 19

The scientific outcomes already realized by AstroSat over the last five years add a wealth of knowledge in the field of ultraviolet and X-ray astronomy, including many new discoveries. This has spurred the Indian scientific community to plan immediately observatories such as XpoSat for X-ray polarization measurements and Aditya L1 for solar studies. Further, discussions are under way among the astronomical community to build a more ambitious AstroSat follow-on mission, that besides enhancing the capability of the ongoing spacecraft in the areas of ultraviolet, x-rays and gamma rays could have better sensitivity, more optimal choice of wavelength bands and much more sophisticated pointing and control system for the spacecraft. Needless to emphasize, the international collaboration and co-operation would certainly be a hallmark of such an ambitious astronomical space mission. The younger generation through AstroSat-1 have been trained, prepared and are confident enough to carry on its rich legacy.

Acknowledgements We use the results of data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). The authors wish to thank all the payload and project team members of AstroSat, and the authors of the scientific publications from AstroSat. The authors also wish to thank the reviewer for useful comments.

References Agrawal P. C. 2006, Adv. Sp. Res. 38, 2989 Agrawal P. C. 2017, J. Astrophys. Astr. 38, 27 Agrawal V. K., Nandi A. 2020, MNRAS 497, 3726 Agrawal P. C., Paul B., Rao A. R. et al. 1997, J. Korean Astronom. Soc. 29, 429 Agrawal P. C., Yadav J. S., Antia H. M. et al. 2017, J. Astrophys. Astr. 38, 30 Agrawal V. K., Nandi A., Girish V. 2018, MNRAS, 477, 5437 Antia H. M., Yadav J. S., Agrawal P. C. et al. 2017, ApJS, 231, 10 Baby B. E., Agrawal V. K., Ramadevi M. C. et al. 2020, MNRAS, 497, 1197 Bala S., Bhattacharya D., Staubert R., Maitra C. 2020, MNRAS, 427, 1029 Bari M., Mudambi S. P., Misra R. et al. 2019, MNRAS, 486, 2964

J. Astrophys. Astr.

(2021) 42:19

Becker P. A., Klochkov D., Scho¨nherr G. et al. 2012, A&A, 544, 123 Bellhouse C., Jaffe´ Y. L., Hau G. K. T. et al. 2017, ApJ, 844, 49 Belloni T. M., Bhattacharya D., Caccese P. et al. 2019, MNRAS, 439, 1037 Beri A., Paul B., Yadav J. S. et al. 2019, MNRAS, 482, 4397 Beri A., Naik S., Singh K. P. et al. 2020, MNRAS, 500, 565 Bhalerao V., Bhattacharya D., Vibhute A. et al. 2017a, J. Astrophys. Astr. 38, 31 Bhalerao V., Kasliwal M. M., Bhattacharya D. et al. 2017b, ApJ, 845, 152 Bhargava Y., Rao A. R., Singh K. P. et al. 2017, ApJ, 849, 141 Bhargava Y., Belloni T. M., Bhattacharya D., Misra R. 2019, MNRAS, 488, 720 Bhattacharyya S., Yadav J. S., Sridhar N. et al. 2018, ApJ, 860, 88 Bhulla Y., Misra R., Yadav J. S., Jaaffrey S. N. A. 2019, RAA, 19, 114 Bhulla Y., Roy J., Jaaffrey S. N. A. 2020, RAA, 20, 98 Chandra A. D., Roy J., Agrawal P. C. et al. 2020, MNRAS, 495, 2664 Chattopadhyay T., Vadawale S. V., Aarthy E. et al. 2019, ApJ, 884, 123 George K., Poggianti B. M., Gullieuszik M. et al. 2018c, MNRAS, 479, 4126 George K., Joseph P., Mondal C. et al. 2018b, A&A, 613, L9 George K., Joseph P., Coˆte´ P. et al. 2018a, A&A, 614, A130 George K., Poggianti B. M., Bellhouse C. et al. 2019, MNRAS, 487, 3102 Goswami P., Sinha A., Chandra S. et al. 2020, MNRAS, 492, 796 Izotov Y. I. et al. 2016a, Nature, 529, 178 Izotov Y. I. et al. 2016b, MNRAS, 461, 3683 Izotov Y. I. et al. 2018a, MNRAS, 474, 4514 Izotov Y. I. et al. 2018b, MNRAS, 478, 4851 Jadhav V. V., Sindhu N., Subramaniam A. 2019, ApJ, 886, 13 Jain R., Vig S., Ghosh S. K. 2019, MNRAS, 485, 2877 Jaisawal G. K., Naik S., Ho W. C. G. 2020, MNRAS, 498, 483 Kasturirangan K., Padmini V. N., Prasad N. L. et al. 1997, A&A, 322, 778 Kumar R., Pradhan A. C., Parthasarathy M. et al. 2019, in Bragaglia A., Davies M. B., Sills A., Vesperini E., eds, Star Clusters: From the Milky Way to the Early Universe, Proceedings IAU Symposium No. 351, 2019 Lalitha S., Schmitt J. H. M. M., Singh K. P. et al. 2020, MNRAS, 498, 3658 Leahy D. A., Chen Y. 2019, ApJ, 871, 152 Leahy D. A., Chen Y. 2020, ApJ Suppl. 250, 23 Leahy D. A., Bianchi L., Postma J. E. 2018, ApJ, 156, 269

J. Astrophys. Astr.

(2021) 42:19

Leahy D. A., Postma J., Chen Y. 2020, ApJ, 889, 131 Maitra C., Paul B., Haberl F., Vasilopoulos G. 2018, MNRAS, 480, L136 Misra R., Yadav J. S., Chauhan J. V. et al. 2017, ApJ, 835, 195 Misra R., Rawat D., Yadav J. S., Jain P. 2020, ApJ Lett. 889, L36 Mondal C., Subramaniam A., George K. 2018, AJ 156, 109 Mudambi S. P., Bari M., Misra R. et al. 2020, ApJ Lett. 889, L17 Mukerjee K., Antia H. M., Katoch T. 2020, ApJ, 897, 73 Mukherjee D., Bhattacharya D. 2012, MNRAS, 420, 720 Mukherjee D., Bhattacharya D., Mignone A. 2013a, MNRAS, 430, 1976 Mukherjee D., Bhattacharya D., Mignone A. 2013b, MNRAS, 435, 718 Navalgund K. H., Suryanarayana S. K., Gaurav P. K. et al. 2017, J. Astrophys. Astr. 38, 34 Pahari M., Antia H. M., Yadav J. S. et al. 2017, ApJ, 849, 16 Pahari M., Yadav J. S., Chauhan J. V. et al. 2018a, ApJ Lett. 853, L11 Pahari M., Bhattacharyya S., Rao A. R. et al. 2018b, ApJ, 867, 86 Pandiyan R., Subbarao S. V., Nagamani T. et al. 2017, J. Astrophys. Astr. 38, 35 Paul B., Agrawal P. C., Rao A. R. et al. 1998, ApJ Lett. 492, L63 Rahna P. T., Das M., Murthy J. et al. 2018, MNRAS, 481, 1212 Ramadevi M. C., Ravishankar B. T., Sitaramamurthy N. et al. 2017, J. Astrophys. Astr. 38, 32 Rani S., Pandey G., Subramaniam A. et al. 2020, accepted in MNRAS Rao V. K., Agrawal P. C., Sreekumar P., Thyagarajan K. 2009, Acta Astronautica, 65, 6 Rao A. R., Patil M. H., Bhargava Y. et al. 2017, J. Astrophys. Astr. 38, 33 Rao N. K., Sutaria F., Murthy J. 2018a, A&A, 609, L1 Rao N. K., Marco D., Krishna S. et al. 2018b, A&A, 620, 138 Roy J., Agrawal P. C., Singari B., Misra R. 2020, RAA, 20, 155

Page 19 of 19

19

Saha K., Tandon S. N., Simmonds C. et al. 2020, Nat. Astron., 24 August, https://doi.org/10.1038/s41550-0201173-5, 2020NatAs.tmp..193S Sahu S., Subramaniam A., Simunovic M. et al. 2019, ApJ, 876, 34 Seetha S., Megala S. 2017, Curr. Sci. 113, 579 Sharma V., Iyyani S., Bhattacharya D. et al. 2020, MNRAS, 493, 5218 Sindhu N., Subramaniam A., Jadhav V. V. et al. 2019, ApJ, 882, 43 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, Proc. SPIE, 9144, 91441S Singh K. P., Stewart G. C., Westergaard N. J. et al. 2017, J. Ap&A, 38, 29 Singh K. P., Girish V., Pavana M. et al. 2020, accepted for publication in MNRAS Sinha S., Sreekumar P., Kasturirangan K. 2001, BASI, 29, 505 Sreehari H., Ravishankar B. T., Iyer N. et al. 2019, MNRAS, 487, 928 Sridhar N., Bhattacharyya S., Chandra S., Antia H. M. 2019, MNRAS, 487, 4221 Staubert R., Klochkov D., Vybornov V. et al. 2016, A&A, 590, A91 Steidel C. C., Shapley A. E., Pettini M. et al. 2018, ApJ, 869, 123 Subramaniam A. et al. 2020, this volume Subramaniam A., Sindhu N., Tandon S. N. et al. 2016, ApJ, 833, 27 Subramaniam A., Sahu S., Postma J. E. et al. 2017, AJ, 154, 233 Tandon S. N., Hutchings J. B., Ghosh S. K. et al. 2017, J. Astrophys. Astr. 38, 28 Vadawale S. V., Chattopadhyay T., Mithun N. P. S. et al. 2018, Nature Astron. 2, 50 Varun B., Maitra C., Pradhan P. et al. 2019a, MNRAS, 484, L1 Varun B., Pragati P., Maitra C. et al. 2019b, ApJ, 880, 61 Verdhan Chauhan J., Yadav J. S., Misra R. et al. 2017, ApJ, 841, 41 Yadav J. S., Agrawal P. C., Antia H. M. et al. 2017, Curr. Sci. 113, 591

J. Astrophys. Astr. (2021)42:17 https://doi.org/10.1007/s12036-020-09678-z

Ó Indian Academy of Sciences Sadhana(0123456789().,-volV)FT3 ](0123456789().,-volV)

PAYLOAD REVIEW

Science with the AstroSat Soft X-ray telescope: An overview SUDIP BHATTACHARYYA1,* , KULINDER PAL SINGH2, GORDON STEWART3,

SUNIL CHANDRA4, GULAB C. DEWANGAN5, NILIMA S. KAMBLE1, SANDEEP VISHWAKARMA1, JAYPRAKASH G. KOYANDE1 and VARSHA CHITNIS6 1 Department of Astronomy and Astrophysics, Tata Institute of Fundamental Research, 1 Homi Bhabha Road, Mumbai 400 005, India. 2 Indian institute of Science Education and Research Mohali, Sector 81, SAS Nagar, Manauli PO 140 306, India. 3 Department of Physics and Astronomy, The University of Leicester, University Road, Leicester LE1 7RH, UK. 4 Centre for Space Research, North-West University, Potchefstroom 2520, South Africa. 5 Inter-University Centre for Astronomy and Astrophysics (IUCAA), P.B. No. 4, Ganeshkhind, Pune 411 007, India. 6 Department of High Energy Physics, Tata Institute of Fundamental Research, 1 Homi Bhabha Road, Mumbai 400 005, India. *Corresponding author. E-mail: [email protected]

MS received 6 November 2020; accepted 4 December 2020 Abstract. The Soft X-ray Telescope (SXT) aboard the AstroSat satellite is the first Indian X-ray telescope in space. It is a modest size X-ray telescope with a charge coupled device (CCD) camera in the focal plane, which provides X-ray images in the  0.3–8.0 keV band. A forte of SXT is in providing undistorted spectra of relatively bright X-ray sources, in which it excels over some current large CCD-based X-ray telescopes. Here, we highlight some of the published spectral and timing results obtained using the SXT data to demonstrate the capabilities and overall performance of this telescope. Keywords. Galaxies: active—novae, cataclysmic variables—space vehicles: instruments—telescopes— X-rays: binaries—X-rays: stars.

1. Introduction The AstroSat Observatory is the first dedicated Indian astronomy satellite (Agrawal 2006; Singh et al. 2014; Seetha & Megala 2017; Singh & Bhattacharya 2017), with the Soft X-ray Telescope (SXT1), which is the first Indian X-ray telescope in space, on board (Singh et al. 2016, 2017a). Cosmic X-rays are reflected by two sets of coaxial nested mirrors in SXT. The first set has This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’. 1

https://www.tifr.res.in/*astrosat_sxt/index.html.

conically approximated paraboloid surfaces and the second set has conically approximated hyperboloid surfaces. This is an approximate Wolter I geometry (Wolter 1952), with a cooled charge coupled device (CCD) camera at the focal plane. Multiple instruments/telescopes of AstroSat can simultaneously observe a source in a wide energy range from optical to hard X-rays (up to  100 keV; Seetha & Megala 2017), and SXT covers the crucial soft Xray band (  0.3–8.0 keV; Singh et al. 2017b) in this range. This telescope has the following modest capabilities (Singh & Bhattacharya 2017; Singh

17

Page 2 of 7

et al. 2017a, b): (1) the maximum effective area of  90 cm2 at  1.5 keV; (2) an energy resolution of 80–150 eV in the 0.3–8.0 keV range; (3) time resolutions of 2.37 s in the Photon Counting (PC) mode and of 0.278 s in the Fast Window (FW) mode; (4) the field-of-view of 40 arcmin square; and (5) the Point Spread Function (PSF) having a full-width half-maximum of  100 arcsec and the half encircled energy radius of  5.5 arcmin. However, SXT has a much smaller pile-up compared to current large soft X-ray imaging telescopes, and hence is ideal to observe bright X-ray point sources. These make SXT capable, as an independent telescope, to study continuum spectra, broad and somewhat narrow spectral lines and variations with timescales of seconds and above of various types of cosmic sources. Furthermore, this telescope, jointly with the Large Area X-ray Proportional Counters (LAXPC) and the Cadmium–Zinc–Telluride Imager (CZTI) aboard AstroSat, or jointly with any other hard X-ray instrument or telescope, such as NuSTAR, can observe the broadband X-ray spectrum of cosmic sources, and can uniquely contribute to the estimation of the hydrogen column density, and the characterization of the soft X-ray spectra. Its good energy resolution and signal-to-noise ratio can be particularly useful for broadband spectral modeling. In this paper, we give an overview of some notable scientific results which required a significant role of SXT. In Sections 2–8, we mention results on black hole X-ray binaries (BHXBs), neutron star lowmass X-ray binaries (LMXBs), neutron star high-mass X-ray binaries (HMXBs), ultra-luminous X-ray pulsars (ULPs), cataclysmic variables (CVs), active galactic nuclei (AGNs) and stars, respectively. In Section 9, we make concluding remarks.

2. Black hole X-ray binaries Astronomical black holes are characterized by two parameters, mass and spin, and hence the measurement of these parameters is essential to probe the fundamental physics of these objects (e.g., Middleton 2016). A black hole X-ray binary, i.e., a stellar-mass black hole accreting matter from a companion star, is particularly useful for this purpose, as well as to test a theory of gravitation and to probe the inflow and outflow of matter and its emission in an extremely strong gravitational field.

J. Astrophys. Astr. (2021)42:17

AstroSat SXT can be very useful to study soft X-ray continuum spectra and spectral lines from BHXBs, to measure their temporal variations and to track their evolution through various states. This telescope is particularly capable to characterize the accretion disk and reflection spectra, which can be used to estimate black hole parameter values. Here, we will give a few examples. A transient BHXB 4U 1630–47 was observed with AstroSat during its 2016 outburst. The source was found in a very soft state with an accretion disk contribution of  97% to the total flux. Such a source state is ideal to measure the black hole spin from the disk inner edge radius estimated from the disk spectral component. SXT was particularly useful to characterize the disk spectrum, having covered its peak and both sides better than other instruments (see Fig. 1). Using the XSPEC2 kerrbb model for the relativistic disk blackbody to fit AstroSat SXT?LAXPC and contemporaneous Chandra High Energy Grating (HEG) spectra, the dimensionless black hole spin parameter was measured to be 0:92  0:04 with 99.7% confidence (Pahari et al. 2018a). Similarly, a characterization of the thermal disk spectrum of the persistent BHXB LMC X–1 with SXT led to a measurement of the black hole spin parameter of  0.93 (Mudambi et al. 2020a). The ability of SXT to characterize both the disk blackbody spectral component and the relativistic Fe Ka emission line of the reflection spectral component was crucial to measure mass and spin parameters of the transient BHXB MAXI J1535–571 (see Fig. 2; Sridhar et al. 2019). More recently, AstroSat (SXT?LAXPC?CZTI) and contemporaneous NuSTAR data were used to characterize another transient BHXB MAXI J1820?070. The black hole mass was estimated to be 6.7–13.9M , and SXT, being the only instrument in this work providing\3 keV data and hence being able to reliably fit the disk spectral component, was crucial for this estimation (Chakraborty et al. 2020). Note that this mass range derived from the spectral fitting is consistent with the results obtained from dynamical measurements (Torres et al. 2019, 2020). In addition to the above examples, several other publications have reported the characterization of BHXB spectral and timing properties with AstroSat science instruments, including SXT (e.g., Maqbool et al. 2019; Bhargava et al. 2019; Sreehari et al. 2019, 2020; Mudambi et al. 2020b; Baby et al. 2020). SXT was also useful to probe the formation of a giant 2

https://heasarc.gsfc.nasa.gov/xanadu/xspec/.

Page 3 of 7

17

10−3

0.01

0.1

1

3. Neutron star low-mass X-ray binaries

Chandra/HEG AstroSat/SXT AstroSat/LAXPC10 AstroSat/LAXPC20

10−4

X−ray flux: keV (Photons cm−2 s−1 keV−1)

J. Astrophys. Astr. (2021)42:17

1

2

5 Energy (keV)

10

20

Figure 1. Joint fitting of X-ray spectra of the BHXB 4U 1630–47 from AstroSat SXT, two LAXPC detectors (LAXPC10 and LAXPC20) and Chandra High Energy Grating (HEG). An absorbed, relativistic, disk blackbody model with Gaussian absorption features and convolved with a Comptonization model is used for fitting. The source was in a soft state with a disk flux fraction of  0.97. It can be seen that SXT covers the spectrum and its peak better than other instruments, which was particularly useful to measure the black hole spin (see Section 2; figure courtesy: Mayukh Pahari; Pahari et al. 2018a).

radio jet base and to measure the orbital period parameters of the X-ray binary Cygnus X–3, for which the nature of the compact object is not yet confirmed (Pahari et al. 2018b; Bhargava et al. 2017).

A neutron star LMXB is a binary system in which the neutron star accretes matter from a low-mass star. In such a system, the magnetic field of the neutron star is typically low (  1079 G), and hence the accretion disk can extend almost up to the stellar surface. Consequently, a number of features, such as thermonuclear X-ray bursts, high-frequency quasi-periodic oscillations, accretion-powered millisecond period pulsations, etc., are sometimes observed from these sources, which can be useful to probe the strong gravity regime and the superdense degenerate core matter of neutron stars (Bhattacharyya 2010). AstroSat science instruments, including SXT, are ideal to study various spectral and timing properties of neutron star LMXBs, and their evolution. For example, the combined spectra from AstroSat (SXT?LAXPC) and XMMNewton (EPIC-PN) observations of the neutron star LMXB SAX J1748.9–2021 suggested the presence of reflection features (Sharma et al. 2020). Figure 3 shows that the spectra measured with SXT and XMM-Newton EPIC-PN match well with each other. Another work on the neutron star LMXB GX 17?2 demonstrated that SXT not only is suitable for spectral and timing studies in soft X-rays, but also can be used for spectro-timing analyses (Malu et al. 2020). In this paper, the cross-correlation studies using SXT and LAXPC light curves showed time lags of the order of a hundred seconds.

Figure 2. AstroSat SXT and LAXPC joint spectrum from the BHXB MAXI J1535–571. The model used to fit is an absorbed disk blackbody plus cut-off power-law, the latter representing a Comptonization component (upper panel). This brings out two prominent features – Fe Ka emission line and Compton hump – of the reflection component of the spectrum in the data-to-model ratio plot (lower panel). It is clearly seen that both the disk blackbody and the asymmetry of the Fe Ka line due to relativistic effects can be measured only with SXT, which are crucial to estimate the black hole mass and spin (see Section 2; figure courtesy: Navin Sridhar; Sridhar et al. 2019).

Page 4 of 7

J. Astrophys. Astr. (2021)42:17 (a)

0.1

0.01

4 2 0 −2

(b)

χ

SXT XMM LAXPC

4 2 0 −2

(c)

χ

keV2 (Photons cm−2 s−1 keV−1)

17

1

2

5

10

20

Energy (keV)

Figure 3. AstroSat SXT and LAXPC, and XMM-Newton EPIC-PN joint spectrum of the neutron star LMXB and accretion-powered millisecond X-ray pulsar SAX J1748.9–2021 (a). Panels (b) and (c) show the residuals for the XSPEC models tbabs(bbodyrad?nthcomp) and tbabs(bbodyrad?nthcomp?xillvercp), respec tively. This figure shows that SXT and XMM-Newton EPIC-PN residuals match well with each other, with the SXT spectrum extending farther in lower energies (see Section 3; figure courtesy: Aru Beri; Sharma et al. 2020).

4. Neutron star high-mass X-ray binaries A neutron star HMXB is a binary system in which the neutron star accretes matter from a high-mass star (e.g., Walter et al. 2015). In such a system, the magnetic field of the neutron star is usually high (  1012 G), and hence the accretion disk is typically truncated far from the star. As a result, the accreted matter is channeled on to the magnetic polar caps, making the source a pulsar. These systems are ideal to study an interaction between the accreted matter and the strong stellar magnetic field. Several neutron star HMXBs, for example, 4U 0728–25, GRO J2058?42, 4U 1909?07, have been observed with AstroSat, and SXT was used to characterize the broadband spectra and pulse profiles in soft X-rays (Roy et al. 2020; Mukerjee et al. 2020; Jaisawal et al. 2020).

5. Ultra-luminous X-ray pulsars Ultra-luminous X-ray sources (ULXs), given their luminosities typically exceeding 1039 erg s1 , could be accreting intermediate-mass black holes, or neutron stars or stellar-mass black holes having accretion with super-Eddington rates. A subset of ULXs have been

confirmed to be accreting neutron stars from the observed spin-induced brightness pulsations, and they are known as ultra-luminous X-ray pulsars (ULPs; e.g., King et al. 2017). The characterization of the first Galactic ULP, Swift J0243.6?6124, can therefore be extremely useful to understand this important class of sources. AstroSat observed the 2017–2018 outburst of Swift J0243.6?6124, and characterized its broadband spectrum, as well as energy-dependent and luminosity-dependent pulse profiles in the energy range of 0.3–150 keV (Beri et al. 2021). SXT was particularly useful to measure the continuum spectrum, as well as pulse profiles (see Fig. 4), in soft X-rays. Besides, recent observations of the Be X-ray binary and pulsar RX J0209.6–7427 with AstroSat SXT and LAXPC have indicated that its spectral and timing properties are remarkably similar to those of ULXs, suggesting that this source could be a ULP (Chandra et al. 2020).

6. Cataclysmic variables AstroSat SXT can characterize spectral and temporal properties of cataclysmic variables (CVs), i.e., accreting white dwarfs. A sub-type of such binary systems, in which the white dwarf accretes matter from a red giant donor star via an accretion disk, can have explosive thermonuclear burning of the accumulated hydrogen rich material. This may lead to an outburst with a massive ejection of the material at velocities  300 km s1 . These are known as Symbiotic Recurrent Novae, and only four such objects are currently known to exist (Schaefer 2010). AstroSat SXT observed one such nova, V3890 Sgr, in two long observations in 2019 from 5th September to 16th September, just  8 days after its third recorded outburst, with the highest cadence monitoring from a low-Earth orbit satellite (Singh et al. 2021). The observations caught the first appearance of Super Soft Source (SSS) emission (\1 keV) on day 8.57 after the outburst, revealing the presence of a very high mass white dwarf. Rapid and highly variable evolution of the SSS, that included its complete vanishing during days 8.6–8.9 and subsequent appearance, followed by another extremely low flux state during days 16.8–17.8, and rising again were observed. A detailed spectral modeling, using white dwarf emission models for the SSS and plasma models for higher energy (1–7 keV) emission, to study the source spectral evolution has been carried out. The rapid spectral evolution (see

J. Astrophys. Astr. (2021)42:17

Page 5 of 7

17

Figure 4. Energy-resolved pulse profile of the first Galactic ultra-luminous X-ray pulsar Swift J0243.6?6124, using the AstroSat SXT data. This figure shows that SXT could sufficiently resolve the X-ray pulsation period of  9.85 s in its FW mode (see Section 5; figure courtesy: Aru Beri; Beri et al. 2021).

Fig. 5) on nearly hourly time scales is not explained by evolutionary models of accretion and ejecta.

7. Active galactic nuclei Active galactic nuclei (AGNs) are supermassive black holes, which accrete gas from the surrounding medium near the centres of galaxies. AGNs can radiate

AstroSat SXT(2019 Sep 5−16)

10

V3890 Sgr

prominently in a wide energy range – from radio to crays, and are rich in observational features in multiple wavebands. Based on these features, these objects have been divided into several subclasses, such as Seyfert galaxies and Blazars. AGNs are extremely important to study the strong gravity regime, accretion/ejection mechanisms, the feedback to the host galaxy and the intergalactic medium, and cosmology (e.g., Netzer 2015).

Day 15.87 Day 16.34

Day 18.44

1

Day 9.64 Day 8.77

0.1

Day 8.26

0.01

normalized counts s−1 keV−1

Day 17.36

SSS (Teff=0.6−0.9 MK) + 2T Plasma(kT=0.76−0.86 & 4−6 keV) NH ~ 1022 cm2

0.5

2

1

5

Energy (keV)

Figure 5. Evolution of AstroSat SXT spectrum of the symbiotic recurrent nova V3890 Sgr. The figure includes 7 of the 79 individual SXT spectra during its 2019 outburst, showing rapid variations. The days indicate the days after the beginning of the outburst. A Super Soft Source (SSS) emission component (with the best-fit effective temperature  0.6– 0.9 MK) and a two temperature plasma emission component (with best-fit temperatures  0.76–0.86 keV and  4–6 keV), with a best-fit hydrogen column density of  1022 cm2 , were used to describe the spectra (see Section 6; Singh et al. 2021).

17

Page 6 of 7

J. Astrophys. Astr. (2021)42:17

Figure 6. AstroSat SXT, Chandra LETGS 0th order and HST light curves of the nearest star Proxima Centauri, which hosts an Earth-like planet in its habitable zone. This figure shows flares and quiet periods of the star, which may affect the habitability of the planet. SXT observations were particularly useful to track the stellar activity throughout the observing campaign, with a flare observed simultaneously with all three instruments (see Section 8; figure courtesy: Lalitha Sairam; Lalitha et al. 2020).

AstroSat has observed several AGNs, and SXT has significantly contributed to the spectral and timing characterization of some of them. For example, SXT has measured the broadband spectrum and variability of the narrow-line Seyfert 1 galaxy RE J1034?396 in soft Xrays, and contributed to the source power spectrum, which, when compared with the power spectrum of the BHXB GRS 1915?105, indicated a supermassive black hole mass of 3  106 M (Chaudhury et al. 2018). In another work, the variability of a Blazar, Mrk 421, was measured with AstroSat SXT and LAXPC, as well as with Swift, which was very promising to establish a way to probe the disk-jet connection (Chatterjee et al. 2018). Besides, SXT was useful to characterize the soft X-ray spectrum of another Blazar, RGB J0710þ591 (Goswami et al. 2020). Simultaneous UV/X-ray observations of a Seyfert galaxy with AstroSat can be useful to probe the thermal Comptonization responsible for the broadband X-ray emission. Joint X-ray spectral analyses of five sets of SXT and LAXPC spectra revealed a steepening and brightening X-ray power-law component with increasing intrinsic UV emission for the Seyfert 1.2 AGN IC 4329A (Tripathi et al., to be submitted). These observations implied that UV emission from the disk indeed provides the primary seed photons for the Thermal Comptonization process, and the X-ray spectral variability is caused by either cooling of the hot corona or increasing optical depth of the corona, each with increasing UV flux. 8. Stars Coronal activity in stars can be studied by observing the variation of its high-energy radiation, including occasional flares (Gudel & Naze 2009). Such a study

can be useful not only to probe the stellar physics and the surrounding environment, but also to understand the impact of the ejected particles and radiation on the plausible planets, including their habitability. AstroSat SXT is capable of tracking the stellar soft X-ray intensity variation, which was demonstrated by an observational campaign on our nearest star Proxima Centauri with SXT, Chandra Low Energy Transmission Grating Spectrometer (LETGS) and Hubble Space Telescope (HST). Proxima Centauri is an M-dwarf with an Earth-like planet within its habitable zone, and several flares and the non-flare emission observed from the star were useful to probe the coronal temperatures, abundance, etc. (see Fig. 6; Lalitha et al. 2020). Particularly, one flare was observed with all three instruments, and showed the Neupert effect, that is the UV emission preceding the soft X-ray emission.

9. Conclusions Recent publications in refereed journals have confirmed that SXT can successfully study spectral and timing properties of a variety of cosmic sources, as a standalone telescope, as well as in combination with other AstroSat science instruments and other satellites, such as Chandra, XMM-Newton, NuSTAR, and even HST. Since the launch of AstroSat in late 2015, SXT has been performing as expected, without any significant degradation of its capabilities. If this continues, SXT can be very useful for deep observations, as well as to track the evolution of relatively bright X-ray transients in a more dedicated manner in the future.

J. Astrophys. Astr. (2021)42:17

Acknowledgements The publications mentioned here used the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). The works mentioned here were performed utilizing the calibration databases and auxiliary analysis tools developed, maintained and distributed by the AstroSat-SXT team with members from various institutions in India and abroad, and the SXT Payload Operation Center (POC) at the TIFR, Mumbai (https://www.tifr.res.in/ *astrosat_sxt/index.html). SXT data were processed and verified by the SXT POC. The authors thank Mayukh Pahari, Navin Sridhar, Aru Beri and Lalitha Sairam for providing some of the figures. References Agrawal P. C. 2006, Adv. Space Res. 38, 2989 Baby B. E., Agrawal V. K., Ramadevi M. C. et al. 2020, MNRAS, 497, 1197 Beri A., Naik S., Singh K. P. et al. 2021, MNRAS, 500, 565 Bhargava Y., Belloni T., Bhattacharya D., Misra R. 2019, MNRAS, 488, 720 Bhargava Y., Rao A. R., Singh K. P. et al. 2017, ApJ, 849, 141 Bhattacharyya S. 2010, Adv. Space Res. 45, 949 Chakraborty S., Navale N., Ratheesh A., Bhattacharyya S. 2020, MNRAS, 498, 5873 Chandra A. D., Roy J., Agrawal P. C., Choudhury M. 2020, MNRAS, 495, 2664 Chatterjee R., Roychowdhury A., Chandra S., Sinha A. 2018, ApJ, 859, L21 Chaudhury K., Chitnis V. R., Rao A. R. et al. 2018, MNRAS, 478, 4830 Goswami P., Sinha A., Chandra S. et al. 2020, MNRAS, 492, 796 Gudel M., Naze Y. 2009, A&ARv, 17, 309 Jaisawal G. K., Naik S., Ho W. C. G. et al. 2020, MNRAS, 498, 4830 King A., Lasota J.-P., Kluzniak W. 2017, MNRAS, 468, L59 Lalitha S., Schmitt J. H. M. M., Singh K. P. et al. 2020, MNRAS, 498, 3658

Page 7 of 7

17

Malu S., Sriram K., Agrawal V. K. 2020, MNRAS, 499, 2214 Maqbool B., Mudambi S. P., Misra R. et al. 2019, MNRAS, 486, 2964 Middleton M. 2016, in Bambi C. ed., Astrophysics of Black Holes - From fundamental aspects to latest developments (ASSL, Springer) Mudambi S. P., Maqbool B., Misra R. et al. 2020b, ApJ, 889, L17 Mudambi S. P., Rao A., Gudennavar S. B., Misra R., Bubbly S. G. 2020a, MNRAS, 498, 4404 Mukerje K., Anti H. M., Katoch T. 2020, ApJ, 897, 73 Netzer H. 2015, ARA&A, 53, 365 Pahari M., Bhattacharyya S., Rao A. R., et al. 2018a, ApJ, 867, 86 Pahari M., Yadav J. S., Chauhan J. V. et al. 2018b, ApJ, 853, L11 Roy J., Agrawal P. C., Singari B., Misra R. 2020, RAA, 20, 155 Schaefer B. E. 2010, ApJS, 187, 275 Seetha S., Megala S. 2017, Curr. Sci. 113, 579 Sharma R., Beri A., Sanna A., Dutta A. 2020, MNRAS, 492, 4361 Singh K. P., Bhattacharya D. 2017, Curr. Sci. 113, 602 Singh K. P., Dewangan G. C., Chandra S. et al. 2017b, Curr. Sci. 113, 587 Singh K. P., Girish V., Pavana M. et al. 2021, MNRAS, 501, 36 Singh K. P., Stewart G. C., Chandra S. et al. 2016, in Proceedings of the SPIE, Vol. 9905, 99051E Singh K. P., Stewart G. C., Westergaard N. J. et al. 2017a, JApA, 38, 29 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, in Proceedings of the SPIE, Vol. 9144, 91441S Sreehari H., Nandi A., Das S. et al. 2020, MNRAS, 499, 5891 Sreehari H., Ravishankar B. T., Iyer N. et al. 2019, MNRAS, 487, 928 Sridhar N., Bhattacharyya S., Chandra S., Antia H. M. 2019, MNRAS, 487, 4221 Torres M. A. P., Casares J., Jimenez-Ibarra F. et al. 2020, ApJ, 893, L37 Torres M. A. P., Casares J., Jimenez-Ibarra F. et al. 2019, ApJ, 882, L21 Walter R., Lutovinov A. A., Bozzo E., Tsygankov S. S. 2015, A&ARv, 23, 2 Wolter H. 1952, Ann. Phys., 10, 94

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:32 https://doi.org/10.1007/s12036-021-09712-8

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

PAYLOAD REVIEW

Large Area X-ray Proportional Counter (LAXPC) in orbit performance: Calibration, background, analysis software H. M. ANTIA1,* , P. C. AGRAWAL1, DHIRAJ DEDHIA1, TILAK KATOCH1,

R. K. MANCHANDA2, RANJEEV MISRA3, KALLOL MUKERJEE1, MAYUKH PAHARI4,5, JAYASHREE ROY3, P. SHAH1 and J. S. YADAV6 1

Department of Astronomy and Astrophysics, Tata Institute of Fundamental Research, Homi Bhabha Road, Mumbai 400 005, India. 2 Centre for Astrophysics, University of Southern Queensland, Toowoomba, QLD 4300, Australia. 3 Inter-University Centre for Astronomy & Astrophysics, Ganeshkhind, Pune 411 007, India. 4 School of Physics and Astronomy, University of Southampton, Highfield Campus, Southampton SO17 1BJ, UK. 5 Department of Physics, Indian Institute of Technology, Hyderabad, Hyderabad 502 285, India. 6 Department of Physics, Indian Institute of Technology, Kanpur, Kanpur 208 016, India. *Corresponding Author. E-mail: [email protected] MS received 23 October 2020; accepted 16 January 2021 Abstract. The Large Area X-ray Proportional Counter (LAXPC) instrument on-board AstroSat has three nominally identical detectors for timing and spectral studies in the energy range of 3–80 keV. The performance of these detectors during the five years after the launch of AstroSat is described. Currently, only one of the detector is working nominally. The variation in pressure, energy resolution, gain and background with time are discussed. The capabilities and limitations of the instrument are described. A brief account of available analysis software is also provided. Keywords. Space vehicles: instruments—instrumentation: detectors.

1. Introduction The Large Area X-ray Proportional Counter (LAXPC) instrument on-board AstroSat (Agrawal 2006; Singh et al. 2014) consists of three co-aligned detectors for X-ray timing and spectral studies over an energy range of 3–80 keV (Yadav et al. 2016a; Agrawal et al. 2017). Apart from LAXPC, AstroSat has three more co-aligned instruments, the Soft X-ray Telescope (SXT, Singh et al. 2016), the Cadmium–Zinc–Telluride Imager (CZTI, Bhalerao et al. (2017)) and the Ultra-Violet Imaging Telescope (UVIT, Tandon et al. 2017). AstroSat was conceived to carry out multiwavelength observations of various sources in the Visible, UV and X-ray bands using these co-aligned This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

instruments. AstroSat was launched on September 28, 2015 and the initial calibration of the LAXPC instrument was discussed by Antia et al. (2017). By now AstroSat has made more than 2000 distinct observations covering a wide variety of sources and a large amount of data are publicly available from the AstroSat Data Archive1. A quick look light-curves of all LAXPC observations are available at the LAXPC website2. A number of science results from LAXPC instrument have been published and a summary of these results is described in the companion paper (Yadav et al. 2021). Each LAXPC detector has five layers divided into seven anodes (A1–A7), with the two top layers having two anodes each. In addition there are three veto 1

https://astrobrowse.issdc.gov.in/astro_archive/archive/Home.jsp. https://www.tifr.res.in/*astrosat_laxpc/laxpclog.lc-hdr.html.

2

32

Page 2 of 21

anodes (A8–A10) on the three sides of the detector. The three faces covered by veto anodes are, the bottom and the two faces covering the long side of the detector as shown in Fig. 2 of Antia et al. (2017). By default the LAXPC detectors operate in the Event Analysis (EA) mode where the timing and energy of each photon is recorded with the time-resolution of 10 ls. The EA mode operation also generates the Broad Band Counting (BB) mode data, which gives the counts with a predefined time-bin, for various energy bins and anodes, including the counts of events beyond the Upper Level Discriminator (ULD) threshold (nominally at 80 keV). The dead-time of the detectors is 42 ls (Yadav et al. 2016b). In addition, there is a Fast Counter (FC) mode with a dead-time of about 10 ls to allow for observation of bright sources. In this mode only the counts with a fixed time-bin of 160 ls, as recorded in only the top layer of the detector in four predefined energy bins are available. However, this mode has not been used for any science observation and no software to analyse the data is available. As a result, this mode is not allowed to be configured by LAXPC users. Thus effectively only one mode covering both EA and BB is allowed. In the xenon gas counters, a large fraction of incoming photons above 34.5 keV emit a fluorescence photon and depending on the cell geometry and filling pressure, the fluorescent photon may generate a second localized electron cloud in a different anode. Therefore, on board electronics is designed to recognise such correlated double events and the energy deposited in the two anodes is added. These are referred to as double events as opposed to single events, where all energy is deposited in one anode. In this article, we mainly focus on the performance of the LAXPC instrument in orbit and its calibration and some software which is available for analysing data. The three LAXPC detectors are labelled as LAXPC10, LAXPC20 and LAXPC30. Currently, only LAXPC20 is working nominally. The rest of the paper is organised as follows: Section 2 describes the performance of detectors during the last five years. Section 3 describes the variation in background with time and some procedures to correct for these. Section 4 describes some capabilities and limitations of the detectors and their sensitivity. Section 5 describes some software for analysing the data. Section 6 describes the various science goals of LAXPC detectors and how they are met by the data obtained so far. Finally, Section 7 describes a summary of calibration and performance of detectors.

J. Astrophys. Astr. (2021)42:32

2. Long-term performance of LAXPC in orbit The health parameters of the detectors, like the temperature, pressure, high voltage (HV) and various energy thresholds are monitored regularly. While the temperature of detectors is steady, the pressure has changed over the last five years from the initial value of about two atmospheres. The HV and thresholds have been maintained, except for some adjustments that were made from time to time. Further, to monitor the stability of detectors, the peak position and energy resolution of 30 and 60 keV peaks in the veto anode A8 from the on board radioactive source Am241, are measured regularly and a log is maintained3. Apart from these, we have also regularly monitored spectrum from Crab observation to check the stability of the detector response.

2.1 Pressure in detectors Figure 1 shows the pressure as estimated from the pressure gauge in all detectors as a function of time. The LAXPC30 developed a leak soon after launch and the pressure was decreasing steadily. The HV of the detector was turned off on March 8, 2018 when the pressure had reduced to about 5% of the original pressure. LAXPC10 also has a fine leak and the pressure has been reducing gradually. Curiously, the pressure gauge of LAXPC20 shows a slow increase in pressure. This is most likely an artefact and shows the limitation of on-board pressure gauge. Because of this, we used other techniques to estimate the density in LAXPC30, as described by Antia et al. (2017) and the results using these are shown in the right panel of Fig. 1. Since these techniques are based on absorption in xenon gas, they yield the density which is assumed to be a proxy for pressure, as the temperature is almost constant during the entire period. The Cyg X-1 observations during the soft state were used to measure the density by calibrating the ratio of counts around 20 keV, observed in different layers of the detector. The soft state was used to ensure very low flux beyond the Xe K-edge to avoid possible contamination from events involving Xe fluorescence photons. The Crab spectra were fitted using responses with different density to get the best fit for density. Other techniques were based on the strength of 60 keV peak in veto anode A8 and the observation of L-edge in the spectrum when the density 3

https://www.tifr.res.in/*astrosat_laxpc/LaxpcSoft_v1.0/gaina8. dat.

J. Astrophys. Astr. (2021)42:32

Page 3 of 21

32

Figure 1. The pressure in LAXPC detectors as a function of time. The left panel shows the pressure as obtained from the pressure gauge for all three detectors. The right panel shows the density for LAXPC30 using various techniques. The dashed-line is the fit to linear part of the curve. Both pressure and density are shown relative to their initial values.

was sufficiently low. It can be seen that results from all these techniques agree with each other. The density in LAXPC30 decreased linearly for some time at a rate of about 4.5% of original density per month. After that it followed an exponential rate, as may be expected for a leak, with an e-folding time of about 200 days. By now the pressure is below the sensitivity of pressure gauge and using the exponential profile it can be estimated to be about 0.05% of the original value. To maintain the gain of the detector in a reasonable limit the HV of detector was reduced from time to time. To allow analysis of LAXPC30 data, the response has been calculated at different densities and the software gives recommendation of which response to use depending on the time of observation.

2.2 Energy resolution and gain of detectors Because of the leak, the gain of LAXPC10 and LAXPC30 was also increasing with time and this needs to be estimated using the 30 keV line in the veto anode A8 from the Am241 source. The calibration source has two lines, one at 30 keV due to K-escape event from Xe and another at 60 keV from the Am241 source, and these peaks can be used to check the drift in gain as well as energy resolution. To correct for the drift in the gain, the HV of LAXPC10 and LAXPC30 was adjusted from time to time. This gives some steps in the gain. After some stage the gain of LAXPC30 had to be adjusted frequently, giving a band in the peak position as seen in the left panel of Fig. 2. On January 22, 2018 the HV of LAXPC30 was reduced to the minimum possible value of about 930 V (as compared to the initial value of about 2300 V), after which the peak channel kept shifting upwards. Even before this stage, the 60 keV peak was

not well defined due to low efficiency and hence its position could not be determined. By the time the HV of detector was turned off, even the 30 keV peak had shifted beyond the ULD and it was not possible to estimate the gain of the detector. The channel to energy mapping is defined by a quadratic (Antia et al. 2017) and it is not possible to estimate the three coefficients of quadratic using the peak position of two peaks. Hence, it is assumed that only the linear term is changing with time and its value is estimated by the position of the 30 keV peak. Responses for different values of the peak position of 30 keV peak are provided and software makes appropriate recommendation based on the time of observation. If the gain was linear the peak channel for 60 keV peak, p2 would be twice that for 30 keV peak, p1 . Hence, the difference 2  p2 =p1 gives a measure of non-linearity in the gain. This quantity is shown in the right panel of Fig. 2. It can be seen that the nonlinearity has been decreasing for all detectors. However, it should be noted that this quantity can also change if the constant term in the gain is changing. Thus it is not possible to correct for this variation. It is advisable to use gain fit command in Xspec to adjust the constant term, and even the linear term, to get the best spectral fit. On March 26, 2018, LAXPC10 showed erratic counts with strong bursts where the dead-time corrected count rate reached 40000 s1 . The cause of this anomaly is not known. To stabilise the counts, the HV of the detector was reduced. Attempts were also made to control the noise by adjusting the Low Level Discrimination (LLD) thresholds of some anodes which were showing low channel noise, but that did not remove the bursts and hence the HV was kept at lower value. After that the counts were stable to some extent

32

Page 4 of 21

J. Astrophys. Astr. (2021)42:32

Figure 2. The energy resolution and peak channel for the 30 keV (p1 , as marked by cross) and 60 keV (p2 , as marked by open squares) calibration peak in veto anode A8 is shown in the left panel. The black, red and blue points show the result for LAXPC10, LAXPC20 and LAXPC30, respectively. The right panel shows the quantity 2  p2 =p1 for the three detectors. The lines mark the best-fit straight lines for the three detectors.

though smaller bursts continued. By looking at the value of counts beyond the ULD it is possible to identify the time intervals when counts are not stable and this has been implemented in the software which automatically removes these time intervals from Good Time Intervals (GTI). This problem occurred a few times after that and every time the HV was reduced to stabilise the counts. The last adjustment was made on April 9, 2019 and since then no bursts have been observed, except for a few days between April 23 and May 3, 2020. Because of these adjustments, the HV of LAXPC10 has been reduced from about 2330 V initially to about 2190 V in March 2018 (before the anomalous behaviour started) and to 1860 V now. Some reduction in HV was also required to compensate for the leak in the detector. About half of the 470 V reduction would have been needed to compensate for the reduction in pressure. The status of LAXPC10 at any time can be checked on the LAXPC website4. Even if the detector remains stable, it would take a few years for it to reach a reasonable gain due to leakage. Because of the reduction in HV the energy thresholds of LAXPC10 have increased. At the time of last adjustment, the LLD was around 30 keV and ULD was about 400 keV. Since then because of the fine leak the thresholds have reduced to some extent, with LLD around 22 keV and ULD around 320 keV. It is difficult to estimate the gain of this detector reliably and to get the corresponding response. As a result, it is not possible to use this detector for spectroscopic 4

https://www.tifr.res.in/*astrosat_laxpc/laxpc10.pdf.

studies. During the period just after March 26, 2018 the gain of the detector was in a reasonable range. As a result, the data obtained during that interval could be analysed if single event mode for only the top layer of the detector is used. It is necessary to reject all double events where the energy is deposited in two different anodes due to Xe K X-ray photons being absorbed in a different anode, as the energy thresholds for choosing these events have not been adjusted due to difficulty in estimating them reliably. The restriction to top layer of detector is needed because the LLD threshold of some other anodes has been increased giving an edge in the spectrum. Even with the low gain, LAXPC10 does detect bursts, e.g., from GRB (Antia et al. 2020a, b). Similarly, it is possible to detect pulsation in LAXPC10, e.g., for GRO J2058?42 during its outburst in April 2019, the pulsation period was determined to be 194:256  0:034 s and spin-up rate was estimate to be m_ ¼ ð1:7  1:0Þ  1011 Hz s1 . This can be compared with P ¼ 194:2201  0:0016 s and m_ ¼ ð1:65  0:06Þ  1011 Hz s1 obtained with LAXPC20 (Mukerjee et al. 2020a). The error-bars represent the 90% confidence limits. The higher error in LAXPC10 is mainly due to lower counts because of higher LLD and low efficiency. This observation was taken at a time when the counts were not very stable in LAXPC10 and only about 7% of exposure time was usable. The pulse profile obtained from LAXPC10 is shown in Fig. 3. The gain of the detector during this observation is uncertain and the LLD was probably

J. Astrophys. Astr. (2021)42:32

Page 5 of 21

32

Figure 3. The pulse profile of GRO J2058?42 as calculated from LAXPC10 data.

around 30 keV. This pulse profile can be compared with pulse profile obtained for energy range 30–40 keV from LAXPC20 (Mukerjee et al. 2020a). The energy resolution of LAXPC10 was stable until March 2018, while that for LAXPC30 was improving with time, probably due to reduction in pressure. The energy resolution of LAXPC20 has been deteriorating with time and currently the resolution at 30 keV is about 20%. The position of peak channel in LAXPC20 has been steady and the last adjustment in HV was made on March 15, 2017. Since then the peak position has reduced by about 12 channels. Since this detector is already operating at higher voltage (about 2600 V) as compared to the other two, further increase in HV has not been attempted. Currently, this is the only detector that is working nominally.

2.3 Fit to spectra of Crab observations AstroSat has observed the Crab X-ray source several times during the last five years and the spectra obtained during these observations have been fitted to monitor the stability of detector response after accounting for known drift in gain and pressure using appropriate responses. The Crab spectra were fitted to a power-law form to obtain the spectral index and normalisation for each observation (averaged over all orbits) and the results are shown in Fig. 4. All fits were performed with 3% systematics in spectra and background, and line-of-sight absorption column density of 0:34  1022 cm2 was used (Shaposhnikov et al. 2012). Due to the anomalous behaviour and abnormal gain change, LAXPC10 and LAXPC30 data were not fitted for 2018 onwards and hence are omitted from respective plots. The gain fit was also used in Xspec v 12.11.1 to allow for small deviations in the gain of responses. The effect of using the gain fit during the last five years for LAXPC20 is shown in Fig. 5 where fitted spectral parameters with and without using gain fit are shown for comparison. It turns out that the slope of best fit for

Figure 4. The results of fit to Crab spectrum observed during the last 5 years is shown as a function of days from launch of AstroSat for the three LAXPC detectors as marked in the figure. The different panels show all fitted parameters as well as the reduced v2 for the fit.

LAXPC20 was always within 2% of unity, which implies that this is largely taken care of in gain shift estimated from the calibration source. However, the offset in gain fit was found to change systematically with time, reaching a value of 0:5 keV by now. This may be expected, as this was not calculated from the calibration source. The inclusion of gain fit improved the fit significantly and is recommended for all spectral fits. It can be seen that the fitted parameters have held steady during the last 5 years, but the v2 for the fits have increased with time. Initially, 1%

32

Page 6 of 21

Figure 5. The results of fit to Crab spectrum observed by LAXPC20 during the last 5 years is shown as a function of days from launch of AstroSat. The results with and without applying the gain fit are shown.

systematics was enough to get an acceptable v2 fit, but now it requires up to 3% systematics for LAXPC20. This could be because of degradation in energy resolution. An example, of the fit for observation during September 2020 is shown in Fig. 6. An estimate of systematic error in fitted parameter for Crab spectra can be obtained by taking the standard deviation over all measurements. For the power-law index we get the values, 2:099  0:041, 2:088  0:013 and 2:136  0:043 for the three LAXPC detectors. Similarly the normalisation is 8:15  0:74, 8:10  0:31 and 8:90  0:89. Some of the variation in normalisation could be due to differences in pointing offset over different observations. It is clear that the fitted parameters for the Crab spectrum are stable over a long time scale. Figure 7 shows the variation over a short time scale of a few days during January 2018 using data for individual orbits. It is clear that there is a diurnal variation in the fitted parameters, which is similar to that seen in the background as shown in the next section. This variation is likely to be due to variation in the background and discrepancy between the background estimated from background model and the actual background. Some of the variation could be due to a shift in GTI with orbit. The period of Earth occultation would drift with time across the AstroSat orbital phase and that may account for some of these variations. As a result some orbits will have more contribution from the region near the SAA passage and these may have diurnal variation. Since the LAXPC spectra often show an escape peak around 30 keV due to Xe K X-rays, we also attempted a fit with an additional Gaussian peak around this energy to account for this

J. Astrophys. Astr. (2021)42:32

Figure 6. A fit to the Crab spectrum observed by LAXPC20 during September 2020 to power-law model with gain fit is shown. The lower panel shows the residuals. The reduced v2 of the fit with 3% systematics is 2.03.

Figure 7. An example of the diurnal variation observed in the fitted Crab spectral parameters from LAXPC10 (shown by triangles) and LAXPC20 (shown by circles) during one observation during January 2018 as a function of time.

feature and the results are shown in the Fig. 8. This resulted in some improvement in the fit and also reduced the amplitude of diurnal variation, but significant variations were still seen. The addition of a Gaussian component has been used in some analysis of LAXPC data to remove the feature in the spectrum around 30 keV (Sreehari et al. 2019; Sridhar et al. 2019). To identify the cause of the observed diurnal variation in the fitted parameters we repeated the exercise for the January 2018 Crab observations, by removing the time intervals that were within 600 s of entry or exit from SAA. This should reduce the background uncertainties. With this modification the diurnal trend is not clear as shown in Fig. 9. These fits included the

J. Astrophys. Astr. (2021)42:32

Page 7 of 21

32

3. Detector background

Figure 8. The figure shows the results of fitting the Crab spectra observed by LAXPC20 during a long observation during September 2020. Marginal improvement of spectral parameters as well the fitting statistics can be noted when a power-law with a Gaussian at Xe K X-ray energy  30 keV (shown by solid circles) is used compared to a simple power-law (shown by hollow circles).

Figure 9. The figure shows the results of fitting the Crab spectra observed by LAXPC20 during a long observation during January 2018 after removing the time-interval near the SAA passage to reduce uncertainties in background estimates. The fit includes gain fit and a Gaussian around 30 keV.

gain fit as well as a Gaussian around 30 keV and hence these results should be compared with those in Fig. 8. However, with reduced exposure due to truncating the GTI, the errors in fitted parameters are larger and the net range of fitted parameters is not substantially reduced. It appears that some of the diurnal variation could be due to uncertainties in background model discussed in the next section. Although, Crab flux is much larger than the background at low energies, at high energies it becomes comparable to background and hence can be affected by uncertainties in background.

To determine the detector background, the instrument is pointed to a direction where there are no known X-ray sources (Antia et al. 2017). Since the background is found to show a variation over a time of about 1 day, all background observations are for at least, 1 day interval. The background counts are found to change during the orbit also with the counts increasing near the SAA passage. These variations are fitted to latitude and longitude of satellite as explained by Antia et al. (2017). To monitor the long term variation in the background, the background observations are repeated about once every month. However, the variation in the background counts and spectrum are too complex to be captured by the models and some of these complexities are described in this section. The gain drift in detector would also result in change of background and for spectral studies the background spectrum is corrected for this shift in the software. Even after correcting for the gain shift, the background counts have been changing with time and the results are shown in Fig. 10. The LAXPC10 results are shown until March 2018 only, as after that the gain has changed significantly. LAXPC30 results are not shown as the counts were decreasing due to reduction in pressure and it is difficult to correct for large variations in the gain. It can be seen that the variation is similar in both detectors during the overlapping time and the counts have been generally increasing with time. The reason for this increase is not clear. Some increase may be expected from induced radioactivity, but it is not clear why it becomes nearly constant over some time intervals. There is also a significant scatter about the best fit curve, which could be due to various factors discussed below. Figure 10 shows the long term variation in the total count rate during the background observations, but there are short-term variations also during each orbit as well as some diurnal variations during the course of a day. To show these variations, Figure 11 shows the variation in the count rate for a long background observation during July 2018. The diurnal variation can be seen clearly in this figure. Figure 12 shows the light curves during a few background observations during the last five years. It is clear that the diurnal variation is present in all observations but there is some evidence that the amplitude of variation has increased with time. Background model used to generate the light-curve for background has an option to remove the diurnal variation with a period of 1 day, which can be applied if the

32

Page 8 of 21

J. Astrophys. Astr. (2021)42:32

Figure 10. The total count rate in background corrected for gain shift as a function of time for LAXPC10 and LAXPC20.

observation covers at least 1 day of observation. For short observations it tends to remove real trend in the light-curve and hence is not applied. For longer observations also it can remove some real variation if it happens to have similar periodicity. During an orbit the counts are generally minimum in between the two SAA passages and tends to increase as the satellite approaches the SAA, as well as when it exits the SAA. However, the magnitude of variation near SAA passage is more complex. In general, for orbits where the counts are near maximum in the diurnal trend, the counts have a sharp spike as the satellite exits the SAA, while the spike when it approaches the SAA is of much smaller magnitude. On the other hand, for orbits where the counts are near minimum of the diurnal trend, the two branches, towards the entry to SAA and while exiting the SAA are comparable in magnitude. It is possible to avoid some of these artefacts by cutting off the interval surrounding the SAA passage from GTI. However, the software does not implement this as it can reduce the exposure time significantly, which may not be desirable in some studies, e.g., study of bursts. For more sensitive studies it may be advisable to remove 600 s on either side of SAA from the GTI. All these figures only show the total background count rate, but the spectrum does not simply scale by this rate and hence to look for the variation in the background spectrum, we selected the background observations during February 2017 and September 2020 and calculated the spectrum during a few parts of the observation. For reference, we use the spectrum obtained during an orbit when the count rate was close to the minimum (referred to as ‘low’) of diurnal variation, and choose another orbit when the count rate was close to maximum (referred to as ‘high’). Here the orbit is defined as the period between two

Figure 11. The light curve for a background observation during July 2018 in LAXPC20 with a time-bin of 32 s.

Figure 12. The light curves for different background observations as marked with the month and year in the respective panels, as observed by LAXPC20 with a time-bin of 32 s.

consecutive passages through SAA. Figure 13 shows the ratio of counts in the spectrum with respect to the average spectrum during the ‘low’ orbit. The red curve in the top panel shows the ratio when spectrum is averaged over the entire ‘high’ orbit, while the red curve in the bottom panel shows the ratio when the spectrum is averaged over the entire observation

J. Astrophys. Astr. (2021)42:32

Page 9 of 21

32

Figure 13. The ratio of spectrum during different times of background observation with respect to that during a ‘low’ orbit for LAXPC20. The left panel shows the result for February 2017 background observation, while the right panel shows that for September 2020 observation. The red line in top panels show the ratio averaged over ‘high’ orbit, while that in the bottom panel shows the ratio for entire observation, including all orbits. The black lines show the ratio during the first 600 s of the orbit, the cyan line shows that during the last 600 s, while the blue line shows that during the middle part of the orbit.

covering all orbits. The black lines show the spectrum during the first 600 s of the orbit just after the satellite comes out of SAA. The cyan line shows the ratio when the spectrum is taken over the last 600 s before the satellite enters the SAA, while the blue line shows the ratio for spectrum during the middle part of the orbit, leaving out 900 s on both sides. The left panel shows the results for February 2017 observations, while the right panel shows that during September 2020. It can be seen that during the ‘low’ orbit the difference is generally within 10%. However, during the ‘high’ orbit all curves show a higher ratio and also show a peak around 30 keV which is the Xe K X-ray energy. Further, during the initial part of the orbit the flux is much higher at all energies with the maximum difference exceeding 50% for September 2020 observation. We have checked that even if we restrict to first 300 s of the orbit the ratio is only sightly higher. Further, for ‘high’ orbit the blue and cyan curves are close, indicating that there is not much difference in the spectrum as the satellite enters the SAA. On the other hand, for ‘low’ orbit the counts increase as the satellite is about to enter the SAA region. It turns out that the ‘high’ orbits are the ones where the passage through SAA occurs when the satellite is near the south end of its range. Although, the result is not shown, the ratio of the spectrum during the two ‘low’ orbits is roughly consistent with the expected ratio from Fig. 10 except for a peak around 30 keV. We have also looked at similar ratio using individual layers of the detector and the behaviour is similar to that in Fig. 13. However, the count

rate in the top layer is about twice that in other layers. Hence, the additional counts are larger in the top layer as compared to other layers. Since the additional counts are larger after exit from SAA, some of these could be due to induced radioactivity, while there could be additional contribution from charged particles coming through the collimator. Thus it is clear that there is a significant change in the background spectrum during later times and most of the increase in background appears to be during ‘high’ orbits and for energy around 30 keV. The background model used in the software does account for the increase in the count rate during the ‘high’ orbit to a large extent, but the spectrum is scaled to the average counts and hence is likely to introduce a bump around 30 keV. Typical rms deviations in the fit to count rate are 10 s1 when total counts in all anodes are considered. For top layer in 3– 20 keV energy range this drops to about 1 s1 , which is comparable to statistical error for a time-bin of 32 s, used in these fits. This may be expected as the deviations are more prominent in energies above 20 keV. Figure 14 shows the residuals in the background fit for the two observations described above using the background model described by Antia et al. (2017). It can be seen that for the restricted energy range using only top layer, the residuals are roughly consistent with the statistical errors, except for the high orbits during November 2020 observations. However, when all events are considered the background model has significant residuals. Thus it is clear that for faint sources it would be advisable to consider only top

32

Page 10 of 21

J. Astrophys. Astr. (2021)42:32

Figure 14. The residuals in the fit to background of LAXPC20 for the two background observations obtained using the background model described by Antia et al. (2017). The left panel shows the residuals in the light curve with a time-bin of 32 s, while the right panel shows the residuals in the energy spectrum.

Figure 15. The residuals in the fit to background for the two background observations obtained using the background model for faint sources described by Misra et al. (2021) which uses only the top layer of the detector. The top panels shows the residuals in the light curve with a time-bin of 100 s, while the bottom panels shows the residuals in the energy spectrum. The left panels show the results for February 2017 background observation while the right panels show the same for September 2020 observations.

layer of the detector with restricted energy range. Figure 15 shows the residuals in the background fit for the same two observations using the background model for faint sources described by Misra et al. (2021) which uses only the top layer of the detector. Since the background model performs better when only top layer of the detector at low energies is used, it is advisable to use this option for faint sources. This is

justified as at low energies a large fraction of counts are registered in the top layers. Figure 16 shows the energy dependence of relative fraction of events in top layer (Anodes 1 and 2) and the top two layers (Anodes 1 to 4) as compared to all layers (Anodes 1 to 7). It can be seen that up to about 10 keV, almost all events are registered in the top layer. Even at 20 keV about 50% of events are registered in the top layer and about 75% in the first two layers. Thus for studies at low

J. Astrophys. Astr. (2021)42:32

Figure 16. The relative efficiency of only top layer and the top two layers of LAXPC detector as compared to when all layers are used.

energies it is advisable to use only the top layer of the detector, as it reduces the background. An alternative is to drop the data during the ‘high’ orbits to avoid this contamination. This would reduce the duty cycle of the observations and if the observation is for a short duration, the entire part may be during the ‘high’ orbits. The fluctuations in the background determine the sensitivity of LAXPC for faint sources, as discussed in the next section. Since the detector background increases close to SAA passage, the contribution is most likely from the flux of charged particles. Although, the detector has veto anode and shield on three sides to protect from charged particles entering from these sides. On the two smaller faces there is only a shield which would offer some protection. However, on the top side there is no veto layer or shield, unlike that in RXTE/PCA which had a propane layer (Jahoda et al. 2006). As a result, there is little protection other than a thin Mylar window, against charged particle coming from the top along the collimator. The opening angle of collimator is about 1 which gives a solid angle of about 0.0003 st. Thus even for a flux of 1 particle cm2 s1 st1 , the detector with an effective area of 2000 cm2 would record 0.6 event per second. This level of flux is entirely possible even outside SAA, while close to SAA the flux could be larger. During a geomagnetic storm the charged particle flux goes up and many more counts are recorded. Figure 17 shows the count rate in detectors during two geomagnetic storms on September 8, 2017 (Kp ¼ 8þ ) and May 28, 2017 (Kp ¼ 7). The Kp index in the parentheses gives a measure of strength of geomagnetic storm. The September 8, 2017 event was the

Page 11 of 21

32

strongest so far during the AstroSat operation. The red points marks the time which would be outside the GTI due to normal SAA definition and would not be considered for analysis. Thus it can be seen that geomagnetic storms can yield significant counts near SAA passage. The most affected regions in this case are those where SAA passage occurs during the north end of the range. For weaker storms, the effect will not be easily seen in light-curve as the net increase would be smaller but such storms can be frequent during high activity part of the solar cycle. Even for smaller geomagnetic storms, of the order of 10 cm s1 can be added before the SAA passages. If such an event occurs during observation of a variable source, it would be almost impossible to separate it out from variation in source counts. Any burst seen close to SAA passage can be suspect and would require further investigation. During the last two years the solar activity has been low and such storms are rare, but in the coming year the solar activity is likely to pick up and more such events would be seen.

4. Detector sensitivity and limitations There is no difficulty in studying bright sources. The flux limit for faint sources is determined by the fluctuations in the detector background and ability of background model to match these variations. Even for long observations where the spectrum is averaged over a long time, the intrinsic variations in background cannot be modelled satisfactorily and it limits the sensitivity of the detector for faint sources. From the discussion in the previous section we can see that at least, a few counts per second from the source would be required to get any meaningful results. The actual limit would obviously depend on the level of details that need to be studied and the fluctuation in background during actual observation. The Crab observation yields a count rate of about 3000 s1 in each detector, which gives a sensitivity limit of about 1 mCrab for faint sources that can be studied. For reference, low energy flux of about 1011 erg cm2 s1 gives a count rate of 1 s1 . The same difficulty arises even for relatively bright sources at high energies where the count rate can be much less than that in the background. The upper limit on energy to which the spectrum can be studied depends on the source and the extent of details that are required. In the following subsections we illustrate some limitations and capabilities of LAXPC for studying different properties of

32

Page 12 of 21

J. Astrophys. Astr. (2021)42:32

Figure 17. The light curve during geomagnetic storms on on September 8, 2017 (left panel) and May 28, 2017 (right panel). The red points show the part of the light curve which would be outside the GTI.

X-ray sources. Instead of giving limit on flux etc., we illustrate the capabilities and limitations by giving some examples. All the results presented in this section are obtained using the software described in Section 5.2 with default parameters, except for choice of energy range and anodes as specified.

4.1 Light curve and spectrum For faint sources typically most flux is observed at low energies and it is better to use only the top-layer of the detector and further restrict the energy range to 3–20 keV. This reduces the background counts by a factor of 10, thus improving the signal-to-noise ratio. To illustrate the limitation we have selected an AstroSat observation of an Active Galactic Nuclei (AGN), NGC 4593, on July 14, 20165. Figure 18 shows the LAXPC light-curve using only top layer and energy range of 3–20 keV. This source shows a count rate of about 5 s1 in all three detectors. For comparison the light curve at low energies from the SXT instrument is also shown. All light curves are with a time-bin of 100 s to reduce the statistical error. It can be seen that all LAXPC detectors and the SXT instrument show similar variation. The SXT being an imaging instrument has low background and further it operates at lower energy of 0.3–10 keV, where the flux is larger. Hence, it is expected to give reliable light-curve for this source. The cross correlation function crosscor of HEASoft v 6.28 tool was used to estimate the time lag between the LAXPC20 (3–20 keV) and SXT (0.5–3.0) keV which is shown in Fig. 19. A standard plot of cross-correlation versus time delay was generated 5

ID: 20160714_G05_219T01_9000000540

using these energy bands with time resolution of 99.86 s with 512 intervals. Two Gaussian models were fitted to the cross-correlation to evaluate the time lag. The resulting fit shows a clear time delay of about 400 s indicating that hard photons lag behind the soft photons (Brenneman et al. 2007). Figure 20 shows the spectrum observed in each of the LAXPC detectors. There is a reasonable agreement between the three detectors at low energies. The LAXPC20 spectrum continues at high energy also, probably because the background is more reliably estimated. To test the spectrum, the combined spectrum from SXT and LAXPC20 covering 0.5–80 keV was fitted using Xspec to the model phabs*(diskbb ? gaussian ? powerlaw) and the resulting fit is shown in Fig. 21. The model fit yielded a reduced chi-squared of 1.45 (639/442) with 1.5% systematics. Also, significant residuals were seen in the 10–30 keV range, which might be due to a broad reflection component. The disk temperature (Tin ) and photon index (C) obtained were 0:13  0:01 keV and 1:56  0:01, respectively, which are consistent with the results of Ursini et al. (2016).

4.2 Pulsation Several X-ray pulsars have been studied by LAXPC and in general there is no difficulty in estimating the frequency and spin-up rate if the change in frequency is significant. To illustrate the performance, we consider the AstroSat observation of SMC X-2 on May 7, 2020 (ID 20200507_T03_205T01_9000003652), when the average count rate from the source was 2.1 s1 . Nevertheless, it was possible to estimate the spin

Page 13 of 21

32

Photons cm−2 s−1 keV−1

5 10 0 0.5

105

0

2×105 Time (s)

(data−model)/error

0 2 4 6 8 2 4 6 8 0

LX30 (c/s)

LX20 (c/s)

LX10 (c/s)

SXT (c/s)

1

J. Astrophys. Astr. (2021)42:32

3×105

10−3 10−4 10−5 10−6 4 2 0 −2 1

10 Energy (keV)

Figure 21. The fit to the combined spectrum from SXT and LAXPC20 for NGC 4593 is shown with the bottom panel showing the residuals.

0.1 0.05 0

Cross Correlation

0.15

0.2

Figure 18. The light curve of NGC 4593 during the AstroSat observation starting on July 14, 2017 in SXT and the three LAXPC detectors with a time-bin of 100 s.

0.01

−1000

0

1000 Time Delay (s)

2000

3000

Figure 19. The cross-correlation between SXT and LAXPC20 light-curve of NGC 4593 is shown as a function of time delay. The red line shows the fit with 2 Gaussians.

0.1 0.01 10−3

normalized counts s−1 keV−1

1

LX10 LX20 LX30

5

10

20 Energy (keV)

50

Figure 20. The spectrum of NGC 4593 during the AstroSat observation starting on July 14, 2017 in the three LAXPC detectors.

Figure 22. The pulse profile of SMC X-2 as obtained by LAXPC20 in the energy range 3–20 keV.

& Naik 2017). Thus, it is clear that even with a low count rate it is possible to study pulsations. Apart from coherent pulsation, LAXPC has been extensively used to study Quasi Periodic Oscillations (QPO) in a wide variety of sources, with frequency ranging from 1 mHz in 4U 0115?63 (Roy et al. 2019) to 815 Hz in 4U 1907?09 (Verdhan et al. 2017). The only problem with detecting QPO arises if their frequency is close to the orbital frequency of AstroSat (  0.15 mHz) or its harmonics. Up to 10 harmonics of orbital frequency can be easily seen in the power density spectrum and need to be accounted for while identifying QPO frequencies. Similarly, for X-ray pulsars there can be interference from pulse frequencies or its harmonics. But this can be easily removed by modelling the pulse including harmonics and removing their contribution, e.g., for GRO J2058?42 (Mukerjee et al. 2020a).

4.3 Cyclotron resonant scattering features period of 2:377441  0:000016 s at the beginning of observation (MJD 58976.575405) and the spin-up rate of ð3:9  1:1Þ  1011 Hz s1 using LAXPC20 in single event mode and energy range of 3–20 keV. The resulting pulse profile shown in Fig. 22 can be compared with other observations (Li et al. 2016; Jaiswal

Detection and studies of Cyclotron Resonant Scattering Features (CRSF) has always been of great interest for direct measurements of magnetic field of the neutron stars and to understand structure of the line forming regions in the accretion column in X-ray

Page 14 of 21

6

ID: 20171107_A04_024T04_9000001670.

0.1 0.01 5–30 solar mass). In the X-ray binaries, the compact objects may be either a neutron star (NSXBs) or a black hole (BHXBs). Among all the wavebands, the most rapid and violent brightness changes are observed in the X-ray emitting sources. X-ray binaries (NSXBs or BHXBs) exhibit intensity variations over a wide range of time scales ranging from millisecond to months and years. In a majority of the neutron star LMXBs, the magnetic field is weak (6109 G) but a small number of the neutron star LMXBs have higher magnetic fields >109 G (  1012 G). The well known binary Her X-1 is an example of the LMXBs in which the neutron star has a high magnetic field as inferred from the presence of the cyclotron feature in its energy spectrum. These systems also exhibit regular X-ray pulsations similar to those in the HMXBs. The LMXBs with low magnetic field neutron stars, are thought to have evolved from high magnetic field neutron star binaries in which the accretion has spun up the neutron stars over a long period during which the magnetic field has significantly decayed. The accreting

J. Astrophys. Astr. (2021)42:40

millisecond pulsars (spin periods  ms) are an example of the spun up neutron stars from their progenitor LMXBs. The LMXBs with weak magnetic field, are characterised by short orbital periods and sporadic or regular short X-ray bursts (strictly not periodic), that have a typical duration of about a few seconds to a minute or in rare cases longer. The matter accretes from the companion star on the neutron star via Roche Lobe overflow leading to the formation of an accretion disk around the neutron star. From the inner part of the accretion disk, the matter falls on the surface of the neutron star. The accreted matter piles up on the surface of the neutron star forming a layer with increasing thickness. As the accretion rate is variable, thickness of the layer is not uniform. When the accumulating matter on the neutron star surface develops density and temperature conditions in a localized region that ignite thermonuclear reactions a powerful X-ray burst occurs. These are called thermonuclear bursts. Initially the flame is localized in a spot but it spreads quickly over the neutron star surface. Due to uneven spread of the nuclear burning, the X-ray intensity is modulated due to rapid spin of the neutron star. The intensity oscillations with frequency of several hundred to about a kHz are detected in the LMXB bursts. This is one of the ways to measure the spin periods of the neutron stars in the LMXBs. Using the RXTE/PCA data, kHz QPOs were discovered from about a dozen LMXBs (usually occurring in pairs) (van der Klis 2011). The precise origin of kHz QPOS is still not well understood. The LMXBs have multi-component continuum energy spectra consisting of a disk blackbody component and a thermal Compton power-law. The thermal Compton part of the spectrum is most likely due to the Compton up scattering of X-rays by hot electrons in a halo of accreted material surrounding the disk. A precise modeling of the continuum is essential to detect weak absorption features known as Cyclotron Resonant Scattering Features (CRSFs) or commonly referred as Cyclotron lines, that originate in the accretion column above the magnetic poles. The CRSFs are detected in the strong magnetic field HMXBs as well as in the spectra of the strong field LMXBs. In the neutron stars with strong magnetic fields, the accreting matter is guided from the disk by the magnetic field lines of the neutron star to its magnetic poles. The X-ray spectrum emitted by the plasma in the accretion column is affected by the magnetic field. This interaction gives rise to the CRSFs. The energy of Cyclotron lines provides a

J. Astrophys. Astr. (2021)42:40

direct measure of the dipole magnetic field of the neutron star in the accretion column region. The HMXBs usually have a pulsating X-ray source. The pulsation periods have a range from  1 s to hours. The slowest pulsar 2S 0114?65 has a spin period of 2.7 hours. The spin rate of the pulsars is variable and depends on the accretion rate. A study of the evolution of the pulsation periods of many pulsars, shows that most pulsars exhibit both spin up as well as spin down episodes. Depending on the accretion rate, a pulsar may make transition from spin up to spin down and vice versa. In a class of X-ray binaries, the optical star is a Be star with a shell or disk of matter ejected by the companion. Whenever the neutron star crosses the shell or disk during the course of its orbital motion, the accretion rate shoots up resulting in dramatic increase in brightness known as X-ray outbursts that occur once or twice in an orbital period. Some times there is a sudden catastrophic release of matter from the disk or shell of the Be star. This may be due to thermal instability that has still not been well understood. This leads to a massive spike in the accretion rate resulting in a giant outburst which is non-periodic and unpredictable. This phenomenon also manifests in many X-ray transients that remain in a dormant state with an exceedingly low accretion rate rendering them invisible. This hibernation may last for years and even decades. Then due to reasons still not understood, they spring back to life with giant X-ray outbursts lasting for tens of days to months. QPOs of  mHz to tens of Hz have been detected in many HMXBs. Study of these QPOs provides insight into the radiation environment closest to the neutron star. In black hole X-ray binaries (BHXBs), accretion disk and relativistic radio jets are integral part on all black hole mass scales and they provide simple scaling of time and length with the mass of the black hole from supermassive black holes in active galactic nuclei (AGNs) to stellar mass black holes in BHXBs in our Galaxy. When accretion rate from the companion star is not sufficient to support continuous accretion flow to the black hole, matter fills the outer disk until a critical surface density is reached and an outburst is triggered. An outburst in the BHXBs may last from  20 days to several months. There are over 20 confirmed BHXBs and many more candidates (Remillard & McClintock 2006). Out of 20 BHXBs, 17 are transient X-ray sources (mostly LMXBs) and the rest three are persistent bright X-ray sources which have a massive O/B type stars as their companion (HMXBs) (Remillard & McClintock 2006). Many new BHXBs have been discovered with time

Page 3 of 20

40

(Mudambi et al. 2020a; Bhargava et al. 2019), and MAXI shows now light curves of around 40 BHXBs. The BHXBs show different X-ray states at different stages of an outburst. Initially, luminosity was the sole criteria to determine X-ray states (van der Klis 1994; Tanaka et al. 1995). With the availability of larger data sample during RXTE/PCA era, it became clear that X-ray states are not simple function of the luminosity (Lewin & Van der Klis 2006; Remillard & McClintock 2006). It is suggested that beside the mass accretion rate, there must be at least one additional parameter which drives X-ray state transition. The hardness-intensity diagram (HID) which shows a q-shaped track, has been used to study the evolution of outbursts and associated X-ray states in the BHXBs systems. The four distinct X-ray states identified in the HID are: (1) (2) (3) (4)

Low Hard state (LH), High Soft state (HS), Hard InterMediate State (HIMS), and Soft InterMediate State (SIMS) (Belloni 2006).

The LS state is associated with relatively low accretion rate (lower than other bright X-rays states) but can be observed at all luminosity. It is a radio-loud state. The energy spectrum is dominated by a hard power-law while power spectrum shows a strong band-limited noise. The HS state is strongly dominated by thermal disk component. Radio quenching is seen during this state. The intermediate states (HIMS and SIMS) have significant contributions from both the hard power law and the thermal disk components and show the most complex variability characteristics including most of the quasi-periodic oscillations (QPOs). Transient radio jets are seen during HIMS to SIMS transition. Remillard and McClintock (2006) provided alternative scheme of X-rays states in BHXBs in terms of Hard State (HS), Thermal State (TS) and Steep Power Law (SPL) state. Here major difference is the SPL state which has significant contributions from thermal as well as non-thermal emissions.

1.2 LAXPC instrument and its features to study temporal and spectral properties Large Area X-ray Proportional Counter (LAXPC) instrument is one of the major instruments onboard AstroSat which is India’s first Space Science Observatory (Agrawal 2006). There are three X-ray

40

Page 4 of 20

J. Astrophys. Astr. (2021)42:40

instruments on-board AstroSat which cover a wide energy band. These X-ray instruments are: (i) LAXPC instrument (Yadav et al. 2016b, 2017; Agrawal et al. 2017) covering 3–80 keV region, (ii) A Cadmium–Zinc–Telluride Imager (CZTI; Bhalerao et al. 2017) array covering 30–100 keV, and (iii) A Soft X-ray Imaging Telescope (SXT; Singh et al. 2016) covering 0.3–8 keV. Beside these X-ray instruments, there is an Ultra-Violet Imaging Telescope (UVIT: Tandon et al. 2017). All the above instruments are co-aligned to provide simultaneous multi wavelength observations from optical to hard X-ray. There is also a Scanning Sky Monitor (SSM) onboard AstroSat. The LAXPC instrument uses three co-aligned identical LAXPC detectors (LAXPC10, LAXPC20 and LAXPC30) to achieve large area. The performance of LAXPC detectors are discussed by Antia et al. (2021) in the same volume. A high detection efficiency in the entire 3–80 keV energy region is achieved by having a 15cm deep detection volume divided in 5 identical anode layers and filling the Xe?CH4 gas at two atmosphere pressure (1520 Torr). The LAXPC instrument was designed to meet the objectives of studying the intensity variations with time, have a high detection efficiency in 3–80 keV band, provide a large effective area over the broad energy range to be able to study weak (  a few milliCrab) sources and have a good spectral resolution to measure the X-ray continuum spectra of X-ray binaries with precision for deciphering the presence of weak cyclotron lines in the NSXBs. A unique feature of the LAXPC is that every detected photon is time tagged to an accuracy of 10 ls to enable investigation of rapid intensity variations over even sub-millisec scale. This feature provides LAXPC with the capability to measure high frequency QPOs of even a few kHz. Good energy resolution (FWHM  15% at 60 keV) was achieved by using an onboard gas purifier system and continuously monitoring of the resolution of the Veto layer by shining 60 keV X-rays on it from an Am241 source. We have done extensive lab as well as in orbit calibration of LAXPC detectors (Antia et al. 2017). Further improvements and updates on LAXPC detector calibration and background are done (Antia et al. 2021; Misra et al. 2021). Since October 2016, AstroSat observatory has operated on the basis of targets selection through scientific proposals during the observing cycles A02–

Figure 1. AstroSat observations during observing cycles A2–A9 showing percentage of time devoted to different instruments as primary instrument.

A09 in the last 5 years. A comparison chart for the observations with different instruments as primary instrument is shown in Fig. 1. Primary instrument is defined in the AstroSat science proposals (as science proposal requirement). It may be noted that simultaneous data from all the instruments will be available for a given source pointing if all the instruments are on during the observation.

2. Highlight of results on neutron star X-ray binaries (NSXBs) Large number of X-ray binaries with an accreting neutron star as the X-ray source, have been observed with the LAXPC instrument to investigate their timing and spectral characteristics. Due to various reasons, briefly discussed in Antia et al. (2021), less than 50% data of the sources have been analyzed so far. Nevertheless, several new and interesting results have emerged from these studies. Here we summarize highlights of some of these results.

2.1 Timing properties: spin periods and their evolution, low and high frequency QPOs The spin periods of the neutron stars are affected by the accretion rate and the resulting accretion torque that varies with times. Usually the low and high frequency QPOs are found in the LMXB pulsars while the low frequency QPOs are more common in the

J. Astrophys. Astr. (2021)42:40

Figure 2. Evolution of the pulsar period of 4U 1909?07 during 2001–2017.

HMXB pulsars. AstroSat has studied several neutron star binaries to investigate their spin rates and their evolution with time. LAXPC has observed a large number of HMXBs to investigate their periodic and aperiodic variations like determination of the spin periods and their time evolution, QPOs and their origin. We summarize some of the significant results derived from AstroSat. The LMXB 3A 1822-371 has an orbital period of 5.2 h and pulsates with 0.59 s period (Jonker & van der Klis 2001). The spin period from 1996 and 1998 data leads to a spin-up rate of 2:85  1012 s s1 . Subsequently measurements of the spin period from several satellite observations have shown that the pulsar is monotonically spinning up implying that the accretion rate has been constant over  20 years. AstroSat LAXPC observations of 3A 1822-371 in September 2016 yielded a barycenter corrected pulsar spin period to be Pspin ¼ 0:5914907  0:0000003 s confirming that the pulsar is still spinning up (Amin & Chakroborty 2020). Making a linear fit to all the measured spin periods gives a spin up rate ¼ ð2:62  0:02Þ  1012 s s1 giving a spin up timescale ðP=PÞ ¼ 7169 yr in agreement with the previous results reported by Jain et al. (2010). AstroSat also studied accretion powered 2.26 millisecond pulsar SAX J1748.9-2021 (Sharma et al. 2020), when a short and faint outburst occurred in it in 2017, with the SXT and LAXPC instruments. From the spectral and timing analysis of the LAXPC data, the best-fitting orbital solution for the 2017 outburst was derived. Using this an average local spin frequency of 442.361098(3) Hz was obtained. The pulse profile obtained in 3–7 keV and 7–20 keV gave a

Page 5 of 20

40

constant fractional amplitude  0.5% in contrast to earlier reported energy dependent profile. The combined SXT?LAXPC spectrum in 1–50 keV showed the source to be in a hard spectral state. This spectrum is best fitted with a single temperature blackbody and thermal Comptonization model. Time-resolved analysis of the bursts revealed complex evolution in emission radius of blackbody for the second burst suggestive of a mild expansion of photospheric radius. The HMXB pulsar 4U 1909?07 studied with the AstroSat-LAXPC observation of 2017 July, showed the presence of 604 s X-ray pulsations. Using the spin period measurements obtained earlier with the various X-ray satellites and the period deduced from the present work, a plot of spin period versus time is shown in Fig. 2 (Jaisawal et al. 2020). Despite a brief episodes of spin-down, there is a clear long term trend of spin-up. The pulse period changed from 604.7 to 603.6 s between 2001–2017 resulting in an average spin-up rate of 1.71 ± 109 s s1 . The pulse profiles show strong energy dependence evolving from a complex broad structure in soft X-rays into a profile with a narrow emission peak followed by a plateau in energy ranges above 20 keV. The change in the pulse profile suggests a change in the beaming pattern. Another HMXB 4U 1907?09 with a spin period of  404 s was studied by by using the LAXPC observations of 4th and 5th June (2020). Timing analysis of the LAXPC data yielded a period of 442.33 ± 0.07 s. Two peaks of the pulse profile of 4U 1907?09 shown in Fig. 3(left panel), exhibited different energy dependence, the pulsed fraction of the main peak increased till about 40 keV and decreased after that while the secondary peak disappeared at energy above about 20 keV. Energy resolved pulse profiles created from combined data of the three LAXPCs are shown in Fig. 3(right panel). The pulsar spin-down was found to continue (Varun et al. 2019b). The Be binary with 194 s pulsar GRO 12058?42 underwent a massive outburst in April 2018 and was observed by AstroSat to investigate its timing and spectral features. The SXT and LAXPC showed strong pulsations with a period of 194:2201  0:0016 s, and a spin-up rate of ð1:65  0:06Þ  1011 Hz s1 (Mukerjee et al. 2020). Pulse profiles in 3–80 keV were found to be energy dependent. The Power Density Spectrum (PDS) of the source revealed a 0.090 Hz QPO and its higher harmonics. Another HMXB pulsar OAO 1657-415 with 10.4 day orbital period, was studied by Jaisawal et al. (2021) with two

40

Page 6 of 20

LAXPC observations in March and July 2019. The observations covered orbital phases of 0.681–0.818 and 0.808–0.968. Despite being outside the eclipsing regime, the PDS from the first data did not show any clear signature of pulsation or quasi-periodic oscillations. However, during the second observation, the X-ray pulsations at a period of 37.0375 s. were clearly detected in the orbital phase range 0.808–0.92. The pulse profile of the pulsar from the second observation consisted of a broad single peak with a dip-like structure in the middle all across the observed energy range. The evolution of the emission geometry was probed by constructing the pulse profile in narrow time and energy segments. The energy spectrum of OAO 1657-415 is approximated by an absorbed power-law model with an iron fluorescent emission line. These results are explained in the frame of stellar wind accretion model. Another HMXB X-ray binary pulsar 3A 0726-260 (4U 0728-25) was observed on 2016 May 6–7 with the LAXPC and SXT, after a gap of almost 20 years. The light curves of the binary show strong X-ray pulsations with a period of 103:144  0:001 seconds in 0.3–7 keV with the SXT and in 3–40 keV with the LAXPC. The pulse profiles are energy dependent, and there is an indication that the pulse shape changes from a broad single pulse to a double pulse at higher energy. At energies above 20 keV pulsations with a period 103:145  0:001 seconds are detected for the first time and a double peaked pulse profile is observed from the source. The energy spectrum of the source is derived from the combined analysis of the SXT and LAXPC spectral data in 0.4–20 keV. The best spectral fit is obtained by a power law model with

J. Astrophys. Astr. (2021)42:40

a photon index ð1:7  0:03Þ with high energy spectral cut-off at 12:9  0:7 keV and a broad iron line at  6.3 keV (Roy et al. 2020). Detection of  1 mHz and  1.7 QPO in the Be Xray binary 4U 0115?63 were reported by Roy et al. (2019) using the AstroSat-LAXPC observation of the HMXB. The QPOs were detected on 2015 October 24 during the peak of a giant type II outburst. Prominent intensity oscillations seen at  1 and  1.7 mHz are shown in Fig. 4. Cyclotron resonant absorption feature at 11 keV and its 3 harmonics are detected in the energy spectra. Possible models to explain the origin of the  mHz oscillations are examined. These oscillations bear resemblance to the intensity oscillations observed from some other neutron star and black hole sources and may have a common origin. Current models to explain the instability in the inner accretion disk causing the intense oscillations were examined and found to be inadequate.

2.2 Energy spectra, detection and study of CRSFs LAXPC has observed a large number of HMXBs to investigate their periodic and aperiodic variations and measure their continuum spectra and their characterization to reveal the presence of cyclotron lines (CRSFs). Thus far AstroSat-LAXPC has discovered cyclotron lines in a few pulsars in which earlier there was no report of the presence of CRSF. Study of cyclotron line energy and its profile enable determination of the magnetic field of the neutron star and at the site of their origin in the accretion columns from the following relation:

Figure 3. The pulse profile of 4U 1907?09 has double peak shape shown in the left panel. The energy resolved pulse profiles created from combined data of the three LAXPCs are shown in the right panel.

J. Astrophys. Astr. (2021)42:40

Page 7 of 20 LAXPC 10

2000

COUNTS/SECOND

1000

LAXPC 20

2000

1000

LAXPC 30

2000

1000

5000

104 1.5×104 TIME [MJD 57319.38701260 + ...]

2×104

2.5×104

Figure 4. The background subtracted light curves from the 3 LAXPCs showing clear intensity oscillations of  1 mHz and  1.7 mHz in the X-rays from the HMXB Pulsar 4U 0115?63.

Ec ¼ 11:6ðB=1012 GÞð1 þ zÞ1 keV:

ð1Þ

Here, Ec is the energy of the cyclotron line, B is the magnetic field of the neutron star in unit of 1012 G and z is the gravitational red shift due to neutron star. There are reports of shift of the line energy with time and correlation of this with the X-ray luminosity. The results on this are conflicting and reality of variation of line energy is still an open question. Her X-1 was the first source in which Truemper et al. (1978) discovered a CRSF usually termed as cyclotron line at  40 keV. The CRSF energy was found to vary with pulse phase, X-ray luminosity, the phase of 35-day precession cycle and with time (Staubert et al. 2014). Using data acquired from several X-ray satellites over a 20-year period, the CRSF energy was inferred to decline by 4.5 keV from 1996 till 2012 (Staubert et al. 2014) as seen in Fig. 5 from

Figure 5. Shift in the energy of the cyclotron line in Her X-1. The 2018 observation with the LAXPC gave a value of 37.5 keV.

40

Bala et al. (2020). After 2015, the trend was reversed and the CRSF energy began to recover reaching a value of 37.4 keV (Staubert et al. 2019). Recent analysis of Her X-1 observations in 2018 with the LAXPC shows that since then the CRSF energy appears to be constant around 37.5 keV (Bala et al. 2020; Staubert et al. 2020). Amin et al. (2020) constructed the energy spectrum of the pulsar 3A 1822-371 using LAXPC10 and LAXPC20 data and fitted it using 3 different continuum models. In all the cases an absorption feature at 23 keV was present which is suggested to be a CRSF and if correct this implies a magnetic field strength of B ¼ ð2:7  3:4Þ  1012 G for the neutron star. The energy spectrum of HMXB 4U 1907?09 (spin period of  404 s) was studied by by (Varun et al. 2019b) using the LAXPC observations of 4th and 5th June (2020). A CRSF was detected in the spectrum at 18:5  0:2 keV . Pulse phase-resolved spectroscopy was carried out with 10 independent phase bins and variations of the CRSF parameters with pulse phase were found to be consistent with previous studies. The pulsar GRO J2058?42 was observed with the LAXPC. Its energy spectrum extracted from the LAXPC data revealed presence of 3 CRSFs (Mukerjee et al. 2020). The cyclotron absorption features were detected in (9.7–14.4) keV, (19.3–23.8) keV and (37.8–43.1) keV, one of which is the fundamental line and the other 2 are harmonics The pulse phase resolved spectroscopy of the source showed phasedependent variation in line energy and relative strength of these features. An important finding from AstroSat is the discovery of a Cyclotron line at  22 keV in the LAXPC energy spectrum by of the HMXB pulsar 4U 1538-52 which has an orbital period of 3.75 days and the spin period is currently  527 s (Varun et al. 2019a). The pulse profile is double peaked at low energy and has a single peak in high energy range, the transition taking place around the cyclotron line energy of the source. The CRSF is detected with a very high significance in the phase averaged spectrum shown in Fig. 6. It is one of the highest signal to noise ratio detection of CRSF for this source. A detailed pulse phase resolved spectral analysis with 10 independent phase bins was performed and parameters of the continuum spectrum and CRSF parameters were derived. These show pulse phase dependence over the entire phase with a CRSF energy variation of  13% in agreement with previous studies. An increase in the centroid energy of the

40

Page 8 of 20

Figure 6. The energy spectrum of the pulsar 4U 1538-52 derived from the LAXPC data. Top panel shows the observed spectrum by the black points and the best fitted spectrum by a red line. The middle panel shows residuals when one fits only the continuum spectrum and the bottom panel shows fit with continuum and a cyclotron line at 22 keV.

CRSF observed between the 1996–2004 (RXTE) and the 2012 (Suzaku) observations, is confirmed affirming that the increase in the line energy was a long-term change. AstroSat observed the Be/X-ray binary pulsar SXP 15.3 in the Small Magellanic Cloud (SMC) during its outburst in late 2017, when the source reached a luminosity level of  1038 erg s1 . Timing and spectral analysis between 3 and 80 keV lead to the pulse profile that exhibits a significant energy dependence. The pulse shape changes from a double peaked profile to a single broad pulse at energies [15 keV. The energy spectrum suggests presence of a Cyclotron Resonance Scattering Feature (CRSF) at  5 keV and is independent of the choice of the continuum model. This feature is also detected in the spectrum obtained by the NuStar. Till now CRSF has been reported in about 36 X-ray pulsars in binaries (Maitra 2017; Staubert et al. 2019). To the best of our knowledge, this is the lowest energy cyclotron line confirmed in any pulsar. This indicates a magnetic field strength of 6  1011 G for the neutron star (Maitra et al. 2018).

2.3 Thermonuclear bursts and QPOs in LMXBs The thermonuclear bursts (or commonly called as Type-1 bursts) occur in the LMXBs having a weakly magnetized (\109 G) neutron star. Detailed timing and spectral studies of these bursts provides information about the spin period, temperature of the

J. Astrophys. Astr. (2021)42:40

thermonuclear burning surface and radius of the neutron star. The event mode data in which each photon is time tagged to an accuracy of about 10 microsecond, enables studies of high frequency phenomenon like kHz oscillations, Some significant results on the Type-1 bursts observed with the LAXPC are summarized below. LAXPC capability for high time resolution studies of phenomenon like detection of coherent oscillations, kHz QPOs etc. has been demonstrated by Verdhan et al. (2017). They analyzed the LAXPC observations of 8th March 2016 for the LMXB source 4U 1728-24. A 3 ks data stretch revealed occurrence of a typical Type-1 burst of about 20 s duration. Dynamical power spectrum of the data in the 3–20 keV band, shows presence of coherent burst oscillation whose frequency increased from 361.5 to 363.5 Hz consistent with an earlier result showing the same spin frequency of the neutron star. Our knowledge of the spin periods of the neutron stars in LMXBs can also be derived from detection of the Coherent Burst Oscillations in the initial phase of the bursts. A kHz QPO, whose frequency drifted from  815 Hz to  850 Hz, was detected just before the burst. The QPO was detected below 10 keV as well as in 10–20 keV band, which was not possible with the RXTE. Even for a short observation with a drifting QPO frequency, the timelag between the 5–10 and 10–20 keV bands can be constrained to be less than 100 microseconds. Beri et al. (2019) studied the LMXB 4U 1636-536 using 65 ks LAXPC observation over 2 days and detected seven thermonuclear X-ray bursts including a rare triplet of X-ray bursts. Time-resolved spectroscopy performed during these seven X-ray bursts suggested the presence of Photospheric Radius Expansion in three of these X-ray bursts. A transient QPO at 5 Hz was also detected. No evidence of kilohertz QPOs or coherent burst oscillations, was found in the bursts and may perhaps be due to the hard spectral state of the source. Effects of thermonuclear X-ray burst on non-burst emissions in the soft state of the LMXB 4U 1728-34 was investigated by Bhattacharyya et al. (2018) to understand if a significant fraction of the burst emission, which is reprocessed, contributes to the changes in the persistent emission during the burst. This is important since it can introduce significant systematics in the neutron star radius measurement using burst spectra. Analyzing the bursts data for 4U 1728-34 in the soft state it was concluded that the burst emission is not significantly reprocessed by a corona covering the neutron star.

J. Astrophys. Astr. (2021)42:40

Devasia et al. (2021) detected 5 Type-1 thermonuclear X-ray bursts and one burst-like event in the neutron star LMXB source Cygnus X-2 using LAXPC X-ray data. An energy-resolved burst profile analysis and time resolved spectral analysis for each of the bursts was performed to characterize the burst properties. An evolution of the blackbody temperature and radius is also observed during each burst. A search for Coherent Burst Oscillations gave only upper limits. A study of the hardness-intensity and color–color diagrams show that during the 2016 LAXPC observation, Cygnus X-2 was in the early Flaring Branch (FB). The LMXB 4U 1323-62 is an interesting source from which periodic X-ray dips were first detected by EXOSAT (van der Klis & Jansen 1985; Parmar et al. 1989). It was extensively studied by PCA/ RXTE and a study of 40 Type 1 bursts showed a recurrence time of 2.45–2.59 h which is probably the orbital period of the binary. The LMXB 4U 1323-62 was observed with the LAXPC and detected 6 Type-1 thermonuclear Xray bursts in  50 ks exposure. The time gap between the successive thermonuclear bursts was found to be consistent with the orbital period. Using the linear and orbital quadratic ephemerides, the value of the Orbital period is estimated to be 2.65–2.70 h. The gap observed between the bursts in two case, is nearly double the wait time of consecutive bursts as 2 bursts were missed in the data gaps. The nearly fixed time gap is the time required to accumulate the accreted matter to reach the level at which the thermonuclear reactions set in producing a burst. The light curve of 4U 1323-62 also revealed the presence of two dips. A known low Frequency QPO (LFQPO) was detected at  1 Hz, from the source. However, no evidence of kHz QPO was found. The radius of the blackbody is found to be consistent with the blackbody temperature and the blackbody flux of the bursts (Bhulla et al. 2020). The spectral and timing properties of the atoll source 4U 1705-44 were studied by Agrawal et al. (2018) using 100 ks data from the LAXPC instrument. The source was in the high-soft state during the LAXPC observations and traced out a banana track in the Hardness Intensity Diagram (HID). From the Power Density Spectra (PDS), a broad Lorentzian feature centered at 1–13 Hz and a Very Low Frequency Noise (VLFN) is detected. The energy spectra are well described by sum of a thermal Comptonized component, a power-law and a broad (FWHM  2 keV) iron line having equivalent width (EW  369–512 eV). Only relativistic smearing in the accretion disc can not explain the origin of this

Page 9 of 20

40

feature. A systematic evolution of the spectral parameters, e.g. optical depth, electron temperature etc. is seen as the source moves along the HID. Search of correlation between frequency of the broad Lorentzian and the spectral parameters seems to suggest that the frequency varies with the strength of the corona. The structure of Corona of the well known Z source GX 17?2 was studied by Malu et al. (2020) in 0.8–50 keV using the Soft X-ray Telescope (SXT) and the LAXPC data. For the first time, cross-correlation studies were performed using SXT soft and LAXPC hard light curves and they exhibited correlated and anti-correlated lags of the order of a hundred seconds. Spectral modeling gave disk radius of  12–16 km indicating that disk is close to the ISCO and a similar value of disk radius was deduced based on the reflection model. Corona size was inferred to be 27–46 km and 138–231 km depending on the model used. Size of the X-ray emitting Boundary Layer (BL) was determined to be 57–71 km. The observed lags and no movement of the inner disk front strongly indicates that the varying corona structure is causing the X-ray variation in the Normal Branch (NB) of Z source GX 17?2. Bhulla et al. (2020) investigated evolution of timing and spectral properties of the bright Z-source GX 5-1 using February, 2017 observations with the SXT and LAXPC instruments. The 0.8–20 keV spectra from simultaneous SXT and LAXPC data at different locations of the hardness-intensity plot is well described by a disk emission and a thermal Comptonized component. The disk flux ratio (ratio of the disk flux to the total flux) increases monotonically along the horizontal branch to the normal one. Thus in the normal branch, the disk dominates the flux while in the horizontal branch the Comptonized component dominates. The disk flux scales with the inner disk temperature as T 5:5 and not as T 4 suggesting that either the inner radii changes dramatically or that the disk is irradiated by the thermal component changing its hardness factor. The PDS reveal a QPO whose frequency changes from  30 to 50 Hz and which is found to correlate well with the disk flux ratio. In the 3–20 keV LAXPC band the rms of the QPO increases with energy (rms/E0.8), while the harder X-ray seems to lag the soft ones with a time-delay of a milliseconds. The results suggest that the spectral properties of the source are characterized by the disk flux ratio and that the QPO has its origin in the corona producing the thermal Comptonized component.

40

Page 10 of 20

J. Astrophys. Astr. (2021)42:40

The results inferred, mainly from the LAXPC data, on various timing and spectral properties of LMXBs summarized here, demonstrate the capability of the LAXPC instrument for probing the bright as well as faint sources It is hoped that in the coming years many more LMXBs will be studied and new results will emerge from these studies.

2.4 Studies of new UltraLuminous X-ray (ULX) pulsars with AstroSat The UltraLuminous X-ray (ULXs) sources detected in the Galaxy and other galaxies were believed for a long time to be Intermediate Mass Black Holes accreting matter from the companion star. This changed when Pulsations with 1.37 sec period were discovered from the ULX in M82. Since then more ULX pulsars have been discovered and at present 8 ULX pulsars have been discovered and their properties are summarized in Table 1 taken from Chandra et al. (2020). AstroSat discovered ULX pulsar nature of a transient X-ray source RX J0209.6-7427 located in the Magellan Bridge when it became active after a gap of 26 years in 2019 and was studied by Chandra et al. (2020) with the SXT and LAXPC instruments on AstroSat. The transient first detected by ROSAT during its 1993 outburst, went into a deep hibernation for 26 years and suddenly sprang back to life in 2019 with a giant outburst in 2019. Using the SXT and LAXPC observations, Chandra et al. (2020) detected strong pulsations in RX J0209.6-7427 with 9.29 s periodicity over a broad energy band covering 0.3–80 keV (first reported by the NICER mission in the 0.2–10 keV energy band). The pulsar exhibited a rapid spin-up during the outburst as can be seen from Fig. 7. Energy resolved

folded pulse profiles were generated in several energy bands in 3–80 keV. This is the first report of the timing and spectral characteristics of this Be binary pulsar in hard X-rays. There is suggestion of evolution of the pulse profile with energy. The energy spectrum of the pulsar is determined and from the best-fitting spectral values, the X-ray luminosity of RX J0209.67427 is inferred to be 1.6  1039 erg s1 . These timing and spectral studies suggest that this source has features of an ultraluminous X-ray pulsar in the Magellan Bridge. A second ULX Pulsar Swift J0243.6?6124, the first Galactic ultra-luminous X-ray pulsar, was observed during its 2017–2018 outburst with AstroSat at both sub- and super-Eddington levels of accretion with Xray luminosities of Lx  7  1037 and 6  1038 erg s1 . A detailed broadband timing and spectral study by Beri et al. (2020) show that X-ray pulsations at  9.85 s have been detected up to 150 keV when the source was accreting at the super-Eddington level. The background subtracted pulse profiles for the the SXT, LAXPC and CZTI instruments are shown in Fig. 8. The pulse profiles are a strong function of both energy and source luminosity, showing a doublepeaked profile with pulse fraction increasing from 10% at 1.65 keV to 40–80% at 70 keV. The continuum X-ray spectra are well-modeled with a high energy cut-off power law (a  0.6–0.7) and one or two blackbody components with temperatures of  0.35 keV and 1.2 keV, depending on the accretion level. No iron line emission is observed at sub-Eddington level, while a broad emission feature at around 6.9 keV is observed at the super-Eddington level, along with a blackbody radius (121–142 km) that indicates the presence of optically thick outflows. Results on neutron star binaries summarized above establish firmly the capability of the LAXPC

Table 1. Summary of the 8 known ultraluminous X-ray pulsars Name of ULX

Host galaxy

M82 X-2 NGC 7793 P13 NGC 5907 ULX NGC 300 ULX1 Swift J0243.6?6124 M51 ULX-7 NGC 1313 X-2 RX J0209.6-7427

M82 NGC 7793 NGC 5907 NGC 300 Milky Way M51 NGC 1313 SMC

Spin period (s) 1.37  0.42  1.13  31.6  9.86  2.8  1.5 9

Orbital period (days)  2:5 64 5.3 –  27.6 2 – –

Spin-up/down Spin-up Spin-up Spin-up Spin-up Spin-up Spin-up Spin-up Spin-up

LX ð1039 ergs s1 Þ 4.9  10  100 4.7 2  10  20  1.6

Residuals

Spin period (s)

J. Astrophys. Astr. (2021)42:40

Page 11 of 20

9.30

9.284

9.29

9.283 9.282

9.28 24.7 25.1 25.5 25.9 26.3 9.27

0.0005 0.0000 −0.0005

0

10

20

30

40

50

60

Days since MJD 58806.98

Figure 7. Evolution of the pulsar period during the outburst of RX J0209.6-7427 in 2019 is shown in this plot. Spin-up of the pulsar during the outburst in clearly seen.

instrument for high resolution timing and spectral studies. Large effective area enables studies of relatively fainter sources like ULX pulsars.

3. Study of black hole X-ray binaries (BHXBs) with LAXPC/AstroSat As has been mentioned earlier, the advantages that AstroSat/LAXPC has over RXTE are: (1) an higher effective area at energies [30 keV, (2) event mode data allowing for temporal analysis in arbitrary user defined energy bins and (3) simultaneous observations

40

from other instruments on board. These have primarily defined and driven the LAXPC study of black hole binaries. It has been known that BHXBs are variable on different time-scales and exhibit both broad band continuum noise and narrow features termed as quasiperiodic oscillations. While there have been several models to explain the dynamic origin of these variabilities, a consensus and conclusive explanation for them remains illusive. One of the possible reasons for this impasse is that the radiative processes that are involved in these variations have not been clearly identified. The time-averaged spectra of black hole binaries are usually represented by physically motivated radiative components such as disk emission, thermal Comptonization, and refection components. The broad band spectrum from AstroSat allows for fitting of complex models such as General Relativistic disk emission and blurred reflection components, which can be used to constrain physical parameters such as the spin of the black hole and the accretion rate. It is also imperative to identify which of these components are responsible for the variability and in particular to identify the spectral parameters which may be changing that cause the observed variability. It is likely that more than one of these spectral parameters are involved and estimating any time difference between the variations of these parameters can provide critical information regarding the illusive dynamic origin of phenomena. AstroSat observations

Figure 8. The pulse profiles derived from SXT (0.5–7.0 keV), LAXPC (3–80 keV) and CZTI (30–150 keV) are shown in the figure for sub-Eddington level (left panel) and super-Eddington level (right panel) for the ULX Pulsar Swift J0243.6?6124.

40

Page 12 of 20

of BHXBs can provide these estimates paving the way for an emerging field where variability is quantified and understood in terms of the active radiative processes of the system. Here we present some of the scientific results obtained so far. It should be emphasised that these results were possible due to the unique capabilities of LAXPC (wide band spectroscopy, higher efficiency in the hard X-rays and fine time resolution) and the simultaneous spectra available from SXT. After decommissioning of RXTE in 2012, AstroSat/ LAXPC is the only X-ray timing instrument best suited for studying spectral-temporal characteristics of black hole X-ray binaries (BHXBs). During last five years, LAXPC successfully observed many BHXBs and very interesting results have come out. Here we discuss some of BHXBs results which were observed on several occasions with LAXPC instrument and indepth analysis were performed. In Sections 3.1–3.5 we present the primary results from LAXPC grouped on the basis of broad topics which highlight the advantage of the instrument. In Section 3.6, we enumerate these results in a different degree of detail, for the individual black hole sources that have been observed by LAXPC.

3.1 Energy-dependent temporal properties Typically the variability is quantified by the power spectrum which is the square of the Fourier transform of a light curve obtained over a certain energy range. If normalized in a certain way, the integration of the power spectrum (over some frequency range) gives the variance of the variation in that energy and frequency range. Another useful quantity is the fractional root mean square (frms) which is the square root of the variance normalized to the mean of the light curve. It is insightful to consider this as a Data Cube where the power is provided as a function of frequency and energy. Collapsing the Data Cube over energy (or a given range of energy) provides the power spectrum in that energy range. On the other hand, collapsing the Data Cube over a given range frequency, provides the frms as a function of energy for that frequency range. Moreover, cross-frequency analysis between energy bands provides time-lag as a function of energy. LAXPC can efficiently provide frms and time-lag as a function of energy for different frequency bands as shown in the early analysis of the black hole systems GRS 1915?105 (Yadav et al. 2016a) and

J. Astrophys. Astr. (2021)42:40

Cygnus X-1 (Misra et al. 2017). Encouraged by these unprecedented results, attempts were made to develop formalisms which could obtain physically meaningful radiative parameter variations and time-lag between them. In the hard state, the spectra of Cygnus X-1 above 3 keV, is dominated by thermal Comptonization. Maqbool et al. (2019) developed a model to predict the variability of such a spectrum when the seed photon flux and the heating rate of the corona varies with a time-lag between them. Fitting the predictions to the observed time-lag and frms as a function of energy, led them to quantify the variability in the seed photon flux and the coronal heating rate as a function of frequency. Mudambi et al. (2020a) used the same analysis to constrain the variability of these quantities for the bright transient X-ray binary, MAXI J1820?070 in its hard state observation by AstroSat/ LAXPC allowing for a comparison with Cygnus X-1. While these analysis are for the continuum variability, Jithesh et al. (2019) applied the model to a QPO observed by AstroSat LAXPC in the hard state of the black hole system, SWIFT J1658.2-4242. The model used in these works mentioned above, is only applicable to the thermal Comptonization component, while a thermal component is often observed in the spectra of black hole systems which are not in a pure hard state. A generalized method which can predict energy-dependent variability properties for arbitrary spectral components is challenging since (1) the response of the spectrum to parameter variation has to be done numerically and (2) the spectral parameters have to be cast into physical ones rather than empirical ones (for e.g., heating rate of corona instead of say the asymptotic power-law index for Comptonization models). An initial step in this direction has been taken by Garg et al. (2020) who have fitted the energy-dependent frms and time-lag of the QPO observed in GRS 1915?105 by AstroSat/ LAXPC.

3.2 Variations of timing and spectral properties and their correlation An important and promising use of AstroSat/LAXPC data is the study of the variation of timing features and their correlation with spectral parameters. The sensitivity of LAXPC to detect small variation in rapid timing properties was proved by the first time detection of the small  7% variation of the high frequency (  70 Hz) QPO in GRS 1915?105 (Belloni et al.

J. Astrophys. Astr. (2021)42:40

2019), which although small has the potential to differentiate between generic models. The QPO has also been reported by Sreehari et al. (2019) who showed that its presence depends on the strength of the high energy spectral component. The correlation with spectral parameters was demonstrated by Bhargava et al. (2019) where a long monitoring of the black hole transient, MAXI J1535571, showed a tight correlation between the QPO frequency with the high energy spectral index rather than the flux. An important expected correlation is that between QPO frequency and the inner radius of a truncated disk, since a physical characteristic timescale associated with the radius may be responsible for the QPO. Using LAXPC and SXT observations of the black hole system GRS 1915?105, Misra et al. (2020) showed that there is such a correlation and identified the QPO frequency with the dynamical time-scale corrected by General Relativity as predicted decades ago. This important result was possible because of the broad band spectral coverage by SXT and LAXPC and the simultaneous rapid timing information provided by LAXPC.

3.3 Long-term variability Long observations of BHXBs by LAXPC, which could be continuous exposure for more than a day, or monitoring of sources on weeks/months time-scale have provided extensive information regarding the long term behaviour of these sources. A good example is the enigmatic black hole system GRS 1915?105 and AstroSat was fortunate to observe it during a transition from a non-variable class to a structured large amplitude one (Rawat et al. 2019). This allowed for tracking of the QPO and its energy dependent property rms and time-lag as the source made the transition. The transient Swift J1658.2-4242 shows ’flip-flop’ state transitions on time-scale of minutes, which is reflected both in the spectra and power density spectrum was studied by LAXPC and other instruments (Bogensberger et al. 2020). Monitoring of an outburst from a black hole binary can be used to track the spectral and timing evolution as has been done for two outbursts of 4U 1630-472 (Baby et al. 2020). The study inferred the appearance of the standard disk after a few hours of the burst and its persistence as the source evolved to the soft state. Monitoring observations of Cygnus X-3 by LAXPC has provided important clues on the formation of the

Page 13 of 20

40

radio jet and its connection to the accretion disk (Pahari et al. 2018b). The observations of Cyg X-3 provides a measurement of the orbital period and the discovery of low frequency mHz QPO whose energy dependent rms and time-lag could be quantified (Pahari et al. 2017).

3.4 Broadband spectral fitting Although LAXPC spectral resolution is lower than that of other instruments such as Nustar, however it does provide a wide energy band especially when combined with SXT data. Hence AstroSat observations have been used to fit complex spectral models to constrain system parameters. For example, after fitting standard models to the AstroSat spectra of the black hole transient MAXI J1535-571, Sreehari et al. (2019), fitted the two component accretion flow model to the data and found that the accretion rate to be nearly at the Eddington value and could further constrain the mass of the black to be around  6 solar masses. For the same system, Sridhar et al. (2019), used the relativistic disk and blurred reflection model to constrain the spin, distance to the source and mass of the black hole to be  10 solar masses. The different mass estimate are due to differences in the model and other assumptions, but what is perhaps important is the ability of these different models to make quantitative estimates given AstroSat data. Black hole mass has also been estimated using two component flow for the black hole binary 4U 1630-472, where the utility of long term monitoring of a source as has been demonstrated. The extra galactic black hole systems LMC X-1 has a well constrained distance and black hole mass and hence AstroSat data has proved useful to constrain the black hole spin (Mudambi et al. 2020b). Relativistic disk fitting of well studied sources like GRS 1915?105, can be used to confirm the black hole mass and to constrain the spin (Sreehari et al. 2020).

3.5 Synergy with other observatories Perhaps the most promising use of AstroSat data is when they are analyzed in conjunction with data from other instruments of observatories. The complementary nature of the different instruments provides a multi-faceted and unprecedented view of the systems. Since black hole binaries are variable simultaneous (or quasi-simultaneous) data would be optimal.

Page 14 of 20

Microquasar) (Fender et al. 2004; Yadav 2006). We have monitored this source regularly. Yadav et al. (2016a) have studied spectra and timing properties of GRS 1915?105 when the source had the characteristics of being in radio-quiet during March 5–7, 2016.The energy-dependent power spectra reveal a strong low frequency (2–7 Hz) Quasi-periodic oscillation (LFQPO) and its harmonic along with broad band noise. The QPO frequency changes rapidly with flux. At the QPO frequencies, the time-lag as a function of energy has a non-monotonic behavior such that the lags decrease with energy till about 15–20 keV and then increase for higher energies. Rawat et al. (2019) have studied GRS 1915?105 during February and March, 2017 when the source underwent a major transition from a non-variable state (close to v class) to a periodic flaring state (similar to q class). It is shown that such transition takes place via an intermediate state when the large-amplitude, irregular variability of the order of thousands of seconds turned into a 100–150 s nearly periodic flares similar to q class/heartbeat oscillations. Figure 9 shows the color–color diagram during this fast transition. It is interesting to note that HR2 remains same when source transits to the intermediate state but when source finally attains the flaring state, it regains the same HR1 but HR2 has changed. Belloni (2006) have analysed 92 ks of data obtained with the LAXPC instrument and they have detected around seven percent variation in High Frequency QPOs in GRS 1915?105. Banerjee et al. (2020) have studied temporal and spectral properties of GRS 1915?105 during h class (Belloni et al. 2000). Sreehari et al. (2020) have studied this source during the soft X-ray state

0.6

0.7

0.8

Non−Variable state Intermediate state Flaring state

Rise

0.5

A good example of using the different capabilities of instruments was the study of the black hole system 4U 1630-47 using the high energy grating spectra from Chandra along with AstroSat’s LAXPC and SXT data (Pahari et al. 2018a). While the latter provided evidence for a highly ionized wind, relativistic disk model fit to the continuum using AstroSat data showed that the black hole is highly spinning, thus making an interesting connection between wind outflow and black hole spin. Another example is when the combination of high resolution Nustar data (3–80 keV) with AstroSat (SXT, LAXPC and CZTI) (0.7–200 keV) of MAXI J1820?070 allowed for the use of more complex and sophisticated models as compared to only AstroSat data (Mudambi et al. 2020a) or just Nustar data. Even when the observations are not simultaneous, the use of multiple instruments has proved extremely worthwhile as in the spectral and timing study of the flip-flop state transitions of Swift J1658.2-4242 using a host of instruments, XMM-Newton, NuSTAR, Swift, Insight-HXMT, INTEGRAL, ATCA and AstroSat (Bogensberger et al. 2020). It is fortunate that at present along with AstroSat, there are two other observatories capable of high time resolution operating namely NICER and INSIGHTHXMT. The three observatories together can provide unprecedented timing information in an energy range (0.2–200 keV) which is about three order of magnitude The improvement in our understanding that would be obtained by such analysis is illustrated by Xiao et al. (2019) where they performed timing analysis of Swift J1658.2-4242 outburst in 2018 with Insight-HXMT, NICER and AstroSat. They found a range of QPO activities detected by the different instruments and quantified their dependence. The potential of such analysis is clear, especially when one notes that the observations used by them were often not strictly simultaneous. It is expected that soon, joint analysis of AstroSat data with one or both of these instruments, will be providing deeper insights into the nature of Black hole systems. The importance having coordinated observations with all three observatories cannot be over emphasised.

J. Astrophys. Astr. (2021)42:40

HR2 (13.0−60.0) keV/(3.0−5.0) keV

40

3.6.1 GRS 1915?105. GRS 1915?105 is a highly variable BHXB (Yadav et al. 1999; Belloni et al. 2000) with frequent radio emission (often referred as

0.4

Fall

3.6 Important results of some of BHXBs with LAXPC Instrument

1.8

1.9

2

2.1 HR1 (5.0−13.0) keV/(3.0−5.0) keV

2.2

2.3

Figure 9. Color–color diagram during fast transition among three X-ray states.

J. Astrophys. Astr. (2021)42:40

Page 15 of 20

8.5

7.5

0.06

6 5.5 5

0.02

6.5

0.04

7

Counts/s/cm2

QPO frequency (Hz) / Mdot (1018 g/s)

15−50 keV Swift/BAT lightcurve of Cyg X−3

χ class IMS HS a=0.97, N=0.22 a=0.91, N=0.35 a=0.99, N=0.17

8

40

4.5 4 3

3.5

4 Radius (rg)

4.5

5

5.5

0

2.5

5.74×104

Figure 10. Color–color diagram during fast transition among three X-ray states. 1 0.1 0.01

X−ray flux: keV (Photons s−1 cm−2 keV−1)

Spectral states of Cyg X−3 as observed by SXT+LAXPC

Soft (12 August, 2016) Hypersoft (01 April, 2017) Very high state (02 April, 2017) Hard (20 November, 2016)

10

3.6.2 Cygnus X-3 and relativistic large radio jets. Cygnus X-3 is an extraordinary HMXBs (a close binary with  4.8 hr orbital period) which produces brightest radio jets in our Galaxy. Although the mass of the central object has not been confirmed using dynamical measurements, the similarity of its spectro-temporal properties with known black hole systems like GRS 1915?105 and XTE J1550-564 favours the binary harbouring a black hole. The source is bright and persistent in X-rays, and the X-ray emission originates as a result of wind accretion from a Wolf-Rayet companion star onto the central compact object. In contrast to the canonical BHXBs, six different spectral states have been reported in Cygnus X-3 by Koljonen et al. (2010) using simultaneous X-ray and radio observations. Spectral states of Cygnus X-3, as observed by LAXPC are shown in the bottom panel of Fig. 11 where unfolded, best-fit spectra from SXT and LAXPC joint fitting are shown. Details of spectral modelling are discussed in Koljonen et al. (2010). With LAXPC, we observed four different spectral states: soft, hard, hypersoft and very high state. The presence of the very high state is anticipated before but observed for the first time with

5.78×104

−3

and have detected HFQPs in the range of 67.96–70.62 Hz with rms  0.83–1.90 per cent. Misra et al. (2020) showed that there is a correlation between QPO frequency and disk radii, and identified the QPO frequency with the dynamical time-scale corrected by General Relativity as predicted decades ago using LAXPC and SXT simultaneous data of GRS 1915?105. Results of their analysis is shown in Fig. 10 which puts a tight limit on the spin of GRS 1915?105.

5.76×104

Time (MJD)

1

2

5 10 Energy (keV)

20

50

Figure 11. Top panel shows a typical 15–50 keV Swift/ BAT light curve of Cygnus X-3 over a timescale of 1.5 year. AstroSat/LAXPC observations during this period are shown by stars and filled circles. Bottom panel shows unfolded spectra using SXT?LAXPC joint fitting in the energy range 0.5–60 keV. Data sets used for spectral fiting are shown with filled circle in respective colours.

LAXPC. Dramatic changes in flux in both soft and hard bands in different states are visible. Very high energetic photons, of the order of GeV, have been detected during the hypersoft state when the radio emission is entirely quenched. AstroSat/LAXPC performed over 20 observations of Cygnus X-3 during last five years. The top panel of the Fig. 11 shows 15–50 keV Swift/BAT lightcurve over 1.5 years and LAXPC pointing observations during this period are shown by stars and filled circles. Using three long observations during 2015–2016 spanning over several orbits, Pahari et al. (2017) determined the binary orbital period of 17253.56 ± 0.19 s, which is consistent with earlier measurements.

f X P (f)

However, using one-year long observation, a slow, low-amplitude variability was observed with a periodicity of  35.8 days which may be due to the orbital precession. During the peak of binary orbital phase, weak but significant quasi-periodic oscillations have been observed at  5–8 mHz,  12–14 mHz and  18–24 mHz. Such detection is significant since no QPO was detected with RXTE/PCA despite its most extensive archival data. Interestingly, 7–15 mHz QPOs from Cygnus X-3 was reported with Exosat/ME observations (van der Klis & Jansen 1985). Detailed analysis with LAXPC observations showed that QPOs were observed only during the flaring hard X-ray state and when the source is brightest (the peak phase of the binary orbital period). It is explained in terms of increased mass accretion rate when the compact object passes through the denser wind section. An enhanced supply of material may temporarily boost the temporal variability features. Until now, investigating the connection between accretion disk and the radio jets in Cygnus X-3 has been partially successful because of the lack of truly simultaneous X-ray and radio data. X-ray and radio monitoring program of Cygnus X-3 with LAXPC provided a rare opportunity to explore the radio/X-ray connection in this X-ray binary. Using long-term X-ray/radio monitoring campaign with Swift/BAT and 11.2 GHz RATAN-600 telescope, it has been observed that Cygnus X-3 used to move into the hard X-ray quenched state where hard X-ray flux decreased by order of magnitude, and subsequently it showed major radio flare ejection event when the radio flux density increases dramatically from few tens of mJy up to 20 Jy. Such conjunction was detected with LAXPC and RATAN-600 telescopes on 1–2 April 2017. Using detailed X-ray analysis, Pahari et al. (2018b) found that Cygnus X-3 undergo spectral state transition from the hypersoft state (HPS) to a harder, more luminous state which was never observed before. We term it as the very high state (VHS). Such a transition occurred within a few hours when the radio flux density increases from  100 mJy to  478 mJy. Using SXT?LAXPC joint spectral analysis, they observed no hard X-rays above 17 keV during the HPS state. Within hours timescale, a flat power law appeared in the spectra with the power-law index of 1.49þ0:04 0:03 and extended up to 70 keV. Such an observation provided direct evidence of synchrotron emission that originates in the radioemitting blob which was caught in the act of decoupling from the accretion disk. Such a detailed radio/X-

J. Astrophys. Astr. (2021)42:40 0.01

Page 16 of 20

5×10−3

40

0.1

1 Frequency (Hz)

10

Figure 12. The frequency times the power spectra of Cygnus X-1 for two energy bands; 3–10 keV band (black points) and 20–40 keV band (red points).

ray coupling event was observed for the first time and made possible due to the LAXPC’s capabilities of higher efficiency for hard X-ray above 30 keV and high time resolution. 3.6.3 Cygnus X-1. Probably one of the best studied black hole systems, the bright black hole system, Cygnus X-1, presents an excellent example to underline LAXPC capabilities and the new understanding that LAXPC data can bring. The higher effective area at high energies has allowed LAXPC to study in detail the energy dependent rapid variability of Cygnus X-1 in the hard state (Fig. 12). LAXPC data analysis by Misra et al. (2017), revealed the rms and time-lag variation as a function of Fourier frequencies in the broad energy range of 4–80 keV, extending the earlier results which were limited to 30 keV. They also showed that the event mode data of LAXPC allows for flux resolved spectroscopy in rapid (  1) second time-scales, which for Cygnus X-1 revealed a correlation between photon index and flux. A more extensive analysis of six observations of Cygnus X-1 by Maqbool et al. (2019), showed that the energy and frequency dependent rapid variability is different for observations which are in the hard state. More importantly, they could now quantitatively fit and explain this variability in terms of a single zone stochastic fluctuation model. 3.6.4 MAXI J1535-571. The bright outburst of this recently discovered black hole binary has been observed by a large number of observatories including AstroSat and has provided one of the most

J. Astrophys. Astr. (2021)42:40

comprehensive view of any such systems. LAXPC detected prominent QPOs at  2 Hz in a wide energy range of 3–50 keV (Sreehari et al. 2019) when the source was in the hard intermediate state. Spectral modelling using standard Comptonization models and physically motivated ones like the two-component accretion flow revealed that source was nearly at Eddington limit and its black hole mass is  6M . A detailed study of the broad band spectrum from SXT and LAXPC using relativistic reflection models provided an estimate of the black hole spin parameter (a  0:7) and a mass of  10M (Sridhar et al. 2019). Correlation between the QPO frequency and spectral parameters was studied by Bhargava et al. (2019) who found that the frequency tightly correlates with the photon index rather than the flux. 3.6.5 4U 1630-472. Among recurrent black hole Xray transients, 4U 1630-472 is unique as it shows frequent outbursts at about 600 days interval. The source was also at the centre of focus due to its ‘superoutburst’ which typically lasts for two years. Because no dynamical mass measurements have been performed, the source has been categorised as the black hole candidate due to its similarity of X-ray spectra-timing characteristics with other confirmed black hole X-ray binaries. A dust scattering halo analysis placed the source between 4.7–11 kpc (Kalemci et al. 2018) while the spectral index vs. mass accretion rate correlation was used to estimate the black hole mass of 10  0:1M (Seifina et al. 2014). Two major monitoring campaigns with AstroSat/ LAXPC took place between 27 August–2 October 2016 and 4 August–17 September 2018 to monitor ‘mini’ outbursts from 4U 1630-472 which typically lasts 5–6 months. Using both outburst observations, Baby et al. (2020) performed detailed spectro-timing analysis. Using broadband (0.7–20.0 keV) spectral modelling, they detected no disk component during the onset of the 2016 outburst while later a geometrically thin accretion disc with the temperature of  1.3 keV was observed. Such behaviour was also accompanied by other observations like the steepening of the power-law index from 1.8 to 2.6 and the decrease of rms variability by  5%. However, all observations with LAXPC during 2018 outburst showed thermal disk emission dominated state with the disc temperature vary between 1:26  0:01 keV and 1:38  0:01 keV. Depending upon spectral characteristics, they observed three distinct spectral states:

Page 17 of 20

40

low hard, intermediate and high soft states. Three states can be distinguished in the ‘C’-shaped hardness intensity diagram track where the hardness is the ratio between the counts in 6–20 keV and 2–6 keV and the intensity is 2–20 keV X-ray flux calculated from LAXPC spectral modelling. Such distinction was also apparent in the flux-rms plot where the flux was calculated from the best-fit spectra while the rms was calculated using 0.005–10 Hz power density spectra. The transition among three states is in agreement with the accretion geometry inferred from the two-component accretion flow model. Due to its high energy coverage, LAXPC spectra of 4U 1630-472 were also found useful in measuring fundamental properties of the accreting compact object when combined with concurrent spectra from other soft X-ray instruments like SXT and Chandra. For example, the spin of a black hole was measured using luminous high soft state observation with AstroSat and Chandra. Using continuum spectral modeling method, Pahari et al. (2018a) determined the presence of a fast-spinning black hole in 4U 1630472. Using relativistic continuum spectral modeling on three independent measurements with Chandra, SXT?LAXPC and Chandra?SXT?LAXPC and applying Markov chain Monte Carlo simulations on fitted spectral parameters, they constrained the spin of the black hole to be 0:92  0:04 within 99.7% confidence limit. Such a measurement is essential for explaining accretion inflow and outflow properties very close to the accreting black hole. 3.6.6 MAXI J1820?070. At a distance of 3.46þ2:18 1:03 kpc (Gandhi et al. 2020), MAXI J1820?070 was discovered as one of the closest and the brightest Galactic BHXBs known till date (Corral-Santana et al. 2016). With the Swift/BAT peak flux of  4 Crab in 15–50 keV, the source showed back and forth transition between hard and soft states. The source was proposed as a black hole candidate depending upon the large amplitude hard X-ray variability observed in the power density spectrum (Uttley et al. 2018). However, with the MAXI data, ‘q’ shaped track is clearly visible in the HID (Chakraborty et al. 2020) strengthening the fact that the source harbours a black hole. AstroSat observed the source on two occasions during the peak of the outburst. Using the two days long observation with LAXPC on 30 March 2018, Mudambi et al. (2020a) observed 47.7 mHz QPO in the power density spectra which is similar to what observed from two other BHXBs:

40

Page 18 of 20

GRS 1915?105 and IGR J17091-3624. Using an SXT?LAXPC joint spectral analysis in the energy range 0.7–30 keV, they observed a flat spectral index of  1.61 and evidence for cool disk at the temperature of  0.22 keV. They noted that the time lag (where the reference band is 4.15–5.37 keV) increasing linearly with the photon energy at 0.1, 1 and 10 Hz. However, the fractional rms decreases with the energy at 0.1 and 1 Hz and independent of the photon energy at 10 Hz. The observed temporal properties, e.g., typical 50–100 ms time lag, can be explained qualitatively by the stochastic propagation fluctuation model (Maqbool et al. 2019). Such a detailed temporal analysis brings out the importance of LAXPC observations. Chakraborty et al. (2020) studied the broadband spectra which include SXT, LAXPC and CZTI in the energy range of 0.3–120 keV and the best-fit spectral modelling shows the presence of a soft excess component in the form of thermal disk blackbody which is significantly below three keV, a thermal Comptonisation model with a very flat spectral index close to 1.4 and two relativistic reflection models to fit broad and narrow iron line complex and a Compton hump. Two relativistic models differ significantly in their coronal electron temperature by a factor of  7. Interestingly, the relativistic reflection component dominates the broadband spectra. Comparing with the detailed NuSTAR spectral study by Buisson et al. (2019), they inferred that the corona in MAXI J1820?070 is inhomogeneous and residing close to the black hole (\3Rg ).

4. Active galactic nuclei AstroSat/LAXPC has the potential to monitor the variability of Active Galactic Nuclei (AGN) and in conjunction with SXT and UVIT to study the their broad band spectra. Blazars are jet dominated AGN which show high amplitude variability. Using AstroSat LAXPC and SXT observations, Banerjee et al. (2020) constructed the power spectrum for the light curve of the blazar, Mrk 421, and detected a break or a characteristic time-scale. Such characteristic time-scale have been detected before for X-ray binaries and regular AGN and are believed to originate in the accretion disk. Hence this result suggests that the jet variability also has an disk origin. For the blazar, RGB J0710?591, Goswami et al. (2020) reported a significant deviation from a power-law shape for the X-ray spectrum obtained from LAXPC and SXT. Such deviation or curvature reflect the underlying

J. Astrophys. Astr. (2021)42:40

shape of the particle energy spectrum that produces the emission. The AstroSat results suggested that particle distribution has a maximum energy cutoff as predicted by models where the particles are shock accelerated and radiatively cooled. Goswami et al. (2020) studied the broad band spectral energy distribution for the blazar 4C?21.35 using multi-wavelength data from AstroSat and other observatories. They modeled the spectrum during flaring and quiescent states with two compact regions which originated at different times and moved away from the central region. For the regular AGN, RE J1034?396, Chaudhury et al. (2018) fitted the LAXPC and SXT spectra to reveal the presence of a soft excess, consistent with earlier results. Studies of faint sources like AGNs by LAXPC is limited by the accuracy of the instrument’s background estimation. It should be noted that most of the results of AGN data analysis using the RXTE/PCA observations were undertaken in the later times of the satellite operations when the response and background were better characterised. With improvements and different estimation techniques for LAXPC background (Antia et al. 2021; Misra et al. 2021), it is expected that in the near future there will be significant spectral and variability studies of AGN using LAXPC data.

5. Summary LAXPC instrument has carried on the legacy of RXTE PCA and HEXTE by being one of the primary instruments to study rapid variability of X-ray systems. As compared to RXTE/PCA, it has enhanced features which are (1) an higher efficiency at hard X-rays (energies  30 keV), (2) event mode data allowing for temporal analysis in arbitrary user defined energy bins and (3) simultaneous SXT data at the soft X-ray band (0.5–8 keV). These advantages have been demonstrated by a number of publications reporting important results discussed in this paper for more than 30 sources. Moreover, the increasing reservoir of archived and future LAXPC data is expected to provide an unprecedented view of known sources as well as those that are yet to be discovered. Improvements in the calibration, especially in the background estimation, would lead to more detailed spectral and timing

J. Astrophys. Astr. (2021)42:40

analysis of the bright X-ray sources and will enable analysis of fainter sources such as a larger number of active galactic nuclei. It is fortunate that at present there are several X-ray missions in operation, which have complimentary capabilities to the instruments on-board AstroSat. Several works have already demonstrated the dramatic advantage of having simultaneous (or even quasi-simultaneous) observations of a source with multiple X-ray missions and with those in other wavebands. There is now an unprecedented opportunity of detailed timing studies for a wide energy range using simultaneous observations of AstroSat/LAXPC and NASA’s Nicer. Despite the challenges of organizing and administrating such coordinated observations by multiple missions (Nicer, Nustar, Chandra and others), it is imperative for break through results, that a large number of them should be undertaken.

Acknowledgements We acknowledge the strong support from Indian Space Research Organization (ISRO) in various aspects of instrument development, space qualification, software development, mission operation and data dissemination. We specially acknowledge ISAC support for electronics development and during space qualification tests. We thank IISU team for providing us bellow pump along with space qualified pump driver. We acknowledge support of the scientific and technical staff of the LAXPC instrument team for their excellent team work as well as staff of the TIFR Workshop who helped us at various level of LAXPC instrument development.

References Agrawal P. C. 2006, Advances in Space Research, 38, 2989 Agrawal P. C., Yadav J. S., Antia H. M. et al. 2017, J. Astrophys. Astr., 38, 30 Agrawal V. K., Nandi A., Girish V., Ramadevi M. C. 2018, MNRAS, 477, 5437 Amin N., Roy J., Chakroborty S. et al. 2020, Private communication Antia H. M., Yadav J. S., Agrawal P. C. et al. 2017, ApJ Supp., 231, 10 Antia H. M., Agrawal P. C., Dedhia D. et al. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036021-09712-8 Baby B. E., Agrawal V. K., Ramadevi M. C. et al. 2020, MNRAS, 497, 1197

Page 19 of 20

40

Bala S., Bhattacharya D., Staubert R., Maitra C. 2020, MNRAS, 497, 1029 Banerjee A., Bhattacharjee A., Chatterjee D. et al. 2020, arXiv e-prints, arXiv:2007.05273 Belloni T. 2006, Adv. Space Res., 38, 2801 Belloni T., Klein-Wolt M., Me´ndez M., Van Der Klis M., Van Paradijs J. 2000, arXiv preprint astro-ph/0001103 Belloni T. M., Bhattacharya D., Caccese P. et al. 2019, MNRAS, 489, 1037 Beri A., Paul B., Yadav J. S. et al. 2019, MNRAS, 482, 4397 Beri A., Naik S., Singh K. P. et al. 2020, MNRAS, arXiv: 2010.08334 Bhalerao V., Bhattacharya D., Vibhute A. et al. 2017, J. Astrophys. Astr., 38, 31 Bhargava Y., Belloni T., Bhattacharya D., Misra R. 2019, MNRAS, 488, 720 Bhattacharyya S., Yadav J. S., Sridhar N. et al. 2018, ApJ, 860, 88 Bhulla Y., Roy J., Jaaffrey S. N. A. 2020, Res. Astron. Astrophys., 20, 098 Bogensberger D., Ponti G., Jin C. et al. 2020, A&A, 641, A101 Buisson D. J. K., Fabian A. C., Barret D. et al. 2019, MNRAS, 490, 1350 Chakraborty S., Navale N., Ratheesh A., Bhattacharyya S. 2020, MNRAS, 498, 5873 Chandra A. D., Roy J., Agrawal P. C., Choudhury M. 2020, MNRAS, 495, 2664 Chaudhury K., Chitnis V. R., Rao A. R. et al. 2018, MNRAS, 478, 4830 Corral-Santana J. M., Casares J., Mun˜oz-Darias T. et al. 2016, A&A, 587, A61 Devasia J., Raman G., Paul B. 2021, NA, 83, 101479 Fender R. P., Belloni T. M., Gallo E. 2004, MNRAS, 355, 1105 Gandhi P., Rao A., Charles P. A. et al. 2020, MNRAS, 496, L22 Garg A., Misra R., Sen S. 2020, MNRAS, 498, 2757 Goswami P., Sinha A., Chandra S. et al. 2020, MNRAS, 492, 796 Jain C., Paul B., Dutta A. 2010, MNRAS, 409, 755 Jaisawal G. K., Naik S., Ho W. C. G. et al. 2020, MNRAS, 498, 4830 Jaisawal G. K., Naik S., Epili P. R. et al. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036021-09701-x Jithesh V., Maqbool B., Misra R. et al. 2019, ApJ, 887, 101 Jonker P. G., van der Klis M. 2001, ApJL, 553, L43 Kalemci E., Maccarone T. J., Tomsick J. A. 2018, ApJ, 859, 88 Koljonen K. I. I., Hannikainen D. C., McCollough M. L., Pooley G. G., Trushkin S. A. 2010, MNRAS, 406, 307 Lewin W., Van der Klis M. 2006, Compact Stellar X-ray Sources, Vol. 39, Cambridge University Press Maitra C. 2017, J. Astrophys. Astr., 38, 50

40

Page 20 of 20

Maitra C., Paul B., Haberl F., Vasilopoulos G. 2018, MNRAS, 480, L136 Malu S., Sriram K., Agrawal V. K. 2020, MNRAS, https:// doi.org/10.1093/mnras/staa2939 Maqbool B., Mudambi S. P., Misra R. et al. 2019, MNRAS, 486, 2964 Misra R., Yadav J. S., Verdhan Chauhan J. et al. 2017, ApJ, 835, 195 Misra R., Rawat D., Yadav J. S., Jain P. 2020, ApJL, 889, L36 Misra R., Roy J., Yadav J. S. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036-021-09734-2 Mudambi S. P., Maqbool B., Misra R. et al. 2020a, ApJL, 889, L17 Mudambi S. P., Rao A., Gudennavar S. B., Misra R., Bubbly S. G. 2020b, MNRAS, 498, 4404 Mukerjee K., Antia H. M., Katoch T. 2020, ApJ, 897, 73 Pahari M., Antia H. M., Yadav J. S. et al. 2017, ApJ, 849, 16 Pahari M., Bhattacharyya S., Rao A. R. et al. 2018a, ApJ, 867, 86 Pahari M., Yadav J. S., Verdhan Chauhan J. et al. 2018b, ApJL, 853, L11 Parmar A. N., Gottwald M., van der Klis M., van Paradijs J. 1989, ApJ, 338, 1024 Rawat D., Pahari M., Yadav J. S. et al. 2019, ApJ, 870, 4 Remillard R. A., McClintock J. E. 2006, Annu. Rev. Astron. Astrophys., 44, 49 Roy J., Agrawal P. C., Singari B., Misra R. 2020, Res. Astron. Astrophys., 20, 155 Roy J., Agrawal P. C., Iyer N. K. et al. 2019, ApJ, 872, 33 Seifina E., Titarchuk L., Shaposhnikov. N. 2014, ApJ, 789, 57 Singh K. P., Stewart G. C., Chandra S. et al. 2016, in den Herder J.-W. A., Takahashi T., Bautz M., eds, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, vol. 9905, Space Telescopes and Instrumentation 2016: Ultraviolet to Gamma Ray, 99051E Sharma R., Beri A., Sanna A., Dutta A. 2020, MNRAS, 492, 4361 Sreehari H., Nandi A., Das S. et al. 2020, arXiv e-prints, arXiv Sreehari H., Ravishankar B. T., Iyer N. et al. 2019, MNRAS, 487, 928

J. Astrophys. Astr. (2021)42:40 Sridhar N., Bhattacharyya S., Chandra S., Antia H. M. 2019, MNRAS, 487, 4221 Staubert R., Klochkov D., Wilms J. et al. 2014, A&A, 572, A119 Staubert R., Tru¨mper J., Kendziorra E. et al. 2019, A&A, 622, A61 Staubert R., Ducci J., Ji L. et al. 2020, A&A, 642, A196 Tanaka Y., Lewin W., Van Paradijs J., Van Den Heuvel E. 1995, Cambridge, p. 126 Tandon S. N., Hutchings J. B., Ghosh S. K. et al. 2017, J. Astrophys. Astr., 38, 28 Truemper J., Pietsch W., Reppin C. et al. 1978, ApJL, 219, L105 Uttley P., Gendreau K., Markwardt C. et al. 2018, The Astronomer’s Telegram, 11423, 1 van der Klis M. 1994, SSRv, 69, 208 van der Klis M. 2001, White N. E., Malaguti G., Palumbo G. G. C., eds, American Institute of Physics Conference Series, Vol. 599, X-ray Astronomy: Stellar Endpoints, AGN and the Diffuse X-ray Background, pp. 406–415 van der Klis M., Jansen F. A. 1985, Nature, 313, 768 Varun Maitra C., Pragati P., Harsha R., Biswajit P. 2019a, MNRAS, 484, L1 Varun Pradhan P., Maitra C., Raichur H., Paul B. 2019b, ApJ, 880, 61 Verdhan Chauhan J., Yadav J. S., Misra R. et al. 2017, ApJ, 841, 41 Xiao G. C., Lu Y., Ma X. et al. 2019, J. High Energy Astrophys., 24, 30 Yadav J. 2006, The Astrophys. J., 646, 385 Yadav J., Rao A., Agrawal P. et al. 1999, The Astrophys. J., 517, 935 Yadav J. S., Agrawal P. C., Antia H. M. et al. 2017, Curr. Sci., 113, 591 Yadav J. S., Misra R., Verdhan Chauhan J. et al. 2016a, ApJ, 833, 27 Yadav J. S., Agrawal P. C., Antia H. M. et al. 2016b, in den Herder J.-W. A., Takahashi T., Bautz M., eds, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 9905, Space Telescopes and Instrumentation 2016: Ultraviolet to Gamma Ray, 99051D

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:20 https://doi.org/10.1007/s12036-020-09685-0

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

PAYLOAD REVIEW

In-orbit performance of UVIT over the past 5 years S. K. GHOSH1,*, P. JOSEPH2, A. KUMAR2, J. POSTMA3, C. S. STALIN2,

A. SUBRAMANIAM2, S. N. TANDON2,4, I. V. BARVE2, A. DEVARAJ2, K. GEORGE2,5, V. GIRISH6, J. B. HUTCHINGS7, P. U. KAMATH2, S. KATHIRAVAN2, J. P. LANCELOT2, D. LEAHY3, P. K. MAHESH2, R. MOHAN2, S. NAGABHUSHANA2, A. K. PATI2, N. KAMESWARA RAO2, K. SANKARASUBRAMANIAN8, P. SREEKUMAR2,6 and S. SRIRAM2 1

Tata Institute of Fundamental Research, Mumbai 400 005, India. Indian Institute of Astrophysics, Bangalore 560 034, India. 3 University of Calgary, Calgary, AB, Canada. 4 Inter-University Centre for Astronomy and Astrophysics, Pune 411 007, India. 5 Ludwig Maximilian University of Munich, Munich, Germany. 6 ISRO Headquarters, Bengaluru 560 094, India. 7 National Research Council of Canada, Herzberg Astronomy and Astrophysics, Victoria, Canada. 8 U.R. Rao Satellite Centre, Bengaluru 560 017, India. *Corresponding Author. E-mail: [email protected] 2

MS received 31 October 2020; accepted 17 December 2020 Abstract. Over the last 5 years, UVIT has completed observations of more than 500 proposals with  800 unique pointings. In addition, regular planned monitoring observations have been made and from their analysis various key parameters related to in orbit performance of UVIT have been quantified. The sensitivities of the UV channels have remained steady indicating no effect of potential molecular contamination confirming the adequacy of all the protocols implemented for avoiding contamination. The quality of the PSF through the years confirms adequacy of thermal control measures. The early calibrations obtained during the Performance Verification (PV) phase have been further revised for more subtle effects. These include flat fields and detector distortions with greater precision. The operations of UVIT have also evolved through in orbit experience, e.g. tweaking of operational sequencing, protocol for recovery from bright object detection (BOD) shutdowns, parameters for BOD thresholds, etc. Finally, some effects of charged particle hits on electronics led to opimised strategy for regular resetting. The Near-UV channel was lost in one of such operations. All the above in-orbit experiences are presented here. Keywords. Space vehicles: AstroSat—telescopes: UVIT—instrumentation: astronomical imaging.

1. Introduction The Ultra-Violet Imaging Telescope (UVIT) is one of the five major scientific payloads on board the first Indian multi-wavelength astronomical satellite mission AstroSat, which was launched on September 28, 2015, with the Indian Space Research Organisation, ISRO’s PSLV-C30 rocket. UVIT consists of two This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

identical telescopes of aperture 375 mm and field of view  280 . UVIT has high angular resolution imaging capability in the Far-UV (130–180 nm) and NearUV (200–300 nm) wavebands using selectable narrow/medium/wide bandwidth filters as well as slit-less spectroscopic imaging. A simultaneously viewing optical band, VIS (320–550 nm) is incorporated to aid implementation of the shift and add algorithm, for getting long exposure images from short exposure frames, on the ground for avoidance of blurring due to

20

Page 2 of 12

drift in the telescope aspect. The details about various sub-systems of UVIT and their respective qualifying tests and calibrations are described in Kumar et al. (2012a, 2012b). After an initial 6 week long in orbit out-gassing phase, gradually individual subsystems of UVIT have been tested for their functionalities and performance with the doors of the twin telescopes still closed. The first light of the entire end to end UVIT system was carried out on November 30, 2015, by imaging the Galactic open cluster NGC 188. This was followed by the nearly six month long Performance Verification (PV) phase when detailed tests, characterization and calibrations were carried out. Many details about results from the early phase of UVIT in orbit have been presented in Subramaniam et al. (2016b) and Tandon et al. (2017a, b). UVIT was thrown open (along with other payloads for X-ray astronomy) for astronomical observations planned as per peer-reviewed scientific proposals, at first under guaranteed time (GT) cycle followed by announcement of opportunity (AO) cycles. During these cycles, additional calibration proposals have also been executed at regular intervals for monitoring the health and quantifying the stability of UVIT’s performance. This article summarizes the journey of UVIT over the first five years in orbit in terms of performance and achievements including unanticipated events and recovery therefrom.

2. Performance verification phase 2.1 Tests prior to opening of UVIT doors The earliest in orbit activities of UVIT pertained to qualification of all functions of the payload other than the optical systems prior to opening of the doors, lasting  50 days post launch. (In these 50 days only 17 days were used for operations with UVIT, rest were used by the other 4 instruments.) These involved electrical and mechanical sub-systems – e.g. communication systems with the spacecraft, various ‘‘states’’ of the detector module, detector read out system, detector safety logic (Bright Object Detection, BOD), generation of high voltages, rotational movements of the filter wheel system etc. Tests were carefully planned in phases with gradually increasing complexity and with live control ensuring options of aborting in case of encountering any abnormality (required scheduling and coordinating operations only during visibility of the spacecraft above main ground

J. Astrophys. Astr. (2021)42:20

station). Cosmic Ray Shower events passing through the detector system provided opportunity for checking certain functionality in the absence of UV photons from the sky. Beginning with operations for one band (among FUV/NUV/VIS) at a time, eventually simultaneous all 3 band operations were qualified. No abnormality was encountered through this phase of commissioning of UVIT’s detector and filter wheel systems prior to opening of the doors.

2.2 First light and lessons from early imaging operations The very first imaging of the sky with UVIT was carried out on 30 November 2015 targeting the Galactic open star cluster NGC 188, whose coordinates (high declination) offered the advantage of optimal visibility through any season allowing long term follow up monitoring. Accordingly, selected stars in this cluster were used as secondary standards for photometric calibration. One of the most critical components of UVIT is the Detector module employing a complex image intensifier system, operable in photon counting mode. Key requirements driving this choice were: (1) to keep drift in pointing to \\100 within an individual frame, the speed of reading out was required to be [ 10 frames/s, (2) given that the total number of UV photons detected in 0.1 s could be as low as 10, a read out noise of \1 (rms) was required. Therefore, read noise of CCD would not be acceptable. Further, a red-block filter with zero red leak would be required with a CCD, to completely block the longer wavelengths. The image intensifier configured for each band of UVIT consists of a photo-cathode deposited on the window where primary electrons are generated by photo-electric effect by incident photons. These are then multiplied by a large factor (gain) using a MicroChannel Plate assembly, MCP, biased to selectable high voltages. The stream of secondary electrons exiting the MCP are made to strike a phosphor acting as anode generating optical light pulses. These pulses are detected by a CMOS imager, Star250, with 512  512 pixels coupled through a fibre-optic taper. This imager is continuously read out as individual frames during imaging operation. The MCP needs carefully planned protection against exposure from bright objects since it can deliver only a limited amount of total charge in its operational life. The areas of MCP having experienced prolonged

J. Astrophys. Astr. (2021)42:20

Figure 1. An example of Bright Object Detect (BOD) trigger due to appearance of an unanticipated non-celestial object in UVIT’s field-of-view. One image frame in VIS band is displayed. The bright streak (near the bottom left corner) caused this BOD trigger.

exposure to high fluxes loses its electron multiplying functionality leading to ineffective areas or even complete damage. In addition, the excessive load on the high voltage supplies could damage them too. Accordingly, two key safety features have been

Page 3 of 12

20

introduced: (i) a Bright Object Detect (BOD) logic has been implemented in the onboard signal processing scheme which triggers a safety shutdown of the affected band and also raises an alarm to the spacecraft for similar action in other bands; and (ii) the initial imaging of any fresh sky field or filter mandatorily begins with a very low gain gradually achieving the optimal setting (by ramping up the high voltages at a selectable rate) which allows the BOD logic sufficient time to process the incoming raw frames read out from the imager. The selectable parameters related to the triggering threshold for the BOD logic are: pixel threshold, 1-D size along faster read-out axis of the frame, and the number of consecutive frames in which the pixel threshold has been exceeded over at least one string of pixels of 1-D size. Given the long period of detector’s operations on ground during extended tests and calibrations, significant experience existed regarding choices for these parameters. In spite of these knowledge base, an accidental BOD operation awaited the UVIT team on first light of VIS channel! It resulted from an erroneous choice of settings for the high voltages – the set corresponding to photon counting mode was configured while attempting to image in Integration mode. This oversight was corrected swiftly. Very soon another BOD trigger occured due to a star in a field thought to be safe, indicating error in estimation of ‘signal per

Figure 2. Example of artifacts appearing in VIS band images. The horizontal stripes occur possibly due to effects of charged particle hits and they disappear after a power reset.

20

Page 4 of 12

photon’ from the ground calibration. The observed signal in orbit was higher, needing adjustment of the trigger threshold. Tweaking of all detector related parameters to optimal values were completed within days of the first light. The selection of the band whose clock would be used as ideal Master Clock, i.e. in effect the clock for all the three bands, as well as a consistence sequence of their configuration for imaging were achieved shortly. This sequence being rather critical, its reliable implemention was achieved by embedding these details in well designed Macros for operations. The initial ramping up of high voltages was enforced for every change of filter. Another important realization during this phase was the need for safeguarding the detector against bright stars while imaging in window mode, i.e. recording data only for part of the full field (selected size smaller than the full size of 512  512). This risk was mitigated by introducing a mandatory full window mode imaging preceding the imaging with a smaller window so as to ensure that BOD is triggered in case a bright object is present in the field though outside the selected window. All such details were implemented in macros for imaging operations. Eventually, this protocol was extended for even full window imaging, to bring in uniformity in operations for simplicity.

J. Astrophys. Astr. (2021)42:20

3. Experiences from long term operations of UVIT UVIT has been in regular use after its commissioning serving scientific as well as calibration proposals. The operations involved imaging as well as slitless spectroscopy. Over the last  5 years, more than 500 proposals have used UVIT. While most of the time UVIT has performed extremely well meeting planned design specifications, at times there have been occasional technical issues, needing urgent review followed by action. The most frequently occurring event has been the automatic shutdown of UVIT due to trigger of Bright Object Detect (BOD) logic. When any band of UVIT encounters a brighter than programmed safe limit, it autonomously parks itself in a safe mode and raises an alarm to the spacecraft bus, which in turn shuts down all the three bands of UVIT following a safe sequence of operations. From the record of which band triggered the BOD, on ground the cause for exposure to such a field is investigated. It may be noted that extreme care is taken while technically approving any sky field to be observed with UVIT, which involves consideration of all cataloged bright optical and UV objects in the target field. In most cases, either a human error or a large offset in pointing have been identified. On a few rare occasions, brightening of a (variable) star of brightness close to the safety limit or passage of a bright non

Figure 3. Example of an artifact due to a stuck bit in the Master Clock. The time stamped on individual frames (Frametime), are plotted against the elapsed time (Tickcount). The fixed amount of jumps at each discontinuity indicates the 20th bit of the Frametime to be stuck at zero.

J. Astrophys. Astr. (2021)42:20

celestial object (shining satellite) is responsible for triggering the BOD. An example of the latter is displayed in Fig. 1. After each instance of BOD triggered shutdown of UVIT, a well defined recovery protocol is followed to normalize the UVIT bands after which regular operations can proceed. Typically about 3–4 incidents of BOD trigger were encountered annually. It is noteworthy that the over-current in high voltage units (described later) was never triggered. This possibly points to the quality/ruggedness of the intensifier and the high voltage generating circuits. An anomaly noticed rather early on (within months of in orbit operations) was appearance of stripes in the raw image frames of VIS band (see Fig. 2). These were diagnosed to be effects due to charged particle hits and complete recovery could be achieved by powering OFF the VIS electronics. Since the processing pipeline on ground was agile enough to ignore such artifacts and generate final products unaffected, the mitigation action was not carried out till the situation demanded powering OFF. At a later phase, when such stripes in VIS appeared more often, along with some additional artifacts in UV bands too, a monthly schedule was drawn up to normalize (power OFF) all the 3 bands FUV, NUV and VIS. The on-board logic was designed to handle five kinds of anticipated emergency situations in UVIT.

Page 5 of 12

20

These are described in order of increasing level of severity. (1) TM acknowledgment error – Initiating an imaging session starting from inactive state, involves configuring the detector following an unique sequence of commands which gradually activate relevant sub-systems ensuring complete safety. Another sequence (reverse order) is used for returning to inactive state. Receipt of any command violating the above triggers this alarm. (2) Failure of the filter wheel mechanism to reach its targeted angle within a stipulated time. (3) Threat to safety of the intensifier on detection of a bright star (BOD). (4) Threat to safety of the high voltage power supplies when current drawn exceeds the set limit. (5) Transition to Fail Safe state on detection of high current indicating radiation induced Single Event Latchup. One critical point for each of these types of emergencies is the final parking state. During pre-launch deliberations, it was decided that the relatively benign ones (first 3) should land the detector electronics (for all 3 bands) to low power state and the remaining 2 severe ones to complete power OFF state through operation of mechanical relays. Being cautious during

Figure 4. Example of multiple bits of the centroid coordinates (along Y-axis) of every photon event to be stuck to zero. The points in the plot correspond to individual photons. The systematic gaps in the values of Y-centroid indicate several successive least significant bits (corresponding to the fractional part) are affected.

20

Page 6 of 12

J. Astrophys. Astr. (2021)42:20

Table 1. Chronology of issues observed in UVIT data* and their resolution. Serial No. 1 2

3 4 5

6

7 8

9

10

Date of appearance

Date of resolution

16-Apr-2017 29-Apr-2017

06-Jun-2017 06-Jun-2017

Type of issue

Remarks

Vertical stripes in NUV images Anomalous size of VIS images

Recovered by RESET Recovered by RESET Data made usable by developing mitigation scheme in ground software 18-Nov-2017 23-Nov-2017 No FUV data Recovered by RESET 06-Feb-2018 22-Feb-2018 Unable to turn on NUV band after Recovery managed after exploration of many a shutdown due to BOD strategies to power on 30-Mar-2018 — Unable to turn on NUV band after No recovery yet, despite repeated attempts a shutdown exploring multiple strategies (periodic attempts continue) 25-Dec-2018 13-Jul-2019 Stuck 20th bit in time stamps on Eventual recovery by RESET (after adopting FUV frames many alternate schemes avoiding RESET, in view of the loss of the NUV band); The data collected during this period made usable by developing mitigation scheme in ground software 05-Mar-2019 04-Jul-2019 X-centroids (FUV) with stuck bits Recovered by RESET (after hesitation for use of the RESET) 03-Nov-2019 16-Nov-2019 Stuck 31st bit of Time Stamp in Recovered by RESET (after exploring FUV band frames (relaying alternatives to RESET) Master Clock from VIS) 26-Nov-2019 13-Dec-2019 Stuck 31st bit of Time Stamp in Recovered by RESET (with understanding of FUV band frames (relaying the damage, future occurrences of the issue Master Clock from VIS) became predictable; devised a periodic RESET plan, which works well) 13-Oct-2020 15-Oct-2020 Y-centroids (FUV) with stuck bits Recovered by RESET

*The many instances of appearance of stripes in VIS band images are not listed here. They were always resolved through RESET. Also mitigating scheme was incorporated in the Level-2 pipeline to handle such artifacts with no loss of functionality, so that data preceding RESET also remain fully usable.

early in orbit operations, this strategy was made more conservative by demanding all 5 situations to lead to OFF state. As a result, BOD used to result in OFF state needing re-boot of the system during recovery operation. During one of the post-BOD recovery operations (January 20, 2018), the NUV channel failed to restart. Based on ground simulations using the engineering model and after prolonged trials based on different well considered strategies (stretched RESET pulses), the NUV band could be recovered (February 16, 2018). Unfortunately, failure of the NUV channel to boot recurred after a routine monthly normalization (March 20, 2018). A very long struggle to revive the NUV channel followed. The various innovative strategies employed to recover the lost channel included: (i) different frequency of normalization trial, (ii) widening of the RESET pulse, (iii) gradual

warming up of the affected electronics and trying RESET at selected higher temperatures (this involved tweaking parameters of the spacecraft’s thermal system settings), etc. While none of the above could recover the NUV channel, the main cause of the failure was eventually understood through deeper study of technical literature. Technical understanding of this failure emerged as follows: the onboard processing of the UVIT detector data is implemented in a FPGA, whose code gets loaded from an EEPROM every time the system is powered on or undergoes RESET. The EEPROM (also the FPGA) used were not of the radiation hardened grade but of MIL spec grade, which is susceptible to damage by cosmic ray radiation. In addition, the EEPROM was of serial type which is known to compromise the reliability. As a further weakness, the code design did not provide any option for re-

J. Astrophys. Astr. (2021)42:20

Page 7 of 12

20

Figure 5. An unexplained feature (streak) observed in images in all 3 bands of UVIT for certain targets. As an example, image of the galaxy M31 in NUV band is displayed. The orientations of the streaks with reference to mountings of FUV/ NUV/VIS detectors imply that they could be caused by some structure within the telescope tube.

programming. Such devices develop weak-cells (due to radiation damage) which are also known to deteriorate further with more read-cycles. Based on the above technical understanding of the reason for loss of NUV band, the practice of monthly normalization of VIS and FUV bands was completely discontinued. In addition, the strategy for on board handling logic post occurrence of BOD emergency state was revised. The parking state on BOD trigger was changed to low power (avoiding any booting action which involves re-loading of the code from EEPROM to FPGA). Attempts for reviving the NUV band continues to date at regular intervals. More recently (since  December 2018), some other types of artifacts due to radiation hits were discovered, which could be recovered. However, their recovery required power RESET. These artifacts were: (a) frozen 20th bit of VIS clock which is the selected master for all bands (see Fig. 3), (b) certain frozen bits of all X-centroids for photon events, etc. Initially, on encountering (a), plans and procedures were set up for change over of the Master Clock from VIS band to FUV band thereby avoiding issuance of any RESET (in view of the loss of NUV band). This involved re-programming all command tables

corresponding to ‘imaging parameters’ for each band. The uploading of these tables to the relevant spacecraft sub-system involved certain new activities. After a review by ISRO experts, this course of action was not found to be advisable. In the mean time, (b) was encountered which forced the use of RESETs which eventually mitigated both these issues. Most recently (October 2020), all bits corresponding to fractional part of the Y-centroid were stuck (see Fig. 4). All instances of appearance of such artifacts (other than periodic appearance of stripes in VIS image frames) and subsequent recovery from them are presented chronologically in Table 1. The lone instance (over 5 years of operations) of UVIT bands remaining in active imaging mode beyond the mission schedule and being exposed during one full bright (sunlit) part of the orbit, was experienced on September 19, 2020. However, no damage or degradation of any performance of UVIT was noticed after this incident. The reason for this anomaly was traced to a logic in spacecraft operations which has been mitigated. One non-recoverable artifact encountered (early November 2019) was the stuck 31st bit of FUV clock relaying Master Clock counter from VIS band (which

20

Page 8 of 12

continued to record correct values). It is fortunate that this bit can change state only after a few weeks and accordingly, this defect could be by passed by an aggressive mitigation plan. The plan involved a periodic RESET of the VIS band electronics every  12 days. Despite the anomalies described above, the two bands of UVIT, viz., FUV and VIS have been serving scientific observation plans leading to interesting research results. Additional effects/events, some unexpected others expected, that were experienced are summarized here. The effects due to Cosmic Rays (CR) on the detector system were anticipated and their handling by the offline data processing pipeline on ground were planned accordingly. The primary CR-energetic charged particle – itself interacting in an imager pixel could corrupt its value corresponding to a large signal. On the other hand, showers of secondary charged particles generated by interaction of primary CR in the proximity of the detector and high electrical fields around MCP, mimic UV photons. While in the low gain operation of the detector (employed for VIS band), only one pixel is affected per primary CR, a large number of randomly located background events are recorded in the high gain operation of the detector used for FUV and NUV bands due to the showers. Fortunately, each shower lasts at most a few microseconds, and hence can affect only one frame. However, the frequency of occurrence of such cosmic ray shower is crucial. The mitigation plan in ground software is to identify frames affected by showers using selectable parameters (statistically determined threshold on number of events) and flagging them for discarding. From long term experience, on average 3.5 showers per sec are observed contributing to  150 events/s for full field operations. The impact of discarding affected frames is rather small (loss of  10% of the observations with full field). Given the uncorrelated nature of the cosmic ray background events, they may be ignored for fields which are UV bright. However, for very deep observations it is important to discard affected frames (Saha et al. 2020). Example of an observed artifact is a bright ‘‘streak’’ due to some very bright object within a few degrees of the telescope axis (though outside UVIT’s 28 arc-min field-of-view). Such streaks have been observed in VIS as well as NUV images. The orientation of the streaks in these images imply a direction fixed to the telescope tube, once the relative angle between the axes of the detectors is taken into account. Such streaks have been observed while

J. Astrophys. Astr. (2021)42:20

observing the targets: M 31, PKS 1510-089 and Crab. One example is presented in Fig. 5 which is an NUV image of M 31. Another artifact presented here relates to the effects due to saturation in photon counting mode of imaging. A deep valley encircling the central peak is observed for a bright star in the star cluster NGC 188 imaged in NUV, which is displayed in Fig. 6. The accompanying plot of the radial profile quantifies this dip. The ground segment software system at ISRO provides absolute time information for the science data by correlating internal clocks of UVIT (one per band) with Universal Time Clock (UTC) in the Level1 (L1) products. This utilizes simultaneous samples of

Figure 6. Effect of saturation is illustrated in the NUV band image of a bright star in the cluster NGC 188 (size: 2800  2800 ). The valley encircling the central peak is due to saturation effect of photon counting mode. The radial profile displays the local dip quantitatively.

J. Astrophys. Astr. (2021)42:20

Page 9 of 12

20

Figure 7. The sensitivity of the FUV band over its operations covering earliest operations till May 2020 has remained unchanged. The plot shows count rates from monitoring of a secondary calibrator, WOCS-5885, a star in the cluster NGC 188. The rms scatter is  1% over 1600 days.

spacecraft’s clock counter with those from UVIT. Accordingly, the UVIT’s Level-2 (L2) processing pipeline was designed using UTC as the primary timing reference. However, often this time correlation in L1 was found to be unreliable. Hence, it was necessary to incorporate new functionality in the L2 pipeline to by-pass UTC and use the Master clock of UVIT which ensured inter-band time synchronization. Even in the UTC by-pass mode a provision for approximate (good to  1 s) absolute time (MJD_UT) for every frame was made based on an intelligent algorithm.

4. Key performance parameters The key parameters of performance of UVIT include the following: (1) The photometric calibration quantified by zeropoint magnitude and the unit conversion factor for all the filters. (2) Effects of saturation. (3) Variation of sensitivity across the field of view (flat field). (4) Point Spread Function, PSF, and its variation over the field. (5) Dispersion, resolution and effective areas in the grating mode. (6) Astrometric calibration including distortion. The procedure followed to arrive at these and related details have been presented in Tandon et al.

(2017c, 2020). Here we summarize the results reported there. Based on the observations carried out during the initial 18 months or so led to the first phase of calibration results (Tandon et al. 2017c). The zero-point magnitudes for all the filters in FUV and NUV band were quantified. The measured sensitivities in FUV and NUV are found to be [ 80% of the expectations based on tests carried out on ground. The spatial resolutions (PSF FWHM  1:3–1.5 arc-sec in FUV and  1:0–1.4 arc-sec in NUV) are found to be better than expected. The variation of PSF across the 280 field is small for the FUV band. For the NUV band an increase of  10% is found in FWHM in the central part of the field compared to the edges. No detectable change in the PSFs in both UV bands have been found to date. The astrometric accuracy over the full field is found to be  0:5 arc-sec RMS. With passage of time (  3 years) additional regular calibration observations with UVIT were carried out. Based on these extensive additional database and improved understanding of the instrument, further refinements to the first phase calibrations were conducted. The results from these studies have been reported in Tandon et al. (2020). These improvements have led to quantification of new photometric calibrations which included subtle effects. The zero-point magnitudes, ZP, for most filters have been revised, and for some with improved precision (e.g. error on ZP for the filter N245M reduced from 0.07 to 0.005). For example, improved determination of ZP for the CaF2 filter

20

Page 10 of 12

J. Astrophys. Astr. (2021)42:20

of the field and when they fall near edge of the field are  0:06. This suggests that the errors on flat-field corrections are no more than 6%. As a result of these improvements, the astrometric accuracy improved to 0.4 arc-sec (rms), indicating uncorrected distortion to be \0:3 arc-sec (rms). The spectral, PSF and astrometric calibrations have also been improved upon. For both the NUV and FUV bands, the FWHM of the PSF is found to be 1.4 arcsec or better within the central 24 arc-min of the field. The new results conclude that there has been no reduction in sensitivities of FUV and NUV bands. The result from monitoring the sensitivity of the FUV band over the entire mission including recent observations is presented in Fig. 7, which shows no detectable change in the count rate for a secondary calibrator star, WOCS-5885, in the cluster NGC 188

Figure 8. Example of typical variation of temperature of the two UVIT telescopes (FUV and NUV/VIS) at respective Tube Top (TT) and Tube Bottom (TB) over one week.

(F148W) in the FUV band and the Silica filter (N242W) in the NUV band led to the values 18:097  0:01 and 19:763  0:002 respectively. The flat fields have been significantly improved by supplementing remainders by analytic functions (third-order polynomial at the central region and linear in radius with azimuthal dependence for the outer regions). The achieved accuracy with this improved flat field correction scheme has been estimated from exposures on multiple fields of Small Magellanic Cloud. The fractional differences in the flat-field corrected counts for sources when they fall near centre

Figure 9. Example of typical variation of temperature of the 3 detectors (FUV, NUV, VIS) in UVIT over one week. The temperature is measured on the Camera Proximity Unit housing the image intensifier and the CMOS sensor STAR 250.

J. Astrophys. Astr. (2021)42:20

Page 11 of 12

20

Figure 10. A plot showing distribution on sky of the astronomical targets observed by the UVIT (total  800 unique pointings).

(Subramaniam et al. 2016a). The stability of sensitivities over long duration has vindicated the care taken over the years on ground towards control of molecular contamination, viz., choice of materials, operations in clean room, cleaning protocols, preassembly baking, purging of the optical cavity of UVIT with pure N2 gas. In addition, the in orbit protocols followed: long in orbit wait for degassing before opening the doors as well as avoiding direct sun light falling on telescope tubes during spacecraft maneuvers also helped. The stability of the PSF is ensured by adequate on board thermal control through the mission. The temperatures of critical elements, viz., telescope tubes and detectors (measured over  a week) responsible for the observed stability in sensitivity as well as PSF size are displayed in Figures 8 and 9 respectively. The temperature of the tubes is stable within 0:5°C, which translates to a geometrical blurring in the image by \0:1 arc-sec rms, implying a change from  0:5 arc-sec to \0:51 arcsec rms for the PSF. The temperature of detector (Star250 imager) is stable to within 1°C.

5. Epilogue UVIT has been performing quite satisfactorily over the years with most of its in orbit specifications close to and a few even better than the corresponding targeted values. A large fraction of AstroSat time has been allocated to UVIT by the Time Allocation Committee based on user driven scientific proposals. The distribution on sky of the astronomical targets

observed with UVIT is displayed in Fig. 10, which includes about 800 unique pointings. A large number of important astronomical results have already appeared in journals with high impact (possibly many more are in the process). All scientific payloads of AstroSat including UVIT, had a baseline design life of 5 years, which has already been achieved. While it is a pity that the NUV band was lost at the mid-point of this life, due to radiation damage of some critical electronic components, the FUV and VIS bands continue to serve the users. It is hoped that these two bands will last many years in future. Although HST allows extremely sensitive imaging in the FUV (STIS, FUV-MAMA; 2500  2500 ), UVIT currently provides the unique opportunity for wide field imaging with 280 diameter field.

Acknowledgements The UVIT project is a result of collaboration between IIA, Bangalore, IUCAA, Pune, TIFR, Mumbai, many centers of the Indian Space Research Organization (ISRO), and the Canadian Space Agency. We thank these organizations for their support. We gratefully thank members of the Ground Segment software and Mission Operations teams of ISRO for their continuing support. We also thank members of the AstroSat Project and the AstroSat Science Working Group for their feedback. This publication uses the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC)

20

Page 12 of 12

References Kumar A., Ghosh S. K., Hutchings J. et al. 2012a, SPIE, 8443, 84431N Kumar A., Ghosh S. K., Kamath P. U. et al. 2012b, SPIE, 8443, 84434R Saha K., Tandon S. N., Simmonds C. et al. 2020, Nat. Astron., https://doi.org/10.1038/s41550-020-1173-5 Subramaniam A., Sindhu N., Tandon S. N. et al. 2016a, ApJ, 833, L27

J. Astrophys. Astr. (2021)42:20 Subramaniam A., Tandon S. N., Hutchings J. B. et al. 2016b, SPIE, 9905, 99051F Tandon S. N., Ghosh S. K., Hutchings J. B. et al. 2017a, CSci, 113, 583 Tandon S. N., Hutchings J. B., Ghosh S. K. et al. 2017b, JApA, 38, 28 Tandon S. N., Postma J., Joseph P. et al. 2020, AJ, 159, 158 Tandon S. N., Subramaniam A., Girish V. et al. 2017c, AJ, 154, 128

J. Astrophys. Astr. (2021)42:49 https://doi.org/10.1007/s12036-021-09691-w

Ó Indian Academy of Sciences Sadhana(0123456789().,-volV)F T3](0123456789().,-volV)

PAYLOAD CALIBRATION

Calibration of AstroSat/UVIT gratings and spectral responses G. C. DEWANGAN Inter-University Centre for Astronomy and Astrophysics (IUCAA), SPPU Campus, Pune 411 007, India. E-mail: [email protected] MS received 6 November 2020; accepted 16 December 2020 Abstract. AstroSat/UVIT carries two gratings in the FUV channel and a single grating in the NUV channel. These gratings are useful for low resolution, slitless spectroscopy in the far and near UV bands of a variety of cosmic sources such as hot stars, interacting binaries, active galactic nuclei, etc. We present calibration of these gratings using observations of UV standards NGC 40 and HZ 4. We perform wavelength and flux calibration and derive effective areas for different grating orders. We find peak effective areas of ˚ for the 1 order of NUV-Grating,  4.5 cm2 at 1390 A ˚ for the 2 order of FUV 18.7 cm2 at 2325 A 2 ˚ for the 2 order of FUV-Grating2. The FWHM spectral resolution of the Grating1, and  4.3 cm at 1500 A ˚ in the 2 order. The 1 order of NUV grating has an FWHM resolution of 33 A. ˚ FUV gratings is 14.6 A We find excellent agreement in flux measurements between the FUV/NUV gratings and all broadband filters. We have generated spectral response of the UVIT gratings and broadband filters that can directly be used in the spectral fitting packages such as XSPEC, Sherpa and ISIS, thus allowing spectral analysis of UVIT data either separately or jointly with X-ray data from AstroSat or other missions. Keywords. Ultraviolet astronomy—ultraviolet telescopes—ultraviolet detectors—calibration.

1. Introduction The Ultra-Violet Imaging Telescope (UVIT; Subramaniam et al. 2016; Tandon et al. 2017, 2020) is one of the four co-aligned payloads on-board the Indian multi-wavelength space observatory AstroSat (Agrawal 2006; Singh et al. 2014). UVIT is a twin telescope system, one of them is designed to observe in the far ˚ and is known as the ultra-violet band (1300–1800 A) FUV channel. The second telescope utilizes a beam splitter that separates near UV and visible light, thus forming two detection channels – near UV (NUV; ˚ and visible (VIS; 3200–5500 A). ˚ The 2000–3000 A) three channels use identically configured intensified CMOS detector systems which only differ in having different photo-cathodes as per the wavelength band, and operate simultaneously. The beam splitter and the mechanical mounting of the telescopes cause the NUV This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

channel field to be inverted and rotated by 33 with respect to the FUV channel field. Each channel is equipped with a set of filters that allow selection of ˚ bands with spectral coverage of Dk  125–500 A ˚ ˚ (FUV), Dk  90–800 A (NUV) and Dk  400–2200 A (VIS). The FUV and NUV channels provide excellent imaging capability with a point spread function (PSF) in the range 1–1.5 arcsec (FWHM). The UVIT is mainly an imaging instrument with limited spectral capability. It can be used for low resolution, slitless spectroscopy in the far and near UV bands. There are three UVIT gratings that are ruled with 400 lines mm1 on CaF2 substrates of 4.52 mm thickness. The dispersion in the detector plane caused ˚ arcsec1 in ˚ arcsec1 and 6 A by each grating is 12 A ˚ One the first and second order, respectively, at 1350 A. of three gratings is mounted on the NUV channel filter wheel while the other two gratings are mounted on the FUV channel filter wheel such that their dispersion axes are nearly perpendicular. Such an orthogonal arrangement of FUV gratings helps in avoiding

49

Page 2 of 10

J. Astrophys. Astr. (2021)42:49

contamination of nearby sources in the grating images. A source along the dispersion arm of the main target in one FUV grating image will cause contamination while the dispersion arms will be well separated in the other FUV grating image due to the orthogonal arrangement. The FUV gratings and the detector are designed to maximize the efficiency in the m ¼ 2 order while the NUV grating and the detector provide maximum efficiency in the m ¼ 1 order. The main parameters of the gratings are listed in Table 1. More details on UVIT gratings can be found in Tandon et al. (2017, 2020). The UVIT gratings have been described with different names, in Table 1 we list all the IDs to avoid any confusion when using different resources. Here we refer the gratins as FUVGrating1 (FUV-G1), FUV-Grating2 (FUV-G2) and NUV-Grating (NUV-G). In this paper, we present calibration of the UVIT gratings. Some of the results derived in this paper are presented in Tandon et al. (2020). Here, we present the calibration of the UVIT gratings, some results of which can be found in Tandon et al. (2020). Our updated work here results in minor changes to the wavelength and flux calibrations, and also includes additional grating orders. We describe the calibration method, derive additional calibration products and discuss cross-calibration between the gratings and broadband filters. The paper is organized as follows. We describe the UVIT data and the reduction in Section 2, extraction of one dimensional (1d) grating spectra in Section 3, and wavelength calibration in Section 4. We derive effective areas and perform flux calibration in Section 5, and discuss cross-calibration between the gratings and broadband filters in Section 6 In Section 7, we derive instrumental response in the form of a product of redistribution matrix and ancilliary response that can be directly used in popular

spectral fitting packages in X-ray astronomy. Finally we summarize our results in Section 8.

2. Calibration observations and data reduction We used UVIT observations of NGC 40 and HZ 4, these observations are as listed in Table 2. The planetary nebula NGC 40 has a rich set of UV emission lines and is well-suited for wavelength calibration. The white dwarf HZ 4 is a well-established spectrophotometric standard with nearly featureless spectrum, and is appropriate for flux calibration (e.g., Bohlin et al. 1990). HZ 4 has also been used for the photometric calibration of UVIT filters (Tandon et al. 2017, 2020) We used the Level1 data from the observations listed in Table 2. We used the UVIT pipeline CCDLAB (Postma & Leahy 2017) to process the Level1 data. The pipeline performs corrections for field distortion, centroiding bias, flat field, pointing drift and accounts for frames rejected due to cosmic rays or missing from the Level 1 data. We used the VIS images to generate drift series which could then correct for the pointing drift in each orbit. We generated cleaned images for each orbit, aligned and merged them for each grating, as shown in Fig. 1. The x, y coordinates in these images represent 1/8 subpixel coordinates. Hereafter, we refer to these 1/8 subpixels simply as pixels which are 0.413 arcsec wide. The images show the undispersed zeroth order image together with the dispersed 1 and 2 order two dimensional spectra. The dispersion axis for the FUVG1 and NUV-G is nearly horizontal, while it is roughly vertical in the case of FUV-G2. The grating images show maximum intensity in the blazed order (m ¼ 2 for FUV gratings and m ¼ 1 for the NUV

Table 1. UVIT grating parameters. Parameter

FUV-Grating1

FUV-Grating2

NUV-Grating

Filter Wheel Slot number IDs in APPS IDs in CCDLAB This paper IDs in Tandon et al. (2020) IDs in Tandon et al. (2017) IDs in UVIT Pipeline m ¼ 1 peak k m ¼ 2 peak k Spectral resolution (FWHM)

4 4–Grating1 (FUV) FUV-Grating1 FUV-G1 FUV1 2nd FUV grating (#66126) F4 – ˚  1400 A ˚ 14.6 A

6 6–Grating2 (FUV) FUV-Grating2 FUV-G2 FUV2 1st FUV grating (#63771) F6 – ˚  1500 A ˚ 14.6 A

4 4–Grating (NUV) NUV-Grating NUV-G NUV NUV grating (#66125) F4 ˚ 2100 A – ˚ 33 A

J. Astrophys. Astr. (2021)42:49

Page 3 of 10

49

Table 2. UVIT/grating observations of NGC 40 and HZ 4. Target

ObsID

Date of observation

Grating FUV-G1 FUV-G2 NUV-G FUV-G1 FUV-G2 NUV-G

NGC 40

C02_010T01_900000

2016-12-07

HZ 4

T01_054T01_9000000

2016-02-02

Exposure time (s) 1194.0 1186.9 2407.8 1337.8 1026.2 3082.9

Window size 350  350 350  350 350  350 512  512 512  512 512  512

Figure 1. FUV/NUV grating images of NGC40. The negative grating orders are marked. The image sizes are  9.8 arcmin on one side.

grating). The FWHM of the intensity distribution along the cross-dispersion direction is a measure of combined PSF due to the telescope, detector and grating. Since gratings introduce some distortions, the PSF is slightly broader (FWHM  2arcsec) for grating images compared to that for the broadband filter images (Tandon et al. 2017). For the calibration and analysis of the UVIT data processed with the CCDLAB pipeline, we have developed a software package UVITTools in the Julia language (Bezanson et al. 2012). We have used this package extensively here. We also used the FTOOLS1, Sherpa (Freeman et al. 2001) and XSPEC (Arnaud 1996) packages for generating response files and fitting the data.

3. Extraction of 1d spectra It is clear from Fig. 1 that the dispersion axes are not exactly aligned to either of the x or y axis. We measured the tilt angle relative to the x-axis and found that the dispersion axes can be represented by the linear relations y ¼ mx þ c where m ¼ tan h, c ¼ yo  mx0 1

https://heasarc.gsfc.nasa.gov/lheasoft/.

and x0 and y0 are the pixel coordinates of the centroid of the zero order image. We list the tilt angles in Table 3. The linear relations define the spectral trace i.e., the centroids of the cross-dispersion spatial profiles at each pixel on the dispersion axis. There is no additional significant distortion in the dispersion direction, so spatial profile fittings to trace the dispersion direction is not required. We define coordinates along the dispersion direction as the pixel coordinates relative to the zero order position. We examined the grating images visually and determined the range of coordinates for different grating orders. These ranges are listed in Table 3. We used a 50 pixel width along the spatial direction centered on the trace defined by the linear relations and summed the counts to generate 1d spectra of a source of interest. Figure 2 shows the region used for the extraction of 1d spectrum of NGC 40 in the m ¼ 1 order. We also extracted a 1d background spectrum from a source-free region in the image using the same relative coordinates in the dispersion direction and the same width along the spatial direction as used for the source. We subtracted the background counts from the source counts and propagated their errors on counts. The net 1d spectra of NGC 40 thus generated are shown in the left panels of Fig. 3.

49

Page 4 of 10

J. Astrophys. Astr. (2021)42:49

Table 3. Parameters for 1d spectral extraction and the coefficients of the linear dispersion relation. Linear dispersion relation Grating FUV-G1 FUV-G2 NUV-G

Tilt angle (h) Order Range in relative pixel coordinate (X) 358:703 358:703 267:531 267:531 358:904

2 1 2 1 1

629 323 624 313 545

Figure 2. NUV grating image of NGC 40 with the grating orders marked. The extent along the dispersion direction and the width along the spatial direction used to extract the 1d spectrum is shown as the rectangular region covering the m ¼ 1 grating order.

4. Wavelength calibration The raw 1d spectra shown in Figure 3 are not in physical units. In order to convert the relative pixel coordinates along the dispersion direction, we used the emission lines from NGC 40. We measured the emission line positions by fitting Gaussian profiles along with low order polynomial for the continuum. We then identified the strong emission lines in the UVIT spectra using the emission lines listed in Feibelman (1999) based on IUE observations. We then fitted the following linear relation between the line wavelengths and the relative pixel numbers. ˚ ¼ a þ bX; kðAÞ ð1Þ where X is the pixel coordinate along the dispersion direction relative to the zero-order position. The bestfit linear relations are shown in the right panels of Fig. 3 for the blazed orders of FUV and NUV gratings. The slope and intercept of the best-fitting linear dispersion relation are listed in Table 3. We converted the pixel coordinates in the raw grating spectra to wavelength in Angstroms using the above dispersion relations.

to to to to to

413 213 426 228 336

Flux calibration is the process of converting the observed count rate to flux density as a function of wavelength. The background-subtracted net count rate

43.4 -18.0 31.2 45.0 45.1

Slope (b) 2:791 5:833 2:812 5:625 5:523

is related to the incident flux density fk nðergs ˚ 1 ) as cm2 s1 A   Z k CðXÞ ¼ RXk Ak fk dk; ð2Þ hc where C(X) is the net count rate at X along the dispersion direction, RkX gives the probability that a UV photon of wavelength k gets detected at X, Ak is effective area of the spectrometer accounting for the collecting area of the telescope, reflectivity of the mirrors, grating efficiency, detector quantum efficiency and attenuation efficiency of any other optical element. Thus Ak is telescope area corrected for all losses. In principle, Ak can be determined from pre-launch lab measurements, however, the various efficiencies generally change with time during and after launch. Hence, Ak is generally determined or corrected based on observations of standard sources with well measured fluxes. Equation (2) is appropriate for the photon counting data from UVIT, and can be written in matrix form with k, X now representing wavelength and spectral bins, CX ¼

X

RXk Ak fk

k

k : hc

ð3Þ

In the case of dispersive spectrometers such as the FUV/NUV gratings with suppression of order-overlap, the response matrix can be assumed to be diagonal. Hence RXk ¼ 1 if k and X follow the dispersion relation. Thus fk ¼

5. Flux calibration and effective area curves

Intercept (a)

Ck ðhc=kÞ ; Ak

ð4Þ

where Ck is the count rate spectrum. To determine the effective areas of the gratings, we used the UVIT grating observations of spectrophotometric standard HZ4 (whose UV spectral flux values were obtained from spectrum file

J. Astrophys. Astr. (2021)42:49

Page 5 of 10

49

Figure 3. The UVIT grating spectra of NGC 40 (left panels) and pixel-wavelength calibration (right panels) of the blazed orders (top panels: FUV-G1, m ¼ 2; middle panels: FUV-G2, m ¼ 2, bottom panels: NUV-G, m ¼ 1). The 1d spectra i.e., count rate vs. pixel numbers from the centroid of the 0-th order, are fitted with a number of Gaussian profiles for the emission lines and polynomial for the continua.

hz4_stis 005.fits available at HST-CALSPEC2). We matched the wavelengths bins of Ck measured with UVIT gratings and the spectrum by linearly interpolating the standard spectrum. One can then calculate the effective area of the UVIT-gratings using Equation (4). However, we note that HZ 4 shows ˚ (Lya) and 1400 A, ˚ strong absorption lines at 1216 A and the spectral resolution of the standard spectrum measured with IUE in the far UV band is superior compared to that of the UVIT grating spectra. We therefore smoothed the HZ4 standard spectrum to match the UVIT grating spectral resolution. We 2

http://www.stsci.edu/hst/observatory/cdbs/calspec.html.

used Gaussian kernels with widths that resulted in smooth effective areas using Equation (4) for different grating orders. The effective areas thus derived are shown in Fig. 4 for the 1 order of NUV-G, and in Fig. 5 for the 2 orders of FUV-G1 and FUV-G2. We have fitted the UVIT-grating effective areas with low order polynomials of the form Ak ¼ P cn ðk  k0 Þn so that the observed count rate spectrum can be converted to the fluxed-spectrum by dividing the best-fitting polynomials using Equation (4). The best-fitting coefficients are listed in Table 4, and the best-fitting polynomials are shown in Figures 4 and 5.

49

Page 6 of 10

J. Astrophys. Astr. (2021)42:49

Figure 4. The m ¼ 1 order NUV grating effective area, the best-fitting polynomial and the residuals.

6. Cross-calibration between gratings and broadband filters The flux densities measured by broadband filters at their mean wavelengths must be the same as those measured by the gratings at the same wavelengths if

the mean wavelengths represent the effective wavelengths of the filters. Small discrepencies may be expected as the mean wavelengths of the UVIT filters are not defined to filfil the above condition (private communication with S. N. Tandon). Nevertheless it is useful to compare the flux measured with the gratings and broadband filters. As mentioned earlier, gratings introduce distortion making the PSF slightly poorer. Hence, the same sizes for the extraction regions i.e., the spatial width along the cross-dispersion direction in the case of grating observations and the diameter of the circular region in the case of broadband filter observations, may not provide the same flux densities. Therefore it is important to cross-calibrate the gratings and the broadband filters. For this purpose, we used the calibration information and count rates already derived by Tandon et al. (2020). For HZ 4, we used their corrected count rates that were derived after applying corrections for flat field and saturation effects. We converted these count rates to flux densities using the unit conversion (UC)

Figure 5. The m ¼ 2 order FUV Grating1 and Grating2 effective areas, the best-fitting polynomials and the residuals.

Table 4. The coefficients of the best-fitting polynomials to the effective areas of UVIT gratings. FUV-G1 Coefficient c0 c1 c2 c3 c4 c5 c6 c7 k0

FUV-G2

m ¼ 2

m ¼ 1

m ¼ 2

m ¼ 1

53940.6148 212:0132 0.3446 0:0003 1:4266  107 3:6369  1011 3:8411  1015 – 0.0

522747:0992 2424.3078 4:8004 0.0053 3:4478  106 1:3509  109 2:9306  1013 2:7159  1017 0.0

15052.4039 49:4344 0.0645 4:1806  105 1:3476  108 1:7298  1012 – – 0.0

1:2976 0.0418 0:00036 1:3620  106 2:2112  109 1:2713  1012 – – 1250.0

NUV-G m ¼ 1 2.3671 0:0362 0.0011 5:1848  106 1:08550  108 1:19142  1011 6:5940  1015 1:4486  1018 1900.0

J. Astrophys. Astr. (2021)42:49

Page 7 of 10

factors which we calculated from the ZP magnitudes listed in Table 3 of Tandon et al. (2020). We derived the UC factors from the ZP magnitude as follows (Tandon et al. 2017), 0:4

UC ¼ 10

ðZPþ2:407Þ k2mean

resolved by the FUV gratings. To further check, we calculated the expected count rates of HZ 4 using Equation (2) based on the effective areas from (Tandon et al. 2020) and the FUV-G1 and NUV-G spectra. The predicted count rates are listed in Table 5. It is clear that the observed and predicted count rates agree very well. This shows the importance of using effective areas as discussed below. Based on the above analysis, we recommend a cross-dispersion width of 50 pixels for spectral extraction of point sources. In case of poor tracking correction and extended sources, a larger cross-dispersion width should be used.

ð5Þ

;

˚ 1 and kmean is in A. ˚ where UC is in ergs cm2 s1 A 1 ˚ is The flux density fk in ergs cm2 s1 A fk ¼ CPS  UC;

49

ð6Þ

where CPS is the count rate in counts s1 . In Table 5, we list the net count rate from (Tandon et al. 2020), the UC factors and flux density calculated using Equations (5) and (6) for HZ 4. In order to check possible calibration differences between the flux densities derived from the gratings and broadband filters, we plot FUV/NUV grating spectra and flux densities derived for different filters in Fig. 6. It is clear that, except for the F148W (CaF21) filter, the flux measurements agree very well between the gratings and the broadband filters. The F148W flux density is  25% lower compared to that measured with the FUV gratings. The discrepancy is at a level of 2:4r. In comparison to the standard spectrum (hz4_stis_005.fits) used for the calibration of gratings, the F148W flux density is also lower by  21%. This apparent discrepancy is most likely due to the presence of sharp spectral features in the standard spectrum near the mean wavelength of the F148W filter. The spectral features are not well

7. Grating spectral responses and multiwavelength spectroscopy Working with data from different instruments onboard AstroSat require tools and techniques to faciliate joint analysis of multi-wavelength data. In particular, the broadband spectral coverage of AstroSat, from near UV to hard X-rays, requires tools for simultaneous fitting of spectral models to the multi-wavelength data. Due to the complex interactions when X-rays go through the detector material, the response function of X-ray detectors are generally complex, and the X-ray spectral data cannot be directly converted to the source spectrum. X-ray spectral analysis begins with an assumed source spectral model which is then folded with the instrument response that results in model spectral data which is then compared with the

Table 5. The unit conversion factor (UC) for UVIT filters, and the observed count rate, flux density and predicted count rate for HZ 4 in different filters.

Filter F148W F154W F169M F172M N242W N219M N245M N263M N279N

CaF2-1 BaF2 Sapphire Silica Silica-1 NUVB15 NUVB13 NUVB4 NUVN2

HZ 4

kmean ˚ (A)

˚ ) (ergs cm2 s1 A

Obs. CPSa

fk b

Pred. CPSc

1481 1541 1608 1717 2418 2196 2447 2632 2792

(2:866  0:026)1015 (3:574  0:033)1015 (4:57  0:0427)1015 (1:143  0:021)1014 (2:3179  0:0043)1016 (4:924  0:091)1015 (7:571  0:035)1016 (8:674  0:080)1016 (3:793  0:035)1015

23.5 20.7 16.2 5.5 127.8 7.4 37.0 27.2 5.4

6:741  0:062 7:392  0:068 7:397  0:068 6:24  0:11 2:9622  0:0055 3:624  0:067 2:799  0:013 2:356  0:022 2:037  0:019

21:5  1:3 19:1  1:4 15:5  1:2 5:7  0:8 131:6  1:2 7:5  0:1 37:8  0:3 27:6  0:4 5:5  0:1

UC 1

˚ 1 . Observed count rate in counts s1 . Errors on the observed CPS are less than 2%. b In unit of 1014 ergs cm2 s1 A c Predicted count rate using the effective areas provided in (Tandon et al., 2020) and the FUV-G1 and NUV-G spectra of HZ 4 shown in Fig. 6. a

49

Page 8 of 10

J. Astrophys. Astr. (2021)42:49

Figure 6. A comparison of flux densities at various wavelengths derived from the FUV/NUV broadband filters and grating observations of HZ 4.

observed spectral data. The source spectrum is thus inferred from the best-fitting model. Grating spectrometers, such as the UVIT gratings, have much simpler response functions, and one generates a fluxed spectrum directly from the observations using the dispersion relation and the wavelength-dependent effective area or sensitivity curve as described in the previous section. The source spectral properties are inferred from the fluxed spectrum after correcting for the instrumental spectral resolution. It is possible to treat the photon counting data from UVIT similar to the X-ray data that are also photon counting by generating appropriate spectral responses in the form of a redistribution matrix (RMF) and ancillary response file (ARF), and use Equation (7) below which is similar to Equation (2) to infer the source spectrum. One can then perform joint spectral analysis of UVIT and X-ray data from SXT, LAXPC and CZTI. In the following, we generate the count spectrum (i.e., distribution of photon counts in different spectral channels), redistribution matrix and ancillary response for UVIT gratings and broadband filters. Such an approach is more accurate than directly converting the count rates to flux densities with unit conversion factors and zero point magnitudes for various filters as the latter quantities depend on the spectral shape. The unit conversion factors and zero point magnitudes are derived for the particular spectral shape of the spectrophotometric standard HZ 4 and are unlikely to be strictly correct for objects with different spectral shape. For the UVIT gratings, we constructed a count spectrum (counts vs spectral channels) from the uncalibrated 1d spectrum (counts vs. relative pixel coordinates) in the same format as the X-ray spectral PHA files. For this purpose, we added a positive integer number to the relative pixel coordinate and

converted to spectral channels (I) which start from 1. This count spectrum is related to the source spectrum in a way similar to Equation (2), Z ð7Þ DðIÞ ¼ T RðI; EÞAðEÞNE dE þ BðIÞ; where D(I) is the source ? background count in spectral channel I, B(I) is the background counts in channel I, T is exposure time. R(I, E) is the redistribution matrix and A(E) is effective area at energy E, NE is the photon spectrum of the source. The redistribution matrix R(I, E) represents the spectral response which is Gaussian with an FWHM that is the same as the spectral resolution of the grating spectrometers. To measure the spectral resolution of UVIT gratings, we fitted Gaussian profiles to the emission lines observed from NGC 40 and derived ˚ (FUV gratings, m ¼ 2), 33 A ˚ the FWHM ¼ 14:63 A (NUV-G, m ¼ 1). In order to generate the response matrix for the NUV-G, we divided the wavelength range in 2341 energy bins. For each energy bin, we generated the Gaussian response in 210 channels which were defined based on the dispersion solution. We also multiplied the effective area with the redistribution matrix as in Equation (7) and created the grating response matrix which is the product R(I, E)A(E). The responses thus created for each calibrated grating order are compatible with the X-ray spectral fitting packages such as XSPEC (Arnaud 1996), Sherpa (Freeman et al. 2001) and ISIS (Houck & Denicola 2000). The source and background PHA files along with the spectral response can directly be used in one of the above spectral fitting packages. This is helpful for joint UV/X-ray spectral modeling. We also created single channel response matrices for the broadband filters using the updated effective areas provided in Tandon et al. (2020). These response files

J. Astrophys. Astr. (2021)42:49

Page 9 of 10

49

Figure 7. Comparison of count PHA spectrum of Fairall 9 with instrument response and fluxed PHA spectrum with diagonal response. Left panels: Count PHA data (black open circles), flux PHA data (red filled circles), the same power-law fitted to the both datasets and the data-to-model ratios. Right panel: Unfolded spectra using a powerlaw of photon index, C ¼ 0 (NE ¼ AEC ) and unit norm at 1 keV. Both the count PHA and fluxed spectra were extracted from the 2 order of FUV-G2 image.

along with the grating or filter spectral PHA data in the fits format give complete flexibility to treat and analyze the photon counting UVIT data in a way similar to the X-ray photon counting data. Another way to use the UV grating spectra and photometric flux in the joint UV/X-ray spectral analysis is to convert the fluxed spectra in PHA format and generate diagonal responses using the FTOOLS package. However, this has the disadvantage of not considering the actual spectral response of the instrument. Hence, the uncertainty associated with the UC factor for the particular spectral shape of the standard star for a given broadband filter will enter into the spectral fitting. In the case of grating spectra, the derived spectral line widths will not be free of instrumental resolution as the instrument spectral response is not being used. We demonstrate the equivalence of these two approaches as follows. For this purpose we used UVIT grating observations of a Seyfert 1 AGN Fairall 9 (ObsID: G06_157T01_9000000). We processed the Level1 data in the same way as in the case of NGC 40 and HZ 4. We generated an m ¼ 2 order FUV-G2 spectrum of Fairall 9 in the PHA format. We also generated the associate background spectrum from the source-free regions as described earlier, and updated the PHA header of the source spectral file to include the grating response and the background spectral file. We loaded these spectral data into XSPEC and plotted the spectral data as shown in Fig. 7. We also generated a fluxed spectrum i.e., fk vs. k of Fairall 9 for the FUV-G2 in the m ¼ 2 order and converted these data to PHA file and diagonal response matrix using

FTOOLS. The two spectra of Fairall 9 are compared in Fig. 7. It is clear that the two approaches agree well.

8. Summary and conclusion We calibrated the two FUV gratings in the orders 2 and 1 and one NUV grating in the 1 order. We derived dispersion solutions and effective areas for the grating orders which can be used for spectral calibration of any source observed with the UVIT gratings. We also checked the cross-calibration between the gratings and broadband filters and found excellent agreement in flux measurements for all broadband filters except the FUV filter F148W. We provide the updated UC and ZP for this filter. We also generated the spectral response files for the gratings and the broadband filters that can be directly used for spectral analysis using XSPEC, Sherpa or ISIS. A software package UVITTools for the analysis of UVIT data processed with the CCDLAB pipeline has been developed.

Acknowledgements The author is grateful to Shyam Tandon for numerous discussions on various aspects of UVIT, and allowing to use the calibration data. The author is thankful to Phil Charles and Shyam Tandon for their suggestions on the submitted version of the manuscript. This publication uses the data from the AstroSat mission of

49

Page 10 of 10

the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). This publication uses UVIT data processed by the payload operations centre at IIA. The UVIT is built in collaboration between IIA, IUCAA, TIFR, ISRO and CSA.

References Agrawal P. C. 2006, Advances in Space Research, 38, 2989 Arnaud K. A. 1996, in Jacoby G. H., Barnes J., eds, Astronomical Society of the Pacific Conference Series, Vol. 101, Astronomical Data Analysis Software and Systems V, 17 Bezanson J., Karpinski S., Shah V. B., Edelman A. 2012, arXiv e-prints, arXiv:1209.5145 Bohlin R. C., Harris A. W., Holm A. V., Gry C. 1990, ApJS, 73, 413 Feibelman W. A. 1999, ApJ, 514, 296

J. Astrophys. Astr. (2021)42:49 Freeman P., Doe S., Siemiginowska A. 2001, in Starck J.L., Murtagh F. D., eds, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 4477, Astronomical Data Analysis, 76–87 Houck J. C., Denicola L. A. 2000, in Manset N., Veillet C., Crabtree D., eds, Astronomical Society of the Pacific Conference Series, Vol. 216, Astronomical Data Analysis Software and Systems IX, 591 Postma J. E., Leahy D. 2017, PASP, 129, 115002 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 9144, Proc. SPIE, 91441S Subramaniam A., Tandon S. N., Hutchings J. et al. 2016, in den Herder J.-W. A., Takahashi T., Bautz M., eds, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 9905, Space Telescopes and Instrumentation 2016: Ultraviolet to Gamma Ray, 99051F Tandon S. N., Subramaniam A., Girish V. et al. 2017, The Astronomical Journal, 154, 128 Tandon S. N., Postma J., Joseph P. et al. 2020, AJ, 159, 158

 Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:70 https://doi.org/10.1007/s12036-021-09765-9

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

PAYLOAD CALIBRATION

Calibration of Scanning Sky Monitor (SSM) onboard AstroSat ABHILASH R. SARWADE1,*, M. C. RAMADEVI1, B. T. RAVISHANKAR1,

BRAJPAL SINGH1, BLESSY ELIZABETH BABY1, DIPANKAR BHATTACHARYA2 and S. SEETHA3 1

Space Astronomy Group, ISITE Campus, U.R. Rao Satellite Centre, Outer Ring Road, Marathahalli, Bengaluru 560 037, India. 2 Inter-University Centre for Astronomy and Astrophysics, Ganeshkhind, Pune 411 007, India. 3 Department of Astronomy and Astrophysics, Raman Research Institute, Sadashivanagar, Bengaluru 560 080, India. *Corresponding Author. E-mail: [email protected] MS received 7 November 2020; accepted 22 April 2021 Abstract. SSM onboard AstroSat is designed to monitor X-ray sky in the energy range 2.5–10 keV to detect and locate X-ray sources in outburst. SSM with its three almost identical 1D-proportional counters mounted on a rotating platform, scans the sky in step and stare mode of operation. It observes the X-ray sky and generates light curves for X-ray sources detected. Here, we discuss the positional calibration to carry out imaging with SSM. Onboard calibration of SSM has been carried out with Crab, the standard X-ray source. SSM observations of Crab are compared with that of MAXI on ISS for cross calibration of the instrument. Keywords. Proportional counter—position sensitive X-ray detector—sky monitor—X-ray sky—codedmask—X-ray transient.

1. Introduction Scanning Sky Monitor (SSM) is a wide field soft X-ray imaging payload on AstroSat. SSM is one of the five payloads onboard AstroSat (Agrawal 2006; Singh & Tandon 2014; Agrawal 2017). SSM with its large Field of View (FoV) scans the X-ray sky continuously in step and stare mode of operation to detect and locate transient X-ray sources. The primary objective of SSM is to monitor the X-ray sky for transient activity in the X-ray in the energy range 2.5–10 keV, both in known sources and new ones. SSM consists of three coded mask cameras, mounted on a rotating platform, so as to enable near complete coverage of the sky. The platform rotates in step and stare mode from 5 to 355 and back. Every stare is for a duration of 10 minutes, by default. After

This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

each stare, the platform steps by 10 to change the FoV of the SSM cameras. The mounting of the three units of SSM cameras is shown in figure 1. The entire assembly rotates about the ?Yaw axis of the spacecraft. A detailed description of the instrument can be found in Seetha et al. (2006) and Ramadevi et al. (2017).

2. Imaging with SSM The coded-mask along with the position-sensitive anodes in the detectors in SSM acts as the imaging element in the instrument. When X-rays enter through the coded mask plate which contains random pattern of slits, shadows of the mask are cast on the positionsensitive anode wires. The position and intensity of these shadows are a function of the incidence angle of the incoming X-rays relative to the mask-detector axis. This information is used in deriving the position of the X-ray source. The position of interaction of the incident X-ray photon in a position sensitive anode

70

Page 2 of 8

J. Astrophys. Astr.

(2021) 42:70

Crab, the standard X-ray source, with SSM at different locations in its FoV.

4. Mathematical model of calibration function

Figure 1. SSM instrument with three cameras mounted on a platform.

wire in the proportional counter is inferred from the ratio of the charge output measured on both ends of a resistive anode wire. The charge ratio of every photon incident on the detector plane is used to obtain the position of incidence on the anode wire in the detector plane. The detector plane comprises eight anodes in each SSM camera. The position of incidence of all the photons is binned in order to get the position histogram on the detector plane which is defined as the Detector Plane Histogram (DPH). The DPH is in principle the shadow of the mask pattern which is illuminated for a particular angle of incidence of the X-ray source in the FoV. In order to derive the position of every photon incident on the detector plane it is required to obtain the relation between the charge ratio and the geometric position along the anode wire in the detector.

3. Position calibration In order to carry out imaging with SSM, the detectors are calibrated for positional response. Position calibration is undertaken to determine the relation between the charge ratio of the photon event incident and the position of its interaction in the detector plane, which is called the calibration function. Each anode is independent from the other and there are eight anodes in each SSM unit for which this calibration has to be done. Positional calibration of SSM detectors involves derivation of the calibration function for each of the eight anode wires in all three SSM units. Imaging is carried out using the shadow generated by binning the position of incident events to get the DPH. Details of ground calibration of SSM can be found in Ramadevi et al. (2011). Onboard calibration involves observing

Figure 2 describes schematic diagram of a single anode wire. The length of the anode wire that is exposed to incident X-rays is 60 mm. However, the electrical length of the anode wires may not be 60 mm. This is because the anode wires are glued, using conductive silver epoxy, to conductive high voltage grooves placed inside the insulator (Kelvin-F) which is fixed into the wire-module of the detector. The glue is applied from one end of the groove, and the position where the wire makes electrical contact may be at the inner edge of the groove (at the geometric end point of wire) or anywhere within the length of the groove. Due to this reason, the start and end points of the resistive anode wire which defines the length, varies from wire to wire. Thus, it is not possible to get the length of the anode wires from mechanical dimensions of the detector. Therefore, it is necessary to know the electrical length of the anode wires (which is the length seen by the charge cloud) to derive the position of photons incident on the detector. We define x ¼ 0 as the mid-point (geometric mid point) of the exposed anode wire as shown in Fig. 2 and x ¼ lL and x ¼ lR as left end and right end of anode wire respectively, which extend till the conductive point inside the Kel-F grooves on either ends of the wire, in the wire module. Note that lL and lR are not necessarily same. There is a calibration wire of diameter 2 mm which is placed above the window of the detector across the anode wires at the geometric centre of the anode wires so that this casts a shadow on the anode wire for incident X-rays, which is used as a reference for geometric centre of the anode wires in the detector plane. When a X-ray photon is photoelectrically absorbed, a charge cloud is created due to the movement of the electrons towards the high voltage applied on the anode wire. The charge cloud is detected on the anode wire at position, say x ¼ x0 on the anode wire, the total collected charge Q is divided on left and right as lR þ x0 ; QL ¼ Q  lL þ lR lL  x0 : QR ¼ Q  lL þ lR As shown in Fig. 2, ends of anode wire are connected to Charge Sensitive Pre-Amplifiers (CSPAs) followed

J. Astrophys. Astr.

(2021) 42:70

Page 3 of 8

• • • • • Figure 2. Schematic diagram of an anode wire.

by post-amplifiers with a finite gain given by amplification factors AL and AR . Therefore, the signals detected at the ends of anode wire are given by lR þ x0 ; lL þ lR lL  x0 : VR ¼AR  QR ¼ AR  Q  lL þ lR VL ¼AL  QL ¼ AL  Q 

R The voltage ratio R ¼ VVLL V þVR can be used as indicative of the position of the X-ray interaction. Simplifying, ratio can be expressed as

R¼

AL ðlR þ x0 Þ  AR ðlL  x0 Þ : AL ðlR þ x0 Þ þ AR ðlL  x0 Þ

Inverting the above equation to express position of interaction of X-ray photon in terms of ratio, we get x0 ¼

RðAR lL þ AL lR Þ þ ðAR lL  AL lR Þ : RðAR  AL Þ þ ðAR þ AL Þ

This is calibration function with lL , lR and AL , AR as unknown calibration constants for each anode wire.

5. Estimation of calibration constants 5.1 Calibration data Onboard calibration of SSM cameras are carried out by observing Crab, the standard X-ray source. In order to estimate the calibration constants, SSM camera was pointed towards Crab with the X-ray source positioned at about the center of camera’s FoV (location of Crab in the FoV of SSM: ðhX ; hy Þ = (0.62 ,0.52 ). The source, Crab was chosen due to its relatively stable intensity in the SSM’s energy range and sufficiently good count rate along with its position in relatively clean portion of the X-ray sky with no other bright sources nearby. The calibration data was acquired for an exposure time of about 15 ks. The data was filtered based on following conditions to generate Good Time Intervals (GTIs): • Crab not occulted by Earth

70

Satellite not in SAA Earth not in the FoV Energy between 2.5 keV to 10 keV Lower particle background (see Section 7) Absence of any attitude variation

Figure 3 shows the data selection according to the GTI generated with the above mentioned conditions for one orbit. Similarly data is filtered for all the orbits as per the GTIs. The effective exposure time, after GTI filtration, is about 10 ks. This data is used to generate the observed DPH for a given source location.

5.2 Monte Carlo simulations For a given source location in the FoV, it is required to simulate the expected DPH for calibrating it with the observed DPH. In order to simulate the shadows registered by the SSM cameras, Monte Carlo methods are employed. All the measured geometric parameters of the cameras are considered to make their model. Among the three SSM cameras, the two edge cameras are physically identical and the central camera has different dimensions. The physical dimensions considered in the model include: (a) the mask plate dimensions including inter-pattern gaps, plate thickness, and the six different patterns, (b) the detector module with individual anode placements in respective wire-cells, (c) the window and the window support rods, (d) calibration wire, and so on. The six mask plates considered for the cameras, joined sideways, each consist of 63 mask elements of open and closed elements which are represented in the model respectively as 1’s and 0’s. In order to simulate the photon strike over the mask plate, random number generation has been employed based on Park Miller Minimal standard algorithm with shuffling (Press et al. 1992). If the random position on the mask plate falls in an open element, for the given incidence angles, hx and hy , it is checked which wire cell it can strike. This is registered only if the photon’s trajectory does not hit any of the structural elements of the camera, either in the mask plate (such as, within the plate thickness or the ribs), window support rods or the calibration wire. For incidences away from the normal, slant-ray effects affecting the photon trajectory are also considered along both x and y directions. Also considered are the quantum efficiency effects which in turn depend on energy-band specific mean-free path value and

70

Page 4 of 8

J. Astrophys. Astr.

(2021) 42:70

Figure 3. Good Time Interval (GTI) selection.

Figure 4. Simulated Detector Plane Histogram (DPH) with each anode wire of 60 mm length divided into 63 bins along the anode wire.

attenuation due to the aluminised Mylar window. After applying all these corrections on the histogram of the detector strike positions of the events, the uncertainty in reading out is simulated by introducing a Gaussian smear of position resolution r varying along the length of the anode wires by  0.5–1.2 mm. The DPH (see Fig. 4) thus generated is used for further modelling. For the better visualization of DPH, each anode’s histogram is plotted side by side (see Fig. 5). The simulated DPH for anode A0 is shown in Fig. 6 for better visualization of shadows incident on anode wires.

5.3 Derivation of calibration constants The observed data is used to generate the ratio histogram of the charge ratios of all the events incident on the respective anode wires. This gives the shadow pattern in ratio domain and the simulated

DPH gives the shadow pattern in position domain. For a given source location, the two patterns are matched to get the ratio to position map for all the eight anode wires. The maximas and minimas of observed ratio DPH are matched with respective maximas and minimas of the simulated position DPH as shown in Fig. 7. This process is carried out independently for each anode wire of all SSM cameras. For the purpose of illustration, we demonstrate the calibration process for one of the anodes, A0 of SSM1. The same procedure is followed for all other anodes in SSM. The simulated DPH for anode A0 is shown in Fig. 6. The calibration constants lL , lR and AL , AR for every anode is derived by obtaining the calibration function using the ratio and position values. Therefore, given the pairs of ratio and position, the calibration equation is fit to estimate the calibration constants. These calibration constants are used to convert the charge ratio of every incident photon on

J. Astrophys. Astr.

(2021) 42:70

Page 5 of 8

70

Figure 5. Simulated Detector Plane Histogram (DPH) of all eight anode wires (named A0 to A7) with each anode’s histogram plotted side by side for better visualization.

Figure 6. Simulated Detector Plane Histogram (DPH) for anode A0.

Figure 7. Comparison of observed calibration data with simulated data (A0).

70

Page 6 of 8

J. Astrophys. Astr.

(2021) 42:70

Figure 8. Comparison between simulated DPH and observed DPH. The observed DPH is obtained using the estimated calibration constants.

the detector plane to position. Position of the events in the respective anodes are binned to get the observed DPH which is then compared with that of the simulated DPH as shown in Fig. 8.

6. Anode response Anodes in SSM are position sensitive resistive wires and the relative detection efficiency from one position bin to the other along the anode wire are not the same. This non-uniform variation along the anode wire is called the anode response and is one of the aspects of the calibration database for imaging with SSM. The observed DPH has the anode response already included, as it is part of the instrument response and this has to be appropriately handled while fitting the observed with the simulated DPH in imaging. The anode response is estimated with uniform illumination of the source along the anode wires for all SSM cameras during ground calibration.

6.1 Anode response from ground calibration Each SSM unit is illuminated with an X-ray source so that the detector plane with eight anodes is uniformly illuminated. The anode responses are derived from this after removal of the source illumination profile as the X-ray source used is a diverging source. The nonuniform anode response along the anode wires are modelled using the ground calibration data with uniform illumination of SSM detectors. The DPH modelled as anode response as shown in Fig. 9.

7. Background modelling The detector background in SSM has two components: X-ray sky background and charged particles. The veto layer inside the SSM detector module is an additional layer of anode wire beneath the main anode wires. This layer is meant to reject charged particles that are incident via the anti-coincidence technique.

Figure 9. Ground Data DPH used to derive anode response.

J. Astrophys. Astr.

(2021) 42:70

Page 7 of 8

70

Figure 10. Background count rate as a function of veto count rate for anode A0 of SSM1.

Figure 11. Background template generated based on observed VCR during calibration observation for anode A0.

The rate of events detected in this layer (called the Veto Count Rate (VCR)) are largely counted as charged-particle background. During the ingress and egress of SAA region, the contribution from charged particles is rather high which is indicated by increase in VCR in SSM. The charged particle counts detected in SSM is modelled with veto count rates detected in every stare. While estimating calibration constants from the Crab calibration data, the X-ray and particle background contribution must also to be taken into account. In order to model the background contribution in the DPH, data from faint field observations were aggregated. A faint field is regarded as one in which the total flux incident on the camera is less than 0.5 photons/cm2 /s so that, it can be considered that all the events registered during these faint field observations are due to background. The background count rate in the field of view is related to that of the veto count rate as shown in Fig. 10.

Background templates are generated using the VCR for every stare and used in the estimation of calibration constants as discussed in this paper. These background templates generated are also being used for imaging. Background template for the calibration data is shown in Fig. 11. Both the anode response and the background model are used to generate the observed DPH and compared with that of the simulated DPH as shown in Fig. 12.

8. Conclusion Different aspects of position calibration of SSM are discussed in the previous sections. The calibration database is generated for all three SSM cameras and are used in imaging with SSM. Calibration observations of Crab with SSM have been extensively studied with respect to various aspects of image processing. The calibration database obtained as per the procedure

70

Page 8 of 8

J. Astrophys. Astr.

(2021) 42:70

Figure 12. Comparison between simulated DPH and observed DPH. The observed DPH is obtained using estimated calibration constants, multiplying with anode response and subtracting background model.

Figure 13. Crab light curve compared with that of MAXI.

discussed here, along with the background modelling, are applied to the data to get the Crab flux consistent to a level of ?/–15% as shown in Fig. 13. The results are compared with that of MAXI onboard ISS as part of cross calibration and are found consistent. SSM data for all other sources observed till date are being reprocessed and the database of the lightcurve for these sources will be updated in the SSM website hosted at ISSDC. A more detailed investigation is in progress and the results will be reported in due course.

Acknowledgements The authors acknowledge Director, URSC and DD, PDMSA and GH, SAG for the constant support and also acknowledge various entities at ISRO who have

contributed their part towards realization of all aspects of the instrument.

References Agrawal P. C. 2006, Adv. Space Res., 38(12), 2989 Agrawal P. C. 2017, J. Astrophys. Astr., 38, 27 Press W. H., Teukolsky S. A., Vetterling W. T., Flannery B. P. 1992, Numerical Recipes in C: The Art of Scientific Computing, 2nd edition, Cambridge University Press. Ramadevi M. C., Ravishankar B. T., Seetha S. 2011, Exp. Astron., 31(2–3), 99 Ramadevi M. C., Seetha S., Bhattacharya D. et al. 2017, Exp. Astron., https://doi.org/10.1007/s10686-017-9536-3 Seetha S., Ramadevi M. C., Babu V. C. et al. 2006, Adv. Space Res., 38, 2995 Singh K. P., Tandon S. N. et al. 2014, Proc. of SPIE, 91442T, 9144-100-100, https://doi.org/10.1117/12.2062667

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:76 https://doi.org/10.1007/s12036-021-09762-y

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

PAYLOAD CALIBRATION

Imaging calibration of AstroSat Cadmium Zinc Telluride Imager (CZTI) AJAY VIBHUTE1,2,* , DIPANKAR BHATTACHARYA2, N. P. S. MITHUN3,

V. BHALERAO4, A. R. RAO5 and S. V. VADAWALE3 1

Savitribai Phule Pune University, Pune 411 007, India. Inter-University Centre for Astronomy and Astrophysics (IUCAA), Post Bag 4, Ganeshkhind, Pune 411 007, India. 3 Physical Research Laboratory, Ahmedabad 380 009, India. 4 Indian Institute of Technology Bombay, Mumbai 400 076, India. 5 Tata Institute of Fundamental Research, Homi Bhabha Road, Mumbai 400 005, India. *Corresponding author. E-mail: [email protected] 2

MS received 2 November 2020; accepted 26 April 2021 Abstract. AstroSat is India’s first space-based astronomical observatory, launched on September 28, 2015. One of the payloads aboard AstroSat is the Cadmium Zinc Telluride Imager (CZTI), operating at hard X-rays. CZTI employs a two-dimensional coded aperture mask for the purpose of imaging. In this paper, we discuss various image reconstruction algorithms adopted for the test and calibration of the imaging capability of CZTI and present results from CZTI on-ground as well as in-orbit image calibration. Keywords. Coded mask imaging—X-ray—AstroSat—CZT imager.

1. Introduction

1.1 Cadmium zinc telluride imager

AstroSat (Singh et al. 2014), India’s first dedicated space astronomy observatory, covers a broad energy band including optical, Ultra-Violet (UV) and soft to hard X-rays, using four different co-pointed payloads, namely Ultra-Violet Imaging Telescope (UVIT; Tandon et al. 2017), Soft X-ray Telescope (SXT; Singh et al. 2017), Large Area X-ray Proportional Counters (LAXPC; Yadav et al. 2016), and Cadmium Zinc Telluride Imager (CZTI; Vadawale et al. 2016). In addition, a Scanning Sky Monitor (SSM; Ramadevi et al. 2017) is mounted perpendicular to other instruments and operates independently to search for X-ray transients. The SXT carries out X-ray imaging of the sky in the soft X-ray band (0.3–8 keV) using reflecting optics and CZTI in the hard X-ray band (20–100 keV) using a Coded Mask. In this paper, Section 1.1 gives an introduction to the AstroSat CZT Imager, ion and Section 2 presents results from the on-ground and in-flight calibration.

Coded Mask imaging (Skinner 1984) with the CZTI is designed to operate in the energy range 20–100 keV. CZTI is a ’box-type’ or ’simple’ coded mask telescope where the size of the coded mask and the detector are the same. The full CZTI detector is illuminated only by the on-axis source. Off-axis sources illuminate only a part of the detector depending on the position of the source, this is called ‘‘partial coding’’. The coded mask for CZTI is designed with seven patterns based on 255-element pseudo-noise Hadamard set Uniformly Redundant Arrays (URA) (Caroli et al. 1987). Each pattern has 16  16 mask elements and used as a mask for an individual detector module. A random arrangement of these patterns into a 4  4 array results in the mask pattern for the first quadrant (quadrant A). The coded masks for the other quadrants (B, C, and D) were obtained by rotating the mask pattern of quadrant A by 90 , 180 and 270 respectively. Figure 2 shows the mask pattern for all four quadrants of CZTI. In the figure, closed mask elements are represented in black and the open ones in white.

This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

76

Page 2 of 16

Figure 1. Assembly of the CZTI.

J. Astrophys. Astr.

(2021) 42:76

except at edge rows which are 2.31 mm wide. Thus, the edge pixels are of size 2:31  2:46 mm, and the corner pixels are of size 2:31  2:31 mm. The CZT detector is sensitive up to 500 keV, and CZTI electronics are tuned to discriminate photons up to 250 keV. A passive collimator wall of height 400 mm separates any two adjacent modules, restricting the view of each detector module to the coded mask directly above it. The collimator thus restricts the Field-ofView (FoV) of CZTI to 4:6  4:6 FWHM at energies below 100 keV. For energies above 100 keV, the collimator walls and the coded mask become progressively transparent, and allows the detection of Gamma-Ray Bursts from all over the sky (Rao et al. 2016). Table 1 lists the main characteristics of CZTI. A detailed description of CZTI and its science goals may be found in Bhalerao et al. (2017). CZTI records the distribution of counts as a function of position on the detector. The observed count distribution does not correspond to a sky image. The latter is obtained by subjecting the observed pattern to a image reconstruction procedure.

2. Imaging calibration of CZTI

Figure 2. CZTI coded aperture mask for all four quadrants designed using 255 element pseudo-noise Hadamard set Uniformly Redundant Arrays. One extra closed element is added to each URA to obtain a square pattern. Here black areas represent closed mask elements and white areas represent open ones.

CZTI has a total geometric area of 976 cm2 achieved using 64 detector modules, divided across four independent quadrants. Each CZT detector module is of size 39:06  39:06  5 mm and contains 256 pixels arranged in a 16  16 array. An individual pixel in the array is of size 2.46 mm2.46 mm,

Like every space astronomy payload, the CZTI flight model underwent a phase of calibration on ground to characterise the pixel behaviour, effective area, imaging and spectral response, etc. The ground-based calibration was carried out by exposing the instrument to various radioactive sources placed at a finite distance from the detector. After launch, the first six months were devoted to carry out in-flight calibration using observations of astronomical sources. CZTI image reconstruction is a two-step process, the first step is to acquire a spatially coded detector data, and the second step involves reconstruction of the observed image by decoding the collected data. The reconstruction is computationally expensive and performed offline, i.e., on the ground. A variety of image reconstruction algorithms are available, from which a choice is made based on the available computing resources, degree of crowding of sources in the FoV, nature of the background and the required accuracy. CZTI is a photon-counting detector and records the time of arrival, energy and position on the detector of each detected photon. CZTI image reconstruction starts by creating a twodimensional map of total counts recorded in each

J. Astrophys. Astr.

(2021) 42:76

Page 3 of 16

76

Table 1. The important characteristics of CZT imager. Detector Pixels Pixel size Imaging method Field of View Angular resolution Energy resolution Energy range Sensitivity

Cadmium–Zinc–Telluride 16,384 (64 modules of 256 pixels each) 2.46  2.46 mm Coded aperture mask 4.6  4.6  8 arc min (18 arc min geometric)  8% @ 100 keV 20–100 keV for collimated FoV (20–380 keV for all sky) 0.5 mCrab (5 sigma; 104 s)

pixel, called a Detector Plane Histogram (DPH). In order to address the impact of inter-pixel sensitivity variation within the detector module on the imaging process, the relative quantum efficiency-area product g of pixels was estimated by shining a radioactive source on the detector from under the coded mask. The array of measured counts per unit area Ci at the pixels was then subjected to a linear fit to the quantity ðh=di Þ2 , where h was the height of the source above the detector plane and di the distance to the source from the centre of pixel i. The fitted relation was used to predict the expected count in each pixel, and the ratio of the observed to the expected count was registered as a measure of g. We accumulated a sufficient number of counts in each pixel to keep Poisson noise well below the RMS count variation seen between pixels. The process was repeated with the radioactive source placed at different locations under the mask. The derived g distribution was very similar at all these instances, and an average of the multiple determinations was used to normalised the DPH, and the normalised count map is called Detector Plane Image (DPI). The relative quantum efficiency (QE) of the pixels are available as a part of the CZTI Calibration Database (CALDB)1. The DPI represents a scaled shadow of the mask on the detector plane cast by the sources in the FoV. The pattern of the DPI depends on the position of the source in FoV and the recorded counts in each pixel depend on the intensities of these sources. If the position or intensity of a source changes, so would the total number of photons recorded by each pixel. The DPI is used as an input to the image reconstruction algorithm. We have used three variants of the Cross-Correlation, namely, mask cross-correlation (Skinner et al. 1987; in’t Zand 1992), shadow cross-correlation, balanced cross1

http://astrosat-ssc.iucaa.in/?q=cztiData.

Figure 3. A schematic representation of the assembly made to test the coded mask imaging procedure using CZTI Qualification Module. In this setup, the source was placed 897.5 mm above the detector while the mask is located at a height of 484 mm above the detector.

correlation (Fenimore & Cannon 1978; in’t Zand 1992) and Richardson–Lucy (Richardson 1972; Bhattacharya 2006; Shankar & Bhattacharya 2003) techniques for CZTI imaging calibration. In this paper, we report the results of imaging calibration of CZTI, both on-ground and in-flight, in the sections to follow.

2.1 CZTI on-ground calibration We performed CZTI ground calibration in two phases, in the first phase, the coded mask and the imaging procedure were validated, and in the second phase, we calibrated the fully assembled CZTI to validate the imaging response.

76

Page 4 of 16

J. Astrophys. Astr.

(2021) 42:76

Figure 4. The expected (left panel) and observed (right panel) shadow of the mask on the detector when the radioactive source 241 Am was placed directly above the centre of the detector, in the arrangement shown in Fig. 3. The observed pattern is normalized and dead/noisy pixels are ignored.

Figure 5. Reconstructed image using Richardson–Lucy (a) and cross-correlation (b) methods, when source is at the centre of the detector.

Figure 6. Image profiles of X (a) and Y (b) cross sections of the image reconstructed using Cross-Correlation and Richardson–Lucy algorithm.

J. Astrophys. Astr.

(2021) 42:76

2.1.1 On-ground calibration using the qualification model. One detector module with 16  16 pixels of a Qualification Model (QM) was used first to verify the imaging procedure. A fixture, schematically shown in Fig. 3, was made to perform the ground calibration. In the fixture, the source and the mask plate were placed at the height of 897.5 mm and 484 mm respectively above the detector. Two radioactive sources, Americium-241 (with a line at 59.54 keV) and Cobalt-57 (with lines at 122 keV and 136 keV) were used to validate the coded mask and the imaging procedure. 241 Am had count rate of 400 counts/s and 57 Co had 350 counts/s. We selected several locations on the fixture, including the centre as well as a few offset locations and carried out multiple observations at these locations. The finite distance of the source from the detector causes the shadow to differ from what would be expected from a distant astronomical source. In the laboratory setup built for CZTI ground calibration, one mask element cast its shadow over 2  2 detector pixels. Hence, only a quarter of the mask illuminated the entire detector module, despite the mask and the detector being of the same physical size. For image reconstruction, we computed a library of the expected shadows of the mask on the detector, spanning source positions over ±19.5 mm from the centre in both X and Y directions, in steps of 0.25 mm. We obtained five different data sets by placing the 241 Am source at different locations, including the centre and a few offset positions. One data set was also obtained by simultaneously placing two sources, 241 Am and 57 Co at two different locations on the fixture. In Fig. 4, the left panel shows the expected shadow pattern, and the right panel shows the observed pattern when the source was placed on a normal passing through the centre of the detector. The CZTI electronics recorded the pixel ID, energy and time of each detected photon, creating an event list. The DPH were constructed by binning the events as a function of pixel number and along with the computed shadow library were used as input to the imaging procedure. For image reconstruction during the ground calibration, we used Richardson-Lucy and cross-correlation algorithms. Figure 5 shows the reconstructed image when 241 Am source was placed at the center using Richardson–Lucy and cross-correlation methods, and image profiles of X and Y cross sections of the reconstructed image is shown in Fig. 6. Figure 7 shows the reconstructed image using cross-correlation

Page 5 of 16

76

Figure 7. Reconstructed image using cross-correlation method for the simultaneous observation of two sources where 241 Am source was placed at the offset of 10.25 mm and 57 Co was placed at the offset of ?10.25 mm from the center in X-direction. Sources were of almost equal strength.

Table 2. Position of source in lab and reconstructed positions using cross-correlation and Richardson–Lucy algorithm. Lab location (in mm)

CC (in mm)

RL (in mm)

(0.0, 0.0) (0.0, –10.25) (0.0, 10.25) (–10.25, 0.0) (10.25, 0.0)

(0.7, 0.75) (0.2, –9.75) (0.5, 11.0) (–9.85, 0.5) (11.0, 0.5)

(0.5, 0.4) (0.25, –9.7) (0.5, 11.0) (–9.75, 0.25) (10.65, 0.35)

Table 3. Lab position of the source during the observation and reconstructed source position using cross-correlation and Richardson–Lucy algorithm. Source 241 57

Am Co

Lab location (in mm)

CC (in mm)

RL (in mm)

(–10.25, 0.0) (10.25, 0.0)

(–9.0, 0.0) (10.7, -1.0)

(–9.0, –0.2) (10.5, -1.25)

for the simultaneous observation of two sources where 241 Am source was placed at offset of -10.25 mm and 57 Co was placed at the offset of -10.25 mm from the center in X-direction. We then compared the reconstructed source location from the image with the source position set in the

76

Page 6 of 16

J. Astrophys. Astr.

(2021) 42:76

Table 4. Absolute difference (in mm) between the source location using a smaller number of events and the full event list. Trial No

105 photons

1 2 3 4 5 6 7 8 9 10

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

104 photons

103 photons

0.0 0.25 0.25 0.0 2.85 2.85 0.0 0.25 0.0 0.0

0.0 2.70 0.35 0.0 0.0 0.25 0.25 0.25 0.25 0.35

Figure 8. Laboratory setup at Tata Institute of Fundamental Research (TIFR) used for the CZTI FM calibration.

Figure 9. Reference positions on the fixture for a quadrant.

102 photons 0.55 0.55 0.55 1.03 0.25 0.75 0.50 0.79 0.25 0.25

Figure 10. Reconstructed source profile using the Richardson–Lucy method when the source is placed at the central reference P3 during FM ground calibration.

lab during the acquisition of the data. Table 2 presents the results of this exercise. In the case of two-source observation, we constructed two separate DPHs, one from photons recorded near 60 keV, and another from photons near 122 keV. Table 3 summarises the reconstruction result for the observation with two sources. The results from cross-correlation and Richardson– Lucy methods agreed within 0.5 mm. The source positions set in the lab and the reconstructed positions were consistent well within 1 mm, except for one case where the deviation was 1.25 mm. This discrepancy in the reconstructed position could be due to the inaccuracy in the manual placement of the radioactive source. The location accuracy of 1.0 mm at a source height of 897.5 mm corresponds to an angular accuracy of 3.8 arcmin. In all the above reconstructions,

J. Astrophys. Astr.

(2021) 42:76

Page 7 of 16

76

Table 5. Laboratory source positions set during FM ground calibration and the offsets in reconstructed source positions by cross-correlation and Richardson–Lucy algorithms. Units are millimetres. Sr. No.

Quadrant ID

Lab location

Source

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 2 2 2 3 3 3 3 3 3 3 3 3 3 3 3 3 3

P1 P3 P4 P5 P3 P3?(–10,0) P3?(–20,0) P3?(10,0) P3?(20,0) P3?(0,10) P3?(0,20) P3?(0,–10) P3?(0,–20) P2 P4 P5 P3 P3?(–10,0) P3?(10,0) P3?(0,10) P3?(0,20) P3?(0,–10) P3?(0,–20) P3 P3?(–10,0) P3?(–20,0) P1 P2 P3 P4 P3 P3?(–10,0) P3?(–20,0) P3?(10,0) P3?(20,0) P3?(0,10) P3?(0,20) P3?(0,–10) P3?(0,–20) P1

41 42 43

3 3 3

P2 P4 P5

the total number of source photons used are more than a million and average background count was 350 counts/s. To examine the reliability of the algorithm, image reconstruction was also performed with a smaller subset of photons. Event lists containing 105 ,

CC (in mm)

RL (in mm)

Am Am Am Am Co Co Co Co Co Co Co Co Co Co Co Co Co Co Co Co Co Co Co Co Co Co Am Am Am Am Co Co Co Co Co Co Co Co Co Co

(3.00, 3.00) (3.00, 2.25) (–0.25, 1.75) (–2.00, 4.00) (–2.00, 2.00) (0.50, 2.00) (0.00, 2.25) (0.75, 2.00) (1.25, 2.25) (0.50, 1.00) (2.00, –1.75) (2.00, –3.00) (–1.25, –2.75) (–1.25, –2.00) (–3.00, 1.75) (–2.25, 1.50) (0.25, –0.50) (0.50, 1.25) (0.00, 0.50) (0.25, 0.25) (0.75, -0.50) (0.00, -1.25) (1.0, -1.00) (1.25, 0.75) (1.00, 0.75) (1.00, 1.00) (-0.75, 3.00) (3.00, 2.25) (0.75, 0.75) (2.25, 1.25) (0.00, 1.25) (–1.00, 0.00) (–0.50, 0.00) (0.75, 0.00) (1.00, 0.00) (0.25, –0.75) (1.50, –0.50) (0.25, –1.00) (1.25, –2.00) (–1.00, 2.00)

Co Co Co

(2.25, 2.25) (0.00, 0.00) (1.00, 0.75)

(2.50, 1.50) (2.75, 2.00) (–0.50, 1.75) (–2.00, 4.00) (–1.75, 1.50) (–0.50, 1.50) (–0.50, 2.00) (0.75, 2.00) (1.00, 2.00) (0.50, 1.00) (1.00, –1.75) (0.00, –3.75) (–2.00,–2.75) (–1.25, –1.75) (–2.50, 1.75) (1.25, 1.75) (0.50, –0.25) (0.00, 1.00) (0.25, 0.50) (0.25, 0.0) (0.50, –0.25) (0.25, –1.00) (0.75, –1.25) (2.25, 0.75) (0.75, 0.75) (0.00, 1.25) (–1.25, 1.00) (3.25, 2.25) (0.75, 0.75) (2.50, 1.00) (1.25, 1.00) (–1.00, 0.00) (–1.50, 0.25) (1.00, 0.75) (1.75, 1.25) (0.25, –0.75) (1.50, 0.50) (0.25, –1.00) (1.00, –2.00) (–2.00, -0.50) (0.00, 2.25) (0.00, 0.00) (1.00, 0.75)

104 , 103 and 102 photons were selected randomly from the full event list and imaging was performed using them. We repeated the imaging procedure with ten different random event sets in each case to estimate the variation in reconstruction results due to

76

Page 8 of 16

J. Astrophys. Astr.

(2021) 42:76

counting statistics. Table 4 lists the absolute difference (in mm) between the source location reconstructed using the subset event list and the full event list. The results show that the reconstruction is quite robust, and the RMS deviation is within 1.3 mm (5 arcmin). 2.1.2 On-ground calibration using the flight module. The results from the qualification model confirmed the working of the coded mask and the imaging procedure. The four quadrants of the CZTI Flight Model (FM) were then assembled and calibrated. The ground calibration of the FM was carried out by shining three different radioactive sources on individual quadrants. In addition to 241 Am and 57 Co, a 109 Cd source with lines at 22 keV and 88 keV was used. Figure 8 shows the laboratory setup used for the FM calibration. The perpendicularity of placement of the source with respect to the detector was ensured using a laser beam and a mirror. The radioactive source was positioned at different locations on the quadrant using the fixture. Five reference positions P1, P2, P3, P4, and P5, directly above the intersection of four adjacent detector modules, were selected to place the source. Figure 9 shows the five selected positions. In addition to the five reference positions, the radioactive source was also placed at a few locations, shifted in the X and Y directions from the central reference P3, as shown in Fig. 9. The height of the source plane was 2 m above the detector. The uncertainty in the horizontal and vertical positioning of the source is estimated to be  2.5 mm and  1 cm respectively. The source illumination axis was kept perpendicular to the detector within an accuracy of about 0.14 . A library of shadows was created by computing expected shadow patterns for a range of source positions, spanning 39 mm on either side of each reference position in both directions in steps of 0.25 mm. The acquired datasets were then analyzed using cross-correlation and Richardson–Lucy algorithms. Figure 10 shows the reconstructed source profile using the Richardson–Lucy algorithm for an observation where the radioactive source is located at position P3. Table 5 shows the deviation, in mm, of the reconstructed locations of the sources from the expected ones. In the flight model calibration, the source was at the height of 2000 mm, whereas during the qualification

Figure 11. Deviation of the source positions recovered using the Richardson–Lucy algorithm with respect to the expected values during the FM ground calibration.

model calibration source was at 897 mm above the detector. Hence, in the flight model calibration, we see more deviation than the deviation seen the qualification model calibration. Figure 11 shows a scatter plot of deviations in the reconstructed locations from those expected. In most cases, the deviation is well within 1.5 mm except for a small fraction extending up to 3 mm. The deviation of 3 mm at a height of 2000 mm corresponds to an angular deviation of 5 arcmin, within the design goal of 8 arcmin for the CZTI.

2.2 In-flight calibration The first six months after the launch of AstroSat was reserved for the Performance Verification (PV) and calibration of all the instruments. During this period, several calibration observations were conducted of the Crab nebula, one of which, observation ID 20160331_T01_112T01_9000000406, was used to characterise the post-launch imaging performance. This observation was performed from UT 2016-03-31 05:39:19 to 2016-04-03 03:55:11, adding up to a total of 114 ks exposure time. The observed data was analyzed using the CZTI data analysis pipeline2. Figure 12 shows the reconstructed sky image of the Crab from all four quadrants of CZTI using the mask cross-correlation technique. In the images, the expected location (shown as a yellow circle) of the 2

http://astrosat-ssc.iucaa.in/?q=cztiData.

J. Astrophys. Astr.

(2021) 42:76

Page 9 of 16

76

Figure 12. (a)–(d) Image of the Crab Nebula reconstructed from CZTI data using mask cross-correlation technique for observation id 9000000406. Note that the angular extent of the source is well within the resolution of the CZTI, so the image profile primarily reflects the angular response of CZTI.

source matches the peak of the reconstructed profile for quadrants A and B, while images from quadrants C and D show a deviation of 0.29 in the Y direction. For quadrant B, although the reconstructed source location is correct, there is an extended tail in the X-direction. To understand the discrepancy between the expected and the reconstructed images, various experiments and simulations were performed. For quadrant A, the reconstruction was as expected, so the experiment started with quadrant B, which showed an extended peak. The following possible scenarios were

investigated to understand the cause of the extension and the shifts observed in the reconstructed image. (1) Extension or shift because of measurement error in the distance between the detector and the mask. (2) Extension or shift because of unstable pointing. (3) Extension or shift because of module dependency. (4) Extension or shift because of misalignment between the detector and mask. 2.2.1 Extension or shift because of measurement error in the distance between the detector

76

Page 10 of 16

J. Astrophys. Astr.

(2021) 42:76

Figure 13. Image reconstructed from simulated DPI with mask plate placed at a height 400 mm (a) and 450 mm (b).

Figure 14. (a), (b) Reconstructed images by introducing planar misalignment between the mask and detector.

and the mask. The distance between the coded aperture and the detector is a key factor in the design of a coded mask instrument. A wrong value for this used in the reconstruction process could lead to distortions in the recovered image. In the case of CZTI, the mask plate is placed 481 mm above the detector as per design. To investigate the effects of a

possible error in this value, we simulated DPIs with the mask placed at heights 470 mm, 460 mm, 450 mm and 400 mm above the detector, and subjected them to the same image reconstruction as employed for the real data. This is a much exaggerated range of heights than the error margin that could be expected in reality. Figure 13 shows the images

J. Astrophys. Astr.

(2021) 42:76

Page 11 of 16

76

Figure 15. Reconstructed image at two time bins for Quadrant B.

detector, including any offset and rotations. If the mask and the detector are misaligned, then it can affect the reconstructed image. To examine the effect of a possible planar misalignment between the mask and detector, we simulated DPIs with planar misalignment of 1 , 3 , 5 , 10 and reduced these DPIs using the image reconstruction method. Figure 14 shows the images reconstructed from simulated DPI with the mask plate tilted with 5 , 10 . The reconstructed images show neither an extension nor a shift. Hence, we discarded the possibility of planar misalignment between the mask and the detector as the reason behind the image distortions observed.

Figure 16. Distortions observed in module level imaging.

reconstructed from simulated DPI with the mask plate placed at 450 mm and 400 mm respectively. The reconstructed images show neither an extension nor a shift. Hence, an error in mask height was ruled out as a possible cause of the image distortions observed. 2.2.2 Extension or shift because of planar misalignment between the mask and detector. The image reconstruction depends strongly on the planar alignment between the coded aperture and the

2.2.3 Extension or shift because of unstable pointing. If the time span over which the data for the DPI ere accumulated saw a significant variation in the pointing of the CZTI, then the reconstructed image could show a smearing. Individual Crab images can be constructed from the CZTI data for integration times as short as 30 s. Doing so with quadrant B data showed very similar extension in individual images (Fig. 15). Examination of the satellite attitude data, collected at 128 ms sampling, revealed that the attitude jitter was no more than 0.045 rms over the time scale of minutes to hours, thus making it very unlikely that the observed image extension could result from it. 2.2.4 Module dependence. Each CZTI quadrant is composed of 16 detector modules, all of which operate independently. At this point it was essential

76

Page 12 of 16

J. Astrophys. Astr.

(2021) 42:76

Figure 17. Simulated 1-d image profiles with zero (a) and 0:5 mm (b) shift of the mask pattern in the X-direction, compared with that obtained from observed data. Table 6. The shift in the mask pattern with respect to the detector for all quadrants. Quadrant

X shift (in mm)

A B C D

0.00 1:45 0.00 0.00

Y shift (in mm) 0.00 0.00 1.68 1.50

Figure 19. Chi-square value as a function of mask shift in the X direction for Quadrant B.

Figure 18. The image reconstructed from simulated DPI with the shifted mask in near field imaging similar to CZTI ground calibration.

We therefore reconstructed images using events from individual modules in all four quadrants. In the reconstructed images, no module in the quadrant A showed any shift or extension. However, in quadrant B, 14 modules showed extension and in quadrants C and D, 14 and 12 modules respectively showed the shift in the reconstructed source position. An account of this is presented schematically in Fig. 16. This collective display strongly indicated that the image extension or shift was associated with the imperfection of the complete assembly at the level of quadrants.

to establish whether the image distortion observed was a quadrant level phenomenon or do individual detector modules in a quadrant display discordant behaviour.

2.2.5 Detector-mask misalignment. Results from module-level imaging lead us to the possibility of misalignment between the detector and the mask. We

J. Astrophys. Astr.

(2021) 42:76

Page 13 of 16

76

Figure 20. (a)–(d) Crab nebula image reconstructed using FFT after application of the phase matrix for observation id 9000000406.

performed various simulations to examine the effect of misalignment on the reconstructed images. Initially, we simulated a DPI by shifting the mask pattern by 0.5 mm in the negative X direction. The source image reconstructed from the simulated DPI showed an extended peak. Next, we shifted the mask pattern at steps of 0.1 mm in the negative X-direction and compared the simulated image profiles with the observed image profile. The one-dimensional image profile of the image reconstructed from observed data, and simulated data without any mask shift showed a strong disagreement. However, the image profile of the image reconstructed

from simulated data with a mask shift of –1.45 mm showed a strong agreement with the observed image profile. Figure 17 shows the one-dimensional image profiles for observed and simulated data generated from the direct cross-correlation image. Then, we simulated DPIs for shifts in the X-direction, from –1.20 mm to –1.65 mm with a step of 0.05 mm and compared the resulting images with that observed. The minimum chi-square was found for a mask shift of –1.45 mm, Fig. 19. Hence, this experiment confirmed that in quadrant B there is a slight misalignment of –1.45 mm between the detector and the mask plate in the X-direction. Shifts for other quadrants

76

Page 14 of 16

J. Astrophys. Astr.

(2021) 42:76

Figure 21. (a)–(d) Reconstructed sky images using the shadow library corrected for mask shifts by Balanced crosscorrelation technique.

were also estimated using a similar methodology. Table 6 below outlines the shift in the mask pattern with respect to the detector for all four quadrants. We have not observed any extension in the reconstructed images during the on-ground calibration. To review the effect of the mask shift on the reconstructed images in near field imaging, we have simulated DPIs for diverging beam radiations shined by a radioactive source under the circumstances similar to the ground calibration. We then subjected the simulated DPIs to the image reconstruction algorithm.

Figure 18 shows the reconstructed image generated using the simulated DPH with the mask shift of – 1.45 mm. The reconstructed images did not show any extension or shift in the source position. In near field imaging, one mask element illuminates more than one detector pixel due to the diverging beam of radiations and possible that the mask shift may not affect the reconstructed image. After calculating the shifts for all quadrants, we corrected for shifts in the imaging algorithm. The section below describes the corrections applied.

J. Astrophys. Astr.

(2021) 42:76

Page 15 of 16

3. Correcting for mask shifts In mask cross-correlation, the DPI is cross-correlated with the coded aperture mask. To correct for the shift in the reconstructed image, we multiply a phase matrix in the Fourier domain, as shown in Eq. (1). F 1 ðF ðDPIÞXF ðMASKÞ  PÞ:

to the detector in quadrant B was –1.45 mm in the X-direction, while those in quadrants C and D were ?1.68 mm and ?1.50 mm respectively in the Y direction. We mitigated the effect of this misalignment by modifying the algorithms employed for image reconstruction.

ð1Þ

Here, F is FFT, F 1 is inverse FFT, and P is a phase matrix as described in Eq. (2), P ¼ ei2pðfx dx þfy dy Þ :

76

ð2Þ

Here, dx is the shift in the X direction in pixel unit, dy is the shift in the Y direction in pixel unit, Nx is the number of pixels in X direction, Ny is the number of 1 pixels in Y direction, fx ¼ 1 and Nx þ    þ Nx 1 1 fy ¼ Ny þ    þ Ny . After multiplication with the phase matrix and performing inverse FFT, the source is found to be at the expected location. CZTI data analysis pipeline is modified to incorporate the phase shift multiplication to correct the images reconstructed using mask crosscorrelation images. Figure 20 shows the reconstructed sky images generated using modified mask crosscorrelation for four quadrants. The phase matrix multiplication compensated for the shift in the reconstructed image, but the extension in quadrant B was not corrected. The extension in the image can be corrected using the shadow cross-correlation method using computed shadows that incorporate the measured mask shifts. Figure 21 shows the sky images reconstructed using the balanced cross-correlation. The reconstruction is performed using the shadow library corrected for the mask shift. In reconstructed images, the source is now at the expected location and also the extension in quadrant B is reduced to a great extent.

4. Conclusions We have described the on-ground calibration and inflight calibration performed for the imaging performance of CZTI. Using the qualification model, we verified the design of the CZTI coded mask and reconstruction algorithms. We verified that the performance of the CZTI meets the design goal of imaging resolution 8 arcmin or better. During the inflight calibration, we found that there is a misalignment between the detector and the mask in three of its four quadrants. The estimated mask shift with respect

Acknowledgements This publication uses data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). The CZT Imager is built by a consortium of Institutes across India including Tata Institute of Fundamental Research, Mumbai, Vikram Sarabhai Space Centre, Thiruvananthapuram, ISRO Satellite Centre, Bengaluru, Inter University Centre for Astronomy and Astrophysics, Pune, Physical Research Laboratory, Ahmedabad, Space Application Centre, Ahmedabad: contributions from the vast technical team from all these institutes are gratefully acknowledged.

References Bhalerao V., Bhattacharya D., Vibhute A. et al. 2017, J. Astrophys. Astr., 38, 31 Bhattacharya D. 2006, Adv. Space Res., 38, 2999 Caroli E., Stephen J. B., Di Cocco G., Natalucci L., Spizzichino A. 1987,45, 349 Fenimore E. E., Cannon T. M. 1978, Appl. Opt., 17, 337 in’t Zand J. J. M. 1992, PhD thesis, Space Research Organization Netherlands, Sorbonnelaan 2, 3584 CA Utrecht, The Netherlands Ramadevi M. C., Seetha S., Bhattacharya D. et al. 2017, Exp. Astron., 44, 11 Rao A. R., Chand V., Hingar M. K. et al. 2016, Astrophys. J., 833, 86 Richardson W. H. 1972, J. Opt. Soc. Am., 62, 55 Shankar B., Bhattacharya D. 2003, Bull. Astronom. Soc. India, 31, 491 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, in Proceedings of SPIE, Vol. 9144, Space Telescopes and Instrumentation 2014: Ultraviolet to Gamma Ray, 91441S Singh K. P., Stewart G. C., Westergaard N. J. et al. 2017, J. Astrophys. Astr., 38, 29 Skinner G. K. 1984, Nuclear Instrum. Methods Phys. Res. A, 221, 33 Skinner G. K., Ponman T. J., Hammersley A. P., Eyles C. J. 1987, Astrophys. Space Sci., 136, 337 Tandon S. N., Subramaniam A., Girish V. et al. 2017, Astronom. J., 154, 128

76

Page 16 of 16

Vadawale S. V., Rao A. R., Bhattacharya D. et al. 2016, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 9905, Space Telescopes and Instrumentation 2016: Ultraviolet to Gamma Ray, 99051G

J. Astrophys. Astr.

(2021) 42:76

Yadav J. S., Agrawal P. C., Antia H. M. et al. 2016, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 9905, Space Telescopes and Instrumentation 2016: Ultraviolet to Gamma Ray, 99051D

J. Astrophys. Astr. (2021)42:61 https://doi.org/10.1007/s12036-021-09748-w

Ó Indian Academy of Sciences Sadhana(0123456789().,-volV)F T3](0123456789().,-volV)

PAYLOAD CALIBRATION

Absolute time calibration of LAXPC aboard AstroSat AVISHEK BASU1,3,* , DIPANKAR BHATTACHARYA2 and BHAL CHANDRA JOSHI3 1

Jodrell Bank Centre for Astrophysics, University of Manchester, Oxford Road, Manchester M13 9PL, UK. Inter University Centre for Astronomy and Astrophysics, Post Bag 4, Pune 411 007, India. 3 National Centre for Radio Astrophysics, Tata Institute of Fundamental Research, Ganeshkhind, Pune 411007, India. *Corresponding Author. E-mail: [email protected] 2

MS received 5 November 2020; accepted 11 February 2021 Abstract. The AstroSat mission carries several high-energy detectors meant for fast timing studies of cosmic sources. In order to carry out high precision multi-wavelength timing studies, it is essential to calibrate the absolute time stamps of these instruments to the best possible accuracy. We present here the absolute time calibration of the AstroSat LAXPC instrument, utilising the broad-band electromagnetic emission from the Crab Pulsar to cross calibrate against Fermi-LAT and ground based radio observatories Giant Metrewave Radio Telescope (GMRT) and the Ooty Radio Telescope (ORT). Using the techniques of pulsar timing, we determine the fixed timing offsets of LAXPC with respect to these different instruments and also compare the offsets with those of another AstroSat instrument, CZTI. Keywords. Pulsars—instrumentation—multi-wavelength astronomy.

1. Introduction Multiwavelength observations are key to unravelling the physical processes ongoing in a variety of astrophysical sources. Such observations commonly involve multiple instruments situated at different locations. The target sources being studied often have time variable intensity and spectrum, so to characterise their properties it is essential to synchronise the intrinsic time stamps at various observatories being used. Even if the reference time information is obtained from a common source like the Global Positioning System, there could be internal delays in data processing electronics which need to be measured to achieve the necessary synchronisation. In this paper we report the result of our attempt to calibrate the timing offset of the AstroSat LAXPC instrument with respect to ground based Indian radio observatories, namely the Ooty Radio Telescope (ORT) and the Giant Metrewave Radio Telescope (GMRT), the This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

Large Area Telescope (LAT) instrument aboard the Fermi gamma-ray observatory. Results of a similar timing offset calibration experiment for another AstroSat instrument, the Cadmium Zinc Telluride Imager (CZTI), have been reported in Basu et al. (2018), and are also included here for comparison. This work is motivated by our ongoing effort to characterise the multi-wavelength properties of the Giant Radio Pulses (GRP) emitted by radio pulsars from time to time. In order to phase align these pulses between radio and X-ray bands an accurate alignment of the time stamps across the instruments is required. We therefore begin by measuring the timing offsets between the instruments involved in our experiment, the results of which we report here. We aim to obtain the offsets precise enough to track every pulse in the radio and X-ray bands unambiguously. The GRP arrives randomly within the  3 ms wide on-pulse window in the average pulse profile (Heiles et al. 1970; Manchester et al. 2005). Our desirable uncertainty on the offset measurement is one-tenth of the pulse width, i.e. less than 300 ls.

61

Page 2 of 7

Our aim can be achieved by timing and monitoring pulsars over a long span of time. Since our experiment involves telescopes operational at low-frequency radio wavelengths to c-rays, we have chosen the Crab pulsar (PSR J0534?2200) as our target source. The Crab pulsar emits pulsed emission across the whole electromagnetic spectrum from radio to very high-energy crays. At L-band (around 1.4 GHz) the Crab pulsar has two distinct components in its profile: the main-pulse (MP) and a relatively weaker inter-pulse (IP). The components are highly aligned across the spectrum except for some intrinsic emission delay. The MP at high energies leads the radio main pulse by 241  29 ls ([30 MeV; Kuiper et al. 2003), 344  40 ls (2–30 keV; Rots et al. 2004), (280  40 ls) (Kuiper et al. 2003), 235  33 ls (10-600 keV; Terada et al. 2008) and 275  15 ls (20–100 keV; Molkov et al. 2009). It is essential to account for these intrinsic delays while computing the time of arrival (TOA) of the pulses (discussed in Section 3.5). The results obtained from our experiment presented in this paper would allow us to time-align the Radio and X-ray time series data, enabling us to search for X-ray photon count enhancements coincident with the GRP. In Section 2 we discuss the instruments used for the experiment, in Section 3 we discuss the methodology adopted to measure the timing offsets and finally conclude the paper by presenting the results in Section 4

J. Astrophys. Astr. (2021)42:61

Monitor (SSM) allowing observations covering a wide ˚ to 380 keV. frequency range from 1300 A 2.1.1 LAXPC. The LAXPC is the prime X-ray detector on AstroSat, operating in the energy range 3–80 keV. LAXPC consists of three proportional counter units filled with primarily Xenon gas at a pressure of about 2 atmosphere, presenting a combined effective area of  6000 cm2 at energies below 20 keV, declining to  2500 cm2 at 80 keV (Antia et al. 2017). This large effective area, along with high timing resolution (10 ls) makes this an excellent instrument for X-ray timing studies, including those of pulsars. A collimator restricts the Field of View (FoV) of LAXPC to approximately 1  1 . LAXPC detectors record event mode data, with each event tagged with an instrument time stamp derived from a System Time Base Generator (STBG) driven by a temperature controlled crystal oscillator. Once every 16 seconds, a synchronising pulse is sent to all AstroSat instruments including the on-board Spacecraft Positioning System (SPS), which provides time in UTC based on the Global Positioning System. All the instruments record their current time stamp at the arrival of the synchronising pulse and these values are collected in a Time Correlation Table which is used in offline analysis to convert, by interpolation, the instrument time stamps assigned to the recorded events to UTC time stamps (Bhattacharya 2017).

2. Instruments and observations We performed multi-epoch, multi-frequency observations using Indian facilities like the Giant Metrewave Radio Telescope (GMRT), the Ooty Radio Telescope (ORT) operational at radio wavelength, and two payloads aboard AstroSat, the Cadmium Zinc Telluride Imager (CZTI), and the Large Area Proportional Counter (LAXPC). We have also made use of the publicly available data from Fermi-LAT operational at high-energy c-rays (  20 MeV– 300 GeV). 2.1 AstroSat AstroSat, India’s first space-based observatory was launched in October 2015 with five payloads on board (Singh et al. 2014). These are the Cadmium Zinc Telluride Imager (CZTI; Bhalerao et al. 2017), the Large Area X-ray Proportional Counter (LAXPC; Yadav et al. 2016), the Soft X-ray Telescope (SXT; Singh et al. (2016)), the Ultra Violet Imaging Telescope (UVIT; Hutchings 2014) and the Scanning Sky

2.1.2 CZTI. The Cadmium Zinc Telluride Imager extends the high-energy coverage of AstroSat to  380 keV, starting from  20 keV. It consists of a solid state, pixellated CZT detector array of total geometric area  976 cm2 , with a collimator and a Coded Aperture Mask situated above it. The Coded Mask and the collimator provide a 4:6  4:6 imaging Field-of-View at energies below  100 keV and gradually become transparent at higher energies. The CZTI records photon events time stamped at 20 ls resolution by its internal clock. These time stamps are converted to UTC time stamps during offline analysis in the same manner as for the LAXPC instrument. The CZTI has carried out extensive studies of the Crab pulsar at high energies, including that of its polarization (Vadawale et al. 2018). 2.1.3 Orbit determination. Comparing time stamps across observatories requires the arrival times to be referred to a common reference system, for which we adopt the Solar System Barycentre. The event time

J. Astrophys. Astr. (2021)42:61

stamps recorded by AstroSat are referred to the corresponding Barycentric arrival times, using the knowledge of the orbit of the satellite. The orbital position and velocity of AstroSat are measured by an on-board 10-channel Spacecraft Positioning System (SPS) unit that operates on signals received from the Global Positioning System (GPS) satellites. These measurements are regularly calibrated against those obtained by ranging from the AstroSat ground station. The housekeeping data stream of AstroSat provides the orbital position values sampled every 128 milliseconds, with an accuracy of better than 5 metres. The error budget in the barycentric correction arising from the uncertainties in the orbital position is thus limited to less than 0.017 ls. We have ignored this contribution in the reported uncertainties in our final results, which are much larger. 2.2 Fermi-LAT The Fermi-LAT is a high-energy c-ray telescope sensitive to the photons with the energy from below 20 MeV to more than 300 GeV (Atwood et al. 2009a). It monitors c-ray pulsars with a cadence of one-sixth of its duty-cycle. Individual photon events are recorded with a time resolution better than 1 ls (Smith et al. 2008). We use data of the Crab pulsar retrieved from the public archive1 of the Fermi mission. 2.3 The Giant Metrewave Radio Telescope (GMRT) The GMRT (Swarup et al. 1991) is an ‘‘Y’’-shaped interferometer with thirty, 45-m steerable dishes operational at low-frequency radio-wavelengths. Fourteen antennas are arranged in a compact array within a radius of 1 km, the remaining antennas are arranged in three arms. The observations were carried out by combining all 14 antennas and the first arm antennas in a tied array with an overall gain of 3.5 K/Jy. The Crab pulsar was observed using the GMRT at seven different epochs (shown with green markers in Fig. 1). The typical observation duration was 1–2 hours. The time-series raw voltage data acquired at 1390 MHz with 16 MHz bandwidth from every antenna were Fourier Transformed to obtain 256 channels voltage data. The instrumental phase lags among the antennas were determined by observing the point source 3C147, which were then compensated in the Fourier domain and added coherently. Further analysis was done offline described in Section 3. 1

https://fermi.gsfc.nasa.gov/cgibin/ssc/LAT/LATDataQuery.cgi.

Page 3 of 7

61

2.4 The Ooty Radio Telescope (ORT) The ORT is a 30 m wide offset parabolic cylindrical antenna in the east-west direction. It is 530 m long in the north-south direction sensitive to a single polarisation and operational at 334.5 MHz (Swarup et al. 1971). The gain of the telescope is 3.3 K/Jy and the system temperature is 150 K. The pulsar observations back-end at ORT is called as PONDER (Naidu et al. 2015), which starts recording the data on arrival of the rising edge of the minute pulse obtained from the GPS system. PONDER performs coherent de-dispersion in real-time and produce time-stamped folded pulseprofiles in ASCII format. PSR J0534?2200 was observed for 15 minutes daily as a part of a larger pulsar monitoring program (Krishnakumar et al. 2018) and the high cadence pulsar glitch monitoring program at the ORT (Basu et al. 2019). In this paper, we have used the data from September 01, 2015 (MJD 57226) to January 14, 2017 (MJD 57767)

3. Methodology The method for the absolute time calibration relies on the technique of pulsar timing. The technique of pulsar timing (Edwards et al. 2006) compares the observed TOAs with the predicted TOAs obtained from a simple rotation model of the pulsars. We perform the analysis in multiple steps in an iterative manner until the best solution is achieved.

3.1 ORT analysis As mentioned earlier in Section 2.4, coherently dedispersed time-stamped profiles are obtained from the ORT. The TOA of a pulse was computed using the software package PSRCHIVE (Hotan et al. 2004) from every profile by cross-correlating with a noisefree template of the pulse profile in the frequency domain described in Taylor (1992). The typical timing accuracy2 at ORT is 318 ls. The noise-free template was created in PSRCHIVE3 by fitting an optimal number of Gaussian waveform to a high S/N observed pulse profile. The TOAs obtained from our high cadence observations were then used to create a phase connected solution. Such a high cadence is especially 2

We refer the median of the TOA errors as the ‘‘typical timing accuracy’’ 3 http://psrchive.sourceforge.net/.

Page 4 of 7

J. Astrophys. Astr. (2021)42:61 F ermi-LAT GMRT ORT AstroSat-CZTI Astrosat-LAXPC

20000 10000 0 −10000

Post-fit Residual (µs)

Post-fit Residual (µs)

61

2000 1000 0 −1000 57300

57400

57500 57600 MJD

57700

57800

Figure 1. The timing residuals from different telescopes. Upper panel: The green, grey, red, black and yellow data points represent the TOAs obtained from GMRT, ORT, Fermi-LAT, LAXPC and CZTI observations, respectively. The systematic pattern is called as timing noise and the parallel tracks indicate the timing offset incorporated by different telescopes back end as explained in the text. Lower panel: The TOA residuals from all the observations after modelling the timing noise, the effect of dispersion measure and clock offsets (detailed description is given in the text).

important for the Crab pulsar to obtain a reliable phase connected solution because of the strong timing noise. The timing noise is observed as systematic wandering in the TOA residuals after removal of the standard spin-down model (Cordes & Helfand 1980). The TOAs were further segmented with a time span of a month to produce the monthly ephemeris from our data by fitting the spin-down model in the high precision pulsar timing package TEMPO24 (Hobbs et al. 2006). The monthly ephemeris with precise rotation parameters was used to re-fold the time-series data to obtain the precise TOAs. The Crab nebula provides a strong scattering screen to the radio waves emitted from the pulsar. The effect of scatter broadening is pronounced at 334.5 MHz and can contribute to the timing residuals as a systematic. Hence, in case of the Crab pulsar, it is difficult to de-couple the effect of timing noise from the scatter broadening, which in our analysis has been taken care by using the publicly available data from Fermi-LAT.

3.2 Fermi-LAT analysis The c-ray pulse profiles are free from the propagation effects, therefore the Fermi-LAT (Atwood et al. 2009b) archival data5 were used to model the timing noise which is a frequency-independent phenomenon. We use all the events in the energy range 0.1 to 300 GeV within a radius of 3 around PSR J0534?2200. The event data were split using Fermi science tool6 into smaller event data each of 7 days duration. These time stamps were referred to the solar system barycentre (SSB) adopting JPL planetary ephemeris DE200 and folded using the Fermi-plugin (Ray et al. 2011) of TEMPO2 using the ephemeris obtained from the ORT timing solutions. The standard template was constructed in a similar manner as explained in Section 3.1. However, to account for the intrinsic delay between the radio and the c-ray pulse profile the templates were aligned with an appropriate shift mentioned in the Section 1 The TOAs were computed from every pulse profile by cross-correlating the standard template. The timing accuracy was 309 ls. The timing analysis was performed using TEMPO2. The timing noise at this band was modelled with the combination of eight sine waves to obtain the white timing residuals using the FITWAVES tool in TEMPO2.

3.3 Re-analysis of the ORT data The timing solution with modelled timing noise from the Fermi-LAT TOAs was applied on the ORT TOAs. At this stage, the TOAs affected from the scatter broadening were removed from our analysis and monthly ephemeris were re-generated. Hence, the rotation parameters were obtained which are free from timing noise and the scatter broadening effects. We refer to this timing solution as the ‘‘iteration-2’’ solution. The iteration-2 solutions were used to re-fold the ORT time-series data and produce TOAs following similar steps as explained in Section 3.1.

3.4 GMRT analysis The raw voltage data obtained from the GMRT were also coherently de-dispersed offline with our 5

https://fermi.gsfc.nasa.gov/cgibin/ssc/LAT/LATDataQuery.cgi. https://fermi.gsfc.nasa.gov/ssc/data/analysis/scitools/overview. html.

6 4

https://bitbucket.org/psrsoft/tempo2/src/master/.

J. Astrophys. Astr. (2021)42:61

pipeline discussed in Naidu et al. (2015). The values of the dispersion measure (DM) were taken from the Jodrell Bank monthly ephemeris7 (Lyne et al. 1993) nearest to the epoch of observations. The de-dispersed time-series were further folded using the iteration-2 monthly ephemeris obtained from the ORT data. The GMRT offline analysis also supplies with de-dispersed 64 channel sub-banded data with 32 sub-integrations. The standard template was constructed from the highest S/N observed profile following the same method explained in Section 3.1. The template of the pulse profile from the GMRT data was aligned with the ORT templates and then the TOAs were computed following the method discussed in Section 3.1. The timing accuracy obtained at GMRT is 162 ls. The dynamic pulsar wind within nebular filaments leads to the variation in the electron density along the line of sight, which results in time variation of the DM. The typical variation in DM is of the order of 0.01 pc cm 3 , which incorporates a change in time of arrival by 21 ls and 370 ls at 1390 MHz and 334.5 MHz respectively. Therefore it is essential to correct for DM variations to obtain reliable and precise estimates of the offsets. The TOAs computed from the ORT data in Section 3.3 and from the GMRT data were used to measure the DM at different epochs. The fixed offset between the data acquisition pipelines at the ORT and GMRT were known from previous measurements (Surnis et al. 2018). Hence, the DM was estimated after accounting for this delay between GMRT and ORT using the JUMP parameter in TEMPO2. It may be noted that only 8 observations were nearly simultaneous between ORT and GMRT. Therefore, 8 different estimates of DM at 8 different epochs were obtained (Fig. 4 of Basu et al. 2018). The DM was further used to perform the offline coherent de-dispersion to obtain the de-dispersed time series data, which were folded using the iteration-2 timing solutions. The TOAs from the GMRT data were finally produced by following the methods described above.

Page 5 of 7

61

converted to SSB adopting the JPL planetary ephemeris DE200 and using the satellite position in the code as1bary8. The barycentre recorded events were folded to construct the pulse profile using iteration-2 timing solution obtained in Section 3.3 using our own codes. Further, the standard template was created following the same method as described in Section 3.1. In case of LAXPC, the event files were created by combining the data from three consecutive orbits, which were then barycentred using the as1bary and folded using the iteration-2 timing solution obtained in the Section 3.3. The standard templates were created following a similar method as mentioned above. The templates obtained from the CZTI and the LAXPC were appropriately shifted with respect to those obtained from the GMRT, ORT and Fermi-LAT to take into account the intrinsic energy-dependent emission delays. Finally, the TOAs were computed for the CZTI profiles and the LAXPC profiles using the templates thus constructed. This method of incorporating the known energy-dependent intrinsic emission delays in the template construction allows us to find the true clock offsets between two instruments directly from the TOA differences between them.

3.6 Offset measurements

The CZTI has four detectors arranged in four quadrants. There are no relative offsets between individual quadrants. Hence data from all the four quadrants were combined and the time tags of the photons were

The TOAs from all the telescopes were collated together to compute the offsets between the different telescopes. These TOAs were analysed using a timing model obtained by merging the pulsar rotation model, which gave the phase connected solution, with a constrained DMMODEL in TEMPO2. The DMMODEL is obtained from the DM time series, fitted from simultaneous observations at the ORT and the GMRT using the procedure described in Section 3.4. Inclusion of DMMODEL terms in the timing model accounts for the DM offsets from the chosen reference DM. The timing residuals obtained by applying this timing model to TOAs from all the telescopes have been shown in the upper panel of Fig. 1. The systematic trend as a function of time in these residuals for all telescopes is due to the timing noise of the pulsar. The residuals represent the difference between the predicted and the observed TOAs. As the TOA from each telescope additionally consists of a clock offset which is fixed with respect to the observation epoch, the residuals of a pair of telescopes are seen as parallel tracks in the diagram. Thus, fitting a constant

7

8

3.5 AstroSat-CZTI and LAXPC analysis

http://www.jb.man.ac.uk/pulsar/crab.html.

http://astrosat-ssc.iucaa.in/?q=data_and_analysis.

61

Page 6 of 7

J. Astrophys. Astr. (2021)42:61

Table 1. Table presents the reference timing solution of the Crab pulsar after considering the effect of DM variation and the timing noise.

Table 2. The table summarises the clock offsets of different telescopes given in the first column with respect to the GMRT.

Pulsar parameter

Instrument

05:34:31.973 ?22:00:52.06 29.6607409(4) -3.6937842(9) E-10 1.1905(3) E-20 57311.000000136 40675 57311.000000136 56.7957 -14.7 2 0.0054325986245627 DE200 57311.000000136 57278 58026

AstroSat-CZTI AstroSat-LAXPC Fermi-LAT ORT

Clock-offsets in ls 4716  50 5689  23 5368  56 29639  50

1 ORT 0 1.0 GMRT 0.5 Normalized Intensity/Counts

RAJ (hh:mm:ss) DECJ (dd:mm:ss) F0 (Hz) F1 (Hz s 1 ) F2 (Hz s 2 ) PEPOCH (MJD) POSEPOCH (MJD) DMEPOCH (MJD) DM (pc cm 3 ) PMRA (mas/year) PMDEC (mas/year) WAVE_OM (year 1 ) Solar system planetary ephemeris WAVEEPOCH (MJD) START (MJD) FINISH (MJD)

Value

1 LAXPC 0 1 CZTI

difference to residuals of a pair of telescope in Fig. 1 determines the timing offsets between the telescopes. The timing model, described above, was merged with the timing noise model obtained in Section 3.2. This reference timing solution after accounting for the timing noise and the DM variation is given in Table 1. Finally, we use the JUMP feature of TEMPO2 to measure the offsets between the telescopes discussed in Section 4.

4. Results TEMPO2 allows one to measure the offsets between different telescopes using the JUMP feature. Utilising this, we estimate the offset between the GMRT and CZTI to be 4716  50 ls and that between GMRT and LAXPC to be 5689  23 ls. The measured offset between the GMRT and ORT is 29639  50 ls, between GMRT and Fermi-LAT is 5368  56 ls. These clock offsets have been further tabulated in Table 2. The phase aligned pulse profile after accounting for the offsets has been presented in the Fig. 2. In the bottom panel of Fig. 1, we present the timing residuals obtained after removing the timing offsets between them. The trend-free residuals imply that all the pulsar parameters and clock offsets have been properly modelled. The clock offset between the LAXPC and the CZTI instruments aboard AstroSat is found to be 969  51 ls. The uncertainties

0 1 Fermi-LAT 0 0.0

0.2

0.4 0.6 Rotational Phase

0.8

1.0

Figure 2. The multi-wavelength pulse profiles of the Crab pulsar. The pulse profiles were aligned after correcting the clock-offsets between the telescopes.

in the offsets are obtained from those of the parameters fitted to the TOA using the JUMP function. The results presented here meet the desired accuracy (see Section 1) for a multi-wavelength investigation of the GRP from the Crab pulsar with the instruments used in this paper.

Acknowledgements We thank the anonymous referees for their valuable suggestions to improve the presentation of the paper. This publication made use of data from the Indian astronomy mission AstroSat, archived at the Indian Space Science Data Centre (ISSDC). The LAXPC data were processed at the Payload Operations Centre at TIFR Mumbai. The CZT Imager instrument was built by a TIFR-led consortium of institutes across India, including VSSC, ISAC, IUCAA, SAC, and PRL. The Indian Space Research Organisation

J. Astrophys. Astr. (2021)42:61

funded, managed and facilitated the project. We thank the staff of the Ooty Radio Telescope and the Giant Metrewave Radio Telescope for taking observations over such a large number of epochs. Both these telescopes are operated by National Centre for Radio Astrophysics of Tata Institute of Fundamental Research. PONDER backend, used in this work, was built with TIFR XII plan grants 12P0714 and 12P0716. BCJ acknowledges support for this work from DST-SERB grant EMR/2015/000515. BCJ and AB acknowledges the support of Department of Atomic Energy, Government of India, under projcet # 12-R&D-TFR-5.02-0200. AB acknowledges the support from the UK Science and Technology Facilities Council (STFC). Pulsar research at Jodrell Bank Centre for Astrophysics and Jodrell Bank Observatory is supported by a consolidated grant from STFC. References Antia H. M., Yadav J. S., Agrawal P. C. et al. 2017, Astrophys. J. Suppl. Series, 231, 10 Atwood W. B., Abdo A. A., Ackermann M. et al. 2009a, ApJ, 697, 1071 Atwood W. B., Abdo A. A., Ackermann M. et al. 2009b, ApJ, 697, 1071 Basu A., Joshi B. C., Bhattacharya D. et al. 2018, A&A, 617, A22 Basu A., Joshi B. C., Krishnakumar M. A. et al. 2019, MNRAS, 491, 3182 Bhalerao V., Bhattacharya D., Vibhute A. et al. 2017, J. Astrophys. Astr., 38, 31 Bhattacharya D. 2017, J. Astrophys. Astr., 38, 51 Cordes J. M., Helfand D. J. 1980, Astrophys. J., 239, 640 Edwards R. T., Hobbs G. B., Manchester R. N. 2006, MNRAS, 372, 1549 Heiles C., Campbell D., Rankin J. 1970, Nature, 226, 529 Hobbs G., Edwards R., Manchester R. 2006, MNRAS, 369, 655 Hotan A. W., van Straten W., Manchester R. N. 2004, Publ. Astron. Soc. Australia, 21, 302

Page 7 of 7

61

Hutchings J. 2014, Astrophys. Space Sci., 354, 143 Krishnakumar M. A., Joshi B. C., Manoharan P. K. 2018, in Pulsar Astrophysics: the Next Fifty Years, Proceedings of the International Astronomical Union, IAU Symposium, Volume 337, pp. 364–365 Kuiper L., Hermsen W., Walter R., Foschini L. 2003, A&A, 411, L31 Lyne A., Pritchard R., Graham Smith F. 1993, MNRAS, 265, 1003 Manchester R. N., Hobbs G. B., Teoh A., Hobbs M. 2005, Astronom. J., 129, 1993 Molkov S., Jourdain E., Roques J. P. 2009, Astrophys. J., 708, 403 Naidu A., Joshi B. C., Manoharan P. K., Krishnakumar M. A. 2015, Exp. Astr., 39, 319 Ray P. S., Kerr M., Parent D. et al. 2011, Astrophys. J. Suppl. Series, 194, 17 Rots A. H., Jahoda K., Lyne A. G. 2004, Astrophys. J. Lett., 605, L129 Singh K. P., Tandon S., Agrawal P. et al. 2014, in SPIE Astronomical Telescopes?Instrumentation, International Society for Optics and Photonics, pp. 91441S–91441S Singh K. P., Stewart G. C., Chandra S. et al. 2016, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Volume 9905 Smith D. A., Guillemot L., Camilo F. et al. 2008, A&A, 492, 923 Surnis M. P., Joshi B. C., McLaughlin M. A. et al. 2018, Astrophys. J., 870, 8 Swarup G., Sarma N., Joshi M. et al. 1971, Nature Phys. Sci., 230, 185 Swarup G., Ananthakrishnan S., Kapahi V. et al. 1991, Curr. Sci., 60, 95 Taylor J. H. 1992, Philosophical Trans: Physical Sciences and Engineering, p. 117 Terada Y., Enoto T., Miyawaki R. et al. 2008, Publ. Astronom. Soc. Japan, 60, S25 Vadawale S. V., Chattopadhyay T., Mithun N. P. S. et al. 2018, Nature Astron., 2, 50 Yadav J., Agrawal P., Antia H. et al. 2016, in SPIE Astronomical Telescopes?Instrumentation, International Society for Optics and Photonics, pp. 99051C–99051C

 Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:24 https://doi.org/10.1007/s12036-021-09700-y

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

MISSION

Challenges of realizing and operating AstroSat in-orbit V. KOTESWARA RAO*

and K. SURYANARAYANA SARMA

U R Rao Satellite Centre, Old Airport Road, Vimanapura, Bengaluru 560 017, India. *Corresponding Author. E-mail: [email protected] MS received 22 October 2020; accepted 3 January 2021 Abstract. AstroSat, a dedicated Space Observatory of India, has completed five successful years of operation in space on 28th September 2020. AstroSat is a quite complex satellite, as it is a multi-wavelength observatory with many scientific instruments. The saga of many agencies, including Indian Space Research Organisation, the lead agency, and many scientists and engineers has resulted in realizing and operating this mission with excellent performance and highly satisfactory results. This mission generated a lot of observations leading to enhanced research activity for Indian astronomers, as well as international astronomers. It has also kindled interest, as expected, in young scientists and science students. The mission still continues in orbit contributing to celestial observations. AstroSat is a collaborative effort of many agencies not only from India but also from international agencies. The managers and the project team had to face many technological and managerial challenges at various stages of the mission. In this paper we present the challenges in conceiving a space science mission in India, and methods adopted to overcome them to make the mission successful. This may help in planning and executing future space science missions, more efficiently, meeting the growing demands from the scientific community involved in the frontier areas of space research. Keywords. AstroSat—astronomy—space observatory.

1. Introduction AstroSat, specialized in simultaneous multi wavelength observations, is the first dedicated Space Astronomy Observatory of India. It was launched on September 28, 2015, with a lift-off mass of 1515 kg, into a 650-km orbit, inclined at an angle of 6 to the equator, by PSLV-C30 (XL) rocket from Satish Dhawan Space Centre, Sriharikota (Seetha & Megala 2017). AstroSat, has completed five successful years of operation in space as a multi wavelength astronomical space observatory. AstroSat is meant to observe galactic and extra- galactic cosmic sources. Complete understanding of these cosmic processes involves enhanced sensitivity, spectral and timing resolution studies which calls for correlation and coordination between various ground-based and This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

space-based observatories. AstroSat observes in a range of wavelengths spanning from UV, soft X-rays and hard X-rays, enabling simultaneous multi-wavelength monitoring with temporal and spectral variability. This multi-wavelength and simultaneous observational capability is the uniqueness of AstroSat. The spacecraft carries five principal scientific payloads (Singh et al. 2014): (i) Large Area X-ray Proportional Counter (LAXPC) has three identical instruments covering 3–80 keV region. (ii) Cadmium–Zinc–Telluride Imager (CZTI) with coded mask aperture sensitive in 20–100 keV. (iii) Soft X-ray imaging Telescope (SXT) uses X-ray reflecting mirrors and X-ray CCD for imaging and spectral studies in 0.3–8 keV. (iv) Scanning Sky Monitor (SSM) for detecting and monitoring new and known X-ray sources in 2.5–10 keV region. It acts as a survey instrument for identifying transient and new sources.

24

Page 2 of 10

(v) Ultra Violet Imaging Telescope (UVIT) consisting of two identical telescopes. One telescope covers Far UV (FUV) band (130–180 nm) and the other covers Near UV (NUV) band (200–300 nm) and visible band (320–550 nm). The science instruments on AstroSat satellite are shown in Fig. 1. Leading institutions responsible for the design and development of the science instruments are indicated in the parenthesis along with each instrument. LAXPC, SXT and CZTI instruments are developed by Tata Institute of Fundamental Research (TIFR) whereas UVIT instrument is developed by Indian Institute of Astrophysics (IIA). SSM payload is developed by ISRO Satellite Centre (ISAC). AstroSat is a complex mission involving many mission elements as shown in Fig. 2. Well-laid down scientific objectives and specifications of the science payloads (Koteswara Rao et al. 2009) have contributed to the success of the mission. Many leading Indian and international agencies/institutions played a key role in building, launching, operating and producing excellent science results by utilizing the observational data. As is the case with many space science satellites, a large number of technological developments were undertaken for both satellite bus and the science instruments.

J. Astrophys. Astr. (2021)42:24

2. Background Aryabhata, India’s first satellite, successfully launched into a near earth orbit on 19 April 1975, from a USSR Cosmodrome (Rao 1978), carried an X-ray astronomy payload. This payload was a gas filled proportional counter with a modest effective area of 15.4 cm2. X-ray observations, on GX17?2 and GX9?9, were made using this payload (Kasturirangan et al. 1976). IRS-P3 satellite, essentially an Earth observation satellite, launched by PSLV-D3 on March 21, 1996 carried proportional counters of the Indian X-ray Astronomy Experiment (IXAE). X-ray observations of the Galactic X-ray transient source GRS 1915?105 were reported, with the pointed payload. which showed remarkable richness in temporal variability (Yadav et al. 1999). The observations were carried out on 1997 June 12 to 29 and August 7 to10, in the energy range of 2 to 18 keV and revealed the presence of very intense X-ray bursts. With this experience of building space grade astronomy instruments and doing advanced research in astronomy with space-based observations, Indian Space Research Organisation (ISRO), being a pioneer in space research, mooted the idea of a dedicated satellite for astronomy. The motive also included the idea of bringing all the Indian astronomers and possibly international astronomers together to build a space observatory and pave way for frontier research in the area of astronomy. The idea also was to kindle interest in young students to pursue science subjects and do good research work in space science. Chairman of ISRO, was firmly behind this idea and vigorously pursued it to take shape.

3. Conceptualization and project formation

Figure 1. AstroSat with instruments.

Prior to the formal organization of AstroSat project, a lot of work was carried out by scientists from ISRO, TIFR, IIA, RRI and IUCAA. Two working groups discussed the proposals and gave recommendations. Subsequently another committee has conducted a detailed study on a mix of scientific payloads that would bring important contributions to high energy astronomy and astrophysics and would fill the gaps in scientific observations by all global missions present and near future. AstroSat configuration committee has evolved the overall satellite configuration. On April 03, 2003, an office order was issued by Chairman of ISRO/Secretary, DOS, on the organization of AstroSat

J. Astrophys. Astr. (2021)42:24

Page 3 of 10

24

government and financial approvals were accorded within a year.

3.1 Satellite and science instruments configuration

Figure 2. Mission elements of AstroSat.

project. Through this order, a Project Director with responsibility to realize the AstroSat spacecraft, its launch and on-orbit operations; a Principal Investigator with responsibility to realize the total scientific payload and coordinate the developments in respective work centres; two Programme Managers and three Project Managers with responsibility for the development and realization of various payloads; were identified. The objective of AstroSat is to focus on high-resolution UV imaging for morphological studies of galactic and extragalactic objects, broad-band studies of X-ray sources and other multiwavelength targets ranging from nearby stars to the very distant active galactic nuclei (Koteswara Rao et al. 2009). Understanding high energy processes in binary systems, searching for black hole sources in the Galaxy, measuring magnetic fields of neutron stars, studying high energy processes in extragalactic systems and detecting new transient X-ray sources are some of the scientific objectives. Accordingly, AstroSat is configured to be a multi-wavelength astronomy satellite for studying the cosmic sources simultaneously over a wide range of the electromagnetic spectrum (from optical, ultraviolet to high energy X-rays). Commensurate with the objectives of AstroSat, the salient features of the mission were set as follows: • Build and operate a multi-wave length space observatory. • Provide opportunity to academia to build instruments. • Nurture space astronomy in the country. • Inculcate scientific temper. • International co-operation. Within a few months of the project organization a detailed project report was prepared. Necessary

In the class of PSLV, XL version has the highest payload delivery capability. A mass of 1500 kg for the AstroSat satellite was contemplated considering the PSLV (XL) capability to place the satellite in the desired orbit. PSLV was a workhorse for all Indian Remote sensing Satellites (IRS). The volume is decided by this class of satellite bus. The mass and size budget for the science instruments (Navalgund et al. 2017) is given in Table 1. As all the instruments with the exception of SSM have to look at the same source at any given time, all the instrument view axes were to be co-aligned and made to look in the same direction. 3.2 Choice of orbit The altitude of the orbit, inclination and mass of the satellite are interlinked and together are dictated by the launch vehicle capability. The science instruments, in operating condition, can get damaged due to the radiations from the South Atlantic Anomaly (SAA) region if the satellite passes over it. From this point of view, a near equatorial orbit is preferred. There is no restriction on orbit height from the science instruments. Altitude selection is mainly based on atmospheric drag which affects the life of the satellite. Considering the PSLV (XL) capability altitude of 650 km and inclination of 8 were contemplated for a 1500 kg satellite. With necessary revisions from time to time the inclination goal was brought down to 6. Altitude of 650 km and inclination of 6 were achieved for the 1515 kg AstroSat by precise launch. In the five years of operation no orbit maintenance was required.

4. Collaborative effort AstroSat project is a fine example of an integrated effort not only by national agencies and institutions but also by international agencies and universities. For the first time, in Indian space research programme, all the payloads, with the exception of SSM, were completely designed, realized, tested, qualified, delivered and used by institutes other than ISRO. The national institutes include Tata Institute of Fundamental

24

Page 4 of 10

J. Astrophys. Astr. (2021)42:24

Table 1. Mass and size of AstroSat science instruments. Instrument

Mass (kg)

LAXPC

415

CZTI SXT SSM UVIT

56 73 75 230

Size (mm 9 mm 9 mm) 1193 9 568 9 690 (each of three detectors) 530 9 510 9 813 382 9 580 9 2482 1200 9 563 9 543 877 diameter 9 3100

Research (TIFR), Mumbai; Indian Institute of Astrophysics (IIA), Bangalore; Inter University Centre for Astronomy & Astrophysics (IUCAA), Pune, Raman Research Institute (RRI), Bangalore and Physical Research Laboratory (PRL), Ahmedabad. The international institutes are Canadian Space Agency, Canada; University of Leicester (UoL), UK and University of Calgary, Alberta, Canada (through CSA). TIFR developed three main x-ray payloads namely LAXPC, SXT and CZTI as well as a Charged Particle Monitor (CPM) instrument. IIA developed the UV payload with twin telescopes, one for far UV and another for NUV and visible, with 375-mm optics for each telescope. Laboratory for Electro Optics Systems (LEOS), a unit of ISRO developed, for the first time in the country, the UV optics with necessary coatings qualified for space application. University of Leicester contributed the detectors for Soft x-ray Telescope (SXT). Canadian Space Agency developed the critical electronics part for the UV payload. TIFR and IIA were funded, for the development of project elements, from ISRO. The costs of manpower, civil works, facilities etc. were borne by respective institutes. In the same spirit of true collaboration there was no monetary transaction of any sort between ISRO and, UOL and CSA. The contributions of the international agencies were acknowledged by means of reserved percentage of observation time. All technical and managerial issues, throughout the project period were amicably solved by mutual discussions and finding suitable solutions. The understanding between scientists and managers from all institutions is definitely one major cause for the grand success of the ASTOSAT mission.

observation or communication satellites. The results from the scientific satellites are neither tangible in immediate future nor have any commercial value. Naturally they get lesser priority. AstroSat schedule was primarily dictated by the realization and availability of all systems. AstroSat had to be realized in parallel with many other missions. Apart from that the main reason for longer realization time is because of the state of art technologies and in a few cases breakthrough technologies that are required for scientific payloads and the satellite bus. The challenges were many in terms of managerial and technical terms. The space of this article permits a few major challenges to be reported.

5.1 Detectors for UVIT The spatial resolution of the UV payload is set at an ambitious level of less than 1.8 arcsec. Galaxy Evolution Explorer (GALEX), a NASA Small Explorer for performing a survey of the sky in two ultraviolet bands was launched on an Orbital Sciences Corporation Pegasus rocket on April 28, 2003 at 12:00 UT from the Kennedy Space Center into a circular, 700 km, 29 inclination orbit. The instrument is designed to image a very wide 1.25 field-of-view with 400 –600 resolution (Morrissey et al. 2005). The resolution of UVIT is set to be better than GALEX by 2 to 3 times. One of the key factors to achieve this resolution is the detector, configured as shown in Fig. 3. The electrons generated from the photocathode have to enter into the pore of the Micro Channel Plate (MCP) just below it without spilling to the neighboring pores. This can only be achieved if the front gap between the photocathode and MCP is very less. The lesser gap creates a technical challenge in terms of arcing, as a voltage of

5. Technical challenges Globally, scientific satellites of the magnitude of AstroSat take longer periods for realization in relative comparison to their counterparts like Earth

Figure 3. UVIT detector configuration.

J. Astrophys. Astr. (2021)42:24

300 V is applied between them, and the possibility of their touching each other due to the vibrations and shocks during the launch of the satellite. The credit for the successful development of these detectors goes to M/s Photek, UK. The challenge was taken by the management of Photek as scientific achievement rather than commercial success. It was a magnificent joint effort by Photek, IIA and ISRO teams.

5.2 Mirrors for SXT Wolter-1 geometry optics is used in SXT. Two sections of the geometry with 4 quadrants in each and with a nesting of 40 layers needs a total of 320 mirrors. The mirrors are thin foils (0.2-mm thickness) of Aluminum with replicated gold surfaces on the reflecting side for each section (Singh et al. 2017). Each mirror is 100-mm long. The process of realizing these mirrors involves fabrication of a highly polished glass mandrel, sputtering of gold on to the glass surface in vacuum, adhering the aluminum foils to gold with epoxy and removing the foils from the glass. Technology and engineering skills with lot of care and patience are required to do such work. The SXT team at TIFR achieved this and produced the optics for SXT, for the first time in India with great success. This type of optics design is better suited for realizing X-ray payloads with less mass. Only a couple of countries, in the world, have produced such type of x-ray optics so far. Final measurements on these ˚ , demonstratmirrors showed roughness of *7–10 A ing the wonderful quality.

5.3 Qualification of detectors for CZTI CZTI uses the large band gap semiconductor device, i.e. Cadmium–Zinc–Telluride (CZT). The detector has 64 modules divided into four quadrants, each quadrant containing a 4 9 4 modules. 256 pixelated contacts arranged in a 16 9 16 array in each module (Rao et al. 2017). The detectors were procured from M/s Orbotech, Israel, mainly producing the detectors for health care instruments. At the end of the year 2010, GE has acquired this company. This was the only source of these detectors and the supplier has no experience in space qualification of the products. The team with TIFR, VSSC (ISRO) and Orbotech scientists has mounted a programme to qualify these detectors and weed out the detectors with bad pixels.

Page 5 of 10

24

The testimony of the qualification is that, so far, no detector has failed on-board AstroSat CZTI.

5.4 Mirrors for UVIT In the two telescopes of UVIT, primary mirrors are of 375-mm diameter. and a central hole of 155 mm. Secondary mirrors are of 140-mm diameter. The surface error for all mirrors is better that k/50 rms and ˚ rms (Kumar et al. micro roughness better that 10 A 2012). Average reflectivity of both the mirrors is better than 60% for the wavelength band of FUV (130–180 nm), better than 70% for the wavelength band of NUV (180–200 nm) and better than 80% for VIS (200–600 nm). Though there are lot of developments in the country for developing optics, all of them are in the visible domain of the optical spectrum and a few are in Infrared domain. There was no institute available within the country which has developed optics in Ultraviolet domain. This challenge was taken up by LEOS (ISRO) team with support from IIA team and produced the required mirrors with better performance than specified. The necessary coating technology was developed inhouse and qualified for rigorous space environment.

5.5 Structural design The five payloads had their own unique mounting requirements on the satellite, with four payloads (UVIT, LAXPC, SXT and CZTI) requiring co-alignment (Navalgund et al. 2017). The proven Indian Remote sensing Satellite (IRS) bus was adopted for AstroSat. The real challenge was to accommodate the payloads with a mass of nearly 60% of the total satellite mass. The problem is further complex because the Field-of-View (FOV) clearance requirements for each payload as the others may obstruct the field or reflect/scatter into the instrument. With the skill, experience, long discussions between different teams and excellent understanding, these have been achieved. As the UVIT payload needed higher stability, it has been accommodated in the central cylinder, necessitating shifting of the propulsion system tanks elsewhere. The rotating platform for SSM payload carried more than 100 electrical wires to be connected to the inside electrical systems. Normal practice is to have these connections through rotating slip rings. But the size and mass of the rotating shaft

24

Page 6 of 10

becomes unwieldy. After careful consideration it was decided to rotate the platform in both clockwise and anticlockwise directions and avoid continuous rotation of more than 360. The bunched twisted harness was qualified for the required number of rotations expected in the life of the mission.

5.6 Thermal design Thermal requirements of UVIT payload are very stringent. Temperature of telescope tubes are to be maintained between 18C and 22C. Axial variation of temperature on telescope tubes is to be within ?/-2C. Circumferential variation of temperature on telescope tubes is to be within 5C (Kumar et al. 2012). The telescopes are of nearly 2 m in length. One end of the telescope may be exposed to the sunlight when the other end is exposed to deep space. The surface areas for any passive control are limited. Prudent and exceptionally brilliant design of thermal team has achieved these stringent requirements and the proof is the on-orbit performance. But this is with a cost of increased number of heaters along the body of the telescopes and their control automatically. Special inorganic black coatings were developed yielding ultra-high emittance and absorptance which are employed for the Invar and aluminum tubes of UVIT payload. SXT and CZT detectors are to be cooled. In CZT case passive cooling with heat pipes and radiator are adequate. In case of SXT in addition to heat pipe and radiator, an active cooling with a thermo electric cooler is used. The large size radiators are realized with honeycomb structures for low mass. The radiators are fixed to the spacecraft body using specially designed flexures for providing high thermal resistance and required structural stability. The DT requirement in case of SXT is beyond the limits of the existing heat pipe technology using ammonia. New heat pipes with ethane were developed for first time and successfully used to meet the requirements.

5.7 Special mechanisms Special mechanisms were developed for payload systems. UVIT and SXT telescope have one-time operated deployable cover mechanisms and SSM has hold down, release and steering mechanism. Both UVIT and SXT employed paraffin wax-based

J. Astrophys. Astr. (2021)42:24

actuators whereas SSM employed pyro cutter-based actuator. Both UVIT and SXT deployable covers are at the top end of the external baffle tubes and act as contaminant covers for the optical elements of the telescope on ground and during launch. In orbit, these covers are deployed exposing the telescope optics for imaging. In case of UV telescopes the deployed covers also act as Sunshades and the covers are deployed to 95 degrees while in SXT payload, the telescope cover deploys to 256 degrees. After deployment the covers are to be locked in position. The door mechanisms were accordingly designed and qualified for these functions.

5.8 Contamination control Neither ISRO nor IIA had previous experience of fabricating, handling, transporting, assembling and testing of UV optics. The surface reflectivity and transmission of UV optical elements degrade with contamination in terms of deposition of particles or molecules. The absorption cross-sections in ultraviolet are very large, and great care needs to be taken to avoid any contamination (Kumar et al. 2012). The particulate contamination can be reduced by control of cleanliness during all phases of operation. The molecular contamination is controlled by using materials with low Total Mass Loss (TML \1%) and Collected Volatile Condensable Material (CVCM \0.1%). The materials passing these criteria are further tested for their potential for contamination before being accepted for use. Purging with high purity nitrogen gas is used at various stages of transport, assembly and even at launchpad. All optical components are open only in clean rooms of class 100. The thermo-vacuum chambers used for testing the payload and the satellite were thoroughly tested for cleanliness before loading the payload/ satellite. All the subsystems delivered to get integrated on to the satellite were baked at higher temperature for sufficient time to avoid any degassing in further tests/operations. 20-mm diameter 9 2 mm thick windows of MgF2 were used as witness samples for measuring possible contamination at every stage. The transmission of the window in UV range (120 nm to 180 nm) was measured and compared with its original transmission. The degradation in transmission gives the measure of potential of contamination. The witness samples were periodically checked for its transmission/reflection in 120–180 nm range. The goal for overall cumulative

J. Astrophys. Astr. (2021)42:24

transmission/reflection loss on a witness sample was set to be \5%. It is gratifying that the witness windows kept with the detectors and the mirrors have not shown any measurable loss of transmission during the life cycle of AstroSat.

5.9 Data storage and handling system The payloads are always ON and continuously obtain science data. The peak data rates vary differently and it depends on mode of observation. The volume and data rate is high for UVIT payload. There is no direct real time transmission of payload data to ground station. The data is always recorded onboard storage and played back. A single ground station, located at Byalalu, Bengaluru, was contemplated to receive the data stored from all the payloads. The data storage and handling systems so far developed by ISRO operated in exclusive mode of either ‘recording’ or ‘playback’. When it came to AstroSat the observations continue during the play back of the data also and this data at any cost cannot be lost. Therefore, the system was newly designed with the feature of simultaneous playback and recording capability. After AstroSat this type data storage system has become normal for almost all the subsequent satellites of ISRO. For the same reason of downloading the data from satellite to the ground station, during radio visibility, a fixed orientation RF antenna would not be sufficient as the observation direction and the ground station direction vary from time to time. Two-phased array antennae were configured, with electronic beam steering capability, each having a coverage of one hemisphere.

5.10 Integration and testing Each model of payload instrument at unit level underwent electrical interface compatibility tests, vibration, and thermal-balance and EMI-EMC tests. A very well chalked out sequence of integration of systems is followed totally in lines of contamination control. During the integration activity and disassembled mode all systems are tested for functional performance. The totally assembled spacecraft underwent functional and performance tests, end to end tests before and after the environmental tests, vibration and thermo-vacuum, autonomous tests, mass properties check, field-of-view and polarity checks, alignment tests, acoustic testing, fit check with launch

Page 7 of 10

24

vehicle adapter, clamp-band release and mechanisms functional tests, solar panel deployment and payload cover deployment tests. Compatibility with communication with ground station equipment is also tested. Thermal-balance tests is carried out to validate the thermal mathematical models and to verify the ability of the thermal control subsystem to keep payloads and spacecraft equipment within specified temperature limits under simulated extreme expected orbital conditions. Minor failures of systems, electrical or mechanical, that occurred during these tests are fixed using special procedures and tools different from conventional methods complying to the contamination control procedures. UVIT, SXT, SADA are continuously purged. The LAXPC and SSM units are bagged and continuously purged. All the tests, including vibration tests and mass properties measurements, are conducted in 10,000 class clean room.

6. Managerial challenges Apart from many technical challenges, some of which are briefed in the previous section, AstroSat mission had its own peculiar managerial challenges also. Again, without going in to the details of all the challenges a few of them are listed in this section.

6.1 Collaboration of many agencies As told earlier, AstroSat is a fine example of collaborative effort involving many Indian and international agencies/institutes. It harnessed and synergized the expertise of the individual agencies. It brought international character to an astronomical observatory in space. It also brought some cost benefits to the project and the project was realized within the budget allocated. During the initial period of the project, there were wide differences of opinions among different teams. The authors’ firm opinion is that they are mainly due to the cultural differences between scientists/engineers from different institutes. The challenge of the project team was to see that they do not grow and act as moderators between teams. This was a delicate and strenuous job requiring lot of patience and balance of mind. This job, nevertheless, was meticulously achieved by the small project team. After AstroSat, handling of such collaborations have become much easier as the cultural differences were narrowed and each team understood them in proper

24

Page 8 of 10

perspective. AstroSat project, in a way, paved a path for managing larger collaborative missions.

6.2 AstroSat payload monitoring committee At some stage, luckily almost in the beginning phase of cutting metal for the payloads, it was felt that monitoring and reviewing of the payloads is not going to be a simple task. Recognizing this, AstroSat Payload Monitoring Committee (APMC) was constituted, under the chairmanship of Dr. George Joseph, Former Director of Space Application Centre, ISRO. This committee had eminent scientists as members. The credit for excellent reviews, controlling and managing the schedules, overseeing the testing and calibration goes to this committee. Constituting ‘Payload Monitoring Committees’ has become a norm in ISRO for subsequent science missions.

6.3 Contamination control committee The need for contamination control has already been briefed in Section 5.7. Though the requirement was known, very little was known about the methodology to achieve it in a systematic manner. In order to streamline this activity, a new committee (this type of committee did not exist for any previous missions of ISRO) was constituted. This committee has done a wonderful job of budgeting the contamination down to each subsystem and taking stock of bill of materials in terms of quantity used and their contribution. The contamination budget was done for the first time. The committee also chalked out procedures for controlling the contamination at various levels starting from component to launch of satellite. This definitely was a new learning and done meticulously. The proof is in seeing the performance of the instruments in space and more evidently from the measurements carried out on the witness samples on ground. Needless to say, some of these practices have now become routine in the U R Rao Satellite Centre.

6.4 Managing the observation proposals The first six months of the AstroSat observations were designated as ‘Performance Verification’ (PV) phase. The next six months were ‘Guaranteed Time’ (GT) phase. A year later, regular observations started with

J. Astrophys. Astr. (2021)42:24

reduced time for GT. The observation time allocation is based on the merit of the proposals, Percentage of time allocated for various groups of observers and technical feasibility. The users are divided in to seven groups: (i) (ii) (iii) (iv) (v) (vi) (vii)

AstroSat instrument teams, Canadian Space Agency, University of Leicester, Indian astronomers, International astronomers, Targets of Opportunity (ToO) and Calibration.

Each group is allocated with a fixed percentage of observational time. As the time allocation and studying the feasibility of observation are quite involved, the need for managing this is quite evident. In order to manage this efficiently and effectively two committees, AstroSat Time Allocation Committee (ATAC) and AstroSat Technical Committee (ATC) were formed. Submitted proposals were reviewed and approved by the ATAC with the help on technical feasibility assessment provided by the ATC. ATC considers payload operation constraints like avoidance of bright objects, constraints on source location with respect to Sun, Moon, etc, and source detectability in various instruments. Based on these approvals, the mission and operations team at U R Rao Satellite Centre and ISRO Telemetry, Tracking and Command Network (ISTRAC) creates the necessary list of targets to be observed. This proposal management system worked very well over the past five years and continues to work with enhanced enthusiasm.

7. Establishment of new facilities Many new ground systems and facilities have been established under the scope of AstroSat mission. A couple of important facilities are listed here.

7.1 Indian Space Science Data Centre AstroSat is operated as Space Observatory. The data received from the ground station is to be validated, archived, stored and disseminated. The data is to be transferred to the corresponding groups as listed in Section 6.4 after a certain level of processing. The user groups process the data further and post it back to the repository. In order to cater to this requirement on a routine basis an Indian Space Science Data Centre

J. Astrophys. Astr. (2021)42:24

(ISSDC) is established under the management of ISTRAC, ISRO. This data centre maintains the lock-in rules provided to the instrument scientists and the proposers. After the lock-in period the data is made available all the interested scientists including university students. Both raw data and the processed are accessible for any user, either on dedicated lines or through internet. ISSDC, now, has grown in stature. Apart from AstroSat, it is catering to all science missions of ISRO including Chandrayaan, Mars Orbiter Mission and Meghatropiques.

7.2 M.G.K. Menon Laboratory for Space Sciences IIA has set up a world class facility, with ultra-high clean rooms including a class 100 room, for assembly and testing of optical instruments especially UV optics. This was necessary for the development UVIT payload of AstroSat. This facility is in Hosakote, now outskirts of Bengaluru. This facility is declared as a national facility. All future optical instruments, where cleanliness is of importance, can be assembled here and tested using the test equipment available. With further improvements by IIA, work is already in progress, for the development and testing of Visible Emission Line Coronagraph (VELC) payload for Aditya-L1 satellite.

8. Results The proof of success of any mission can be evaluated, to a large extent, by means of the results achieved. By this standard, AstroSat mission achieved a grand success. More than 1500 scientific papers/announcements are published just prior to the completion of five years of the satellite launch. Out of this, at least 150 papers are in refereed journals. Currently, AstroSat has nearly 1500 users from 48 countries around the globe. It is heartening to note that half of them are from India. These users also include, academicians and students. To the best of our knowledge, 15 PhD theses, so far, are based on AstroSat data. Over 900 unique fields were observed by AstroSat over the last five years. There are numerous headlines, some of them are sensational like ‘‘India’s AstroSat makes rare discovery’’, in both print and electronic media in India and abroad. Many of them surely are/will be reported by means of scientific papers.

Page 9 of 10

24

9. Conclusion AstroSat mission is conceived, designed, managed, executed and operated satisfactorily. The mission was managed without any cost over runs in spite of a little time over run. The technical, managerial and technomanagerial skills employed in this mission are unique for Indian Space Science mission. It is a great experience to the project team with several lessons learnt. Some of these skills, tools and techniques may be useful for next generation of scientists, engineers and others if they can adopt, improvise and employ them. AstroSat, though designed for an operational life of five years, is in still good health. Hardly any fuel is used for the orbit maintenance., AstroSat is expected to serve for many more years to come, subject to no component failures. The mission accomplished all its objectives. It is heartening to note it energized the Indian astronomy community and kindled the interest of science in many young students.

Acknowledgements We thank the entire AstroSat community (the list is very long) including scientists, engineers, managers, operators, planners and the supporting staff for contributing their mite wholeheartedly for the success of AstroSat mission. We also thank the Principal Investigators, Prof. P C Agrawal and Dr. S Seetha, for their major contribution to the scientific activities. We are, indeed, deeply indebted to the managements of ISRO, TIFR, IIA, IUCAA, RRI, CSA, and UOL without whose support this successful mission would not have been possible achieve.

References Kasturirangan K., Rao U. R., Sharma D. P. et al. 1976, Nature, 260, 226 Koteswara Rao V., Agrawal P. C., Sree Kumar P. et al. 2009, Acta Astronautica, 65, 6 Kumar A., Ghosh S. K., Hutchings J. et al. 2012; Proc. SPIE 8443, Space Telescopes and Instrumentation, Ultraviolet to Gamma Ray, 84431N, https://doi.org/10. 1117/12.924507 Morrissey P., Schiminovich D., Tom A. et al. 2005, Astrophys. J., 619, L7 Navalgund K. H., Sarma K. S., Gaurav P. K. et al. 2017, J. Astrophys. Astron., 38, 34, https://doi.org/10.1007/ s12036-017-9455-8 Rao U. R. 1978, Proc. Indian Acad. Sci., C1, 117

24

Page 10 of 10

Rao A. R., Bhattacharya D., Bhalerao V. B. et al. 2017, Curr. Sci., 113, 595, https://doi.org/10.18520/cs/v113/ i04/595-598 Seetha S., Megala S. 2017, Curr. Sci., 113, 579, https://doi. org/10.18520/cs/v113/i04/579-582 Singh K. P., Tandon S. N., Agrawal P. C. 2014, Proc. SPIE 9144, Space Telescopes and Instrumentation, Ultraviolet

J. Astrophys. Astr. (2021)42:24 to Gamma Ray, 91441S, https://doi.org/10.1117/12. 2062667 Singh K. P., Stewart G. C., Westergaard N. J. et al. 2017, J. Astrophys. Astron., 38, 29, https://doi.org/10.1007/ s12036-017-9448-7 Yadav J. S., Rao A. R., Agrawal P. C. et al. 1999, Astrophys. J., 517, 935

 Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:41 https://doi.org/10.1007/s12036-020-09681-4

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

MISSION

Contamination control of UVIT S. KATHIRAVAN1,* , S. N. TANDON2, B. RAGHAVENDRA PRASAD1,

S. SRIRAM1, A. PRADEEP1, T. VISHNU1, P. K. MAHESH1, P. U. KAMATH1, S. NAGABHUSHANA1 and AMIT KUMAR1 1

Indian Institute of Astrophysics, Bangalore 560 034, India. Inter University Centre for Astronomy and Astrophysics, Pune 411 007, India. *Corresponding Author. E-mail: [email protected] 2

MS received 29 October 2020; accepted 6 December 2020 Abstract. Ultra Violet Imaging Telescope (UVIT) is one of the 5 instruments on AstroSat satellite, which was launched on September 28, 2015. UVIT was designed to make images with a resolution of \1:800 , simultaneously in two ultraviolet channels: Far Ultraviolet (130–180 nm) and Near Ultraviolet (200–300 nm). Images are also made in visible region (320–550 nm) for tracking drifts in pointing. The shortest wavelengths to be observed with UVIT can be heavily absorbed by mono-molecular deposits/contamination on the optical surfaces. Keeping contamination under control in UVIT was a major challenge and it required a variety of actions: (i) strict control of the payload materials and process, (ii) mechanical configuration, (iii) baking of all the parts to release all the adsorbed molecules etc., (iv) assembly in ultra cleanrooms, (v) pre-inspection and auditing of all the areas, in which UVIT was placed, for any potential for contamination, (vi) continuous purging, with ultrapure nitrogen gas, till a few days before the launch, etc. In order to minimise any possible cross contaminations from the other payloads/satellite, the doors of UVIT were opened 2 months after the launch. The high performance in the orbit and high stability of the sensitivity over 4 years in the orbit shows that the contamination was negligible. This paper presents the processes and protocols followed during the integration and testing phase to minimise the contamination in order to prevent any performance degradation. Keywords. Molecular contamination—ultraviolet optics—space optics—transmission measurements— ultra high pure (UHP) nitrogen purging.

1. Introduction This paper gives the general cleanliness and contamination control requirement for UVIT payload. It also brings out the overall contamination control requirements during the design, fabrication, assembly, integration, testing and launch of the UVIT on AstroSat spacecraft and the methods to achieve the required cleanliness, in order to ensure a successful mission. Further, it also gives the contamination estimation methodology for particulate as well as molecular contamination for UVIT payload,and the methods adopted for limiting contamination throughout all This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

phases of the mission have been presented. The methods to estimate and monitor contamination, laboratory investigations, clean-room and hardware monitoring have also been addressed. This paper covers the importance of controlling contamination for UVIT, elements of contamination control, mechanical configuration, selection of materials for sub systems, Ultra-pure nitrogen (99.999% pure) purging, optical witness window monitoring plan, facility used for integrating and testing the instrument, instruments used for monitoring contamination,and contamination monitoring plan. Contamination control during storage, transport, environmental tests, and all other operations till launch are also discussed. A contamination budget for different stages is also presented.

41

Page 2 of 12

2. General overview of Ultra Violet Imaging Telescope (UVIT) UVIT is composed of two nearly identical focussing telescopes of 375 mm aperture, one for far ultraviolet (FUV) (130–180 nm) wavelengths and one for near ultraviolet (NUV) (200–300 nm) and visible (VIS) (320–550 nm) wavelengths. The second telescope has a dichroic mirror to split the light into NUV and VIS channels. The field of view for both telescopes are circular with diameter 280 . The FUV, NUV and VIS channels each have a number of filters with different band passes, as described in Tandon et al. (2017). UVIT was developed through a collaboration between several Indian institutions: IIA, ISRO, IUCAA, TIFR and Canadian Space Agency. AstroSat has five instruments, covering NUV and FUV with the Ultra Violet Imaging Telescope (UVIT) telescope and soft through hard X-rays with the SXT, LAXPC, CZTI, and SSM instruments. For details please refer to the paper by Agrawal (2017). A drawing of UVIT is shown in Fig. 1 and a photograph of AstroSat is shown in Fig. 2.

3. Contamination Contamination may be simply defined as any foreign matter. In general, contamination is grouped into two broad categories labelled as molecular and particulate.

Figure 1. UVIT model.

J. Astrophys. Astr.

(2021) 42:41

Molecular contamination refers to the cumulative build-up of individual molecules of foreign matter. Molecular contamination may occur during ground processing and in the orbit. Particulate contamination refers to the deposition of visible conglomerations of matter. These particles, which are deposited mainly during ground operations, will transfer from the air on to exposed surfaces or be left as residues after some process.

3.1 Importance of contamination control for UVIT The performance of the UVIT payload can be affected due to particulate as well as molecular contamination. Studies reveal that many of the molecular contaminants are orders of magnitude more absorptive in the ultraviolent than in the visible or infrared. For this reason, UV payloads are much more sensitive to molecular contamination. For 130 nm wavelength, a \1 nm thickness of many materials can reduce transmission/reflectance by an order of magnitude.In addition to absorbing the signal, molecular contamination may also cause an increase in scattering from the mirror surface. Both of these effects may give rise to additional noise, and decrease the signal noise ratio (SNR) Similarly, any particulate contamination which resides on a mirror or a focal plane array would prevent the optical element from transmitting the signal and reduce its signal strength in proportion to the percentage area coverage (PAC). In addition to surface obscuration, which effectively reduces signal strength, the presence of particles on primary mirror can induce other consequences like scattering into an

Figure 2. Assembled AstroSat.

J. Astrophys. Astr.

(2021) 42:41

Page 3 of 12

41

Table 1. UVIT contamination sensitive components. Component FUV primary and secondary mirrors CaF2 filter BaF2 filter Sapphire filter Silica filter FUV detector

Specification

Wavelength

375 mm dia. and 150 mm

130–180 nm

50 9 2.5 mm thick 50 9 2.4 mm thick 50 9 2.0 mm thick 50 9 2.7 mm thick 300  300 25 micron CMOS star 250 imager

156 nm 158 nm 160 nm 170 nm 130–180 nm

optical train which could be disastrous for high contrast imaging.

using door of the telescope tube as sun-shield and by control of pointing.

3.2 UVIT contamination sensitive components

4.2 Selection of materials and processes

The components of UVIT which are most sensitive to contamination are listed in Table 1.

There are primarily seven categories of materials and components used in the payload:

4. Elements of contamination control The approach for the UVIT design was to minimize the sources and effects of contamination through hardware design, material selection and process such as Isolation of the mirrors, detectors/filters/high voltage units and electronics, avoidance of organic materials/glues as far as possible, finishing of the surfaces to minimise dust collection and easy washability, testing of materials and components like cables/ heaters, the baking and cleaning with verification tests for the specific materials. 4.1 UVIT configuration Each telescope is sectioned in 3 parts with a view to minimise cross contamination. Thus, there were three sections of each telescope which are isolated from each other: (1) Focusing optics consisting of the primary and secondary mirrors within its tube (2) The detectors (with front end electronics and high voltage supply) and filters within an enclosure behind the tube (3) The main electronics which is outside the preceding two sections. In addition to these precautions, any possible leakage of direct solar radiation or solar radiation scattered by earth on the mirrors is avoided as far as possible by

• • • • • • •

metals coatings optical filters glues potting compound mirrors cables, cable ties and heaters

Of these above seven classes, coatings, glues, potting compound, cables and cable ties are potential sources of contamination as these are likely to carry organic materials. 4.2.1 Metals and processes. All the metal components were fabricated with smooth surface finish. After receiving the parts from the workshop, these were cleaned to remove particulate and molecular contamination using HEPA (High Efficiency Particulate Air) filter vacuum cleaning, solvent wiping (soap solution and acetone), ultrasonic cleaning with soap water and ultrasonic cleaning with commercial grade acetone. After ultrasonic cleaning components were given two dips in de-ionised water to remove the traces of the surfactant and dried in class 100 drying chamber. Finally, the components were wiped with 99.9% pure spectroscopic grade iso propyl alcohol (IPA) After cleaning the parts were packed in the cleanroom environment with black PE bags (which were wiped with IPA) as primary cover and wrapped with multi layer cushion materials to take care of transport accelerations. Packed items were transported to other facilities in a paper carton box for inorganic black

41

Page 4 of 12

coating which was developed by ISRO (Indian Space Research Organisation) for this payload. The black-coated parts were again cleaned with acetone in ultrasonic bath in low power mode and given two dips in de-ionised water to remove traces of the surfactant and dried. The black coating was inorganic and it was developed by ISRO for avoidance of contamination. After the final cleaning/drying, the part is examined in bright and in black-light (UV) to check for any leftover deposits etc. If any contamination were found, the process was repeated till it is certified as clean. Next, the metal parts were baked in baking chamber for 24 hr at 100 C in high vacuum (\104 mbar) to drive out the molecular contamination. After the bake, degassing is checked, with thermal quartz crystal micro balance (TCQM) kept at –10 C, to be less than 1  107 g cm2 hr1 . The chamber has two ports for thermal quartz crystal micro balance (TQCM) and residual gas analyser (RGA) to monitor the contamination levels and species. After baking the components were tagged as precision cleaned and after this very great care was taken to avoid contact with notclean surface of any kind. Baked components were stored in class 1000 area with two layer covers to protect from particulate contamination till it got integrated with the payload. The primary cover was UHV grade aluminium foil and the second layer was black poly ethylene (PE) cover. 4.2.2 Non-metallic materials (e.g. adhesives, potting elements, cable ties etc.). As far as possible only those materials were selected which have a collected volatile condensable mass (CVCM) to be less than 0.01%. The materials were further screened by ensuring that a 24 hours exposure in vacuum of a MgF2 -window witness sample (kept at roomtemperature) to the material (heated to 100–120 C) does not reduce its transmission, at 150 nm, by more than 5% of the original value. In most cases the sample was preconditioned by heating in vacuum at 100–120 C for 24 hours before exposing witness sample to it. In case the material cannot be heated to 120 C, a lower temperature was chosen and duration of the exposure were increased by a factor of 2 for each 20 C decrease in the temperature. After cleaning the parts were preconditioned by baking in vacuum as described above and packed in Electro Static Dissipative (ESD) multi-layer covers

J. Astrophys. Astr.

(2021) 42:41

and stored in class 1000. Small components were stored in dedicated SS containers. 4.2.3 Optics (e.g. mirrors, UV filters, detectors etc.). For control of particle contaminants, mirrors and optical components were handled in clean and monitored environment of cleanliness class 100 or purged with ultra-pure nitrogen. Witness samples of MgF2 windows (see Section 6.2.1 for more details) were used to assure that control of the environment are maintained during shipping, storage, groundoperations, and testing. The purge plans had the flow rates that will maintain some positive pressure inside the optical cavity to avoid any environmental ingression. Uncoated optics were ultrasonically cleaned before bonding with mirror cells. The glue used to bond the optics to the mechanical cell was tested and cleared for contamination. The glued optics were vacuum baked to remove any molecular contamination deposits. Figure 3 shows the UVIT primary mirror in a special transport container with the provision for optical witness coupons All the optical parts were stored in containers purged with ultra-pure nitrogen. Witness samples of MgF2 -witness samples were kept in the containers for monitoring any contamination leading to reduction in transmission at 150 nm of these samples. In general, at least two witness samples were used as a set: one for monitoring differential contamination, say every 15 days called periodic witness coupon, and the other to monitor total contamination over the period of storage called cumulative witness coupon. 4.2.4 Mechanisms (e.g. door and filter wheel mechanism). The mechanisms were designed to minimize particle generation and molecular out gassing, and have provision for purging. Design features included the use of materials with low out gassing, and certification with thermal vacuum bakeout. Only approved, low-out gassing lubricants, glues, and potting compound were used. The finished parts were preconditioned by baking in vacuum as described in Section 4.2.2. After the preconditioning, where possible, the parts were approved after ensuring that the vacuum baking test with MgF2 -window witness sample met the criterion for transmission as described in Section 6.2.1. The baked parts were packed in black PE bags sealed with ulta-pure nitrogen and stored in clean desiccators.

J. Astrophys. Astr.

(2021) 42:41

Page 5 of 12

41

purging. Witness samples of MgF2 windows (see Section 4.3.2 for more details) were used to monitor cavity of the instrument as well as the exiting purginggas, to assure that control of the environment were maintained during shipping, storage, groundoperations, and testing. The purge flow rates were ensured to maintain some positive pressure inside the optical cavity to avoid any environmental ingression.

Figure 3. UVIT primary mirror in container.

4.3 Integration and assembly During the assembly and integration of the instrument the key elements for elimination of contamination were use of suitable cleanrooms and purging with ultra-pure nitrogen. Figure 4 shows the purging arrangement of UVIT detectors during integration at MGKML. 4.3.1 Ultra-pure nitrogen purging. UVIT instrument vent-path, door, and aperture configuration and locations were controlled to exclude molecular contaminants during ground-operations and on-orbit, and to provide an environmentally controllable zone in the interior of the instrument. Storage and shipping containers for individual optical components and subsystems were designed to provide clean, controlled, and monitored environments with ultra-pure nitrogen

Figure 4. Purging of UVIT during integration.

4.3.2 Optical witness window monitoring plan. Optical witness samples (MgF2 -window) were used as incremental indicators of condensed material when direct sampling is not possible. Witness samples were accompanied each mirror through all phases, from initial fabrication/coating to integration with spacecraft and launch. Optical witness samples (OWS) were analysed for transmission by UV at baseline and after exposure, and the difference in transmission/reflectance was calculated and used as an indicator of the presence of condensed molecular contaminants from the sampled environment. Optical witness samples were also used in purge-systems, thermal test cycles, and thermal vacuum bake outs as appropriate for monitoring and/or out gassing verification. In particular, the witness samples were also used to monitor purity of the outgoing purging gas, which samples environment inside the cavity being purged. In general two witness samples made a set: one being used to monitor changes in transmission after each stage of operation and the other being used to monitor total change in transmission through all the stages starting from cleaning/receiving to the latest stage when the sample could be taken out. The transmission of the window in UV range (120–180 nm) was measured and compared with its original transmission. Three bands of average wavelengths 130–140 nm average,160–170 nm average and 130–180 nm average were used for contamination monitoring. 4.3.3 Facility for integration and calibration. Ultra cleanroom starting from class 100 to class 3,00,000 were required to cater various activities involved in the assembly, integration and calibration of UVIT. They require 99.9997% efficient ULPA filters, temperature control 21–22 C and relative humidity 30–60% with positive room pressure of 5 Pa minimum and 20–200 air changes per hour. UVIT components were assembled, integrated, verified and stored at Prof. MGK Menon Space Science Laboratory (MGKML) to meet the contamination budget as mentioned in the Table 2 and Table 3.

41

Page 6 of 12

J. Astrophys. Astr.

(2021) 42:41

The lab is designed and built as per ISO 14644-1, 14644-2 and ISO 14644-4 standards consists of Class100 to Class-300000 cleanrooms. It is on par with international space instrumentation facilities with an approximate total area of of 400 m2 . Latex/nitrile gloves were used for handling, assembly and installation of coarse/fine/precision cleaned parts. Gloves were changed when proceeding to handle components at different stages of processing. Frequent IPA washing of gloves were ensured while handling flight hardware. Tools and fixtures which may contact cleaned parts in assembly or transport were cleaned ultrasonically in IPA solution and air-dried. Figure 5 shows the picture of FUV telescope and NUV telescope during integration at class 100 area.

or further assembly were stored in bags or wrapping material made of contamination-free UHV grade aluminium foil and electro static discharge (ESD) PE covers. Small parts were stored in stainless steel or glass containers, cleaned and prepared in the same way as vacuum equipment. During down times when payload is not actively being worked on, or for weekends and other non-operational times, it was covered with black removable PE cover with purging on. The use of removable covers provide protection during storage, transportation and waiting for work period.

4.3.4 Storage and integration The optics were stored in desiccators that are pressurized with ultrapure nitrogen to meet consistent cleanliness and low humidity levels. Processed parts awaiting installation

Budgeting of contamination at major mile stones ensured that the cleanliness of the optics and instruments will remain within designated optical requirements for operations in space.

5. UVIT contamination budget

Table 2. Particulate contamination budget for UVIT. Estimated time

Activity

Air cleanliness

Assembly of NUV/VIS telescope Assembly of FUV telescope Thermal management work Satellite adapter integration

Class 100 Class 100 Class 1000 on component level Class 100/Class 1000 with both telescopes Class 1000 Class 1000 Under Cover Class \30000a Class \30000a

Thermal work on both telescopes Alignment and checks on twin telescopes Packing and transportation Vibration and Thermo vac @ ISRO Facilities Satellite integration Total a

Estimated particle fall out (mm2 =m2 )

65 days 52 days 50 days 4 days

130 104 800 40

30 days 5 days 4 days 45 days 60 days

150 50 Under cover 3000 4000 5000

Payload covered.

Table 3. Molecular contamination budget for UVIT.

Telescope NUV/VIS Primary mirror NUV/VIS Secondary mirror FUV Primary mirror FUV Secondary mirror

Storage at MGKML

Assembly at MGKML

Transport to ISRO

Storage assembly at ISRO

Storage at SHAR till launch

Total molecular contamination budget

\5% \5% \5% \5%

\5% \5% \5% \5%

\5% \5% \5% \5%

\5% \5% \5% \5%

\5% \5% \5% \5%

\10% \10% \10% \10%

J. Astrophys. Astr.

(2021) 42:41

Page 7 of 12

41

cleanroom environment. In Fig. 6, 1 represents the 45 mm  45 mm glass plate in which 2 is the 15 mm circular area used for contamination measurement. PFO plates were used to monitor periodically for the surface coverage of the plates and evaluated by a PFO meter. After a fixed time (24 hours) the sample plates were collected and analyzed by PFO meter. In general, the levels are expressed in parts per million (mm2 =m2 ) Table 4 gives an approximate correlation between cleanroom classes and particle fallout (PFO).

Figure 5. UVIT telescopes during integration.

5.1 Particulate contamination budget Particulate contamination budget is given in Table 2 based on the estimated time of each activity, cleanliness of the environment and obscuration rate of the witness coupons (Particle Fall Out (PFO) plates defined in Section 6) which were kept very close to the primary mirror during that activity. Estimated particle fall out rates (PFO) were calculated using the relation: PFO ¼ 0:069 ðNcÞ0:72 ;

ð1Þ

where Nc is number of particles with respect to the cleanliness class. The coverage of 5000 mm2 =m2 is acceptable as for a total of 5 surface in the optical train it gives a total loss of \5% . 5.2 Molecular contamination budget Molecular contamination budget is given in Table 3 in terms of reduction in transmission in the witness samples as described in Section 4.3.2. Wave length range for transmission and reflectance loss is 130–180 nm average.The accuracy of measurements of transmission for the witness samples is a few percent and a limit of less than 5% for any individual stage could lead to false negatives. A limit on total transmission of 10% implies a reduction of 20% in sensitivity and this is considered acceptable.

6.1.2 Laser particle counter. On daily basis particle counts were taken by particle counters to verify the cleanliness in all the areas. Along with particle count other vital cleanroom parameters like temperature, relative humidity and positive pressure were also verified and recorded.

6.2 Molecular contamination monitoring Figure 7 shows the picture of different witness coupons used at MGKML. Apart from that combination of a TQCM and a mass spectrometer were also used for the identification of different condensed species. 6.2.1 Optical witness windows

(1)

MgF 2 witness window. The material of the optical window used as witness sample is magnesium fluoride (MgF2 ). These windows were exposed to the material or environment or equipment or the place where critical activities take place in the cleanroom laboratory under different conditions. As the contamination is very sensitive to UV photons, the degradation in transmission measured on optical windows quantifies the possible contamination of the material or the environment where the window is exposed.

6. Methods and euipments used for contamination monitoring 6.1 Particulate contamination monitoring 6.1.1 Particle fall out method. Particulate fall out (PFO) plates shown in Fig. 6 were used to monitor the

Figure 6. PFO plate.

41

Page 8 of 12

J. Astrophys. Astr.

(2021) 42:41

Table 4. Correlation of airborne particles and PFO for cleanrooms. Airborne particle cleanliness class 100 1,000 10,000 100,000

PFO (mm2 =m2 ) 2 10 52 275

Figure 8. 50 mm reflective mirror coupon with primary mirror.

sensitive hardware to monitor contamination levels of the environment. Before and after the exposure, reflectivity measurements were done by a photometer. Any change in the original values of reflectivity will indicate the contamination potential of the environment. Figure 7. Optical witness coupons (MgF2 windows and mirror coupons).

The window was exposed to the material, which can degas contaminating molecules or particles for certain period under certain condition like under vacuum or vacuum baking. The material degassed from contaminant deposits over the sample. The transmission of the window in UV range (130 nm to 180 nm) was measured and compared with its original transmission using a spectrophotometer. The degradation in transmission gives the measure of potential of contamination of the particular material. In MGKML the windows were used for finding the possible contamination of environment like cleanrooms, vacuum chamber, optics storage containers, ultra pure nitrogen lines, thermal baking chambers etc., spectrophotometer and thermal desiccators were the test setups being used for measuring and monitoring contaminations. There are two sets of MgF2 windows were kept near critical optics for periodic and cumulative monitoring. The results of periodical and cummulative MgF2 measurements are presented in Section 10. (2) Optical mirror coupons. Small size mirror coupons with coating similar to primary mirror were used as witness mirror coupons as shown in Fig. 8. Mirror Coupons were exposed near the contamination

The results of periodical mirror coupon measurements are presented in Section 10.

6.3 Thermal desiccators setup Thermal desiccators are the glass vacuum chambers (150 mm diameter) evacuated with vacuum pumps used to find out the contamination potential of a material is shown in Fig. 9. This chamber was placed over heating furnace and can be heated to 120 C. The material (sample), which can out-gas the contaminant, is kept inside the chamber and the witness sample was placed inside the chamber close to the sample. The witness sample was mounted on an aluminium rod which projected out of the chamber and its top was kept at room temperature. The chamber was evacuated to 104 mbar pressure and heated to 120 for 24 to 48 hours depending on the out gassing of the material. The molecules come out of the sample deposits over the witness sample. The witness sample were taken out and its transmission was measured in the spectrophotometer.

6.4 Thermal quartz crystal micro balance (TQCM) method The method of measurement is based on a quartz crystal piezoelectric characteristics. All the non-metallic

J. Astrophys. Astr.

(2021) 42:41

Page 9 of 12

41

Figure 9. Thermal dessicator setup.

components were checked for the TQCM values during baking and cleared for payload integration. Every time empty chamber is baked and TQCM values were verified before new component loading. Df : ð2Þ Sensitivity of the crystal ¼ ðDm=AÞ

Figure 10. Packed UVIT loaded on to the base of the inner container.

If the out gassing requirement of the component was not met, it was subjected to further long bake out.

7. Packing and transport of UVIT The payload was packed with multi PE covers which were wiped with IPA in class 1000 area as shown in Fig. 10. A two-layered container was designed for transportation of UVIT and the materials were carefully selected to avoid any particulates/molecules. Inner container which encloses UVIT was made of stainless steel sandwich material to maintain clean environment. The payload was completely packed inside the container, closed in the clean environment and purged with ultra purity nitrogen, as shown in Fig. 10. Then the inner container was rolled in to outer container and fixed with the outer container base with wire rope isolators which reduces the transportation shock loads on UVIT which is shown in Fig. 11. Outer container protects the inner container from the dusty outside ambience during the transport and online shock watch monitors were used to monitor acceleration levels of UVIT during transport.

Figure 11. UVIT packed in inner container.

table was cleaned with IPA and checked for cleanliness visually before UVIT integration. The UVIT was covered and purged with ulta-pure nitrogen all the times except during testing. Figure 12 shows the picture of UVIT on the vibration table with purging. During the thermo-vacuum tests all the protocols for cleaning and monitoring contamination, which were used during the assembly and testing earlier, were followed. Figure 13 shows the loading of UVIT into the thermo vauum chamber for thermal cycling test. The same precautions and procedures were followed during satellite tests with integrated UVIT.

8. Vibration and environmental test facilities contamination control

9. Contamination control at launch site

It was ensured that the particle counts are less than 5000 in the facility when UVIT is exposed for vibration and thermo-vacuum tests. The shaker

Contamination control measurements were taken at different cleanrooms at launch site before the arrival of AstroSat. Witness windows were employed for

41

Page 10 of 12

J. Astrophys. Astr.

(2021) 42:41

and at the time of heat shield closure, witness windows fixed in the thermal cover of UVIT were removed and purging also was removed.

10. Summary of transmission and reflectance loss in witness windows in FUV (130–180 nm) 10.1 Transmission loss in MgF2 windows

Figure 12. UVIT on vibration shaker.

Figure 13. UVIT loaded in to thermo vacuum chamber.

molecular contamination monitoring and particulate levels were checked for acceptance. Purging was continued till the satellite is integrated with launcher

Calibrated MgF2 windows were introduced to monitor the contamination levels immediately after the primary mirrors were received at MGKML. These MgF2 windows were accompanying the FUV primary mirror during all the process of testing and integration. Once the mirror was integrated with the telescope, these witness windows were fixed in the FUV telescope door and the optical cavity was purged with ultra-pure nitrogen. After the payload was integrated with satellite, the MgF2 window and purging were re located to thermal cover, which is located at base of the payload and encloses the cavity in which the detectors and the filters are mounted, for easy accessibility. The details of results are summarised in Table 5. Please note that errors are not mentioned; the estimated error on each individual measurement is 2%. Transmission loss of the MgF2 periodic witness windows which accompanied FUV primary and secondary mirrors from July 2011 to January 2015 are given in Fig. 14 and Fig. 15. Other MgF2 coupons accompanied NUV and collimator telescopes followed the similar pattern with transmission loss of \1% in 130–180 nm average. 10.2 Reflectance loss in MgF2 windows Reflectance loss on the primary and secondary mirror witness coupons which accompanied FUV telescope

Table 5. Transmission loss at various stages. Location MGKML ISRO cleanroom SP1B cleanroom at launch site SP2 cleanroom at launch site MST cool air In UVIT optical cavity

No. of days of exposure

Transmission loss in 130–140 nm

Transmission loss in 130–140 nm

2200 155 25 3 2.7 812

2% \2% 6.6% 3.9% 1.3% 1.3%

\1% \1% 3.5% 1.9% 0.5% 1%

J. Astrophys. Astr.

(2021) 42:41

Page 11 of 12

41

Figure 14. FUV primary mirror. Figure 16. FUV primary mirror periodic.

Figure 15. FUV secondary mirror. Figure 17. FUV secondary mirror periodic.

from April 2011 to August 2015 are given in Fig. 16 and Fig. 17. Reflectance loss on the primary and secondary mirror witness coupons which accompanied NUV telescope from May 2011 to December 2014 are given in Fig. 18 and Fig. 19.

11. Results and discussion The transmission loss observed on various windows during integration and testing phase of UVIT is given in Table 5. At MGKML the transmission was less than 1% for almost 2200 days. The windows which were kept in the UVIT optical cavity was removed 9 days before the launch had transmission loss of 1.3%. Maximum of 7.8% reduction was noticed on the reflective mirror witness coupons which accompanied FUV primary mirror. This could be largely because of ageing effect of FUV coating and contamination is negligible as otherwise it would have shown in MgF2 windows which was kept next to it. The on-orbit measurements do not show any measurable degradation of the sensitivity in 130–180 nm range over 4

Figure 18. NUV primary mirror periodic.

Figure 19. NUV secondary mirror periodic.

41

Page 12 of 12

years. For more details, please refer Additional Calibration of UVIT (Tandon et al. 2020).

12. Conclusion Stringent contamination control perspective, UVIT on board AstroSat which was the 1st Indian space mission to include a demonstration of the effectiveness of the systems contamination control plan. The cumulative results of both particulate and molecular contaminations were in support of the plan’s effectiveness. UVIT contamination control plan was well thought out and implemented throughout the design, fabrication, assembly, testing and integration of the payload with spacecraft. The effectiveness of the plan was also measured by the on orbit sensitivities of the instruments while performing their scientific objectives.

J. Astrophys. Astr.

(2021) 42:41

Acknowledgements UVIT project was a collaborative effort of IIA (Bengaluru), IUCAA (Pune), TIFR (Mumbai), and ISRO from India, and CSA of Canada. The performance obtained was a result of hard work over many years by engineers and scientists from these institutions.We thank Prof. Ashok Pati for his contribution in designing the layout of MGKML.

References Agrawal P. C. 2017, J Astrophys. Astr., 38, 27 Tandon S. N., Subramaniam A., Girish V. et al. 2017, AJ, 154, 128 Tandon S. N., Postma J., Joseph P. et al. 2020, AJ 159, 158, arXiv:2002.01159

 Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:62 https://doi.org/10.1007/s12036-021-09730-6

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

MISSION

Innovations in the AOCS of AstroSat spacecraft JASVINDER SINGH KHORAL*, R. ADITYA, HARISH JOGLEKAR, PRASHANT KULSHRESHTHA, MANOJ KUMAR, DEEPA PRAMOD and A. M. NAGALAKSHMI U R Rao Satellite Centre (URSC), Old Airport Road, Vimanapura Post, Bengaluru 560 017, India. *Corresponding author. E-mail: [email protected] MS received 4 November 2020; accepted 6 February 2021 Abstract. AstroSat is India’s first mission dedicated to space-based Astronomy and carries a complement of instruments sensitive over a wide spectral region covering visible, ultraviolet, soft X-ray and hard X-ray bands. The spacecraft platform provides the required stability, pointing, power, thermal control, and interface requirements. Among the critical functions of the Attitude and Orbit Control System (AOCS) were, manoeuvre the spacecraft from one target source to another in an optimal way while avoiding the sun, compensate for the rotational disturbance torque created by the scanning of the SSM payload on the other payloads, primarily the UVIT payload, provide for calibration of all the payloads, which involved manoeuvring through 2-dimensional target patterns. All the above novel control functions, along with the routine house-keeping operations, were designed, developed, and tested extensively on ground, before declaring them as flight-worthy. The payload performance and the resulting science are a testimony to the excellent performance of the on-board Control System. Keywords. Attitude and Orbit Control System—On-Board Computer—Sun avoidance manoeuvre— disturbance compensation—payload calibration—rotation mechanism—stepper motor device.

1. Introduction AstroSat is a unique space observatory satellite with simultaneous multi-wavelength continuous observations of cosmic sources over UV (FUV, NUV, VIS) spectrum and soft and hard X-ray bands which has never been attempted in any earlier similar purposed space craft across space faring nations.

Attitude: 0.05 Drift rate: 5.0e-04/s Jitter: 0.5 arcsec/s, frequencies [2 Hz The control system consisted of sensor and actuators along with the On-Board Computer (OBC) that housed the various control algorithms and logics. By definition, jitter frequencies lie outside the S/C control bandwidth and hence its effects/mitigation are addressed by appropriate design of the structural elements.

1.1 Attitude and Orbit Control System (AOCS) The main objective of the AOCS is control of orientation of the spacecraft while minimizing the drift rate about each of the three axes. This objective was codified as a specification on the following parameters: This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

1.2 Control system The attitude control was implemented as a PD controller with a Kalman filter for estimation of the attitude, rate, and disturbance torque. The controller bandwidth was of the order of 0.005 Hz, because of the fine pointing requirement. The controller has a feed-forward torque feature to accelerate the time

62

Page 2 of 8

needed for manoeuvring the spacecraft. The Kalman filters are designed for estimation of attitude and rate and are designed based on the measure of noise specified by the sensors.

1.3 Sensors An Inertial Reference Unit (IRU) measures the inertial angular rates, sun sensors and star sensors measure the inertial attitude, and magnetometers are used for measuring the earth’s magnetic field.

1.4 Actuators A set of 4 reaction wheels are used for fine attitude control and 8 thrusters for coarse attitude control, seldom used, to avoid contamination. A set of three magnetic torquers are used for momentum management of the reaction wheels.

2. Spacecraft manoeuvres and control We provide in this section the various types of manoeuvres planned on the AstroSat. This is followed by subsections providing the performance results of each of these. Next, we discuss the compensation for the SSM induced disturbance on the spacecraft and normal mode control for imaging.

2.1 Sun Avoidance Manoeuvre (SAM) The S/C is manoeuvred from one source to another quite frequently and since the payload optics is quite sensitive to the sunlight, the rotation should ensure that the Sun is away from the payload boresight during this manoeuvre. Mengali and Quarta (2004) and Hablani (1998) describe complex algorithms to reorient a S/C while avoiding bright objects during the manoeuvre. However, these methods were not suitable for on-board implementation and therefore a novel manoeuvre scheme exploiting the S/C configuration (payloads pointing along ?Roll axis and ?Yaw axis), constraints (avoid sun exposure on a 45 cone about ?Roll and in the entire ?Yaw hemisphere) and attitude definitions (?Roll pointing to source etc.) were developed ab initio (i.e., without recourse to any similar work done elsewhere). The main inspiration for the algorithm implemented on board AstroSat was

J. Astrophys. Astr.

(2021) 42:62

derived from a series of simulated sources located on critical points in the S/C Roll-Yaw plane for AstroSat. The insights gained from these ‘‘virtual’’ sources provided the critical clues to design the SAM. The sun avoidance algorithm generates an attitude profile that meets this constraint. It also ensures that the wheel momentum is within safe limits. 2.1.1 Performance of SAM. The first SAM was carried out successfully on 15-Oct-2015 with Yaw rotation of 170. Figures 1, 2 and 3 below, show the sun vector with respect to the S/C body axes, wheel speeds and body rate during a typical SAM operation. The performance of the SAM was as designed and met all the mission/payload requirements/constraints.

2.2 Payload calibration manoeuvres All the payloads were calibrated against the known sources in the initial phase of operations. The concept of this calibration was to make the reference source to fall on various points on the camera and use the registered data for calibration. This needed the spacecraft to rotate about yaw and pitch axes and trace well define patterns. Some calibrations needed constant rate scans. This calibration was a compact algorithm that could trace any arbitrary pattern and worked effectively. The rate scan was based on a novel idea of pure rate control while propagating the attitude in background. All the calibration exercises worked very well in space. 2.2.1 Performance of payload calibration manoeuvres. The CZTI was the first payload to be calibrated. This was a pure attitude calibration. Subsequently rate scanning was carried out for LAXPC and the calibration algorithm worked flawlessly and has been carried out many times to trace different patterns for all other payloads, like UVIT. The rate scanning accuracy is typically 0.010/s.

2.3 SSM disturbance torque compensation The SSM is the only rotating payload on-board the AstroSat. It rotates by 10 in 10 s and stops (stare) at the new position for 10 minutes and this is repeated. The pulsing of this mode is at 8 ms frequency. During the rotation it disturbs the other payloads. A disturbance compensation scheme was designed to reduce the transient response time of attitude and rate. The disturbance is modelled and the net momentum due to

J. Astrophys. Astr.

(2021) 42:62

Page 3 of 8

62

deployment of the solar panels. The Reaction Wheels were switched on in the second orbit and spun at 3000 rpm for two orbits. The star sensors were also switched on in the second and third orbit. In the 4th orbit, the Spacecraft was commanded to Reaction Wheel control with SSQ mode of feedback. The Star Kalman Filter and Rate Kalman Filter were initiated in the background. The Star Kalman filter estimates gyro drift and spacecraft attitude quaternion. The spacecraft was commanded to SKF mode after convergence of drift and attitude estimate. The performance of the SKF control mode was nominal.

3. On-Board Computer (OBC)

Figure 1. Sun in body frame during a large 170 SAM, notice that all constraints are met throughout the manoeuvre.

the rotation is compensated. A single torque pulse is applied at the start and at the end, the amplitude and duration of this pulse are based on the net momentum imparted due to the SSM rotation. This has significantly improved the transient response of the attitude. 2.3.1 Performance of SSM disturbance compensation. SSM was operated in from Oct. 29, 2015. Figures 4, 5 and 6 below show the performance with and without SSM compensation. The performance of the SSM disturbance compensation scheme was as designed and met all the mission/payload requirements/ constraints.

On-Board Computer (OBC) used in ISRO satellites achieves cost-effective spacecraft engineering by integrating the various elements of the spacecraft bus that realize the functions of attitude and orbit control, command processing, data acquisition and processing, telemetry, and thermal management into a single system. As OBC is the heart of the spacecraft, it is built as a robust system with features such as four converters to provide adequate redundancy, software with programmable and remote programming features, robust design with no single point failures and ability to go for a modular extension if required to meet additional system requirements. The above features have enabled OBC to meet the complex requirements from each payload which has resulted in excellent payload performance and the resulting science is a testimony to the excellent performance of the on-board Control System. Standard OBC systems available until AstroSat realization could cater to a wide range and fixed set of requirements but in AstroSat with a total of five payloads, catering to huge requirements was a challenge. Hence, apart from the standard designs, a project specific board was developed to cater to payload specific requirements and additional interfaces for the mainframe bus (Fig. 7).

4. AstroSat specific interfaces 2.4 On-orbit control

4.1 Payload data commands

AstroSat was launched from SHAR on September 28, 2015. The attitude stabilization was initiated after

The five payloads on AstroSat namely LAXPC, UVIT, SSM, CZT and SXT had to carry out various

62

Page 4 of 8

J. Astrophys. Astr.

(2021) 42:62

Figure 3. Typical S/C body rates during SAM. Figure 2. Typical wheel speeds during SAM.

operations and observations for which there was huge commanding requirements from OBC. OBC had to send out commands through serial data command interface so that payload datasets can be set by the payload interfaces for various complex operations. Though standard data commands were available to serve the mainframe, additional logic was built to decode command and telemetry requirements of different payloads.

4.2 Telescope door opening for UVIT and SXT UVIT has identical and co-aligned telescopes which collect celestial radiation and feed to the detector system via the selectable filters on the filter wheel drive mechanism. Similarly, SXT also has a soft X-ray imaging telescope which is used to study the absorption properties of X-ray sources. These two telescope doors had to be deployed after launch.

The hold down and release mechanism for the SXT and UV telescope doors consist of paraffin actuator (HOP) based hold down release mechanism. This actuator works on the principle of converting volumetric expansion of the paraffin wax to do mechanical work and release the hold down bolt of the door mechanism. 280 seconds was the estimated time calculated for the hold down release to act. OBC used a provision in commanding called the Variable Extended Pulse Width (VEPW) commanding for the same. By up linking the number of frames the VEPW logic inside OBC can be retriggered as many times as the uplinked frames to extend the pulse width of the command, and thus actuator can be heated for the estimated time. However, the number of frames had to be pre-decided on ground and the command execution would stop only after the uplinked frames would get over payload requirement of terminating the heating of paraffin actuator immediately after confirmation of hold down release was achieved by ground encoder at control centre. This mode of terminating the

J. Astrophys. Astr.

(2021) 42:62

Page 5 of 8

62

Figure 5. Spacecraft response without compensation.

Figure 4. Motor and SSM response.

command-by-command control at ground was used for the first time during AstroSat door deployment. 4.3 Dump resistor interface There was also a specific requirement with regards to battery used in AstroSat. Lithium-ion batteries of 36AH is used here which is prone to damage in case of an overcharge. In a scenario where the bus load becomes less, voltage from the panel goes up thus, overcharging the battery. This indication of the battery voltage going beyond a threshold value is available with OBC in telemetry. This overcharge had to be dumped, for which dump resistors catering for 100 W on each battery were provided on the spacecraft deck. For dumping this charge, a project specific interface circuit was realized. OBC software on meeting certain

pre-conditions operates two relays of the dump circuit which start dumping the 100 W of power for excessive charge of batteries. 5. Software aspects On-Board Computer (OBC) carries out functions such as sensor electronics control, TC and TM, attitude and orbit control, thermal management, ampere hour meter etc. It uses Mil-Std-1553B protocol for interfacing with other sub-systems of the spacecraft. 5.1 OBC software OBC software has fault tolerant features: safe mode, fault detection, identification and recovery, battery safety logic, soft emergency, remote programming. These are general features of the OBC software, across missions. However, the AstroSat OBC system is built with special logic to meet the needs of the astronomy instruments.

62

Page 6 of 8

J. Astrophys. Astr.

(2021) 42:62

Figure 7. Block diagram showing the control function of OBC.

Figure 6. Spacecraft response with compensation. START

5.2 Salient features of AstroSat OBC software Uplink Macro Definion through Remote uplink

Mission objective is to manage six different and independent scientific payloads, and all the design, autonomy and fault handling logics are built around them. These are addressed in the following sections.

Issue Macro Definion update command

Uplink Dataset Structure through Remote uplink

5.3 Macro based P/L sequencer Macro-based payload sequencer is implemented to take care of the operations of all 6 payloads. This new design approach allows to plan independently all the scientific instruments, and other main frame applications in a fully configurable manner. It has the following steps shown in Fig. 8 for each independent chain.

5.4 Payload time marker scheme Time correlations for each astronomical payload is required. A unique scheme is designed with the help of SPS Standard Time base generator and OBC’s Event interface. This time information captured from

Issue Payload Dataset update command

Issue Master Sequencer Enable command

Issue Macro Enable command with macro no

Issue Macro Init command which contains macro no and dataset no through OBT-TT

STOP

Figure 8. Typical payload sequencer plan.

J. Astrophys. Astr.

(2021) 42:62

various sources, is transferred as consistent pair in telemetry.

5.5 Payload calibration A simplified scheme was designed to meet the various requirements of Payload for calibration. For some payloads it was unique pattern, for some it was rate pattern. On-board software is designed to handle both in a generic way to cater to any future plans of Payload calibration.

5.6 Safety feature for LAXPC purification LAXPC payload uses Xenon as the filling gas, which is susceptible to electronegative impurities like oxygen, water vapour etc. Therefore, periodic purging of the xenon gas is required after launch to remove these impurities. Purification is initiated based on feedback from the payload teams on the quality of calibration source spectrum. A 100 Torr difference may be maintained between inlet and outlet. Purification is stopped after specified hours of purification and confirmation from the payload team again based on calibration spectrum. The LAXPC detector pressure (analog channel) is monitored by OBC and whenever the pressure falls below a certain value (100 Torr, this value could be different for each detector, command up loadable) purification will be terminated with three samples check of the pressure.

5.7 UVIT safety logics UVIT (UV Imaging Telescope) is sensitive to bright objects, wrong states of operations etc. In such conditions, UVIT needs close monitoring and appropriate actions to handle the situation to safeguard the instrument autonomously. There are 6 types of safety features incorporated. OBC software is designed to detect the safety warning issued by UVIT and associated logics for subsequent recovery.

5.8 Sky Scan monitor (SSM) platform rotation SSM is a scanning payload, mounted on the platform to carry out scanning operation. SSM platform is rotated using stepper motor. BMU software generates the

Page 7 of 8

62

desired scanning sequence and other platform motion operations. Major requirement of SSM was Step-Stare Operation: This is continuous operation; rotation takes place between specified Start Position and End Position. Stare operation takes place for the specified duration, then Step (rotation) takes place for specified Step Angle (combination of Mode-I and Mode III).

6. Rotation mechanism and drive electronics for SSM The steering mechanism consists of a stepper motor and a pancake type harmonic drive reducer. The motor is fixed onto the housing which is connected to the platform through a pair of diaphragm housing. The housing also supports the harmonic gear drive and resolver. The resolver is placed on harmonic output gear drive to obtain the absolute angular position. A micro-switch is provided for initial point indicator and a hard stop prohibits 360 rotations to avoid entangling of harness running from the three SSM detectors to the processing electronics package. The nominal range of positions between which the system is designed to rotate from two ends is 355. The SSM drive electronics package has been designed to drive the SSM platform with OBC providing the drive pulses for the stepper motor; nominal pulse width being 8 ms. The gear ratio is 1/157. The angle information received from the resolver is processed by the Resolver to Digital Converter (RDC) hardware and transferred to OBC for telemetry. On ground, the 1X and 8X resolver angles are processed to compute fine-angle of the platform position after applying required correction. Look up table (LUT) for resolver raw angle correction was generated after extensive testing of the motor by recording the resolver angle and absolute position angle through servo table and theodolite. Initial tests showed that resolver along with processing electronics was having repeatable error pattern up to ?/–8 arcmin with respect to the servo table angle as shown in Fig. 9. Hanselman (1991a, b) describes requirements for high accuracy Resolver to Digital Converters and the techniques for improving resolverto-digital conversion accuracy. Based on the error data, cubic spline-based curve fitting was carried out to generate an error LUT. After LUT correction, resolver angle accuracy was brought within ?/–0.5 arcmin, as shown in Fig. 10, thus improving SSM platform measurement accuracy.

62

Page 8 of 8

Figure 9. Resolver measured angle vs. error and cubic spline curve fitting.

J. Astrophys. Astr.

(2021) 42:62

from Position 1 (5) to Position 2 (355) and counterclockwise (reverse) rotation from Position 2 to Position 1. Three modes of operation include: (a) stare and step, (b) open-loop and (c) parking. nominal mode of operation is stare-and-step where in the platform moves to a specified start-position, stays there for specified stare-time, then steps by specified step-angle to next position; this continues till specified end-position after which the direction of rotation is reversed, and operation is continued. The nominal stare-time is 10 min, step-angle 10, starts with Position 1 (5) and end with Position 2 (355). Until stop command is issued the platform continues to be in stare-and-step mode between these start and end positions. Openloop mode is also provided to rotate by a specified angle and stop to allow SSM to undertake longer observation of sources of interest. Parking command brings the platform to Position 1 (5). SSM has achieved required observation by using all the above modes as and when required.

7. Conclusion After the completion of 5 years of AstroSat it is safe to conclude that • All AOCS features have been exercised and have worked very well meeting the control objective of providing a stable platform • All OBC features (hardware and software) have worked as per required specifications • A few lessons have been learnt, especially w.r.t the impact of star sensor measurement gaps and these are addressed in future missions

References

Figure 10. Resolver corrected angle error vs. motor step angle.

6.1 Nominal mode of operation The sense of rotation as one looks into the rotation axis (along ?Yaw) is clockwise (forward) rotation

Hablani Hari B., AIAA Course on Spacecraft and Interceptors Guidance, Navigation and Control, Chapter 13, 13-167–13-190 Hanselman D. C. 1991a, IEEE Trans. Ind. Electron. 38, 501 Hanselman D. C. 1991b, IEEE Trans. Ind. Electron. 37, 6 Mengali G., Quarta A. A. 2004, Spacecraft control with constrained fast reorientation and accurate pointing, Aeronaut. J., 85–91

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:21 https://doi.org/10.1007/s12036-021-09731-5

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

MISSION

AstroSat mission operations management AMIT KUMAR SINGH* , M. DEEPAN, LEO JACKSON JOHN and B. N. RAMAKRISHNA ISRO Telemetry Tracking and Command Network, Peenya Industrial Area, Bangalore 560 058, India. *Corresponding author. E-mail: [email protected] MS received 4 November 2020; accepted 11 February 2021 Abstract. AstroSat is India’s first dedicated astronomical observatory in space with multi-wavelength payload on a single platform. It enables simultaneous observations in the desired wavebands. The increase in the number of payloads has also led to an increase in the complexity in space segment design, ground segment design, and mission operation management. Each payload instrument and mainframe has its own constraint for the operation, which needs to be satisfied to ensure the safety of the systems. In this paper, we explain the challenges in ground operations for mission management with various constraints. Also, we list the various constraints, both geometric and otherwise, with respect to the mainframe systems and payload instruments of AstroSat. Keywords. AstroSat—mission operation—on-board constraint.

1. Introduction To study the cosmic sources, satellite-based space observatories provide us a better vantage point than ground-based observatories. This is because measurements made by space-based observatories are free from errors arising due to atmospheric effects. Since the intensity of most cosmic sources varies with time, the variability being wavelength-dependent, it is necessary to make simultaneous observations in different wavebands. However, there are several logistical problems in making simultaneous and coordinated studies of a specific object from different satellites and groundbased telescopes. As a result, despite many such multiwavelength observation campaigns, very few sources have so far been studied over wide spectral bands leading to poor understanding of the underlying physical processes. With these problems in mind, a dedicated satellite called AstroSat was designed with multiple co-aligned instruments covering the desired spectral bands so that simultaneous observation in all the desired wavebands is possible. This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

AstroSat is designed and developed with five astronomy payloads (UVIT, LAXPC, SXT, CZT, SSM) covering optical, Near-Ultraviolet (UV), FarUV, and X-ray frequency bands and was launched on 28th September 2015 by PSLV-C30 (XL) from Sriharikota, India. The spacecraft was put into a 650 km near-equatorial orbit with an inclination of 6 degree to meet mission objectives.

2. Mission operations and planning AstroSat is an inertially pointed satellite with the instrument view axis oriented towards the celestial source for observations in multi-wavelength. The AstroSat payloads are designed to sense the faint signal from the distant objects and are capable of identifying the signal source at arc-second angular separation. Conforming to the above requirement resulted in design of payload system which are very sensitive to the signal intensity, attitude deviation and jitters on the platform. This requirement imposes the constraint on on-board control system and mission operation management.

21

Page 2 of 7

Different payloads of AstroSat, operates in different frequency bands and each payload has its own operating constraint. To ensure the safety of on-board instruments, it is of utmost importance to verify that the commands, that are up-linked to the satellite conform to the constraints of all the payloads and mainframe sub-system. The validation of the geometric constraints for payload operation starts from the proposal submission on-wards till the generation of final command list, at different levels by different software with different time period and accuracy level. The mainframe subsystem constraint is validated by operations team before uploading of commands to the satellite. The AstroSat payload operation is based on a proposal driven system. The proposal submission starts six months in advance with Announcement of Opportunity for users. The users first check the feasibility of their intended source using the Astroviewer portal, which check the observation feasibility for a duration of up to one year over specified start and end time. After the feasibility check with respect to geometrical constraint, the proposal is submitted for observation through the AstroSat Proposal Processing System (APPS). The Submitted proposal is validated for science merit by AstroSat Time Allocation committee (ATAC), based on recommendations on technical feasibility by the AstroSat Technical Committee (ATC). The final list of the proposal for next six months/one year is stored into a Mission Control and Proposals (MCAP) database for further processing. The next level of scheduling of proposal happens for fifteen days in sliding window pattern. This task is executed by the AstroSat Scheduler Software for operations (ASSORT) with latest orbit parameter and geometric constraints validated with finer accuracy. A Payload Operation Plan scheduler uses the aforementioned files and generates time lined command sequence outputs for a pre-planned observation. The scheduler creates an observation sequence plan with constraints check using orbit related events such as eclipse, South Atlantic Anomaly (SAA) entry/exit, Earth occult entry/exit, X-Band data dump feasibility, station visibility, star sensor availability, earth limb brightness viewing etc.. The corresponding commands are generated and posted in the operation area for uplink. The operations team runs the final validation (Mamidi et al. 2020) for payload and mainframe constraint check before uplinking of the command. The constraint validated by the operation team is explained in detail in next section. When observation

J. Astrophys. Astr.

(2021) 42:21

is in progress, the two critical constraints namely Inertial sensor drift and separation angle between the Sun and spacecraft payload axis are monitored at each orbit interval to meet the mission specifications. Corrections are carried out from ground whenever the observations deviate from the desired limits.

3. Constraints in mission operations The AstroSat is a three axis stabilized spacecraft. the axis definition of Astrosat is shown in the Fig. 1. The proposal which are selected for the observation are validated for conformance to the following geometrical constraints: (1) Payload axis should never see the bright object in their respective operational wavelength (visible, near UV, Far UV ) during operating condition. (2) Payload axis and the Sun vector angle should not be less than 65 degrees. (3) Payload axis and Velocity vector angle should not be less than 12 degrees. (4) Cross axis-1 should avoid sun by 100 degrees. (5) Cross axis-2 should maintain 90?/–0.5 degrees with Sun. The celestial sources which satisfy the above constraints are selected for observation for corresponding payload planning cycle. Since, the attitude of the observatory is held fixed in inertial space throughout the observation, it causes payload axis and attitude sensor bore sight axis to observe the Earth every orbit. The average blockage duration by the earth per orbit varies based on the location of the celestial source in inertial space. The source located near equatorial plane cause the maximum blockage duration of approximately 35 minute per orbit. The sources with declination of greater then 66 degrees will not face payload axis blockage by the Earth. The X-ray payloads are safe during earth occult duration even though they do not observe the source, while the optical payload is sensitive to scattering due to Earth’’s albedo and needs to be put OFF whenever payload axis to Earth albedo angle goes below 12 degrees. AstroSat has two star trackers for attitude measurement mounted with sufficient separation angle to ensure attitude measurement from at least one sensor during observations. The field-of-view of the Star trackers need to be away from Sun, Moon and Earth albedo by certain angles to provide measurement. Hence while planning orientation for any source, the

J. Astrophys. Astr.

(2021) 42:21

Page 3 of 7

21

Figure 1. AstroSat axis definition.

Star tracker availability needs to be predicted in advance and necessary actions to be taken on ground accordingly. The part of the orbit when star trackers are occulted, inertial sensors (gyros) control the platform. Gyros are having inherent drift, which needs to be estimated on ground, based on post facto attitude analysis and corrected periodically to meet the pointing requirement of the payload. Since the AstroSat attitude is inertially held throughout the observation period, it becomes necessary to validate the number of stars available in FoV of star tracker for each observation. Only those sources are planned for observation for which at least one star tracker is available. Most of the AstroSat payloads are operated at high voltages during observation time. The X-ray payloads are prone to failure when the spacecraft passes through the South Atlantic Anomaly (SAA) region, where charged particles are also detected, leading to very high count rate which can lead to high voltage breakdown. AstroSat crosses the SAA region every orbit. In order to safeguard the systems from high energy particles, payloads are operated with low voltage configuration during SAA region and those data are not used for scientific purpose. The UVIT payload is very sensitive to light intensity, which puts operating constraint during the sunlit part of the orbit. To conform to this requirement, UVIT payload is always planned in eclipse, non-SAA region and Earth occult free zone. All payload instruments are continuously ON storing the data in on-board memory. On-board has the

memory capacity of 144 Gb. The expected data to be recorded per orbit is around 40 Gb. The recorded data is downloaded every orbit over dedicated ground station situated in Bangalore. As per the data volume generated per orbit and on-board memory storage capacity, the data download opportunity exist for nearly 3 orbits. These payload data are very precious and unique especially in case of Astro Events and is of high importance. Since these data cannot be recreated unlike in IRS satellite, where multiple opportunities exist to capture the same area. therefore, to avoid the overwriting of useful data, the quality of downloaded data needs to be checked for every download orbit. In case the data quality is below acceptable threshold, redump needs to be planned within certain time periods. The re-dump operation is highly unpredictable and to be performed by the operations team based on realtime observation. AstroSat has two-phased array antennas providing omni-coverage for X-band data download. One of the antenna provides a clear FoV and there is partial blockage for the second one by on-board appendages. The blockages are predicted on ground a priori and mitigation plan is executed by the operations team to ensure error free data download.

4. Mission operation challenges and management The AstroSat provides a platform for the user community to perform simultaneous observation of the celestial source in multi-wavelength band. This system was realized by increase in the number of on-board payloads,

21

Page 4 of 7

which has also led to an increase in the complexity in space segment, ground segment and mission operation. Each payload instrument and mainframe has their own constraint for the operation, which need to be satisfied to ensure the safety of the systems. The AstroSat payload operation related commands are uploaded on-board one day in advance, after the validation of all the constraints at different level. There are certain critical operations which need to be performed by operation team in the real-time pass during the observation period. These operations depend on the response of payload instrument and mainframe system for that observation source and corresponding attitude. The requirement and frequency of these operations is decided by the post facto analysis. In the following sections, the operation scenario and challenges are discussed in further detail. 4.1 On-board programming constraint Though AstroSat payload planning is done 15 days in advance by the scheduler, the final command sequences cannot be uploaded at once due to multiple operational limitations. The payload operation includes On-Board Time Tag (OBT) based commands related to various events such as enabling/disabling of Star tracker updates, SAA Entry/Exit, Data download, X-Band On/Off, attitude maneuver sequence commands, Solar panel Offset, UVIT On/Off and other configuration commands. The maximum number of such OBT events which can be stored on-board is limited to 512. Since the average number of required events uploaded to spacecraft over a day is around 250, this calls for daily upload of commands and cannot be programmed for more than 2 days, definitely not weeks in advance. Apart from planned observations, there is provision to submit Target of Opportunity (ToO), with short notice. This also puts limitation for long term on-board programming. 4.2 On-board timer drift Since, the AstroSat observation are very time critical and on-board operation depends on OBT timer. It becomes necessary to maintain the drift of the on-board timer within mission specification, which calls for frequent OBT drift estimation and correction from ground. Since the drift is directly related to the temperature of the on-board clock generator, the frequency of this operation varies based on the source orientation.

J. Astrophys. Astr.

(2021) 42:21

4.3 Long-term observation planning constraint In AstroSat, the source pointing attitude is derived based on the Sun position at start of the observation. Subsequently, all the geometric constraints are validated against this attitude and AstroSat is inertial pointing to the defined attitude during full observation period. Since, AstroSat has stringent constraint limit of ?/–0.5 degree variation for the cross axis-2 with respect to the Sun, and the apparent motion of the Sun is 0.9856 per day, the observations which are planned for long duration, i.e. more than two days puts overhead on ground operations. Hence frequent attitude corrections maneuver are needed to compensate the apparent Sun motion. This operation is done for observation duration exceeding 130 kilo-seconds and corrections are to be performed when no other sequencer or UVIT operation or payload data dump is in progress.

4.4 Gyro drift correction Attitude pointing requirement for the UVIT payload is very stringent and any deviation may lead to switching OFF of the instrument by on-board safety logic. The gyro drives the platform in absence of Star tracker update and is prone to drift causing attitude deviation. During such periods, the drift is estimated at the ground and correction commands uploaded. This is a regular operation in AstroSat and frequency of operation varies depending on the orientation in inertial space.

4.5 Payload reset operation To meet the requirement of periodical UVIT instrument reset which are time critical, ground planning needs to be done to ensure that no UVIT related events are programmed during this operation. Also, critical monitoring is required after every such reset to assess the instrument status and take appropriate actions for any deviations.

4.6 Critical health monitoring AstroSat has stringent real-time health monitoring requirement for mainframe and payload systems. In every orbit due to SAA crossing, the payload

J. Astrophys. Astr.

(2021) 42:21

instruments transit from high voltage to low voltage and vice versa and these transition events are preprogrammed from ground. Quick action needs to be taken from ground during the limited contact time with the satellite (maximum of around 15 minutes) in case of any abnormal behavior. The assessment of proper functioning of mainframe systems to satisfy the mission specification during real-time pass and take corrective action for further operation is very time critical task, mainly the attitude control and power systems. This is the herculean task to be performed by operation team in real-time pass. To reduce the realtime health monitoring load, OOL (Out Of Limit) based anomaly detection software is deployed in operational environment. The expected limit of the parameter has been configured based on the on-board system performance during the initial phase of the mission. This software has played the significant role in safeguarding the satellite.

Page 5 of 7

(2) (3)

(4)

(5)

21

favourable for the maneuver, planning of new source to satisfy the maneuver requirement through Sun Avoidance Maneuver (SAM). Planning of the orbit maneuver and creating corresponding command sequence. Disabling of all the planned operation in on-board memory which are clashing with new maneuver time. Execution of the collision avoidance maneuver operation as per the plan provided by the Flight Dynamics team. Normalization of the routine operation.

This operation disrupts the full chain of the routine sequence, which needs to be restored after the execution of the maneuver operation. Till date, one collision avoidance maneuver is executed for the AstroSat mission.

4.9 Anomaly and contingency handling 4.7 Operation management during lunar shadow The maximum load on the bus is around 1.6 kW and maximum solar power generated is around 2 kW. In case of lunar shadow which occurs few times in a year, the power generation reduces and appropriate load shedding operation is planned and executed from ground to maintain positive power margin.

4.8 Operational overhead As explained in earlier sections, the final planning of the payload operation and corresponding commands sequence are uploaded to on-board, one day in advance. As per the mission definition and selected orbit requirement, no orbit raising maneuver is required throughout the mission life of AstroSat. But to safeguard the satellite from debris, collision avoidance maneuver (Singh et al. 2014) is to be planned as per the requirement. The collision avoidance maneuver warning comes to the operation team in 12-hour window and which is mandatory operation. In view of the collision alert, orbit maneuver operation needs to be planned and executed within 3 orbits, which involves: (1) Finding the suitable time for maneuver for the ongoing source with respect to thrust axis to velocity vector angle. In case if angle is not

The operation team is responsible for health monitoring of the spacecraft and detection of any deviation in on-board system. The OOL based real-time health monitoring softwares are deployed in operation area to detect any deviation from the expected range. If any anomaly alarm is observed during the real-time pass, the spacecraft controller will validate the alarm, initiate preliminary action to safeguard the spacecraft and alert the concerned team for further actions and normalisation. This operation involves coordinated effort of the sub-system designer, mission team and operation team. As part of post recovery operation, based on the implication of the anomaly, on-board configuration and operation strategy are modified to avoid future occurrence of same class of anomaly.

5. Software developed for operation management AstroSat operation is unique in nature and different from other on-orbit satellites. The management of such complex and dynamic mission operation scenario with manual decision making is a Herculean task for the operation team. To handle all these monitoring and operational requirement various software modules are developed by the operations team and implemented for regular operations. The utilities developed facilitate the operation team with an independent system to perform following operations:

21

Page 6 of 7

J. Astrophys. Astr.

(2021) 42:21

Figure 2. AstroSat operations portal data flow diagram.

(1) Validate (Mamidi et al. 2020) the planned operation with respect to the defined mission constraint. (2) Generates the visual display of the planned operation timeline. (3) Keeps track of up-link events and validate the execution of operation at desired time through telemetry data. (4) Checking re-dump requirement and generates command for the same. (5) Estimate the inertial sensor drift and generates the correction command using offline telemetry data and in-build algorithm. (6) Prediction of attitude maneuver requirement and time of correction operation for each planned observation. (7) Maintains the database of daily payload operation and on-board events. (8) Provides the user interface to access the payload statistics through intranet. This data is made available to the operations team in graphical form and tabular form like event clash,

regime of operation planned, efficiency of the payload operation carried out, duration of the payload and regions covered. Figure 2 represents the overall data flow of the operations portal. The three core modules of operations portal are as follows: (1) The validation and verification of payload based on the constraints as discussed earlier are carried out by a software module. All the events are represented by the different colored bars. the overlapping of the events can be visually verified with ease by the spacecraft controller in addition to automated constraints verification. (2) The validated and verified information about payloads are then collected and transferred through FTP to the server network. This collection of information happens periodically at defined time and the database is updated after verification. (3) The database is accessed through a web interface internal for ISRO users for report generation and analysis.

J. Astrophys. Astr.

(2021) 42:21

Page 7 of 7

21

6. Conclusion

References

This paper was brought out to emphasize the challenges involved in the mission operation management. In spite of various constraints and limitations, operating the spacecraft as per the plan has been achieved with an aim to provide maximum science data to the users.

AstroSat Online Resource. https://www.isro.gov.in/Space craft/AstroSat AstroSat Proposal Submission System [online]. Available at: https://www.issdc.gov.in/astro.html Mamidi A. V., Singh A. K., Seth A., John L. J., Madaswamy S. and Ramakrishna B. N. 2020, On-Orbit Geometric Constraints Computation and Conformance for Astrosat Class of Spacecraft, 2020 IEEE International Conference on Electronics, Computing and Communication Technologies (CONECCT), Bangalore, India, pp. 1–4. https://doi.org/10.1109/CONECCT50063.2020. 9198531 Singh A., Srivastava V. K., Harish K., Chandrasekhar B. S. et al. 2014, Orbit Maneuvers for In-plane Altitude Corrections for IRS Satellites—Collision Avoidance, Conference: National Conference on Space Debris Management and Mitigation Techniques-ISRO-HQ, pp. 3–4, available: https://www.researchgate.net/publica tion/ 262488558

Acknowledgements We would also like to extend our appreciation to ISTRAC operations team including spacecraft controller, ground station operation teams, ISSDC team for their sincere effort in managing the operations in 24  7 time line. We would be remiss to not mention the substantial support received from mission team of U. R. Rao Satellite Center, other ISRO centers and payload instrument teams.

 Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:26 https://doi.org/10.1007/s12036-021-09728-0

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

MISSION

Special calibrations during AstroSat mission operations towards meeting observation and pointing requirements K. SUBBARAO, N. HARI PRASAD RAO, V. MAHADEVAN, R. PANDIYAN and J. S. KHORAL* U R Rao Satellite Center, Bengaluru 560 017, India. *Corresponding author. E-mail: [email protected] MS received 23 November 2020; accepted 6 February 2021 Abstract. AstroSat, India’s first astronomical observatory was launched on 28th September 2015 by PSLV into an orbit of altitude 650 km and 6o inclination. The simultaneous observations made by the payloads allow imaging of celestial source in various wavebands and has resulted in scientific achievements published in various fora. AstroSat is designed to carry 6 (5 ?1 auxiliary) astronomy payloads namely, UVIT, LAXPC, SXT, CZTI, CPM and SSM. The payloads are mounted such that the boresight of all the 4 main payloads except SSM are along the ?Roll direction in body frame. To meet all requirements of different payloads carried on-board AstroSat, various mission planning, analysis, and operations specific to this mission are carried out. These include post launch observations and mitigations planned and implemented onboard. Keywords. AstroSat—alignment—calibration—pointing.

1. Introduction The AstroSat payload configuration is shown in the Fig. 1. It can be observed that all the 4 main payloads except SSM are mounted on spacecraft deck such that the boresight of payloads is along the ?Roll axis of the body. Amongst all payloads the most stringent pointing and stability requirements are imposed by the UVIT payload. Table 1 shows the pointing and stability requirements to be met by the spacecraft. The scientific objectives call for stringent requirements in terms of orbit selection, attitude definitions and attitude stability, when compared to other satellites. To meet the stringent pointing and stability requirements, it is necessary to improve the payload alignment, mounting measurement and the alignment stability to withstand launch vibrations. Towards this a unique approach is followed for measuring the mounting matrices of each payload and their coalignment angles, refer Sedlak et al. (2003) for details of calibration techniques. This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

In this paper, we bring out the various calibration activities carried out on orbit to estimate the alignment of each payload with respect to spacecraft axes and also characterisation of the spacecraft mainframe, mainly in terms of star sensor and gyros to meet the pointing and stability.

2. Main frame characterization Post launch, spacecraft is acquired in three axis stabilized mode with Star sensor, gyro and reaction wheels in loop. The attitude followed in this spacecraft is inertial pointing with selected payload bore sight pointed towards the Source and to meet the constraints as shown in Fig. 2. When the S/C is pointed towards the source, sun should be constrained from ?Roll to -Roll axis about –Yaw axis with minimum angle with ?Roll by 65 deg and –Roll by 60 deg. The ?ve Roll constraint is from both payload and sensor, whereas the –ve Roll constraint is placed in order to ensure the thermal environment is stable and within limits irrespective of the source to which the

26

Page 2 of 7

J. Astrophys. Astr. + ROLL Axis

- Yaw Axis ( EP-03 Deck)

UVIT

LAXPC

SXT CZT

-Pitch Axis (Solar Panel On EP-04)

SSM

+Pitch Axis (Solar Panel On EP-02) - ROLL Axis

+Yaw Axis (EP-01 Deck)

Figure 1. AstroSat payload configuration. Table 1. Pointing and stability requirements. Spacecraft orientation

Defined inertial pointing

Absolute pointing Drift rate (\ 0.2 Hz) Max satellite rate

\0.05 deg (3r) \3.0910–4 deg/s ±0.24 deg/s (Maneuvering) \0.3 arcsec \5.0910–5 deg/s

Jitter ([0.2 Hz) Design goal for body rates

-Yaw 30 deg

-Roll

Gyro calibration is carried out in order to estimate the scale factor, misalignments and drift, such that in absence of star sensor updates to platform, the attitude of the spacecraft is maintained within the specifications. The scale factor and misalignments could be estimated to the accuracy of star sensor (\40 arcsec). However, gyro drift variations are observed during mission and as the attitude pointing is inertial pointing with no rate on the body, drift stability is more essential in this mission to achieve the pointing accuracy over the period of source observation. As the attitude is held to source pointing inertially, star sensor gets occulted by Earth during every orbit around Earth and hence it is very much required to maintain the residual gyro drift values within 0.05 deg/hr. The drift variation is sensitive to the base plate temperature and it is around 0.05 deg/hr/oC. Hence it is required to control the gyro base plate temperature within ?/–1 for all conditions. As the design is not meeting the stringent requirement, a strategy is worked out to update the gyro drift values based on the temperature observed and a procedure is established for doing this correction onboard. Figure 3 shows the gyro base plate temperature variation with the Earth vector (local vertical) angle with spacecraft body pitch variation. Whenever the body pitch comes close towards the Earth vector, due to albedo effect the gyro base plate temperature is rising and when body pitch is away, the temperature reduces. By studying this behaviour, a strategy is worked out by predicting the temperature variation w.r.t. this angle and drift corrections are carried out to meet the pointing requirements.

65 deg. 23 deg.

40 deg.

SS-2

(2021) 42:26

Roll SS-1

Yaw

Figure 2. Constraint for source pointing.

spacecraft is oriented. And also, the solar panels are oriented such that panel normal is also made to track sun through on-board solar panel drive mechanism, thereby ensuring the maximum power generation from the panels.

3. Payload alignment calibration On-orbit, based on the payload observations, the alignment of payloads is estimated with respect to body axes by keeping one of the star sensor’s mounting fixed with respect to the ground measured numbers and aligning the other star sensor to the first one. In order to estimate the alignment of payloads the following approaches are followed.

3.1 SXT alignment estimation In order to estimate the SXT alignment, the payload is pointed to source 2E 0102-7217 (February 2016) and

J. Astrophys. Astr.

(2021) 42:26

Page 3 of 7

26

Figure 5. SXT image post alignment correction.

Figure 3. Gyro base plate temperature variation. + PITCH (0, 599)

C (300, 300)

A (340, 266) B

- YAW (0, 0)

(599, 0)

Figure 4. SXT image coordinates with respect to body axes.

observed that the source is not at the centre of the image plane. Figure 4 gives the definition of pixel and line numbers in terms of spacecraft body axes and image coordinates. In the current attitude of ASTROSAT, the source is seen by the pixel A (340, 266). To move the central pixel C (300, 300) towards source, following rotations are required, • a –Yaw rotation (X axis of SXT) of 34 lines (140.08 arcsec rotation to move from C to B) and

• a ?Pitch rotation of 40 pixels (Y axis of SXT) (164.8 arcsec rotation to move from B to A) After the correction of –140 arcsec about X- axis and 164 arcsec about Y-axis (X, Y are axes in SXT image plane) are carried out and with the updated alignment, again the SXT payload is made to observe the same source. With this updated mounting, when the pointing is carried out to the same source, Figure 5 clearly shows that the image of the source of observation carried out on 10 February 2016 falls closer to the centre of the image plane. With this correction, we have almost achieved the centre pointing for the SXT! On 01 May 2016, along with SXT, combined observations are planned with main instrument as CZTI, source GX 301-2 with RA: 186.656 and DEC = –62.77 deg. Two sources are observed by the SXT payload with the following details and the SXT image plane coordinates are marked as follows: Source-1: RA: 186.656 deg, DEC = –62.77 deg; PX = 268 and PY = 314, Source-2: RA: 186.656 deg, DEC = –63.099 deg; PX = 77 and PY = 104, where PX and PY are the coordinates in SXT image plane. By using the post-dynamic alignment matrix of SXT, the attitude of the spacecraft body from the telemetry data for the CZTI pointing to source GX 301-2 on 01 May 2016 and the two source coordinates that are observed by SXT, the following coordinates

26

Page 4 of 7

J. Astrophys. Astr.

(2021) 42:26

Figure 6. SXT observation.

Figure 8. SXT image coordinates for two sources with updated alignment

Source-2: RA = 186.656 deg, DEC = –63.099 deg; PX = 580 to 600 and PY = 400 to 430. When updated alignment matrix of SXT is used along with the telemetry data of attitude for the same observation, the following coordinates are expected in the SXT image plane (X, Y) shown in Fig. 8, The coordinates for the two sources are expected to be: Source-1: RA = 186.656 deg, DEC = –62.77 deg; PX = 310 to 340 and PY = 250 to 270, Source-2: RA = 186.656 deg, DEC = –63.099 deg; PX = 535 to 565 and PY = 440 to 460.

are expected in SXT image plane (X, Y) shown in Fig. 6 From Fig. 7, the coordinates for the two sources are expected to be:

From these (Figures 6 and 8) observations, it is clearly understood that there is a rotation of around 180–6 = 174 deg about the boresight axis which needed to be accounted. The coordinates PX and PY in the SXT image plane for the above sources with the above correction are expected to be as shown in Fig. 9. From Fig. 9, the coordinates for the two sources are expected to be:

Source-1: RA = 186.656 deg, DEC = -62.77 deg; PX = 340 to 380 and PY = 240 to 200,

Source-1: RA = 186.656 deg, DEC = –62.77 deg; PX = 320 to 350 and PY = 260 to 290,

Figure 7. SXT image coordinates for two sources with post dynamic alignment.

J. Astrophys. Astr.

(2021) 42:26

Page 5 of 7

26

Figure 10. Convention followed during calibration.

Figure 9. SXT image coordinates for two sources with updated alignment.

Source-2: RA = 186.656 deg, DEC = –63.099 deg; PX = 50 to 80 and PY = 115 to 140. The coordinates shown in Fig. 9 and the picture of the observation given in Fig. 6 are in close match. The following is concluded on SXT alignment usage: • Whenever the prime instrument is not SXT, this matrix can be used in the pipe line and observations can be correlated. • Whenever the prime instrument is SXT, the rotation about boresight does not affect, however if any secondary source like above one is observed then the coordinates have to be interpreted in a different way.

3.2 UVIT alignment estimation In order to estimate the UVIT alignment with respect to body axes, a 9-point calibration scheme is used, in which the same source is made to point at 9 locations in the UVIT image plan for specified exposure time before moving the spacecraft from one point to another. Figure 10 shows the convention followed during the 9-point calibration. Figure 11 shows the planned vs. observed 9-point calibration profile before the misalignment correction.

Figure 11. Planned vs. observed 9-point calibration profile for FUV and NUV channels.

In Fig. 11, suffix ‘f’ denotes FUV (Far Ultra Violet tube) and suffix ’n’ NUV (Near Ultra Violet tube). The radius of the planned circle is 10 arc min and shown in yellow colour. The approx. circle with red line indicates the actual achieved points for NUV, the approx. circle with blue line indicates the actual points achieved for FUV. The centre computed by taking average errors of both FUV and NUV is (316, 268). With reference to this centre, a pitch error of ?60 pixels and yaw error of ?12 pixels is observed. Thus, for moving this centre towards ideal centre (256, 256), a negative pitch rotation of 60 pixels and a negative yaw rotation of 12 pixels is required.

26

Page 6 of 7

J. Astrophys. Astr.

(2021) 42:26

is applied on the post dynamic alignment matrix. It is observed in Fig. 12 that the corrected centre appears to be in-between the centres of the FUV and NUV circles.

4. Algorithm to compute the Right Ascension (RA) and Declination (DEC) of source from the SXT image plane coordinates (X, Y)

Figure 12. Corrected vs. observed 9-point calibration profile for FUV and NUV channels.

Figure 13. SXT frame orientation with respect to spacecraft body frame.

UVIT FoV is 28 arcmin and it is divided into 512 pixels, which implies 1 pixel = 3.28125 arcsec. Therefore, 60 pixels = 196.875 arcsec and 12 pixels = 39.375 arcsec. Thus, for correcting the misalignment, a negative yaw rotation of 39.375 arcsec and a negative pitch rotation of 196.875 arcsec is required. This correction

SXT image plane is defined by 600 9 600 pixels of each width 4 arcsec field-of-view. The total field of view of the SXT is 0.66 (which in other words is ?/– 0.33 about boresight). The origin of the SXT image plane has the coordinates of (300, 300). Any source which makes an angle of 0.33 with respect to boresight in the X-direction of SXT image plane will have a coordinate = (600, 300) or (0, 300). Similarly, any source which makes an angle of 0.33 with respect to boresight in the Y-direction of SXT image plane will have a coordinate = (300, 600) or (300, 0). Figure 13 shows the SXT frame orientation with respect to spacecraft body. • S/C body Roll and SXT Z are in same direction • S/C body Yaw is along the negative SXT-X direction • S/C body Pitch is along the SXT-Y direction • Let the coordinates of the source in SXT image plane by (x1, y1) • Then the following steps are followed to determine the source RA and DEC • HalfFovSxt = 0.33o • Let the body attitude in inertial frame be QInrToBody • Let QBodyToSxt be the Qs for transforming from Body to SXT frame. • Length = SQRT ((300 - y1)2 ? (x1 - 300)2) • Az = atan2((x1 - 300), -(300 - y1)) • El = Length * (HalfFovSxt / 300.0) • SourceInSxtFrame = [sin(Az) * sin(El) cos(El) - cos (Az)*sin(El)] • QInrToSxt = QInrToBody X QBodyToSxt • InrToSxtDcMatrix = Quaternion_TO_DC (QInrToSxt) • SourceIner = (InrToSxtDcMatrix)T * Src1InSxt • SourceEme = Convert_From_TrueOfDate_To_ Eme(SourceIner) • RA = atan2(SourceEme(2), SourceEme (1)) * (180.0/pi)

J. Astrophys. Astr.

(2021) 42:26

Page 7 of 7

• • • •

DEC = asin(SourceEme(3)/norm(SourceEme)) if RA\0:0RA ¼ 360:0 þ RA End Azimuth (Az) definition is from –Y to ?X in the SXT image plane with center at (300, 300) • Elevation (El) is defined as the angle from Boresight direction towards the X-Y plane of SXT Verification of the same is carried out with following case study: • Target: Mrk501, Date of observation: 2016-0815-05-33-49-294 UTC • Raw pixel position observed in SXT image: (300, 291) • Body attitude w.r.t Inertial = [0.9029055480 – 0.1978787120 -0.2308991880 0.3037456540] • Actual Source coordinates in EME frame: RA = 253.46 deg, DEC = 39.76 deg • The following are the results obtained using above algorithm:

Sl. No.

Matrix used

Source RA (deg)

Source DEC (deg)

1 2 3

Post-dynamic Updated Final

253.505 253.469 53.479

39.697 39.750 39.768

26

5. Conclusion With above payload alignment corrections uplinked onboard for improving the pointing and the algorithm for SXT payload during coordinated observations along with other payloads, the specifications of the payload are met. And the gyro drift correction procedure worked out made operational as explained is meeting the drift specifications (mainly for UVIT payload observations). Even after 5 years, the alignments provided for star sensors, gyro calibrations and gyro drift correction procedures are proving correct by meeting the payload pointing and stability requirements.

Acknowledgements We acknowledge R. Pandiyan, Mission Director and V. Mahadevan, Assoc. Mission Director, AstroSat, for the support provided during operations.

Reference Sedlak J., Welter G., Ottenstein N. 2003, Towards Automating Spacecraft Attitude Sensor Calibration, 54th IAC, Bremen

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:27 https://doi.org/10.1007/s12036-021-09703-9

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

MISSION

AstroSat proposal processing system C. BALAMURUGAN1,* , SACHIN NARANG1, PRADNYA BHOYE2,

MANDAR HULSURKAR2, GULAB DEWANGAN2, DIPANKAR BHATTACHARYA2 and B. N. RAMAKRISHNA1 1

ISRO Telemetry Tracking and Command Network (ISTRAC/ISRO), Bengaluru 560 058, India. Inter-University Centre for Astronomy and Astrophysics (IUCAA), Pune 411 007, India. *Corresponding Author. E-mail: [email protected] 2

MS received 4 November 2020; accepted 1 January 2021 Abstract. AstroSat Proposal Processing System (APPS) is a mission critical software solution designed and developed by the joint efforts of Indian Space Research Organization and the Inter-University Centre for Astronomy and Astrophysics. It facilitates the participation of global scientific community to submit scientific proposals for observations with India’s first multi-wavelength space observatory. The software systematizes the proposal submission and review process before the successful proposed observations are scheduled to carryout scientific observations. This paper describes the overall architecture of the system, implementation stratagems, administrative aspects and security aspects of the software. The paper also describes the techniques adopted for seamless day-to-day operations in meeting the mission requirements. As a future direction, the paper summarizes the roadmap for development of a generic multi-mission proposal handling model from the experiences gained through APPS. Keywords. AstroSat Proposal Processing System—Indian Space Science data centre—multi-wavelength astronomy—security assessment process.

1. Introduction Indian Space Science Data Centre (ISSDC) is primarily responsible for Ingestion, Processing, Archival and Dissemination of payload data from space science, lunar and interplanetary missions of ISRO. AstroSat (Singh et al. 2014; Navalgund et al. 2017) is India’s first multi-wavelength astronomy mission which enables simultaneous observations of celestial sources in X-ray, UV and optical bands. This is achieved using five science instruments on-board the spacecraft namely, Ultra-Violet Imaging Telescope (UVIT) (Tandon et al. 2017), Soft X-ray Telescope (SXT) (Singh et al. 2017), Cadmium Zinc Telluride Imager (CZTI) (Rao et al. 2017), Large Area X-ray Proportional Counters (LAXPC) (Agrawal et al. 2017) and Scanning Sky Monitor (SSM) (Ramadevi et al. 2017). AstroSat This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

operates like any other space observatory in the world. The specific support requirements for AstroSat mission encompasses payload data reception, Level-0 and Level-1 processing, dissemination of processed data to Payload Operations Centre, Proposers, Instrument teams, and Public Observers. This is the first Indian space mission for which a novel approach is being followed where the global astronomical community/researchers worldwide participate directly in the generation of proposals specifying the astronomical object or event to be observed and the duration of observation. The proposals submitted by astronomers are reviewed by the AstroSat Technical Committee (ATC) for technical feasibility with respect to the spacecraft constraints. The proposals are finally evaluated for the scientific merits and approved for command generation by the AstroSat Time Allocation Committee (ATAC). The approved proposals are then translated as spacecraft commands and scheduled for operations. To realize this complex

27

Page 2 of 9

workflow scenario, a proposal processing pipeline has been established at ISSDC for the first time. Section 2 describes the completely automated proposal processing pipeline solution at ISSDC covering the proposal submission and approval workflows. Section 3 specifies the usage scenarios of APPS. Section 4 discusses the system architecture and development strategies in depth. Deployment and operational challenges to ensure seamless day-to-day operations are discussed in Section 5. Various aspects on how the system was deployed at ISSDC during the pre-launch phase for qualification, the testing strategies including security aspects of the software, challenges faced during the initial phase of operations and regular day to day operations are brought out in this paper. Section 6 describes the path ahead for handling the future proposal driven science missions of ISRO. The need for a generic system for proposal processing is envisaged from the experiences of APPS to address future needs. 2. Proposal processing pipeline at ISSDC This section outlines the sequence of activities involved in the conversion of approved user proposals to spacecraft commands. The spacecraft payload operations are based on the proposals for astronomical observations by scientists, both from India and the International community. The activities of the pipeline start with the submission of proposals by Astronomers through the APPS software. The APPS software caters to different groups of users which include proposers, payload scientists, guest observers, reviewers, committee members and approval chairs. The proposals submitted by the astronomers are evaluated by an expert committee for their scientific merit and technical feasibility in terms of various mission constraints. The final approved proposals for carrying out observations are transferred to the core mission network in a secure way. The Mission Control And Proposals (MCAP) database (Pandiyan et al. 2017) is a centralized repository store for approved proposals from APPS. The constraint check and scheduling software accesses the approved proposal from MCAP database and arranges proposals in slots based on priority, geometrical and spacecraft constraints. The command generator accesses the sorted proposals and generates a sequence of spacecraft commands, which are identified by an observation-id. As a final step, the commands are uplinked to the spacecraft and the MCAP database is updated with the observation details.

J. Astrophys. Astr. (2021)42:27

A few more web applications as mentioned below are deployed along with the APPS software to facilitate the proposal preparation activity. Astroviewer is a web based software utility, used to generate the probable visibility periods of the celestial bodies which can be observed by AstroSat (Nagamani et al. 2010). AstroSat Exposure Time Calculator (ETC) is a simulation tool to predict the exposure time required for the various payloads of AstroSat from source characteristics. The visibility periods and Exposure time estimation are mandatory for proposal submission. The pipeline elements are distributed over multiple networks spanning different security domains spread across geographically separated locations, as indicated in Fig. 1. The web based applications APPS, Astroviewer, Exposure Time Calculator are hosted in the external network of ISSDC. The MCAP software, constraint check and scheduling software are located in the internal mission network layers at ISSDC and Mission Operations Complex (MOX). All the above mentioned software forms the pipeline suite for proposal processing. The interaction between the pipeline entities are through files and are prioritized for guaranteed transfer of necessary inputs. The software operations are completely automated and are monitored through the alerts generated.

3. APPS usage scenarios The APPS automates the proposal submission and review process for the AstroSat mission. The APPS is a web-based tool and is the core of the proposal processing pipeline. Proposals are submitted under various streams, which include Announcement of Opportunity (AO), Guaranteed Time Observation (GTO), Target of Opportunity (ToO), Calibration (CAL), and Performance and Verification (PV). Each of these streams has different evaluation, review, revision and approval workflows. There are multiple cycles opened for proposal submission with only one for each stream. Each cycle can have different types of proposals viz. regular proposals, monitoring proposals and anticipated ToO proposals. Considering the above, different classes of users have been defined like registered observers, payload scientists, instrument scientists, piggyback science team members, calibration scientists, validator, committee members and committee chairperson. The entire functioning of APPS is co-ordinated, monitored

J. Astrophys. Astr. (2021)42:27

Page 3 of 9

27

Figure 1. Proposal Processing Pipeline at ISSDC.

and managed by the APPS administrator. The completed workflow of APPS is described in the following sections and shown in Fig. 2.

3.1 Proposal submission The proposal submission process begins when a proposer login into the APPS (after successful registration) and clicks on the ‘‘create a new proposal’’ button. While some proposal streams like ToO are always ‘‘open’’, others such as AO or GTO proposals can only be submitted during the period when the administrator announces a proposal cycle as ‘‘open’’. A proposal can have more than one investigator. The Principal Investigator (PI) can add more registered users as Co-Investigators (Co-I) for the proposal. Only the PI of a proposal can edit, submit, withdraw and revise the proposal. The Co-I can only view the proposal irrespective of what state the proposal is in. The proposal creation form incorporates details like the title, abstract, sources to be observed, scientific and technical justification, payload configuration details etc. The proposers can access external tools like Astroviewer, Exposure time calculator by using links provided by AstroSat Science Support Cell (ASSC)

Figure 2. APPS work flow/usage scenario for AO cycle.

website. The output from these tools are attached and uploaded as a part of proposal submission process, depending on the source and instruments. They will also have the provision for specifying time constraints in case of co-ordinated observation with other

27

Page 4 of 9

observatories. A proposal can only be submitted after it passes validation tests to ensure that the specified parameters in the proposal are within the limits.

3.2 Proposal review and evaluation The submitted proposals for every cycle is subjected to an evaluation process by the cycle specific AstroSat Time Allocation Committee (ATAC). If a cycle is configured to have a two level hierarchy then, members of the ATAC may chair subcommittees called Sub-TACs. Reviewers are domain specific experts and are invited to review and grade the proposals by the ATAC chairperson. The ATAC chairperson is the final authority and is assisted in the evaluation process by the members of the ATAC and ATC. The final approved proposals for observations are transferred to the MCAP software for subsequent scheduling operation.

4. Architecture design and development strategies This section describes the overall architecture of APPS software and the development strategies adopted.

4.1 Product perspective At the input end, APPS interacts with various types of users like Proposers, Chairpersons, ATAC members, etc. On the output end, APPS consolidates and provides the information on proposals that are approved by ATAC in a format suitable to be ingested into MCAP database for further processing.

4.2 Design and architecture The following are the main modules of APPS software: (i) Authentication and authorization: This module provides mechanism for user login, password hashing, salting, and CAPTCHA code verification during login and profile change. The software uses local CAPTCHA to avoid dependency on third party services. Role based access to dynamic pages is provided by this module.

J. Astrophys. Astr. (2021)42:27

(ii) Interface Engine: This module is a repository of classes implementing JSP pages designed and constructed as per the workflow requirements. This also keeps tracks on the state of proposal and presents the page dynamically based on roles and state. (iii) Notification module: Based on role, users are sent notification regarding successful registration, successful proposal submission, during profile change, call for revision of partially accepted proposals, successful completion of revision, rejection, call for piggyback setting, validating a cycle for approval and change, cycle creation, cycle closure, setting of instrument configuration and validation of instrument configurations. The notifications sent are logged in database. A template message body and subject are made available to the reviewers to offer their comments regarding acceptance/ rejection/revision of proposals. The chairman can send relevant messages to users by editing the template message provided. (iv) Content check module: This module validates the submitted proposals for syntax checking of attributes, allowed parameter ranges, attachments from utility software and proposal completeness. This allows the proposers to detect errors. This module also does the periodic auto commit of parameter values entered by the proposers during proposal submission. (v) Configuration module: This module allows administrator to dynamically configure instrument configuration of proposal for different cycles which include setting the UI type (check box, text box and drop-down list), data types, allowed range of values and default values. This is set for every payload, detector and exposure wise. This module allows user to select detectors and set values for exposure based on the configuration by the administrator. (vi) Report generation module: This module generates necessary outputs for proposal viewing and evaluation which includes cover sheets in PDF format for proposers, administrator and the committee members. It allows downloading of the complete set of documents uploaded during proposal submission both at individual proposal level and as consolidated set. This module also generates cycle wise reports on proposal submitted by proposers, details of observation, target lists and overall statistics both in PDF

J. Astrophys. Astr. (2021)42:27

and in CSV format for management and reviewers. (vii) Xml engine: This is responsible for generation of MCAP input files in xml format. All the proposal details including the general information, target and instrument configurations are generated as xml file and validated. The xml files are then posted to trigger the next activity in the proposal pipeline

4.3 Development strategy The requirements for the software were provided by ISRO and IUCAA. A review-based approach to software development was adopted and the development was carried out in phases as per the ISRO Software Process Document guidelines. During the requirements gathering phase an inter-centre committee was formed which included members from ISRO, IUCAA and Payload teams. The following describes the review based approach used during the software development. (i) Contract establishment: A series of reviews by the committee were conducted to analyse and finalize the requirements after which the contract was established with the third party for software design and development. (ii) Review Phase: The development was monitored by the inter-centre committee and was subjected to ISRO level review process including the ground segment design-reviews. (iii) Testing and acceptance: The software was deployed at ISSDC and the testing was carried out by the formal ISRO level test and evaluation committees which included the participation from the instrument scientist and the Payload Operation Centres (POCs). (iv) Maintenance phase: The regular operations are carried out from ISSDC and the software is administered by the ISSDC team. Versioning, bug and release management activities are carried out in a co-ordinated manner by ISSDC team at ISRO and the ASSC at IUCAA.

5. Deployment and operational challenges The software used for proposal submission activity is deployed at the external network of ISSDC which is isolated from other internal networks. The following

Page 5 of 9

27

section describes the deployment and operational challenges and how it was overcome to ensure smooth operations at ISSDC.

5.1 Safe hosting of applications The software APPS, Astroviewer, AstroSat Schedule Viewer, SSM Web are hosted in ISSDC over the Internet for access by proposers from all over the world. Hence, these software have to be checked for security measures before deployment. To prevent any security breach, ISSDC has devised a mechanism in accordance with the ISRO-DOS Information security policy. The software that is to be hosted at ISSDC is subjected to a complete vulnerability assessment process to identify the vulnerabilities in the software before hosting them at ISSDC for access via Internet. The vulnerabilities identified are shared with the designers for necessary corrective action. A second round of assessment is carried out after the vulnerability issues are fixed by the designers. Test setup for assessment: An isolated test environment has been set up at ISSDC to carry out the assessment activity as shown in Fig. 3. Local accounts for testing, local file systems for hosting and configuring the applications are created on the server. All libraries used by the application and dependencies are identified and the same is configured on the test server. A test database suite at ISSDC serves as the database for all applications being tested. Exactly the same schemas as used by the application are created with dummy values set for all attributes in the database. A test mail server is also set up to receive and drop the message without actually sending them. The application that is under testing should ensure that the authentication mechanism in the application bypasses the CAPTCHA verification process to avoid manual intervention during an automated scan. The software is made accessible over the intranet if the assessment is internal. If the assessment is by an external certification agency, then the application is made accessible only to the identified nodes on the certification agency’s network. Suitable modification in firewall configuration like disabling intrusion detection and prevention was done to prevent the firewall from hindering the assessment process. Internal assessment process: This activity is carried out by an internal team. The process flow has been illustrated in Fig. 4. The following are the stages in the assessment process:

27

Page 6 of 9

J. Astrophys. Astr. (2021)42:27

Figure 3. Test environment for vulnerability assessment. Figure 4. Process flow for internal assessment.

(i) Pre-briefing: The requirements and the test case document of the application are circulated to the team. A briefing on the need and functional aspects of the software is done to the committee to have a better understating of the system. A demonstration of the usage scenarios of the system are provided to the team. The team assesses the requirements and comes out with an assessment plan which includes the requirements for setting the test environment, schedule of the activity and the expected deliverables of the assessment. (ii) Preparation: Based on the assessment plan, a test environment is setup at ISSDC as described in the previous section. The servers are connected to the intranet. This enables designers from other ISRO Centres to access the application while the assessment is in progress and the attacks can be demonstrated to the designers. This would help the designers understand the vulnerability and take quick actions to resolve or mitigate them. (iii) Automated tool test: The automated tool test is done by exposing the web application to the vulnerability assessment tool. A login sequence covering all pages and all actions is recorded. The automated testing is initiated based on the login sequence. The tool classifies the

vulnerabilities as high, medium, low and informational. A developer summary and a compliance report are generated at the end of the test. (iv) Manual mode testing: The reports generated during the previous phase are taken as reference for this test. This gives an insight into the vulnerabilities found in the application and approach that needs to be adopted for manual mode of testing. BURPSUITE, FIREBUG utilities are used for intercepting the requests and responses from the client/server over the network by setting up suitable proxy. As a first step the vulnerabilities found using the automated tool testing phase are analyzed to identify true positive from false positives. In the second step, the application is tested for Open Web Application Security Project (OWASP) top ten vulnerabilities. Additional tests are carried out based on the outcomes from the previous phases of testing. The observations are recorded in the deliverable. (v) Vulnerability fixes management: The final list of vulnerabilities after ruling out false positives from both manual and automated mode of testing is provided to the designers for vulnerability fix. An elaborate briefing on how to fix or mitigate the vulnerability is provided to

J. Astrophys. Astr. (2021)42:27

Page 7 of 9

27

designer upon request. The designer fixes the vulnerabilities and re-submits the application for assessment. The processes iii, iv and v are repeated if the vulnerabilities are still noticed in the application. (vi) Clearance for hosting: Once the vulnerabilities are fixed, the software is cleared for hosting. This is only applicable for applications which are hosted in the Internet like application for viewing schedules, data searching etc. which are not mission critical where the input from the users are not translated to critical information to control the mission activities. Mission critical applications need certification as mentioned in the next section. Certification process: ISSDC follows a certification process by Indian Computer Emergency Response Team (CERT-In) empaneled agency when hosting applications that are mission critical where the user input directly controls the mission activities. The advantage of certification process is that, it detects the vulnerabilities, which are left unnoticed during the internal manual and auto mode of assessment. The steps in the certification process are depicted in Fig. 5. (i) Contract establishment: The process begins by establishing the contract with the external agency for certification. Before establishing the contract, the certification agency evaluates the application based on the audience for the application, technologies used, architecture, use of content management modules, hosting domain, number of dynamic pages, data entry points, file upload/download feature, authentication mechanism used, number of roles in the authorization matrix, use of payment gateway/ digital signatures, security incidents in the past, test environment, links to live database and target date for completion of the activity. The deliverables include the vulnerability reports and a ‘safe to host’ certificate. (ii) Audit kick off: Once the contract is established, ISSDC team formally convenes a meeting to describe the usage scenarios of the application. This is followed by the demonstration of the application for all the roles. (iii) Preparation: A test environment is established as described in the previous section. The systems are deployed in an isolated network and access to the system is given for only a specific identified node of the certification agency over the Internet. The activities on the network are

Figure 5. Process flow for certification.

monitored regularly during the assessment. Before the software is made available for external audit, it is ensured that the application qualifies the internal assessment as described in the previous section. (iv) Cycle-wise audit: The assessment is carried out in a number of cycles as agreed in the contract. Normally this is done in two cycles. The audit involves both: the automated tool testing and manual mode testing for OWASP top ten vulnerabilities and ruling out false positives. (v) Follow up: The audit report generated by the certification is then shared with the designers. Upon request, ISSDC team briefs them about possible solution to fix the vulnerabilities. This phase also involves internal assessment before the next cycle of assessment by the certification agency. If the software fails in the internal assessment process, the designers are informed about the vulnerabilities and the activity is repeated. The application is re-tested to ensure that the vulnerabilities found during the first cycle of audit by the external agency are fixed.

27

Page 8 of 9

J. Astrophys. Astr. (2021)42:27

Figure 6. Flexible testing scenario.

(vi) Reassessment and certification: On successful internal assessment, the software is subjected to the second level of assessment. The audit agency carries out the next cycle of assessment and if the vulnerabilities found in the first level of scan are fixed, then a ‘safe to host ‘certificate is issued for that version of the software which has undergone the audit. The certificate is valid only for that version or until a new OWASP top ten vulnerabilities is released.

5.2 Issue resolution through scenario based testing A test-bed setup has been implemented at ISSDC to resolve/simulate issues related to the usage of various AstroSat web applications. The database from the production database is replicated to the test database. Repeated simulation of various usage scenarios were carried out with the active participation of the identified proposers from the instrument team. Remote/ restricted access was given to the proposers and reviewers during the test and evaluation of the software. This set up had great significance during the initial months of Performance Validation (PV) phase after launch to have a quick look into the proposals and test the evaluation and approval scenarios.

5.3 Flexible single sign-on/identity management using AstroSat Authentication Manager (AAM) The web applications for proposal submission: APPS and Astroviewer, the web applications for payload schedule access: AstroSat Schedule Viewer (ASV), the

application for transient alerts: SSM Web and the software for processed data access: Astrobrowse have authentication mechanism to gain access to them. These software had their own authentication schemes and had forced the proposer/users to have multiple accounts for accessing the above mentioned web applications. To have a common authentication/identity mechanism, AstroSat Authentication Manager (AAM) was developed by ISSDC and is currently operational. The software allows different web applications to register for authentication service. The user base from the proposal submission web application APPS was taken as the reference user base for all other AstroSat web application. The users who wish to access AstroSat web-applications need to register with APPS Software. Upon registration to APPS, users can access the various AstroSat web applications using the single sign-on user name and password. The user base of APPS includes the user-id, one-way hashed password, salt and other user details. The user-id, one-way hashed password and the salt from the APPS database are replicated using an intermediate schema with read only access. This ensures that user base of APPS is protected from unauthorized access. The software provides Application Programming Interfaces (API) for various web applications for authentication. All AstroSat applications use this software for authentication. Hence, the availability of AAM is of utmost importance. The software ensures availability in the event of denial of service attack. The applications listens for the connect requests and if the rate of the requests is more than the threshold, the AAM software blocks the application from accessing the authentication service. This prevents the starving of other application, in the event of misbehaviour by one application.

J. Astrophys. Astr. (2021)42:27

This application has been extremely useful when web applications are hosted on the Internet for public. This also gives the flexibility to switch between production and test database with users being transparent to the replication of user scenario during testing. The flexible testing scenario established at ISSDC is shown below in Fig. 6.

Page 9 of 9

27

been established at ISSDC to simulate and resolve issues reported by the public users at the earliest. Steps have been taken up by ISSDC to improve the evaluation and scheduling process using machine learning techniques.

Acknowledgments 6. The path ahead Handling mission critical proposal based system, interaction with public users from all over the world towards proposal submission, issue resolution and data access has given ISSDC a wide range of new experiences. Taking this experience forward, a single platform to service all activities from proposal submission till data access has been envisaged and the support for the near future proposal based missions shall be done by the new system. A model encompassing deterministic and machine learning based approach has been planned to bring in improvements to the proposal evaluation and planning process. This would facilitate quick and optimal convergence on arriving at the observation plans for maximizing the spacecraft utilization and science outcome.

7. Conclusions Currently, the proposal based AstroSat operations are being handled smoothly from ISSDC. The varied experiences gained right from development, deployment, maintenance of web applications, issue tracking and timely resolution, ensuring safe hosting of web applications has provided more confidence to flawlessly manage future proposal based mission of ISRO. ISSDC has also evolved a set of guidelines for web application developers to ensure safe hosting of applications over the Internet. A test environment has

The authors would like to thank and acknowledge the support provided by ISSDC team, ATAC and ATC members, ISTRAC FDO, Mission operations, Scheduling teams, URSC teams, Space Science Programming Office, ISRO HQ and IUCAA-AstroSat Support Cell for their extensive support in establishing and maintaining the proposal processing pipeline at ISSDC.

References Agrawal P. C., Yadav J. S., Antia H. M. et al. 2017, J. Astrophys. Astr., 38, 30 Nagamani T., Bharadwaj N., Dakshayani B. P., Pandiyan R. 2010, ASTROSAT Software Tool to Aid Celestial Source Viewing, 61st International Astronautical Congress, Prague, CZ, Paper IAC-10-A3.4.6 Navalgund K. H., Suryanarayana Sarma K., Gaurav P. K. et al. 2017, J. Astrophys. Astr., 38, 34 Pandiyan R., Subbarao S. V., Nagamani T. et al. 2017, J. Astrophys. Astr., 38, 35 Ramadevi M. C., Seetha S., Bhattacharya D. et al. 2017, Exp Astron., 44, 11 Rao A. R., Bhattacharya D., Bhalerao V. B. et al. 2017, Curr. Sci.. 113, 595 Singh K. P., Tandon S. N., Agrawal P. C. 2014, SPIE, 9144, 1 Singh K. P., Stewart G. C., Westergaard N. J. et al. 2017, J. Astrophys. Astr., 38, 29 Tandon S. N., Hutchings J. B., Ghosh S. K. et al. 2017, J. Astrophys. Astr., 38, 28

 Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:23 https://doi.org/10.1007/s12036-021-09727-1

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

MISSION

Novel payload planning and scheduling approaches implemented for AstroSat mission ANSHU CHAUHAN*, T. NAGAMANI, NITIN BHARDWAJ, NARESH KUMAR, S. V. SUBBA RAO and JASVINDER S. KHORAL U.R. Rao Satellite Centre, HAL Airport Road, Bengaluru 560 017, India. *Corresponding author. E-mail: [email protected]; [email protected] MS received 5 November 2020; accepted 7 February 2021 Abstract. AstroSat is an astronomical observatory from India which is in a low earth orbit of 650 km altitude with a 6-degree inclination and has been functioning successfully in orbit for the last 5 years. The simultaneous observation by the payloads allows imaging of celestial sources in various wavebands that has resulted in scientific achievements which are published in various forum. Uniqueness of its orbit, and scientific objectives calls for a different type of scheduling approach and a hands-free automated tasking of payloads, on this satellite. Several software architectures involving a Centralized data-base called Mission Control And Proposals (MCAP) with the Scheduler, AstroSat Scheduler TeRminus for PAyLoads (ASTRAL), AstroSat Scheduler Software for OpeRaTions (ASSORT), Command Sequence Generation (CSG), AstroSat Schedule Viewer (ASV), Astroviewer and AstroSat Long Term Planner (ASPlanner) were developed and used for celestial source observations. This paper describes briefly the scheduling approaches, their goals, utilization, lessons learnt and improvements carried out over the last 5 years. Keywords. Operational scheduler—planning—South Atlantic Anomaly—short-term planner—commands.

1. Introduction Image planning of celestial sources involves taking into account user requirements such as instrument to be used, length of time of observation, instrument configuration etc. and considering all geometrical & spacecraft constraints for the safe operation. This scientific mission has the requirement of longterm plan in advance specifically for the coordinated observations using similar other satellites missions. This is to be followed with regular operationalisation closer to the actual observations. To cater to these stringent and variant observation requirements a novel approach based on software architecture was designed and operationalized for AstroSat Mission.

This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

User submits the observation proposal through the AstroSat Proposal Processing System (APPS). The proposal goes through a review process and once approved, the observation requirements are fed to the Mission Control And Processing (MCAP) system MCAP database, and used by all other ground software elements for planning for the particular observation cycle. The AstroSat Scheduler TeRminus for PAyLoads (ASTRAL) software provides feasibility of image opportunity duration orbit wise for each source after checking various geometric constraints for the given cycle span of six months/one year. Based on ASTRAL input, tentative target sequence is generated and populated in MCAP database. The AstroSat Scheduler Software for OpeRaTions ASSORT software is used to finetune the observation timings closer to the actual operation which is decided to be a fortnight for operational efficacy. These parameters are translated into satellite

23

Page 2 of 10

time tagged command sequences by CSG software with respect to each payload while considering various mission operation constraints and the requested payload configuration and exposure requirements and observation duration. These commands are uplinked and stored in the memory bank of the satellite for subsequent operations in orbit. It accounts furthermore to servicing the actual exposure time which user has requested. Figure 1 depicts the interfaces of various planning software. All elements were planned to be operated from different network and environment. With the help of data driven approach and file interface via common database, each element could be independently operated and still in a coordinated manner.

2. Astroviewer The viewing tool ‘Astroviewer’ is a web-based tool to check for viewing conditions/opportunity for a celestial source. This tool is available on Indian Space Science Data Centre web page, to any astronomer who is registered with Indian Space Science Data Centre (ISSDC) and wishes to submit proposals to observe with instruments of AstroSat. This tool was updated based on the operational requirements with tuned South Atlantic Anomaly (SAA) boundary model and to provide day-wise and day–night viewable duration as requested by payload team and has been in use till date to submit proposals. The Principal Investigator (PI) uses this tool to study the availability of source of her/ his interest, by providing RA and Dec of the source and

J. Astrophys. Astr. (2021)42:23

the duration over which visibility has to be checked. Time constrained proposals are those which are coordinated for observations with other observatories or for specific time related events of the source. This demands the Astroviewer’s predicted table of view timings for a year based on which the coordinated view will be planned. The tool has been used extensively for the last five years allowing the PIs to put forward their proposals in a systematic way (Nagamani et al. 2010). The approach of developing the portal using Java based web technologies offers all the benefits to achieve the objectives of fast accessibility, reachability on the desktops with minimal environmental and software support. The key components in the data generation and presentation design are the request, data fetching and presentation mechanism. The request from user through web page has been packed into the file and this file is fed to data fetcher application over FTS (File Transfer System) while the sensor/servlet waits for the response. The sensor at the computer system end, senses the return files and initiates the necessary presentation process for request under consideration which is then presented to the user in the form of text files and plot images. Architecture of the software is captured in Fig. 2.

2.1 Data plotting The plots are generated only if the user opts for them. The different plots such as Roll–Sun vector, Roll– Moon Vector, Yaw–Sun vector have been generated over the period of time in the popular standard file formats like PDF. In the later stage, plots as given in Fig. 3 and text file for the source visibility duration have been added into the package based on the user feedbacks. The java plot package has been upgraded to generate the plots for enhanced clarity and time efficiency.

2.2 Data in text Text files are combined and put together as PDF format and given to user for viewing and download.

2.3 Approaches to realize the package

Figure 1. Software architecture diagram for AstroSat source scheduling.

The trigger/sensor based architecture is upgraded to realize the web application with better security measures.

J. Astrophys. Astr. (2021)42:23

Page 3 of 10

23

Figure 2. Architecture diagram for Astroviewer. JSP: Java Server Pages, FTS: File Transfer System. WWW: World Wide Web.

The enhanced version of application is realized with improved security measures, removing account linkage and data transfers and • Isolation of data generation modules and presentation tools. • Improved downloadable kit. The turnaround time in generation of data (as plot and text files) improved with reducing the multiple process communication and implementing pure request/response system.

3. AstroSat proposal processing system (APPS) AstroSat observation time is made available to astronomers worldwide. To facilitate proposal submission process across the world, the Web based APPS software was designed (Balamurugan et al. 2021). Proposals are reviewed by scientists and approved by AstroSat Time

Allocation Committee (ATAC). Once approved, the requirements are fed to MCAP from APPS and further, MCAP controls the complete flow of data among all planning and data processing software.

4. Mission control and processing system Centralized database approach was worked out for AstroSat Planning Process. The centralized database is referred to as Mission Control And Processing System (MCAP). MCAP contains complete details of user approved Target Sources, requested payload configurations, exposure requirement and observation duration. All proposed sources, out of which sources that are selected for observations, sources that are actually planned and observed are all captured and archived for the complete Mission life in MCAP database. Furthermore, MCAP helps to obtain statistics of observed target sources and their satellite time utilization details

23

Page 4 of 10

J. Astrophys. Astr. (2021)42:23

Figure 3. Plots Roll–Velocity (RAM) and Roll–Sun (ROSV) angle variation.

over the entire Mission life. Exchange of various planning software is carried out through MCAP.

5. Planning approaches Planning process has been categorised into two parts. Long-term planning phase for the cycle duration and short-term operational phase for the actual duration closer to the operations usually 15 days in advance. ASTRAL and ASPLanner are the main software used

for Planning Phase. Operational software set comprises of ASSORT and CSG software. In the planning phase, sequence can be either for the complete cycle which is one year or shorter. For initial two years after guaranteed time observations, the sequence was for full one year and the sequence was modified only for ToO (Target of Opportunity) requests. Later, the sequence has been made generally for 2 to 6-months span based on requirements. First year of AstroSat, the sequence was made ready with more of coordinated manual approach using ASTRAL

J. Astrophys. Astr. (2021)42:23

software outputs as reference. Graphs were generated separately using ASTRAL input and Sequence was arrived for the complete one year. With first year of experience, and consultations with science team all activities were packaged together as a single tool and the AstroSat Planner was designed in December 2016 as an interactive software. This planner tool was given to the source sequencing team to ease the Planning Phase. The software allows to check the constraint free and best slot for any target for the complete cycle in an interactive manner. In addition to the visibility provided in Astroviewer, it also provides star sensor visibility and antenna availability for data download. It is a manual tool but it eases the process of planning. It absorbs tasks of multiple handshaking of ASSORT and CSG. Following sections depict functionality of each and every software and enhancement done over time.

5.1 Flight dynamics scheduler, ASTRAL Realizing AstroSat Scheduler Software, ASTRAL (AstroSat Scheduler TeRminus for pAyLoads) was of vital importance, as scheduling 5 payloads during operations was a challenging task. ASTRAL is the tool used during planning phase interfacing with MCAP and ASSORT. ASTRAL is intensively used for long-term scheduling of targets which allow proposers to know in advance when the proposals/targets are likely to be scheduled. Initially, the architectural design of ASTRAL (Pandiyan et al. 2017) was to collect, validate and segregate various types of proposals and targets, compute and sort the un-obstructed view period for each target, taking into consideration the priority of targets as assigned by the AstroSat Time Allocation Committee. Further accommodation of time-constrained proposals, repeated observations of monitoring proposals called for enhancement of design. The experience obtained during first year after launch, motivated to update this design with parallel computing approach using threads. The main function of the tool is to compute view period for multiple sources accounting for the availability of star sensor, phased array antennae (PAA) field-of-view, Earth occultation, eclipse and ground station visibility. Unlike other events’ computation that usually depend on the spacecraft geometry, celestial source view computation depends on the location of the source in the celestial sphere with respect to Sun, Moon and other planets in combination

Page 5 of 10

23

with subsystem availability. Spacecraft Ephemeris along with the ephemeris of Sun, Moon, planets (Jupiter, Uranus, Neptune, Pluto and Saturn) are stored and retrieved for one year. This has introduced computational complexity which was overcome by parallel computing approach. The complexity of accommodating these constraints among multiple targets is resolved beneficially by this tool with multithreading approach as shown in Fig. 4, resulting in minimizing the execution time drastically. As the number of sources to be planned and its computation are mutually exclusive with one another, parallel computing was a straight-forward approach to implement. Assigning a set of targets to a particular slot which contains all regular, time constrained and one instant of pointing of a monitoring proposal is made flexible by this tool. The challenges faced during scheduling operations are (a) Assigning scheduling priority so that highest priority observations are included first. (b) To evaluate a scientific merit of an observation. Or to include the scientific priority assigned for an observation (c) Tuning of the spacecraft attitude to align precisely to the interested celestial sources in the instrument field-of-view. (d) Apart from these proposals, monitoring proposals are of different type where repetitivity is expected based on the frequency with which proposer wants to monitor. However, it is to be noted that the maximum view period during day and night does not follow the same pattern throughout the observation cycle. The enhanced design of the tool was put in the system during year 2016 and since then it is exhaustively used for planning of proposals of AstroSat cycles enriching mission operations to expose the spacecraft to different parts of the celestial sphere.

5.2 ASPlanner ASTRAL provides Visibility details of all targets for full one-year span satisfying all constraints and are fed to MCAP. ASPlanner takes MCAP as backend and provides a front-end to planning scheduler where the scheduling team is provided with information of observable period for scheduling of targets. Planner has also accounts for constraints required for the short-term

23

Page 6 of 10

J. Astrophys. Astr. (2021)42:23

downloaded as a file and further parameters of the outputs loaded in MCAP. Once MCAP is loaded with the sequence, short Term planning (operational) software works in automated manner. MCAP transfers generated sequence and configuration parameter to CSG software. CSG makes the final operational plan in terms of command. CSG also has the provision to manually control the sequence in cases of emergency. The graphs in Fig. 5 and Fig. 6 depict view period for two different target of different cycles A06 and A09, how the constraints free region is displayed to the Planning team. Furthermore, it is to be noted UVIT and SXT payload observation are always restricted to eclipse times only, and hence the observation time per orbit is less than CZTI/LAXPC.

6. Short-term scheduling/operational chain of AstroSat

Figure 4. Parallel processing of view processing.

planning. Once the target is cleared by the planner, probability of failure at uplink scheduling is minimized. Planner also has the provision to show day-wise free imaging region for each payload though the Proposal might have any other payload as the primary instrument. Planner has the provision of checking target to target manoeuvre duration and limit on spacecraft momentum constraint using Sun avoidance manoeuvre algorithm (SAM) and accounts for constraint on momentum and manoeuvre duration. With operational experience, more constraints have been introduced to the Planner. To name a few, Planner has been integrated with Catalogue Level Star sensor availability and the number of stars available for tracking in each sensor is also made available while fixing the view time of the target. Initially graphical outputs had the provision for only visibility period for the complete cycle in pictorial view but in the next cycle Phased Array Antenna (PAA) constraints were also incorporated in graphs so that Target can be scheduled with respect to on-board data dumping constraint as well. Using the Planner, the scheduling team can make the complete target sequence and the sequence can be

This is the regular automated chain which is operational for last five years for AstroSat Mission. Complete oneyear cycle span is split into smaller time span of 15 days’ cycle by the mission. It requires a target sequence to be available in MCAP for the fortnight which is ensured by the ASPlanner. Through the available flow, MCAP makes available this fixed sequence of Targets to ASSORT and all events are fine-tuned with current updated orbit and fed to CSG. CSG software fine-tunes and generates the plan while ensuring both user requirement and on-board system safety. As there might be small differences in the times due to refined orbits and checks in this short-term planning chain, there is provision of overlapping previous to next chain while providing the targets. This chain is capable of handling any urgent request which was not incorporated during Planning phase. With experience gained in initial phase, at CSG end, more control has been given to the user to alter the sequence provided by Planning phase to incorporate any urgent request and also to extend and shorten the exposure time as per requirement.

6.1 Operational scheduler, ASSORT During the first six months called Performance Verification Phase, calibration exercises of various payloads individually and then in combinations were performed through proper scheduling to avoid any attitude loss of the spacecraft. Since the commencement of calibration of payload CZTI on day 9th from

J. Astrophys. Astr. (2021)42:23

Page 7 of 10

23

Figure 7. Older SAA boundary (rectangular boundary) X-axis longitude, Y-axis latitude. Source: ROSAT SAA Guest Observer Facility (NASA Goddard Space Flight Centre).

Figure 5. Plot for Target A06_009 T02 over one-year period from October 2018 to October 2019 (for X-axis, date format is dd/mm/yy).

Figure 8. Operational SAA boundary (red triangular region within black contour).

Figure 6. Plot of A09_147 T01 Target over 5 months from May-2020 to September-2020 (for X-axis, date format is dd/mm).

launch date 28 September 2015 with the first observed celestial source being CRAB, ASSORT provided the payload operations with view information accounting for all geometrical constraints. This is continued for

successive payload calibration for SSM from day 15th, LAXPC from day 22nd, SXT from day 29th, and finally UVIT from day 63rd day after launch. SSM is mounted with its view axis along positive yaw axis and the view axes of all other payloads along positive roll. As the payloads are sensitive to Sun, there is a necessity of a minimum angle to be maintained between payload field-of-view and Sun. When positive roll axis was brought closer to the Sun by *48, the drift of gyroscopes and outputs of 4p Sun sensors showed some anomalous behaviour affecting pointing accuracy and therefore, the Sun angle constraint was decided to be always greater than 65 till today. In case of attitude loss of the spacecraft, the spacecraft was designed with safe pointing attitude viz. positive roll axis pointing the North pole to

23

Page 8 of 10

J. Astrophys. Astr. (2021)42:23

Figure 9. SAA duration for 4 days.

safeguard sensitive payloads and spacecraft Earth occult constraint that occurs every orbit for most of the roll pointing, were computed accounting for Earth albedo angle which was substantial and were accounted in planning total requested observation time. Spacecraft pointing is obtained with true of date quaternions corresponding to celestial source view start computed using Sun and celestial source position. Star Sensor 1 (SS1) and Star Sensor 2 (SS2) by virtue of its sensitivity to bright objects such as Earth, Moon can malfunction and needs to be disabled suitably for operations. Observation of behaviour of star sensors (SS1/SS2) onboard during Earth occult and, moon in FOV eventuated into uplinking of SS1 or SS2 occult information that called for ground computation. The process of appropriately disabling either SS1 or SS2 before albedo start and enabling after albedo end began from 12th November 2015 onwards. Spacecraft operations were planned to avoid positive roll axis seeing bright Earth terminator by Sun Avoidance Manoeuvre (SAM) for each orbit, where the terminator region considering Earth limb angle of 12 degrees could be avoided. This was carried out before UVIT door by tuning its electronics. Further to that when UVIT doors were opened, extensive analysis was carried out to study the behaviour of the payload during bright Earth passage. Based on the initial analysis for a period of about one month data observation over various sources carried out by payload team, it was concluded that the bright earth avoidance need not be exercised for future operations as there was no degradation on payload observed during bright

Earth. Two Phased Array Antennae (PAA)’s are available on board for data download, that should be appropriately selected from ground based on the availability depending on the current roll pointing. There are situations when the download can be changed from PAA1 to PAA2 and these situations were properly accounted through ground commands. ASSORT accommodates Target of Opportunity (ToO) sources along with nominal schedule giving flexibility to science community to ingest any emergency observation that arises based on celestial phenomena.

6.2 South Atlantic anomaly (SAA) South Atlantic Anomaly (SAA) is an area where the Earth’s inner Van-Allen radiation-belt comes closest to the Earth’s surface dipping down to an altitude of 200 km. This leads to an increased flux of energetic particles in this region and exposes orbiting satellites to higher than usual levels of radiation. The shape of the SAA changes over time. Many astronomy satellites in low Earth orbit are adversely affected by SAA. The orbit of AstroSat is inclined at 6 degrees, due to which the satellite periodically passes through the SAA region. Since each satellite orbit traces a different ground path, the duration of the time spent in SAA region varies across the day. This necessitates the prediction of SAA passages to bring down the payload voltage henceforth to avoid degradation of the payload. A monitor for particle count rate, called Charged Particle Monitor (CPM), is an auxiliary

J. Astrophys. Astr. (2021)42:23

Page 9 of 10

payload on the AstroSat satellite. Before launch, an initial rectangular boundary as shown in Fig. 7 for an altitude of approximately 650 km where the SAA spans from –60 to –2 geographic latitudes and –90 to 30 geographic longitude for which the fluxes would be potentially harmful to the payloads was provided. CZTI payload uses the CPM data to reduce voltage when passing through the SAA. The other 4 payloads depend on uplinked predictions of SAA entry/exit times based on ephemeris data provided by the Scheduler software. Once in orbit, the size of the SAA model was tuned and a safe contour was obtained. On the other hand, celestial source view timing has to be increased to optimise the science returns by having the best possible model of the SAA boundary. An extensive effort was done by payload teams of LAXPC and SSM on post launch data to utilize

23

charged particle monitor data counting rates and LAXPC and SSM data to refine the model of the SAA boundary. Figure 8 provides the new SAA boundary with an optimal contour which resulted in increase in payload observation time to 100 min per day approximately. This new refined boundary was incorporated in ASSORT and is used for regular spacecraft operations till date. SAA occurs every orbit varying from 18 minutes to 31.5 minutes over a day as given in Fig. 9.

6.3 Command sequence generation Command sequence generation software is the main operation software which ensures that operation is safe with respect to on-board resources and also ensures that all user requirements are met during the operations.

Figure 10. AstroSat Schedule Viewer.

23

Page 10 of 10

CSG maps user requirement to the unique observation. It ensures that respective payloads are switched ON during safe period and for sufficient time like UVIT is operated only during eclipse. CSG also configures payloads as per the configuration provided through MCAP. It arranges targets as per the Sequence provided from Planning phase but has the provision of handling any ToO requests also. Over time, CSG has been enhanced for better management of Time constraint and monitoring request. Also, it has been enhanced to allow increasing/restricting the observation period in case of contingency. 7. AstroSat schedule viewer (ASV) This is a web-based tool which helps user to know what is the current schedule of AstroSat. It has the provision to display the Payload wise observation which have been carried out over time in the present cycle. ASV has the provision to view long term planning well in advance. It also has the feature to show proposal wise statistics (Fig. 10). 8. Conclusion After five years of operation, planning and scheduling chain is well stabilized and has been upgraded over time based on experience gained during initial 2 years. There are well defined methods available to fill the gaps or handle contingency requirement. It has given

J. Astrophys. Astr. (2021)42:23

the base knowledge of having a system which can generate optimal sequence for the complete cycle, and the experience will be used for Future Astronomy Missions. Over the five years, total 8 general public cycles from A02 to A09 have been scheduled and planned by the Planning software. Around 200–300 targets are submitted in each cycle. Currently A10 cycle is in progress which contains maximum target sources.

Acknowledgement We acknowledge Dr Ramalingam Pandiyan, Mission Director (Retired), Mr A. S. Shastry, Group Head, SHMSG and Dr M. P. Ramachandran, Group Director, Flight Dynamics Group for the support provided during design and development activities.

References Balamurugan C. et al. 2021, AstroSat Proposal Processing System, J. Astrophys. Astr., 42, https://doi.org/10.1007/ s12036-021-09703-9 Nagamani T., Bhardwaj N., Dakshayani B. et al. 2010, AstroSat Software Tool to Aid Celestial Source Viewing, 61st International Astronautical Congress, Prague, CZ, Paper IAC-10-A3.4.6 Pandiyan R., Subbarao S. V. Nagamani T. et al. 2017, Planning and Scheduling of Payloads of AstroSat during Initial and Normal Phase Observations, J. Astrophys. Astr., 38, 35

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:43 https://doi.org/10.1007/s12036-021-09747-x

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

MISSION

Data processing, archival and dissemination pipeline for AstroSat: Challenges and strategies C. BALAMURUGAN1,* , SACHIN NARANG1, U. PRIYANKA2, NAVITA THAKKAR3,

HIMANSHU PANDEY1, AMIT PUROHIT1, A. L. SATHEESHA2, P. GUNASEKHAR1, T. P. SRINIVASAN3, A. S. SHASTRY2 and B. N. RAMAKRISHNA1 1

ISRO Telemetry Tracking & Command Network (ISTRAC/ISRO), Bengaluru 560 058, India. UR Rao Satellite Centre, Airport Road, Bengaluru 560 017, India. 3 Space Applications Centre, Ahmedabad 380 015, India. *Corresponding Author. E-mail: [email protected] 2

MS received 4 November 2020; accepted 7 February 2021 Abstract. The ground data processing, archival and dissemination of the science data for India’s first multi wavelength astronomical spacecraft, AstroSat is carried out from Indian Space Science Data Centre (ISSDC). The proposals submitted by users worldwide are grouped into observations and are observed in multiple orbit segments. Large volume of data in various configurations from different payloads and the inherent diversity of data products resulting from observations had demanded intricate procedures in handling the data processing, archival and dissemination scenario at ISSDC. The paper describes the layered approach followed in implementing a software stack for establishing and operating a completely automated pipeline for data processing, archival and dissemination of astronomical data. The operations of the automated pipeline, challenges faced during the implementation and the strategies adapted to overcome them are also summarized. Keywords. AstroSat—data pipeline—archive management—dissemination management—event management—automation—good time interval—bad time interval.

1. Introduction AstroSat (Singh et al. 2014; Navalgund et al. 2017) is a proposal-driven multi-wavelength first Indian space observatory capable of simultaneous observation over a wide range of electromagnetic spectrum. The primary science objectives of the mission are met with 5 science payloads namely Ultraviolet imaging telescope (UVIT) (Tandon et al. 2017), Cadmium Zinc Telluride Imager (CZTI) (Rao et al. 2017), Soft X-ray Telescope (SXT) (Singh et al. 2017), Scanning Sky Monitor (SSM) (Ramadevi et al. 2017), and Large Area X-Ray Proportional Counter (LAXPC) (Agrawal et al. 2017). The

This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

astronomical community and researchers worldwide submit proposals for the astronomical object/event to be observed. The proposals submitted by users for different targets are grouped into observations. The data for a single observation would be recorded and received at the ground stations in several data dumps. Each dump session refers to the availability of the station for data collection from the spacecraft during a single orbit. Hence the data products generated are dump-orbit wise and when the data belonging to a complete observation are available on ground, observation wise products are generated. The data processing at ISSDC includes Level-0 and Level-1 processing. The Level-0 process carries out time corelation, segmentation, attitude filtering and orbit corrections. The Level-1 processing focuses on applying filters and corrections to properly interpret the data.

43

Page 2 of 9

The data products of different payloads of AstroSat are archived and disseminated to the respective Payload Operation Centers (POCs) on regular basis immediately after generation at ISSDC. POCs qualify the data and generate higher level data products and send them back to ISSDC for archival and dissemination to different classes of users worldwide. The dissemination during lock-in and after lock-in period is handled in a flawless manner without any deviation to the defined data policies of AstroSat mission. In Section 2, the overall pipeline for ground operations at ISSDC has been elaborated. The details of Level-0 and Level-1 processing are outlined in Section 3. Section 4 describes the specifics of archival and dissemination pipeline. Section 5 describes the layered approach adopted for automation of data processing, archival and dissemination. Section 6 describes how this pipeline caters to the observation cycle specific - proposal based data dissemination of AstroSat mission through Astrobrowse and Generalized Access and Dissemination Software (GADS).

2. AstroSat pipeline at ISSDC The intrinsic diversity of data products, the distributed network and storage system over which the pipeline is established has imposed challenges in implementing and completely automating the activities of the payload data pipeline at ISSDC.

2.1 Network architecture The architecture of the ISSDC consists of different layers namely, Mission layer (MIS), Archive layer (ARC), Exchange layer (EXH), External Network (EXT) layer. MIS layer interfaces with all the supporting ground stations of AstroSat mission and provides access to the payload and auxiliary data. In addition to interaction with the ground elements, it also provides support for quick look processing and display, data ingest and Level-0 and Level-1 processing. The ARC layer encompasses the ISSDC archives and higher levels of processing. The support provided in this layer includes ingest, media handling, data product generation, migration, integrity and security for data archives. The EXH layer stages and controls exchange of data products between the internal layers (MIS, MOX, and ARC layers) and the EXT Network. Ingestion of

J. Astrophys. Astr.

(2021) 42:43

higher level data products from POCs to the ISSDC archives and dissemination to POCs is carried out in this layer through the Ingest and Dissemination software. This layer is separated from other internal layers by high end security devices. Main functionality of EXT network is to host Web, Mail and File transfer services. EXT layer provides an entry point for users to access data and submit proposals through various web based applications. Figure 1 shows the set of software elements that form part of the pipeline at ISSDC for AstroSat.

2.2 Pipeline elements The proposal processing pipeline has elements for proposal submission, evaluation, scheduling and command generation. The AstroSat Proposal Processing System (APPS) is the main software in the proposal processing pipeline. Proposers, Guest Observers and Instrument teams can submit the proposal through the proposal processing system. The proposals are submitted under various categories which include calibration, opportunity based observation and emergency requests to observe active targets and transient events. The proposals submitted are evaluated for their merit and are accepted or rejected. Various other web applications are deployed along with the APPS software to facilitate the proposal submission activity. Astroviewer is a web based software utility used to generate the probable visibility periods of the celestial bodies which can be observed by AstroSat (Nagamani et al. 2010). AstroSat Exposure Time Calculator is a simulation tool to predict the exposure time required for the various payloads of AstroSat considering parameters like count rate and energy. The visibility periods from Astroviewer and Exposure time prediction are mandatory for proposal submission. The Mission Control and Proposals database (MCAP) is the repository of all the data from the approved proposals. This is accessed for command generation of the payloads. The data received from the spacecraft are termed as raw data. (Pandiyan et al. 2017). The raw payload data collected at the ground station is ingested to ISSDC for further processing, archival and dissemination. The ingested data is subjected to Level-0/Level-1 processing. The Level-1 processed data at ISSDC is disseminated to the Payload Operation Centres (POCs) for quality checks and higher level product generation.

J. Astrophys. Astr.

(2021) 42:43

Page 3 of 9

43

Payload Data Pipeline

Proposal Pipeline Ground Station Gr on

Commands

Data Reception & Segregation S/W

Ground Station –Station Computer

Health Monitoring Software

Data Da

Mission Operations Complex (MOX)

Command Sequence Generation

Ground Station Network

Mission Layer (MIS)

Payload Data Verification –Quick Look Display Software

Scheduling

Payload Data Processing Level-0, 1 Spacecraft Constraint Check

Payload Data Storage –Archival Software

Mission & Proposals Database

Archive Layer (ARC)

Higher Level Processing

File Exchange Service

File Exchange Service

Ingest Software

Dissemination Software

Exchange Layer (EXH)

EXT Network

Bright Source Detection Tool

Common Services: Web Servers, Mail Servers, File Transfer Servers.

Exposure Time Calculator

Astroviewer

Astrobrowse

Proposal Processing Software

Proposals

Data PI/ POC /GO / Registered Users

Figure 1. Proposal processing pipeline software stack.

POCs send back the higher level products for dissemination to Principal Investigators during the lockin period and to all registered users after the lock-in period. The validated and processed data are also archived at ISSDC. The dissemination of data to the POCs is through Virtual Routing and Forwarding (VRF) over the public internet. The dissemination of qualified data to proposers is through the web based software Astrobrowse.

3. Data pipeline for Level-0 and Level-1 processing Level-0 products: Level-0 products are generated from the raw data ingested by the Data Acquisition software. Level-0 products are segmented time-tagged science data in binary format with auxiliary information, for each observation-id. Level-0 products are input to Level-1 pipeline for generation of Level-1 products.

43

Page 4 of 9

J. Astrophys. Astr.

Level-1 products: Level-1 products are mode-segregated science data files along with Time Correlation Table data, Orbit, Attitude, MKF (Make Filter File), and GTI (Good Time Interval) and BTI (Bad Time Interval) data provided in Flexible Image Transport System (FITS) format for Astronomical data analysis. Level-1 Products are provided to POCs for generation of higher level products for public release. AstroSat Level-0/Level-1 data pipeline is a complete automated system. Orbit-wise data products as well as observation-id wise product generation are done in automation without any manual intervention. Observation-id is made available in Mission and Proposals database (MCAP) by constraint check and scheduling software which is fetched by processing software to generate observation-id wise products. Auxiliary data are processed to update Time database and SPICE database (SPICE stands for S = Spacecraft ephemeris, P = Planet, I = Instrument information, C = C-matrix, E = Events used for planning and interpreting scientific observations) for Orbit, Attitude and Time data. Payload chain generates time-tagged, formatted, mode-wise segmented data files along with Time Correlation Table data and Orbit Attitude files also in FITS format. These data files are bundled together and provided to the secure file transfer system software for archival at ISSDC.

(2021) 42:43

At Level-1 processing, MKF consisting of HouseKeeping (HK)/health parameters and Auxiliary data are generated. Level-1 software also provides the GTI and BTI data which takes a combination of parameters from each payload and their range to mark the data interval as good or bad. Figure 2 illustrates a simplified version of one of the payload (LAXPC) processing at Level-1 for AstroSat mission. All the data received are Level-0 processed and kept in separate observation-id wise folders. Once the given observation is completed, a separate processing chain gets triggered to generate merged Level-0 product. This chain merges all the data available in the observation-id folder after removing duplicate records. It also triggers Level-1 processing chain and thus an observation-id merged product gets generated.

4. Data archival and dissemination pipeline at ISSDC The complete proposal based operations of AstroSat takes place through different proposal cycles like Performance Validation (PV), Guaranteed Time (GT), Calibration (CAL), Announcement of Opportunity (AO) and Target of Opportunity (ToO). The data products generated is in the FITS file format and is a bundled data set. The archival of these

Figure 2. Typical Level-1 pipeline for LAXPC payload.

J. Astrophys. Astr.

(2021) 42:43

Page 5 of 9

data products is done by the organizer software based on observation sequences along with the meta-information from the APPS. The dissemination policy and the lock-in period for the data observed and processed during various cycles is specific to every cycle. A lock-in defines the period within which the rights over the access to data are restricted only to the proposer of the data. The data dissemination strategies have been devised to handle such diversity. The Level-1 data products of UVIT, SXT, CZT and LAXPC, Level-0 data products of SSM are generated at ISSDC and disseminated to the POC on regular basis immediately after generation. POCs post back the merged and processed Level-1 data and Level-2 data along with the quality feedback. Observation-id wise Level-1 and Level-2 data products for CZTI, UVIT and LAXPC payloads and orbit wise products for SXT and LAXPC are received at ISSDC. Figure 3 shows the elaborate dissemination scenario at ISSDC. The processed data along with the quality feedback are transferred to the scheduler software for compression and archival. The AstroSat Data Products Tracking Tool (ADAPT) software interfaces with the scheduler and configures the public release date for ingested data in the database. The Astrobrowse software accesses the compressed archive, lock-in information from the database and makes it available for download based on the data dissemination policy. The notification to users and POCs regarding the

Payload Operation Center (POC)

Data Processing System

Proposal Processing System

Level 0/1 data Respective instrument data to identified POC All cycles, immediately after generation at ISSDC Level 1/quality / level2 data - Instrument data from POC for all cycles (except ToO) immediately after generation. ToO data within 2 days of observation

43

availability of data for download is done through the ADAPT software as shown in Fig. 4.

5. Strategies and challenges 5.1 Information and data exchange between data pipeline entities The pipeline elements are distributed over multiple networks spanning different security domains across geographically separated locations. The web based applications APPS, Astroviewer, Exposure Time calculator, Bright Source Detection Tool are hosted in the external network of ISSDC. The Mission-proposals database, flight dynamics, scheduling and command generation software are located in the internal mission network layers at ISSDC and at Mission Operations Complex. The network architecture at ISSDC and the distribution of pipeline entities across various network locations posed an inherent challenge in transferring the required information and data in a reliable, automated and secure manner between the pipeline entities.The interactions between the pipelines entities are through files. The layered approach as described below is used for transfer of files between entities and had ensured a prioritized, guaranteed transfer of necessary inputs for entities in the pipeline.

Proposers

Pipeline automation system

Level 1, quality, Level 2 data Only Configured Instruments data for: *AO, GT Cycles (during lock-in) *Anticipated ToO Cycle (during lock-in)

Level 1, quality, Level 2 data Only Respective piggybacked instrument Piggy back data for AO & GT Cycles. scientists Level 1, quality, Level 2 Calibration data for CAL Cycle

Processed L0/L1 for archival & dissemination

Calibra Calibration Scientists Scienti

Meta Information for archival *Level 1 from DP system *Level 1, 2, quality data from POC *Meta information from Proposal Processing System ISSDC Storage

Level 1, quality, Level 2 data (after lock in) for AO, GT and anticipated ToO. Without lock-in for anticipated ToO Registered Users

Figure 3. Astronomical data processing, archival and dissemination scenario.

43

Page 6 of 9

J. Astrophys. Astr. POC

(2021) 42:43

Proposers/ piggy back scientist /calibration scientists (Access via Internet)

(Access via AstroSat VRF) L1/Qal/L2

File available event

Level 1 data

Data download

Mail notification

ISSDC mail service

ADAPT

Web display for download

Meta info DB

Browse L1 /Qal/L2

Meta info Ingestion

Scheduler ISSDC –External

Compressed archive

L1/Qal/L2 L2 from POC OC

Level 0 / Level 1 Processing

L data for L1 SXT, LAXPC S

ISSDC Internal archive

ISSDC –Internal

Figure 4. AstroSat archival and dissemination pipeline.

5.2 Layered approach: Automation of astronomical data processing, archival and dissemination A layered approach to handle the Automation of processing, archival and dissemination of diverse categories of AstroSat data has been implemented at ISSDC, as shown in Fig. 5. This section describes the function of different layers.

Data products audit layer

Archive management

Dissemination Management

Event Management

Notification Layer

File Management layer

Figure 5. Layered model for astronomical data processing, archival and dissemination.

File management layer: This layer verifies the completeness and correctness of the files received from Level-0 and Level-1 processing software and the POCs based on the check sum file sent along with the data products. The software layer interfaces with the Level-0/Level-1 processing software ingestion of raw data received from different ground stations and collects the processed data for archival and dissemination to the POCs. This layer also handles the reception and ingestion of final products from POCs for archival at ISSDC and for dissemination to the proposers and public release. Event management layer: Events are defined to track every activity in the pipeline, which include events for data ingestion, reception of incomplete/ incorrect data sets, unavailability of data sets, data product generation and successful ingestion of data products for archival. This layer tracks and transfers the events to the relevant modules for necessary action.

J. Astrophys. Astr.

(2021) 42:43

Notification layer: This layer receives the events related to the successful ingestion of data products to the archival software. Then it accesses the database for proposer related information like the emailid and user-id, generates file availability messages, and transfers them to the mail server for notifying the proposer. It receives the events related to incomplete/incorrect files received from the POC and sends notification to POC to repost the incorrect data sets. This layer interfaces with the Exception Notification System (ENS) to alert and send logs to the designers of data processing software upon encountering a software exception for quick resolution of issues. Data products audit layer: This layer receives the events related to data products generated at ISSDC, data products posted from POC, their availability and dissemination status. It performs an audit to find the missing data products and keep track of the different versions of the data products. It periodically generates statistics on the data status at ISSDC and generates messages for notifying the same. It also receives the feedback from POCs regarding the duplicate data and data-sets which cannot be processed at POC and records them in the database. Archive management layer: This layer receives the files for archival from Level-0/Level-1 and POCs, interfaces with the proposal processing software and generates meta information file for every observation like the co-ordinates of the celestial sources, source name, category, mode of observation, science values of the observation, quality of data, percentage of usable data etc. It organizes the data products in the storage based on the observation- id, proposal-id and instrument-id along with the meta-info file. Dissemination management layer: This receives the events regarding the successful ingestion of data products and sets the public release dates for every data set in the database. If the ingestion event corresponds to a ToO data, then it sets the release date as the ingestion date. The Astrobrowse software reads the release date information from the database and presents the authorized view to the proposers and general users for downloading the data. A check sum file is generated along with the actual data to ensure the integrity of data during transmission. Older versions of the same data products are removed from the storage based on the data retention policies.

Page 7 of 9

43

5.3 Data processing at ISSDC At the data processing level, the main challenge has been that every payload in AstroSat can be considered a separate mission in itself considering the varying time correlation, payload data handling and observation-id wise data product generation requirements. The challenges in handling large volumes of data from different payloads have compelled the system to be more robust. Observation-wise product generation has the complexity of knowing when the observation has ended as the order of dumping data is based on station availability and spacecraft constraints. Observation which have spanned for a number of days (7–10 days) has provided data volume in the order of *50 GB (especially UVIT payload) to be processed. This has called for the optimal use of compute and storage resource to process high volume data without any failure in reasonable timelines. Another challenge in designing the software was to develop highly reliable software system to successfully process 14 orbits of data per day. In spite of all the above, there have been backlogs due to late update of attitude information, any minor malfunction of a payload component necessitating reset etc. A dedicated processing infrastructure has been established at ISSDC to process the backlog data without affecting the regular operations. Taking into consideration the redundancy in data acquisition at the ground stations, there are multiple streams of data available on ground for processing at the same time. Availability of datasets from multiple data ingestion system triggers the data merging software which merges the data reception at any of the ground station chains. These merged data are provided to the Level-0/1 data pipeline which takes care of the payload specific processing. A Filter File which is basically collection of required HK and Auxiliary data information is also generated. Every payload has its own science data format, processing requirement and product definition. For e.g., UVIT data are organized as data blocks with header providing payload configuration information, while LAXPC and SXT data records have mode information in every payload line whereas CZTI payload processing requires to consider quadrant information along with mode and SSM payload processing requires event information decoding. Each payload has its own clock which needs to be decoded and correlated with Spacecraft Positioning System (SPS) time accurately. UVIT even has a separate

43

Page 8 of 9

clock for each detector, with one of them set as master, which added further complexity to the system. At Level-1 processing, MKF which is basically collection of required HK and Auxiliary data are also generated. Along with this, there has been a requirement of generation of Good Time Interval (GTI) and Bad Time Interval (BTI) data which takes a combination of parameters from each payload. For BTI files, qualify flags are also provided to mark which parameters are at fault for the bad data indication. To take care of different payload parameters and data processing conditions, ‘‘expression evaluation’’ techniques have been employed to make the software configurable and adaptive. The data are processed inmemory to avoid multiple disk access and to provide better throughput. 5.4 Target of Opportunity (ToO) data dissemination challenges and strategies Target of Opportunity (ToO) proposals require the immediate observation of an astronomical event. These proposals are subjected to fast-track review process which is different from normal procedure for Announcement of Opportunity (AO) proposals. Hence, as a policy, the dissemination of the Target of Opportunity data has a turnaround time of 7 days from the time of observation. This turnaround time includes any delays in processing from both ISSDC and POCs. Once the quality feedback, Level-1 and Level-2 products are available from POC, the ADAPT ingests the same to the archival software and sets the release date of the data products as per the ToO data dissemination policy. The Astrobrowse software configures the release of data as per the release date set by the dissemination management layer. The challenge encountered in this scenario is that in cases where the turnaround time is not met and the data are not available for public download within 7 days; this would result in delay of analyzing critical celestial events, especially when there are observations carried out in co-ordination with other international observatories. To overcome this challenge, the ToO data immediately after generation at ISSDC are also made available for public download with a disclaimer on the quality of data. This is done using the Generalized Access and Dissemination System (GADS) software. GADS provides an authorized access to data at ISSDC through customized views, while maintaining transparency from

J. Astrophys. Astr.

(2021) 42:43

various underlying system, storage and network complexities. After the qualified data are available from POC, the data are made available to the public users through Astrobrowse and the unqualified ToO data from GADS are removed. This allows users to effectively download and use the data observed under the ToO category 6. Future work on pipeline operations Data mining techniques based on resource usage heuristics and archived meta-information is planned to be implemented to bring about intelligent resource planning. Natural Language Processing (NLP) based agents for autonomous event generation, pipeline monitoring, pipeline configuration, alert notification and self-recovery from pipeline failures through backtracking is also envisaged. 7. Conclusions The strategy adopted for handling the challenges in astronomical data processing, archival and policy based dissemination for AstroSat has proved to be effective and has provided a platform to handle future proposal based missions. The strategy has led ISSDC to envisage a unified intelligent framework for having access to download data against the serviced proposals. Acknowledgements The authors would like to thank and acknowledge the support provided by Indian Space Science Data Centre-ISTRAC, Space Application Centre (SAC), IUCAA, IIA, TIFR, RRI, Space Astronomy GroupURSC, PPAD – URSC, Spacecraft Operations team – ISTRAC and Space Science Programming Office, ISRO HQ for their extensive support in establishing and maintaining the AstroSat Processing, Archiving & Dissemination pipeline at ISSDC.

References Agrawal P. C., Yadav J. S., Antia H. M. et al. 2017, J. Astrophys. Astr. 38, 30 Nagamani T., Bharadwaj N., Dakshayani B. P., Pandiyan R. 2010 AstroSat Software Tool to Aid Celestial Source

J. Astrophys. Astr.

(2021) 42:43

Viewing, 61st International Astronautical Congress, Prague, CZ, Paper IAC-10-A3.4.6 Navalgund K. H., Suryanarayana Sarma K., Gaurav P. K. et al. 2017, J. Astrophys. Astr. 38, 34 Pandiyan R., Subbarao S. V., Nagamani T. et al. 2017, J. Astrophys. Astr. 38, 35 Ramadevi M. C., Seetha S., Bhattacharya D. et al. 2017, Exp Astron 44, 11

Page 9 of 9

43

Rao A. R., Bhattacharya D., Bhalerao V. B. et al. 2017, Curr. Sci. 113, 595 Singh K. P., Tandon S. N., Agrawal P. C. 2014, SPIE 9144, 1 Singh K. P., Stewart G. C., Westergaard N. J. et al. 2017, J. Astrophys. Astr. 38, 29 Tandon S. N., Hutchings J. B., Ghosh S. K. et al. 2017, J. Astrophys. Astr. 38, 28

 Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:28 https://doi.org/10.1007/s12036-021-09706-6

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

MISSION

AstroSat science support cell J. ROY1,* , MD. S. ALAM1, C. BALAMURUGAN2, D. BHATTACHARYA1,

P. BHOYE1, G. C. DEWANGAN1, M. HULSURKAR1, N. MALI1, R. MISRA1 and A. PORE1 1

Inter-University Center for Astronomy and Astrophysics (IUCAA), Ganeshkhind, Pune 411 007, India. ISRO Telemetry, Tracking and Command Network (ISTRAC/ISRO), Bengaluru 560 058, India. *Corresponding Author. E-mail: [email protected] 2

MS received 7 November 2020; accepted 8 January 2021 Abstract. AstroSat is India’s first dedicated multi-wavelength space observatory launched by the Indian Space Research Organisation (ISRO) on 28 September 2015. After launch, the AstroSat Science Support Cell (ASSC) was set up as a joint venture of ISRO and the Inter-University Centre for Astronomy and Astrophysics (IUCAA) with the primary purpose of facilitating the use of AstroSat, both for making observing proposals and for utilising archival data. The ASSC organises meetings, workshops and webinars to train users in these activities, runs a help desk to address user queries, provides utility tools and disseminates analysis software through a consolidated web portal. It also maintains the AstroSat Proposal Processing System (APPS) which is deployed at ISSDC, a software platform central to the workflow management of AstroSat operations. This paper illustrates the various aspects of ASSC functionality. Keywords. AstroSat—AstroSat science support cell.

1. Introduction AstroSat is the first Indian multi-wavelength astronomy observatory, operational for the last five years. It was launched by the Indian Space Research Organisation (ISRO) on 28 September 2015 at 10:00 AM and placed at an altitude of 650 km above the Earth into a circular orbit at an inclination of 6 and an orbital period of 98 min. AstroSat operates in specific bands over the energy range from  1 eV to 100 keV by means of 4 co-aligned instruments, one UV imaging telescope and three co-aligned X-ray astronomy instruments. The Ultra Violet Imaging Telescope (UVIT) telescope (Tandon et al. 2020; Subramaniam et al. 2016) is capable of simultaneous observations in the optical, near and far UV bands. A Soft X-ray Telescope (SXT) (Singh et al. 2016, 2017) observes in the 0.3–8 keV band. The hard X-ray instruments Large Area X-ray Proportional Counter (LAXPC) This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

(Antia et al. 2017; Agrawal et al. 2017) and Cadmium–Zinc–Telluride Imager (CZTI) (Rao et al. 2017; Chattopadhyay et al. 2016; Vadawale et al. 2015) are sensitive in the 3–80 keV and 10–150 keV bands respectively. There is also a Scanning Sky Monitor (SSM) (Ramadevi et al. 2017a, b) to monitor variability of cosmic X-ray sources in 2.5–10 keV. With five specialized instruments on-board, AstroSat’s mission is to observe a wide variety of astronomical sources such as star-forming regions, violent explosions like gamma-ray bursts and high energy emission from systems harbouring black holes, neutron stars, white dwarf. For the last five years, AstroSat has been producing unprecedented highquality data from cosmic sources. It is a proposaldriven, observatory class mission, with a large fraction of the observing time open to national users. AstroSat has already observed  1200 distinct targets in  2100 pointings. Moreover, 239 observations of Target of Opportunity (ToO) have been carried out. These observations are providing a unique opportunity to a large number of scientists from all over the

28

Page 2 of 6

country to carry out front-line research using a state of the art indigenous national facility. AstroSat has huge potential to attract young students and researchers in Indian Universities and institutions and engage them in high-quality astronomical research. Moreover, young teachers in Universities and Colleges would also be interested in using AstroSat observations. However, since this is the first time such an observatory is available in India, the young scientists would necessarily require initiation, training and mentoring, if they have to accurately utilize the AstroSat opportunity. There are already  1369 users AstroSat from 42 different countries. There is a critical need to address the queries from these users and to facilitate optimal usage of AstroSat data. To address these requirements, the Indian Space Research Organization (ISRO) and IUCAA have established the AstroSat Science Support Cell (ASSC).1 The cell which is hosted at IUCAA started its operations in May 2016. With most of the AstroSat data made available publicly and more such releases expected in the near future, the role of ASSC is critical in providing the science support. We describe the main objectives and functionalities of ASSC in assisting Indian and International community to participate in observations and using data from AstroSat.

2. Functionalities of ASSC 2.1 Overview of the web-portal ASSC web page serves as a web repository of all the software, latest updates, proposal cycles, publications and link to various web pages relevant for proposing with AstroSat and data analysis. ASSC has a centralized web portal which provides AstroSat users with links to tools developed at ISRO, Payload Operation Centres (POC) as well as documents/tools useful for scientific analysis. The document section in Proposers page allows user to access the Proposers Guide which explains the details of the AstroSat proposal preparation and submission procedure using the APPS software, it also acts as the proposers manual for the APPS software. There is a red book available in the Document section which gives the details of all the accepted AO and GT proposals along with the abstract. The redbook helps to check if there is already an observation proposed or completed for the source of interest that the proposer wants to

J. Astrophys. Astr.

propose. The home page of the web portal provides the AstroSat schedule viewer link2 to check the status of a proposed observation or the planned scheduling of the observations. Further, AstroSat hand book is provided to the users explaining the technical details about the design and characteristics of each of the payloads on-board the satellite, instrument calibration, selection of filters, different modes of operations, the primary scientific objectives intended with these payloads. The relative angle between the payload boresights3 (UVIT, SXT, LAXPC) before and after alignment corrections is also available which enables the proposer to account for the offset of secondary instruments for the proposed observation with the choice of primary instrument. After a successful observation of a proposed target the procedure of the data processing, data rights, proprietary period, etc. are provided in the document ‘‘AstroSat data usage guidelines’’. The ASSC web portal maintains up to date information and online tools to help users propose for AstroSat observations (explained in details in the next subsection on software and other resources for proposal writing). The proposers page provides the pieces of information on the upcoming, ongoing and past announcements of opportunities cycles, scientific and technical proposal templates, online tools such as exposure time and visibility calculators, simulation and necessary documents such as the proposers’ guide. For data processing and analysis, the latest available pipelines for all four instruments are maintained including sample data which allows the user to actually try out the analysis (explained in details in the next subsection of software and other resources for data analysis). The results of the analysis of the sample data are made available allowing the new user to evaluate whether his/her analyses match with the standard ones or not. Additional important software such as the AstroSat orbit file generator, the barycentric correction code and the AstroSat time converter developed in-house at the ASSC are made available through the portal. ASSC has also developed advanced scientific tools such as to compute frequency and energy-dependent time lag from LAXPC data. The ‘‘Recent Updates’’ panel in the right side of the page informs all the users about the recent updates and announcements relevant to AstroSat for example recent software updates, proposal cycle deadlines, health of any instrument, etc. The home page of the web page displays all the publications based on 2

1

http://AstroSat-ssc.iucaa.in/.

(2021) 42:28

https://webapps.issdc.gov.in/MCAP/. http://AstroSat-ssc.iucaa.in/?q=documents.

3

J. Astrophys. Astr.

(2021) 42:28

Page 3 of 6

observation. This tool tells about the safety of UVIT filter based on the modelling from visible magnitude. There are other tools to check the safety of VIS and FUV/NUV channels of UVIT separately. UVIT VIS filter checking tool (Theia)7 is used for VIS filter whereas a tool UVIT FUV/NUV filter checking tool (Gaia)8 is used for FUV and NUV filters. Both are developed by UVIT POC. ASSC also provides a UVIT 9-point coordinate generator script, developed by ASSC itself. This tool generates 9 points around the proposed source to quickly scan the area around the proposed target. This is essential for those fields for which prior UV observations do not exist. • Exposure time calculator for UVIT instrument: ASSC provides two separate online tools, i.e. UVIT exposure time calculator at IIA9 and UVIT exposure time calculator at Calgary (Leahy 2014),10 to estimate the exposure time. This tool calculates the exposure time by taking source characteristics as input. It gives the exposure time to achieve a particular signal to noise ratio. If a user provides the exposure time as input then, in that case, it will give the signal to noise ratio in the output. • WEBPIMMS: For observations of X-ray sources, if a user has prior knowledge of some spectral characteristics of the source with other space-based telescopes like XMM-Newton, Chandra, etc. or with the same telescope AstroSat, the online tool WEBPIMMS,11 developed by ASSC based on HEASARC tools predicts the observable count rates based on the spectral properties provided by the user. This tool can be used for instruments SXT, LAXPC, CZTI and SSM. It uses some simple inbuilt models.

AstroSat data from all the instruments. Home page also announces the AO cycle opening from time to time as advertised by ISRO. ASSC web page also displays the picture of the month based on AstroSat results which is regularly updated.

2.2 Software, web tools and resources 2.2.1 Software and resources for proposal preparation. ASSC provides the software, online/ offline tools, documents, etc., necessary for the proposal writing. It is a single platform which contains proposal preparation materials for all the instruments UVIT, SXT, LAXPC, and CZTI onboard AstroSat. Using these materials, users can simulate energy spectrum, power spectrum, image, etc. They can find the optimum value of exposure time to achieve a scientific goal with AstroSat observations. As the satellite time is very precious, it is a crucial step in proposal writing and acceptance of the proposal. These proposal preparation tools are described below in more detail: • AstroSat visibility: ASSC provides AstroSat visibility calculators, i.e. ASTROVIEWER developed by ISRO4 and AVIS5 developed by ASSC, which tell the user whether a target of interest will be visible to AstroSat for observations during specific periods. There may be earth occultation when user wants to observe the source. ASTROVIEWER incorporates visibility constraints arising from the Sun, the Moon, the Earth and the telescope ram angle in the orbit. AVIS, in addition, also provides instrumentspecific visibility. • Bright source checking tools: For the safety of the UVIT detector, it is recommended to check for the presence of bright sources in the Field of View (FoV) which may damage the instrument irrespective of whether these bright sources are the subject of the proposal or not. ASSC provides link to a tool, called Bright Source Warning Tool (BSWT),6 developed by UVIT POC. This tool generates the list of bright visible sources with a magnitude which are near the proposed position. It also mentions that whether any source is too bright to observe or not and which filter is not safe for the 4

https://webapps.issdc.gov.in/astroviewer/jsp/UserInput.jsp. http://AstroSat-ssc.iucaa.in:8080/AstroVisCal/. 6 https://uvit.iiap.res.in/Software/bswt. 5

28

If a user wants to simulate spectrum of a source with any of the AstroSat instrument, ASSC provides two softwares, i.e. AstroSat.sl and Event_Simul, developed by ASSC, for spectral and temporal data respectively. For UVIT, ASSC also provides a tool UVIT simulator, developed by UVIT POC, to simulate image. All these software tools can be found 7

http://uvit.iiap.res.in/Software/theia/. http://uvit.iiap.res.in/Software/gaia/. 9 https://uvit.iiap.res.in/Software/etc. 10 http://uvit.ras.ucalgary.ca/cgi-bin/UVIT.cgi. 11 http://AstroSat-ssc.iucaa.in:8080/WebPIMMS_ASTRO/index. jsp. 8

28

Page 4 of 6

J. Astrophys. Astr.

• LAXPCsoftware FORMAT A software to do all tasks separately. • laxpc_soft FORMAT B software to do all tasks in one step. • LAXPC Level 2 Data Pipeline Ver 3.1 is another package to analyse LAXPC data.

at the section ‘‘downloadable resources’’ of the web page ‘‘proposal preparation’’.12 • Spectral simulation code AstroSat.sl: An ISIS code AstroSat.sl, can be used to simulate the energy spectrum using any model. It is a single code which can work for all instruments UVIT, SXT, LAXPC and CZTI. Users can get the expected count rate for input spectral parameters for all the instruments. However, a user can take response files from this place and perform the simulation on another platform like XSPEC, etc. also. • Timing simulation code Event_Simul: For the simulation of temporal properties like light curve, power spectrum, detection of quasiperiodic oscillations (QPOs), etc., ASSC provides a simulation code Event_Simul. It works for the LAXPC instrument. • UVIT Image simulation code: An image simulation code, UVIT_simulator to simulate the image of the source. In this software, a user provides the image of the same source with other telescopes and some parameters like exposure time, size of the field-of-view in arcsec, effective area of UVIT detector, etc. as input. Finally, a user can prepare a proposal using the tools mentioned above. The user needs to prepare the proposals using scientific justification and technical justification templates and submit the proposal through the AstroSat Proposal Processing System (APPS). 2.2.2 Software and resources for data analysis. Since photons from an astronomical source can interact with the instrument, background photons can also reach on the detector. We should have some procedure to differentiate between source’s genuine feature and unwanted signals. We also need some methods to convert the observational data into a particular format, which can be used to generate spectra, images, etc. For this purpose, ASSC brings necessary software tools like pipeline software, calibration files, etc., from the respective POC of instruments. It also develops some tools. It also has some data analysis tools. The ASSC web-portal makes available these tools and related documents. All these materials can be found on the webpage ‘‘Data & Analysis’’.13 ASSC team updates the materials from time to time. There are three software that can be used to analyze LAXPC data. 12 13

http://AstroSat-ssc.iucaa.in/?q=proposal_preparation. http://AstroSat-ssc.iucaa.in/?q=data_and_analysis.

(2021) 42:28

LAXPC gain record and HV record are also updated regularly. For SXT data analysis, other than pipeline software, ASSC provides software to merge orbit wise event files and to generate the ancillary response file (ARF) file formation.14 These softwares are also developed by SXT POC. ASSC also provides a higher-level resource for data analysis. A software COMPT-Time-Lag-RMS, developed by ASSC, is used to study the dependence between spectral and temporal properties. It can compute the energy-dependent time lag when seed photon temperature is varied or coronal heating rate is varied. Other than this, ASSC provides an online tool to generate an orbit file15 which is useful in the barycentre correction. Software related to the barycentre correction (as1bary) is also provided.16 An online tool for the conversion between AstroSat time and other time (like Julian date, modified Julian date, etc.) and vice versa is also available here.17 Documents related to the installation of pipeline software and analysis procedure are either attached with the software or separately available. As for UVIT data analysis, two documents UVIT_Pipeline_ Cookbook_v5.pdf and UL2P_quick_install ation_and_output_product_help_v9.pdf18 can be used to understand the procedure. At the same time, the SXT pipeline guide is attached to the pipeline software. For LAXPC data analysis software LAXPCSOFT, readme files for all the tasks are attached to the software. A user can follow these files to understand the analysis. For the CZTI, ASSC provides the user guide for the same.

2.3 APPS APPS19 is an online web page assisting scientists in proposal preparation, submission, scientific and technical review and selection process. APPS caters to 14

http://AstroSat-ssc.iucaa.in/?q=sxtData. http://AstroSat-ssc.iucaa.in:8080/orbitgen/. 16 http://AstroSat-ssc.iucaa.in/?q=data_and_analysis. 17 http://AstroSat-ssc.iucaa.in:8080/AstroSattime/. 18 http://AstroSat-ssc.iucaa.in/?q=uvitData. 19 https://apps.issdc.gov.in/apps/auth/login.jsp. 15

J. Astrophys. Astr.

(2021) 42:28

different types of users including the General or Guest Observers, Payload Operation Centre team members, payload scientists and proposal reviewers. APPS is deployed at the Indian Space Science Data Centre (ISSDC), ISRO, Bangalore. APPS has been successfully used for proposal preparation, submission and selection for observations in the (i) performance verification (PV) phase in the first six months of AstroSat operations, (ii) six Guaranteed Time (GT) cycles, (iii) ten Announcement of Opportunity (AO) Cycles, Target of Opportunity proposals, Calibration proposals, and Legacy proposals. Details of the APPS is provided in Balamurugan (2021). Over the last five years, the ASSC at IUCAA has played a crucial role in fixing of bugs and security issues, documentation, and feature enhancements. They have visited ISSDC, Bangalore and coordinated with the ISRO team to make APPS software secure and on managing resources and memory leakage in APPS software.

2.4 AstroSat helpdesk and visitor programmes The ASSC team at IUCAA manages AstroSat helpdesk20 to address a large number of queries from AstroSat users on proposal preparation, software installation and usage and data analysis. These queries are based on the issues faced by the proposers in procuring data based on the announcement of opportunity and target of opportunity observation cycles, issues related to software installation and analysis procedure, addressing software related bugs. Helpdesk re-directs the more involved inquires to the respective Payload operation centres. AstroSat helpdesk facilitates the Indian and international proposers with proposal based queries and ensures smooth proposal submission during the proposal cycles announced at regular intervals from ISRO. Some of the queries which are generic to all the users are listed in the frequently asked questions (FAQ21) list available at ASSC web page. In the last 4 years, the ASSC hosted more than  300 visits by Ph.D. students, University and College teachers, as well as a few experts who interacted with the ASSC staff and used its facility. Help was provided to these visitors, who came from different parts of the country, to analyze the data they had from their own proposals or given to them.

20 21

[email protected]. http://AstroSat-ssc.iucaa.in/?q=faq.

Page 5 of 6

28

2.5 Workshops and meetings organized by ASSC ASSC organizes national workshops and meetings in various locations across India since July 2016. These workshops were a set of extensive demonstration sessions where the participants were instructed on how to propose for AstroSat observations, in particular the choice of how to configure each instrument, exposure times, filters etc., in order to optimize the science output. Proposal writing techniques, simulation of expected science results were explained. Other than the scientific and technical justification details and types of proposals to be sought for during an announced cycle, these workshops focus on hands-ontraining sessions of data analysis techniques of different instrument onboard AstroSat. The content of the webinars and a few talks are also available in the web-portal. ASSC also conducts refresher course in Astronomy and Astrophysics for teachers at Indian universities and colleges for observational and theoretical aspects of astronomy and state-of-the-art methods of data analysis, especially with AstroSat. ASSC trained more than 489 participants to enable them to propose with AstroSat and to handle data analysis of AstroSat during its 4 years of existence. A Workshop was conducted where experts guided M.Sc./M. Phil. students to analyze and interpret AstroSat data, which directly led to the publication of the results with the students as co-authors Mudambi et al. (2020) and Jithesh et al. (2019). Details of completed and upcoming workshops are regularly updated in the web page http://astrosat-ssc.iucaa.in/ ?q=workshops. The distribution of participants’ affliations and location of the workshops organized by ASSC is shown in Fig. 1. AstroSat related meetings such as those of the time allocating committee and the Science Working group are also organized by the ASSC.

2.6 Resources relevant to publication and calibration The calibration web page of ASSC website provides the list of calibration sources observed by the AstroSat satellite. The list mentions the effective exposure of each of the observation instrument wise. It is highly beneficial for the user to use this data for calibration purposes. Some of these datasets are open access and some proprietary of the proposer or the instrument teams. This page also provides the list of publications based on calibration and performance of the payloads onboard the AstroSat satellite.

28

Page 6 of 6

J. Astrophys. Astr.

(2021) 42:28

Figure 1. Pie chart (a) shows the distribution of participants affliations of undergraduate colleges affiliated to universities, universities and research institutions, and pie chart (b) presents the locations of different workshops organized by ASSC.

With a lot of the AstroSat data being released to the public and more such releases expected in the near future, the ASSC will be critical in providing the opportunity and means for the larger scientific community to take part in optimal utilization of AstroSat data. Acknowledgements ASSC team would like to thank AstroSat mission of the Indian Space Research Organisation (ISRO), the Indian Space Science Data Centre (ISSDC) team. We thank all the instruments POC teams for their support for solving queries and time to time help with new software updates. References Agrawal P. C., Yadav J. S., Antia H. M. et al. 2017, J. Astrophys. Astr., 38, 30 Antia H. M., Yadav J. S., Agrawal P. C. et al. 2017, ApJS, 231, 10 Balamurugan C. et al. 2021, J. Astrophys. Astr., 42, https:// doi.org/10.1007/s12036-021-09703-9 Chattopadhyay T., Vadawale S. V., Rao A. R. et al. 2016, Proceedings of SPIE 9905, Space Telescopes and

Instrumentation 2016: Ultraviolet to Gamma Ray, 99054D Jithesh V., Maqbool B., Misra R. et al. 2019, The Astrophys. J., 887(1) Leahy D. 2014, Astronomical Data Analysis Software and Systems XXIII, Proceedings of a meeting held on 29 September–3 October 2013 at Waikoloa Beach Marriott, Hawaii, USA, edited by N. Manset and P. Forshay, ASP Conference Series, 485, 69 Mudambi S. P., Maqbool B., Misra R. et al. 2020, The Astrophys. J. Lett., 889(1) Ramadevi M. C., Ravishankar B. T., Sitaramamurthy N. et al. 2017a, J. Astrophys. Astr., 38, 32 Ramadevi M. C., Seetha S., Bhattacharya D. et al. 2017b, Exp. Astron., 44, 11 Rao A. R., Bhattacharya D., Bhalerao V. B. et al. 2017, Curr. Sci., 113(4), 595 Singh K. P., Stewart G. C., Westergaard N. J. et al. 2017, J. Astrophys. Astr., 38, 29 Singh K. P., Stewart G. C., Chandra S. et al. 2016, Proceedings of SPIE, 9905, id. 99051E pp Subramaniam A., Tandon S. N., Hutchings J. B. et al. 2016, SPIE, 99051F Tandon S. N., Postma J., Joseph P. et al. 2020, ApJ, 159(4), id.158 Vadawale S. V., Chattopadhyay T., Rao A. R. et al. 2015, A&A, 578, A73

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:45 https://doi.org/10.1007/s12036-021-09755-x

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

MISSION

AstroSat science operations and userbase V. GIRISH1,*, LEO JACKSON JOHN2, C. BALAMURUGAN2 and K. R. SANGAMESH2 1

ISRO Headquarters, New BEL Road, Bangalore 560 094, India. ISRO Telemetry and Tracking Center, Peenya, Bangalore 560 058, India. *Corresponding author. E-mail: [email protected] 2

MS received 7 November 2020; accepted 19 April 2021 Abstract. AstroSat, India’s first observatory class satellite, operates as a proposal based observatory. AstroSat carries five science payloads capable of simultaneous observations in a broad energy band. In this article, we will summarize the science operations of AstroSat, AstroSat user distribution, user support available for AstroSat users and publications related to AstroSat and AstroSat data. Keywords. AstroSat—X-ray—ultra-violet—space observatory.

1. Introduction AstroSat is India’s first multi-wavelength observatory class satellite dedicated for astronomy. AstroSat, carrying five scientific payloads (Singh et al. 2014) and one auxiliary detector charge particle monitor, was launched into a low earth orbit with low inclination on 28th September 2015 from Sriharikota Range (Seetha & Megala 2017). The near equatorial orbit was chosen to reduce the effects of enhanced charged particle concentration over the South Atlantic region popularly referred as South Atlantic Anomaly (SAA). Four payloads onboard AstroSat, viz., Ultra Violet Imaging Telescope (UVIT), Soft X-ray Telescope (SXT), Large Area X-ray Proportional Counter (LAXPC) and Cadmium Zinc Telluride Imager (CZTI) are capable of simultaneous observations over a broad energy band from UV to high energy X-rays. The fifth science payload is an all sky monitor called Scanning Sky Monitor (SSM). The broadband coverage of AstroSat is shown in Fig. 1. AstroSat is a multi institutional mission. Five of the payloads onboard AstroSat were built by collaborating institutes which are Tata Institute of Fundamental Research (TIFR, Mumbai, India), Indian Institute of This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

Astrophysics (IIA, Bengaluru, India), Inter-University Center of Astronomy and Astrophysics (IUCAA, Pune, India) in collaboration with Center for Space Astronomy (CSA, Canada) and Leicester University (UK). The fifth payload SSM was built by Indian Space Research Organisation (ISRO, India). Several centers of ISRO were involved in the realization of these payloads (see Tandon et al. 2017; Agrawal et al. 2017; Singh et al. 2017; Rao et al. 2017; Ramadevi et al. 2017 for payloads realization and calibrations). After the launch, the payloads were switched on in a pre-determined order starting with the Charged Particle Monitor (CPM), followed by CZTI, SSM, LAXPC, and SXT. The last payload to be switched on was UVIT. UVIT was turned on almost after two months of launch. This was to avoid contamination due to out-gassing from the components and harness used in the different payloads and overall satellite. Currently all the payloads are working satisfactorily except for near UV part of UVIT, one of the proportional counters of LAXPC and the second detector of SSM. LAXPC3 was switched off due to abnormal gain changes in the second year of operation, while LAXPC1 has been operating at a lower gain since the second year. NUV failed to recover from one of the regular resets during the third year of operation of AstroSat. Though frequent attempts are made to restart NUV, the probability of recovering NUV seems

45

Page 2 of 7

Figure 1. Schematic showing the broadband coverage of AstroSat. The figure clearly shows the overlapping of LAXPC coverage with that of SXT and CZTI.

to be negligible. SSM-2 was switched off after a gas leak from the second year of operation. 2. AstroSat proposal cycles The first six months after the launch of AstroSat constituted the performance and verification phase (PV). PV phase was followed by the guaranteed time (GT) for the payload teams for another six months. The regular Announcement of Opportunity (AO) proposals started from the end of the first year. Initially the duration of the AO was six months which was later switched to full year. The call for proposals for regular observations are typically issued in the months of February/March every year for the observation cycle starting from October of that year. The proposals are reviewed by an AstroSat Time Allocation Committee (ATAC) while the proposals are verified for technical feasibility by another committee, AstroSat Technical Committee (ATC). ATAC and ATC are constituted by ISRO and are made up of eminent scientists from India and engineers from ISRO. The proposals can be submitted for observations using all the four co-aligned payloads with one of them as the primary or a subset of the payloads or even only for the primary payload. ATAC requests reports from external referees for grading and selection of the submitted proposals. In addition to AO, AstroSat also services proposals under calibrations of the payloads (CAL) and target of opportunity (TO). TO proposals typically have preference over other observations and the turn-around times for a TO can be as short as two days. AstroSat is currently servicing proposals approved under A10 AO. Under A10, ISRO received a total of 105 proposals, 68 of which were from the Indian community and 37 from the international community. While the ratio of proposals for SXT and UVIT as prime instrument were similar for the Indian research community (42% and 47% respectively), the international community preferred UVIT as the primary instrument compared to SXT (60% and 30%

J. Astrophys. Astr.

(2021) 42:45

respectively). The remaining proposals were for LAXPC as the prime instrument. The international community also submitted twice the number of proposals for Anticipated Target Of Opportunity (ATOO) with 17 proposals compared to 8 ATOO proposals from India. Note that, for estimating the above ratios we have not considered the ATOO proposals. The requested observing time under A10 were 6.11 Ms and 5.78 Ms from India and international community respectively. The maximum requested from India was for SXT as the prime instrument (51.9%) followed by UVIT (36.4%). From the international community the requested time for SXT and UVIT as prime instruments were 42.4% and 54.1% respectively. The observations obtained under AO have a lock-in period of twelve months and they are made public as archival data hosted on AstroBrowse at ISRO Space Science Data Center (ISSDC, Bangalore) and maintained by the ISRO Telemetry and Tracking Center (ISTRAC, Bangalore). ISSDC hosts both calibrated data (Level 1) and scientific data (Level 2) and everybody is allowed to download the archival data after registering with their e-mail. The TO data are made public immediately after the clearance from the respective payload operation centers (POC), archived at ISSDC and released simultaneously to the public and to the PI of the proposal simultaneously. For streamlining AstroSat operations from proposal submission till scheduling of the targets along with planning, ISRO uses a software tool ‘‘AstroSat Proposal Processing System (APPS)’’. APPS and the planner software used for sequencing the targets are discussed in detail in another article (Anshu Chauhan et al. 2021). For completeness, the end to end science operation process is shown in the flow chart in Fig. 2, starting from proposal submission by the principal investigator (PI) till the data reaches the PI. For the safety of AstroSat, safety checks are carried out at multiple stages and finally by the scheduling team before uploading the commands. These final checks are carried out manually and the commands are uploaded 2–3 days before the actual observations by the scheduling team at ISTRAC.

3. Science operations As mentioned earlier, AstroSat operates as a proposal based observatory. Proposals by the PIs are selected after a critical review followed by technical feasibility

J. Astrophys. Astr.

(2021) 42:45

Page 3 of 7

45

Figure 2. Flow chart of AstroSat science operations from proposal processing to data dissemination.

study of each source. The technical evaluation mainly checks for UVIT safety like bright object in the field of view, visibility and safety of UVIT in the selected filters. AstroSat observations at ISSDC are generally planned for two weeks at a time, called as observation cycle. The selected targets for the two week cycle are sequenced based on the visibility and feasibility of observations. The sources sequenced in a two week cycle are not disturbed unless there is an approved TO or in special contingencies. TO sources do not follow this regular sequencing and based on requirement, a TO source can be scheduled for observation. TO sources do not follow this regular sequencing and based on requirement, a TO source can be scheduled for observation within one day of its approval by the TO committee. In case an on-going observation is interrupted for a TO target, the observations of the original target may be continued after the TO observations by either sliding or re-arranging the earlier sequencing. The feasibility of scheduling a target involves checking for stars in the field of view of the two star sensors onboard AstroSat and checking for geometrical constraints. For the safety of the satellite and different scientific payloads, the following constraints are imposed on each observation. The body axis and different angles with respect to AstroSat body are shown in Fig. 3.

(1) Angle between positive ROLL axis and Sun is always maintained at a minimum of 65 . (2) Angle between ?YAW axis and Sun is always more than 100 . (3) Angle between ?ROLL axis and velocity vector (RAM) should not be less than 12 . This constraint is to avoid contamination of UVIT telescope as ploughing through charged particle region results in collection of more particles which if deposited on the primary mirror of UVIT will lead to reduced efficiency. (4) Angle between star sensors and Sun is always maintained at a minimum of 50 . In addition to these constraints, a target is deemed to be observable by AstroSat only if the position of the target satisfies the conditions: (1) Sun Avoidance Maneuver (SAM) feasibility: When moving towards the target, the telescope should always satisfy the Sun angle constraint. (2) A minimum of five stars available in at least one of the star sensor’s Field-of-View (FoV). If the prime instrument is UVIT, this criteria needs to be satisfied for both star sensors. (3) The Phased Array Antenna (PAA) should be available for dumping the data to the ground. This criterion may be relaxed if the observations include only the X-ray payloads, as typically

45

Page 4 of 7

J. Astrophys. Astr.

(2021) 42:45

(a) SAA entry and exit command sequence initiation. (b) Star sensor updates for enable/disable command initiation. (c) Commands for UVIT payload imaging operation and SAM sequencer operation. (d) Downloading science data. The POP macros for scheduling of sources are uploaded usually on daily basis too, unless the scheduled observations of the target exceeds one day. In the case of observations longer than one day, the POP macros corresponds to the next target is uploaded one day before the scheduled observation.

3.2 AstroSat data pipeline Figure 3. Figure showing different angles for estimating the constraints.

UVIT observations forms a significant portion of the overall data volume as compared to X-ray payload data. The mission operation teams at ISTRAC checks for all these constraints and only when the target position satisfies these checks, the target is sequenced for scheduling. Once the spacecraft is oriented towards the target, the spacecraft is inertially held with the spacecraft roll axis pointed to the source.

3.1 Spacecraft operations Once the target is acquired by the observatory, the following spacecraft operations are carried out during the observations. (1) Attitude (orientation of the satellite body frame with respect to the orbit frame at each point) corrections are done regularly for maintaining Sun-pitch angle of 90 ± 0.5 for continuous power generation. (2) Drift corrections for the onboard timer are carried out to keep the drift within 100 ms/day. (3) For maintaining accurate pointing, the Gyro drift compensation values are tuned to maintain the pointing error is within 0:05 /hr. (4) AstroSat orbit coefficients are uplinked to the satellite on daily basis for updating the onboard orbit model. (5) Commands generated by Payload Operation Planner (POP) macros are usually uplinked daily for

AstroSat data pipeline for converting satellite data into Level 1 and Level 2 science products of the four coaligned instruments were developed completely inhouse by Satellite Application Center (SAC) and U R Rao Satellite Center (URSC, Bangalore) with the help of respective payload operation centers (POC). The data from the satellite are first split into individual payload data and combined with house keeping (HK), auxillary data (AUX) and timing data, We call this as Level 1 data. The Level 1 data are sent to POC for validation and for generating next level science data, called Level 2. After validation, both Level 1 and Level 2 are hosted on ISSDC and are available for downloading to the PI and the public depending on the lock-in rules. The processing from the satellite data up to Level 1 data is completely automated and manual intervention is needed only in case of contingency. The generation of Level 2 from Level 1 varies for different payloads. For the different processes involved in Level 2 generation of SXT and CZT, readers are referred to Singh et al. (2017) and Bhalerao et al. (2017). The LAXPC level 1 to level 2 data generation involves generating a standard Good Time Interval (GTI) file for removing earth occultation and SAA part of the data and extracts the lightcurve and spectra along with background lightcurve and background spectra. The UVIT pipeline mainly involves correcting for the satellite drift and produces sky images corresponding to individual orbits as well ascombined image using data from all the orbits. In addition, images corresponding to exposure and errors are also generated.

J. Astrophys. Astr.

(2021) 42:45

Page 5 of 7

45

Figure 4. Global distribution of AstroSat users as on September 2020.

Figure 5. Distribution of sources observed by AstroSat. Different colours represent sources observed as with the corresponding instrument as the prime instrument.

The Level 1 and Level 2 AstroSat data are in standard FITS format and the Level 2 data are compatible with popular analysis tools like HEASOFT, XSPEC, IRAF etc. The Level 2 pipeline software is designed to run on Linux platforms.

The pipeline for processing SSM data was developed by the SSM team. The final science products, lightcurves and hardness ratio of selected sources observed with SSM are archived at ISSDC (Ramadevi et al. 2017).

45

Page 6 of 7

J. Astrophys. Astr.

(2021) 42:45

Figure 6. GRBs detected by CZTI till September 2021. Credit: Varun Bhalerao, private communication.

an AstroSat user is that of registering within the APPS through email. At the end of five years of operations, AstroSat now has 1438 registered users (AstroSat users from nowon) from 48 countries. Figure 4 shows the distribution of AstroSat users. India contributes close to half of these users. For clarity the countries with fewer than 11 users are clubbed under ‘Others’.

5. AstroSat observations and publications

Figure 7. Distribution of publications till September 2020 from AstroSat and AstroSat data over time. For clarity, all the publications before 2014 are grouped together under 2014. The Ph.D. theses includes both awarded and at various levels of submission.

The AstroSat calibration database (CALDB) is hosted at the AstroSat Science Support Cell. ASSC also hosts the software required for analysing AstroSat data and tools required for data analysis. The major software hosted at ASSC are the Level 1 to Level 2 pipeline software of the co-aligned UVIT, SXT, LAXPC and CZTI payloads, and software utilities like routines for barycentric correction, AstroSat orbit file generator etc.

4. AstroSat users From the second year after launch, AstroSat has been servicing proposals received from users (Principal Investigator, PI). The only requirement for becoming

As of September 2020, AstroSat has completed five years of its designed life. We summarize the nature of these observations and resultant publications.

5.1 Targets observed In the first five years of its operation, AstroSat has observed a total of 909 proposals under multiple AO and 264 targets under ToO observations. In addition to these, AstroSat has observed 111 targets during PV phase, serviced 32 targets under GT and 116 targets under calibration. Figure 5 shows the distribution of the positions of all the targets observed by AstroSat till September, 2020. Different colours represent the prime instrument used for the observation. From Fig. 5 we can see that the whole of RA-Dec space is covered by UVIT and SXT whereas most of LAXPC observations are limited to the galactic plane. UVIT sensitivity is susceptible to contamination. To reduce this contamination, the RAM angle of AstroSat observations are confined to a minimum angle of 6 , that can be seen as a gap around the celestial equator

J. Astrophys. Astr.

(2021) 42:45

in Fig. 5. The SSM pointings are not shown in the figure as SSM is a scanning sky monitor and the pointed observations by SSM are performed only for calibration purpose. CZTI has a primary FoV of 4:6  4:6 defined by the collimator. But beyond 100 keV, the collimator and the support structure of CZTI becomes increasingly transparent making CZTI instrument capable of detecting bright transient events like GRBs. CZTI till September 2020 has detected a total of 325 GRBs (Sharma et al. 2021). Figure 6 shows the distribution of CZTI detected GRBs with well localised positions.

5.2 Publications The data from AstroSat has resulted in 153 refereed publications till September 2020. Note that this includes publications based on instrumentation as well. In addition, AstroSat has resulted in close to 1500 non-refereed publications including 270 GCN circulars and 26 Astronomer’s Telegrams. The distribution of publications over the years is shown in Fig. 7. AstroSat and data from AstroSat has been used in fifteen Ph.D. theses which are at various stages of being awarded.

Acknowledgements The authors would like to acknowledge all the POCs, ISRO centers involved in running the operations for their continuous support in the mission operations and

Page 7 of 7

45

data validation. We are thankful to all the people involved from conceiving the satellite, design and developing of the payloads and ISRO for providing a world class observatory for astronomers and supporting the continuous operation of AstroSat. We are thankful to different committees and AstroSat science working group for helping in the optimal utilization of AstroSat.

References Agrawal P. C., Yadav J. S., Antia H. M. et al. 2017, J. Astrophys. Astr., 38, 30 Anshu Chauhan, Nagamani T., Bharadwaj N. et al. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036021-09727-1 Bhalerao V., Bhattacharya D., Vibhute A. et al. 2017, J. Astrophys. Astr., 38, 31 Ramadevi M. C., Ravishankar B. T., Sitaramamurthy N. et al. 2017, J. Astrophys. Astr., 38, 32 Rao A. R., Bhattacharya D., Bhalerao V. B. et al. 2017, Curr. Sci., 113, 579 Seetha S., Megala S. 2017, Curr. Sci., 113, 25 Sharma Y., Marathe A., Bhalerao V. et al. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036021-09714-6 Singh K. P., Stewart G. C., Westergaard M. J. et al. 2017, J. Astrophys. Astr., 38, 29 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Proceedings of SPIE, Volume 9144, 91441S Tandon S. N., Annapurni S., Girish V. et al. 2017, AJ, 154, 128

 Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:35 https://doi.org/10.1007/s12036-021-09694-7

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

DATA PIPELINE

JUDE (Jayant’s UVIT Data Explorer) pipeline user manual P. T. RAHNA1,2,* , JAYANT MURTHY2 and MARGARITA SAFONOVA2 1

CAS Key Laboratory for Research in Galaxies and Cosmology, Shanghai Astronomical Observatory, Shanghai 200030, China. 2 Indian Institute of Astrophysics, Bengaluru 560 034, India. *Corresponding Author. E-mail: [email protected] MS received 28 October 2020; accepted 13 December 2020 Abstract. We have written a reference manual to use JUDE (Jayant’s UVIT Data Explorer) data pipeline software for processing and reducing the Ultraviolet Imaging Telescope (UVIT) Level 1 data into event lists and images—Level 2 data. The JUDE pipeline is written in the GNU Data Language (GDL) and released as an open-source which may be freely used and modified. GDL was chosen because it is an interpreted language allowing interactive analysis of data; thus in the pipeline, each step can be checked and run interactively. This manual is intended as a guide to data reduction and calibration for the users of the UVIT data. Keywords. Ultraviolet—data reduction—pipeline.

1. Introduction This manual describes the JUDE software system for processing and understanding the Ultra-Violet Imaging Telescope (UVIT) data. UVIT was launched on 28 September 2015 on a PSLV-C30 into a 650 km, 6 inclination low Earth orbit by Indian Space Research Organization (ISRO), as part of the multi-wavelength Indian AstroSat mission. The UVIT has seen its first light on 30 November 2015 by observing the open cluster NGC188. UVIT operates in three channels: visible, near-ultraviolet (NUV), and far-ultraviolet (FUV), covering a broad range of UV to the visible ˚ , dividing it spectral region from about 1250 to 5500 A into 15 narrow and broad wavelength bands. The UVIT instrument has been continuously observing since 2015, and the data, including the PV (performance verification) phase observations used to characterize the instrument (Tandon et al. 2017b) with in-flight calibration and verification (Rahna et al. 2017; Tandon et al. 2017a), are being released to the observers. The Announcement of Opportunity (AO) This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

Cycle was initiated in April 2017 for the Indian astronomy community. Though at present (2020) only the FUV and VIS channels are operating (NUV channel has failed in March 2018), the archive is large and open to the public—more than 1000 datasets of UVIT data were made public. The data may be accessed at the AstroSat data archive1 (Astrobrowse) at the Indian Space Science Data Centre (ISSDC), Bangalore. Further data will be released upon completion of the respective proprietary periods (12 months). AstroSat has now completed its 5 years of operation. The official UVIT pipeline is developed and available for download at the Science Support Cell (http://astrosat-ssc.iucaa.in). It is designed to produce scientifically usable images from the raw data— Level 2 data, which are flat-field, distortion, and driftcorrected images with absolute calibration. Canadian UVIT pipeline called CCDLAB is also available for UVIT data reduction (Postma & Leahy 2017). However, the source codes of these two pipelines are proprietary and difficult to modify. 1

https://astrobrowse.issdc.gov.in/astro_archive/archive/Home. jsp—one needs to register and create an account there.

35

Page 2 of 8

We have written the software package JUDE (Jayant’s UVIT data Explorer) entirely in the GNU Data Language (GDL) (Coulais et al. 2010) to reduce Level 1 UVIT data into Level 2 image files and event files. JUDE is released under an Apache License2, which may be freely used and modified. GDL was chosen for the development environment because it is an interpreted language which lends itself to the interactive analysis of data. This was invaluable in the development of the pipeline, where we could check each step and run commands interactively. GDL is an open-source version of the Interactive Data Language3 (IDL) and runs on a wide variety of systems, just as IDL. GDL will run most IDL programs without modification (and vice versa), allowing access to the rich library of utilities developed for IDL over the last four decades, thus easing the development of JUDE. We have tested JUDE using both GDL (version 0.9.6) and IDL with identical results on multiple operating systems. In the remainder of the manual, GDL and IDL may be freely interchanged. JUDE starts with the Level 1 data, provided by the Indian Space Science Data Centre (ISSDC), and produces photon lists and images suitable for the scientific analysis. JUDE is archived at the Astrophysics Source Code Library (Murthy et al. 2016), and the latest version is available on GitHub at: https://github. com/jaymurthy/JUDE. We will continue to update the pipeline as we find errors and also welcome feature requests to improve the scientific utility of the programs.4

1.1 UVIT instrument details The UVIT is one of the five payloads on AstroSat spacecraft. It consists of two identical 37.5 cm Ritchey–Chre´tien telescopes, one is reserved for FUV channel, and the second one sharing NUV and visible (VIS) channels through a beam splitter (Table 1). The VIS channel is not intended for science purposes and is used solely to track the spacecraft (s/c) motion. Each channel is equipped with a 5-filter filter wheel, providing spectral coverage in several passbands from the FUV to the visible. The basic properties of UVIT filters are given in Table 2. The 2

http://www.apache.org/licenses/LICENSE-2.0. https://idlastro.gsfc.nasa.gov/. 4 As was communicated to us by a user, the use of a /notime keyword in the interactive.com code was giving an error message about exposure times. We have corrected this in the code; Nov. 2020. 3

J. Astrophys. Astr.

(2021) 42:35

effective bandwidth Dk in Table 2 is the integral of the normalized effective area, and k0 is the central (or ‘mean’) source-independent wavelength, weighted by the normalized effective areas; these quantities were calculated from the ground calibration values (Rahna et al. 2017). See Tandon et al. (2020) for the latest updated corrections to the ground calibration values.

1.2 Status of JUDE • • • •

The development of JUDE began in June 2016. The production release of JUDE is June 2017. JUDE update: June 2019. The last update of JUDE: November 2020.

2. UVIT data definitions 2.1 Level 0 data Level 0 data is the raw data from the AstroSat spacecraft. This data is sent to the AstroSat Data Center at the ISSDC, where the data is separated by instrument and written into Level 1 data.

2.2 Level 1 data The Level 1 data from ISSDC is sent to the payload operations center (POC) at the Indian Institute of Astrophysics (IIA), Bengaluru. The Level 1 data are distributed as a single zipped archive for each observation. The files within the archive are in a number of sub-directories organised by the orbit number and the type of file, all under a single top-level directory. All data files are FITS binary tables. More details about the data are described in Murthy et al. (2017).

2.3 Level 2 data Once the refined coordinate information is brought into the FITS binary extension tables, it is ready to be converted to an image of processed data from a single scan of the sky. These images are termed Level 2 data. This is the final data product from the pipeline software. Level 2 data are FITS images, readable by any of the standard astronomical data processing packages.

J. Astrophys. Astr.

(2021) 42:35

Page 3 of 8

35

Table 1. Summary of characteristics of UVIT instrument Parameter

Value

Typea Diametera Field of Viewa Detectora Mode of operationa Channelsa Spatial resolutionb Frame ratea Sensitivitya Time resolutionc Pixel scale JUDE pipeline image format a

Ritchey–Chre´tien 37.5 cm 280 CMOS, 512 9 512 Photon counting mode (FUV and NUV) ˚ NUV (2000–3000 A) ˚ FUV (1300–1800 A), .1.500 29 frames/s 20 AB mag. in FUV for 200 s \35 ms 3.2800 (for 512 9 512 window size) 4096 9 4096 pixels with 0.4100 pixel size

Kumar et al. (2012); bRahna et al. (2017); cMurthy et al. (2017).

Table 2. Properties of different UVIT UV filters

Type

Filter name

Passband (nm)

Effective bandwidth Dk (nm)

Central k0 (nm)

CFa ˚ 1 cnt1 ) (erg cm A

ZPb (mag)

FUV

CaF2 -1 BaF2 Sapphire Silica

F148W F154W F169M F172M

125–179 133–183 145–181 160–179

44.1 37.8 27.4 13.13

150.94 154.96 160.7 170.3

3.8689e-15 4.2036e-15 5.4399e-15 1.4273e-14

17.73 17.58 17.22 16.05

NUV

Silica NUV15 NUV13 NUVB4 NUVN2

N242W N219M N245M N263M N279N

194–304 190–240 220–265 220–265 273–288

76.9 27.1 28.17 28.23 8.95

241.8 218.5 243.6 262.8 279.0

1.9459e-16 5.8360e-15 1.0327e-15 6.9611e-16 2.7988e-15

19.95 16.48 18.12 18.39 16.75

Filter channel

a

2

Conversion factors from Rahna et al. (2017). bZero-point magnitudes derived from conversion factors.

3. Installing and setting up the JUDE pipeline (1) Download and install GDL/IDL: JUDE runs under either IDL or GDL. You have to download and install either of them to run JUDE software system. (2) Download the IDL libraries: JUDE uses many library routines from the IDL Astronomy Users’ Library5 and from the Markwardt IDL Library6 that can be downloaded and installed. These routines should be placed in one of the directories pointed to by the !PATH variable. Our recommendation is to create a separate directory for each library and unzip the files into that directory. Update the !PATH system variable to include these directories. 5

https://idlastro.gsfc.nasa.gov/ftp/astron.zip. http://www.physics.wisc.edu/*craigm/idl/down/cmtotal.zip.

6

(3) Download JUDE programs from GitHub.7 Add the JUDE library to the IDL path i.e,: !path = path to jude/: ? !path (4) Download Level 1 data of your target from ISSDC Astrobrowse. Unzip and save the downloaded raw data in a folder called Level1(dname). Note that JUDE will find all relevant files under the dname directory. The individual files may be gzipped to save considerable space. 4. Running JUDE pipeline 4.1 Automatic mode JUDE comprises a number of individual routines that together can be run as a pipeline. A program 7

https://github.com/jaymurthy/JUDE.

35

Page 4 of 8

(process_uvit.com) is included in the JUDE distribution, which may be modified for local use and can be run using the syntax: Step 1. @process_uvit.com (from IDL) or idl process_uvit.com (from terminal). This program chains together the following commands (Note that, if desired, these statements may be entered in sequence at the GDL prompt). You may have to set the path before running process_uvit.com: (1) Setting path (as per above setup): • !path = ‘‘\ path to jude folder [ :’’þ!path; Add JUDE routines to path. • !path = ‘‘\path to cmlib folder[:’’ ? !path; Add Markwardt routines. • !path = ‘‘\ path to idl libraries [ :’’ þ!path; Add IDLASTRO routines. • dname = ‘‘\ path to level1 folder [’’; Location of Level 1 data. (2) Produce Level 2 data: • • • • • •

jude_driver_vis, dname; jude_driver_uv, dname,/nuv,/notime jude_driver_uv, dname,/fuv,/notime jude_verify_files, dname jude_uv_cleanup, /nuv jude_uv_cleanup, /fuv

Description: jude_driver_vis will process the visible files (VIS) and jude_driver_uv will process the UV files to produce event lists and images. When running jude_uv_driver for the first time, it is faster to use the /notime keyword. The time per pixel will be approximate at the edges. Program jude_verify was introduced because there are crashes in GDL for long runs, and it will go through each of the files and ensure that they are, at least, readable. jude_uv_cleanup will process the event lists and images, merge the data, and register the images. Finally, it creates Level 2 data. There are random crashes caused by memory issues which can affect datasets with many files. These errors may occur in any module and at different locations. It was found that 4 GB RAM is enough but, obviously, more RAM is better. In most cases, the program may be restarted and will pick up from where it was left off, skipping files already processed. If the memory problem occurs at a critical point, it is possible that the FITS file may not be properly written. We have added a program (jude_verify_files) which should check the integrity of each FITS file and delete it if there is a problem. If this does not work, it would be best to

J. Astrophys. Astr.

(2021) 42:35

delete any suspect files and start process_uvit.com again. Files already completed will be skipped. If the verification (from jude_verify_files) is successful, a file named jude_VERIFY_FILES_DONE will be created. There will be no further Level 1 processing if this file exists. It should be deleted if processing from the scratch is really desired.

4.2 Interactive mode After the initial run of JUDE using process_uvit.com, you can run it interactively with interactive.com using the syntax: Step 2. @interactive.com (from IDL) or idl interactive.com (from terminal). It will open a window with the image and ask a few questions: • Enter new image scaling (Enter if ok)? If image is too messy or crowded, you can change the contrast (scale), e.g. 0.001. • Parameter to change? -1 to continue, -2 to exit, -3 to debug: It will display default parameters on the screen: The explanation of these parameters is given in Table C.5. in the appendix to the JUDE paper (Murthy et al. 2016). You can change the parameters based on your data by entering the corresponding parameter number and change the values. If you don’t want to change the default parameters, type –1 and hit Enter. • Run centroid (y/n/v/r)? (y – yes, n – no, v to use VIS, r to reset offsets): – If you want it to run automatically using JUDE centroid program, press ‘y’ and Enter. – If you want to correct for s/c motion using VIS data interactively, press ‘v’ and Enter. – If you choose ‘n’, it will run with default offset values and gives you the final output with default s/c correction. – If you press ‘r’, it will reset offsets (make them full nulls). • Run using JUDE centroid: First, it will ask to change the image scaling, then it will ask to choose the individual star from the displayed image. Select the isolated bright star in the field within the binned image by clicking the cursor

J. Astrophys. Astr.

(2021) 42:35

on the display image. It is better to avoid stars on the edges of the field. • ‘Enter new box size (or press Enter):’ You can specify a box size here, and the program will perform the s/c correction by following the selected star within the box size in each set of frames, and display the co-added image. It will also display the PSF of the image. • ‘Write files out (this may take some time)?’: If you are satisfied with the PSF and the image, you can write the FITS file by pressing y to this question. • ‘Do you want to run with different parameters?’: If you want to improve the images, you can run the code again by entering y, otherwise it will automatically go to the next image. Follow the same procedure. After finishing all NUV, it automatically goes to FUV. Follow the same procedure. Important note: When you run interactive.com, you can use /notime keyword with jude_interactive.pro in interactive.com to run fast. But, in order to get the correct exposure time and flux (CPS) in the final output, remove the /notime keyword from interactive.com and run it again, or run jude_apply_time.pro afterwards for all the files. Calling sequence: jude_apply_time, L2_data_file, uv_base_dir, params = params where L2 data file is the directory of Level 2 events list folder (/events) and uv base dir is The top level UV directory (/fuv or /nuv). 5. Data products • Events list: A FITS binary table with N extensions, where N is the number of frames in the data. Each extension has a list of all the photons detected in that frame. The format of the file is described here: in the first stage of processing, one event file will be created for each Level 1 data file. Duplicate files will be removed in the second stage (jude_uv_cleanup) and the data will be merged if they are continuous. There will thus be fewer event files than Level 1 data files. Display events: make_movie.pro will provide a movie of the frames from the events files, which can be run using make_movie, events_file. • Images: A FITS image file will be created for every event file. Each frame will be shifted and

Page 5 of 8

35

added to correct for spacecraft motion. The first extension in the FITS image file will be the flux in counts/second (CPS). The second extension in the FITS image file will be the exposure time for each pixel in seconds. If /notime keyword was set, then the array contains the total number of frames. The final image from JUDE will be 4096 9 4096 pixels (280 9 280 ) with the pixel size of 0.4100 . The total exposure time will be given in the fits header of each images. Changing exposure time: The exposure time of a final image can be changed by changing parameters ‘‘minframe’’ and ‘‘maxframe’’ when running interactive.com. This can be helpful if the user is looking for flux variability at different time scales. • Diagnostic files: Corresponding quick-look PNG image will be created for every event file. This image will additionally contain the following plots: a plot of the data quality index (DQI) for each frame, a histogram of the number of counts per frame, and a plot of spacecraft motion derived from calculating shifts between the frames in both X and Y. For details on DQI and calculation of shifts see Murthy et al. (2017). • CSV files: final_obslog.csv: Observation log for Level 2 data; obs_times.csv: Observation log for Level 1 data with duplications; merge.csv: Log of files merged to produce the final Level 2 data. After running interactive.com, jude_obs_log.pro can be run to create the observation log of final images. Calling sequence: jude_obs_log, data_dir, output_file, params. 6. Verifying the output In principle, JUDE should automatically produce usable science products: photon lists and images. However, there are times when the registration fails, and it is best to verify the images manually. Program interactive.com will run through each Level 2 file and give options for reprocessing the data, if desired. The main purpose of the program is to run jude_interactive.pro for each event file. As a general rule, there are three parameters to check:

35

Page 6 of 8

• The integration time of the image should be close to that of events file. If it is much less than the actual time observed, it is worth checking why. • The PSF depends on the quality of the correction for the s/c motion, and is typically 1.2–1.5 arcseconds. If the PSF is much greater than that, it means that automatic registration failed. • The appearance of the image depends on the s/c motion, and this may be seen as tails in the sources. jude_checkpsf.pro can be used to check for the PSF of a star in the field. The program gives PSF in x and y direction (unit of both pixels and arcsec) and also displays 3D surface profile and a contour plot of the star. Calling sequence: jude_checkpsf, im, res where im is the Level2 image file of UVIT, and res is the resolution (number of sub-pixels in one physical pixel) of the image. By default, JUDE produce images with 4096 9 4096 (res=8) pixels.

7. Data registration Data registration corrects for spacecraft motion. The first estimate comes from the VIS images. The program jude_match_vis_offsets.pro only samples every 1 sec in time, and thus the derived spatial resolution is poorer. The second estimate comes from self registration—program jude_centroid.pro. As long as there is a bright star in the UV channel, this will give better resolution than getting offsets from the VIS channel. Most fields have stars that are bright enough to follow along. If the automated registration fails, there are several steps to follow: • Increase the number of bins (the amount of time) to be binned over. The default is 20 frames (0.6 s). The box size usually has to be increased because the s/c motion is larger in longer times. • Decrease the resolution, which effectively increases the S/N per pixel. It is usually easier to find bright sources suitable for registration in the NUV channel. We have written jude_match_nuv_offsets code to match the NUV offsets to the FUV channel. 8. Astrometric corrections We have found that the astrometric information provided by the s/c star sensor is not accurate enough for good astrometric calibration of the UVIT images, and we have implemented several alternatives.

J. Astrophys. Astr.

(2021) 42:35

8.1 Astrometry.net This is our recommendation and it gives excellent solutions for most NUV fields. Though it is less likely to work well for the FUV files because the stellar magnitudes are much more different from the optical. There are several different ways to use Astrometry.net. 8.1.1 Directly on the website • Submit images to the website (http://nova. astrometry.net). Anonymous uploads (without signing in) are allowed but then the data becomes public. Creating an account allows more customization. • The solution will be faster if we restrict the field size to the UVIT FoV. • The FITS file with the astrometric solution will be on the ‘‘SUCCESS’’ page and is named newimage.fits. • Note that this file will take the first extension from the original file (the data) and will not include the exposure time per pixel. • We have written a program (jude_copy_astrometry_hdr.pro) to merge the astrometry from the new-image.fits into the original file. Calling sequence: jude_copy_astrometry_hd, astrometric, l2_file 8.1.2 Python script to upload data files to Astrometry.net • Python and client.py are required to run this. Typing python client.py will print the arguments to the program. • A key (API_KEY) to Astrometry.net (obtained through creating an account) is also required. • jude_call_client.py is the interface to client.py. The key may be stored in the system variable AN_API_KEY; entered as a keyword, or entered when requested. The path to client.py should be changed in the program (if it does not exist, the user is prompted for input). The client.py is called with preselected parameters. The Level 2 image is uploaded to the Astrometry.net website, processed online, and then downloaded. Finally, We update the astrometry in the Level 2 file. The call would be (with default parameters) jude_call_client_py,inp_ dir,out_dir. inp dir is the directory containing

J. Astrophys. Astr.

(2021) 42:35

image files to be corrected and out dir is the directory to put files from astrometry.net. 8.2 Independent astrometric corrections There are times when the astrometric correction fails, more often in the case of the FUV where there are not enough recognizable stars. Program jude_fix_astrometry takes an image with known astrometry (typically the NUV image) and compares it with the test image. We have tried to automate the procedure as much as possible, but user intervention is necessary at times. At least two points are required for an astrometric solution. If two or three points have been defined, the code starast.pro is used. If 6 or more points have been defined, the code solve_astro.pro is used. In case not enough points are found, they can be added by hand to a reference file. The name of the reference file is the same as the image file with .FITS.gz replaced by .ref. The reference file has 4 lines, each with at least two elements. The lines contain the x and y positions (in pixels) of each star with its RA and DEC (in degrees). The order of the lines is unimportant, but the variable names (newxp, newyp, newra, newdec) are required, e.g., newxp ¼ ½3494:0319; 3099:7382; 1987:0005; newyp ¼ ½1579:9278; 1483:6838; 2929:903; newra ¼ ½9:1011347; 9:1398875; 9:4078713; newdec ¼ ½39:927001; 39:963891; 39:913863: Calling sequence: jude_fix_astrometry, new_file, ref_file, where new file is the Level 2 image file to be corrected astrometrically, and ref file is file with accurate astrometry.

Page 7 of 8

35

jude_coadd.pro will take a set of images, place them on the same coordinate frame, and add together. The images must have astrometric information and the quality of the addition will depend on the correction. The result is an exposure time-weighted sum of the inputs. Note that the output pixel is larger than the input pixel, so the counts/pixel will be necessarily larger. Calling sequence: jude_coadd, images_dir, out_file, ra_cent, dec_cent, fov, pixel_size, where ra cent and dec cent are coordinates of the image center, fov is the image field of view in degrees, and pixel size is in arcsec.

10. Flux calibration The flux calibration of UVIT is described in Rahna et al. (2017). The flux unit of the final image derived from JUDE is counts/second (CPS). CPS can be ˚ -1 by converted into physical flux in erg s1 cm-2 A multiplying by the conversion factor CF (erg s1 cm2 ˚ 1 cps1 ), given in Table 2 (Rahna et al. 2017): A FðkÞ ¼ CF  CPS :

ð1Þ

Errors in CPS can be calculated from the square root of the total number of counts divided by square root of the total exposure time, pffiffiffiffiffiffiffiffiffi CPS CPSerr ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : ð2Þ exposure time To obtain errors in flux FðkÞ, multiply CPS errors by the corresponding conversion factors. 11. Saturation correction

jude_apply_astrometry.pro will take an astrometrically corrected image file and apply the astrometry back to the event list. The x and y of each photon (in the detector plane) will have an associated RA and DEC.

At high count rates, intensified CMOS detectors are subject to non-linearity and saturation. Corrections for saturated bright stars in the UVIT field are discussed in Section 3.2.3 of Rahna et al. (2017). We modeled the non-linearity using the formulation of Kuin and Rosen (2008). The observed UVIT counts are nonlinear after 15 CPS (9.7% roll-off) and saturated count rates above this value (till 29 CPS) can be corrected using empirical relation,

9. Co-addition of images

C obs ¼ 29  ð1  eðaCinc =29Þ Þ ;

The final step is the co-addition of the images from different orbits. We used jude_coadd.pro routine to produce final co-added image. Program

where a ¼ 1:24 (determined empirically), Cinc is the number of events incident on the detector, Cobs is the number of events detected, and there are 29 frames in

8.3 Coordinates for every photon

ð3Þ

35

Page 8 of 8

a second. After 29 CPS, the counts saturate and the true value cannot be recovered. 12. Conclusion We have described here how to install and operate the independent UVIT data pipeline JUDE. The pipeline is free and open-source and consists of modules where users can control and change the parameters. We have successfully used JUDE in the last three years; first to independently characterize the in-flight performance of the UVIT (Rahna et al. 2017) and, secondly, in our scientific programs, of which several are completed (Rahna et al. 2018; Kameswara Rao et al. 2018a, b; Rubinur et al. 2020) and few are still running on (Safonova et al. 2020; Rahna et al. 2021; Yadav et al. 2021). Attribution and Acknowledgements If you use this software, please cite Murthy et al. (2017). We have used GDL very heavily and here is the attribution for that: Coulais et al. (2010). The IDL Astronomy User’s Library reference is Landsman (1995), the Markwardt IDL Library: Markwardt (2009), and the Astronomy.net reference is Lang et al. (2010). This publication uses the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). The UVIT is built in collaboration between IIA, IUCAA, TIFT, ISRO and CSA. The authors gratefully thank all the members of various teams of UVIT project. RPT acknowledge IIA (Indian Institute of Astrophysics, India) and SHAO (Shanghai Astronomical Observatory, China) for providing the computational facilities. This research has been supported by the Department of Science and Technology under Grant No. EMR/ 2016/00145 to IIA.

References Coulais A., Schellens M., Gales J. et al. 2010, in Mizomoto Y., Morita K.-I., Ohishi M., eds, Astronomical Society of

J. Astrophys. Astr.

(2021) 42:35

the Pacific Conference Series, Vol. 434, Astronomical Data Analysis Software and Systems XIX, 187 Kameswara Rao N., De Marco O., Krishna S. et al. 2018a, A&A, 620, A138 Kameswara Rao N., Sutaria F., Murthy J. et al. 2018b, A&A, 609, L1 Kuin N. P. M., Rosen S. R. 2008, MNRAS, 383, 383 Kumar A., Ghosh S. K., Hutchings J. et al. 2012, in Space Telescopes and Instrumentation 2012: Ultraviolet to Gamma Ray, Vol. 8443, 84431N Landsman W. B. 1995, in Shaw R. A., Payne H. E., Hayes J. J. E., eds, Astronomical Society of the Pacific Conference Series, Vol. 77, Astronomical Data Analysis Software and Systems IV, 437 Lang D., Hogg D. W., Mierle K., Blanton M., Roweis S. 2010, AJ, 139, 1782 Markwardt C. B. 2009, in Bohlender D. A., Durand D., Dowler P., eds, Astronomical Society of the Pacific Conference Series, Vol. 411, Astronomical Data Analysis Software and Systems XVIII, 251 Murthy J., Rahna P. T., Safonova M. et al. 2016, JUDE: An Utraviolet Imaging Telescope pipeline, Astrophysics Source Code Library, ascl:1607.007 Murthy J., Rahna P. T., Sutaria F. et al. 2017, Astronomy and Computing, 20, 120 Postma J. E., Leahy D. 2017, PASP, 129, 115002 Rahna P. T., Das M., Murthy J., Gudennavar S. B., Bubbly S. G. 2018, MNRAS, 481, 1212 Rahna P. T., Murthy J., Safonova M. et al. 2017, MNRAS, 471, 3028 Rahna P. T., Zhenya Z., Das M., Murthy J. 2021, FUV excess emission in the outer disk of SO/ETS galaxies, manuscript in preparation Rubinur K., Kharb P., Das M. et al. 2020, MNRAS, 500, 3908 Safonova M., Kashyap U., Rahna P. T. et al. 2020, Testing the nature of the ULX by AstroSat: UV–X-ray variability in ultra-luminous X-ray source Holmberg II X-1, manuscript in preparation Tandon S. N., Subramaniam A., Girish V. et al. 2017a, E-print arXiv:1705.03715, arXiv:1705.03715 Tandon S. N., Hutchings J. B., Ghosh S. K. et al. 2017b, JAA, 38, 28 Tandon S. N., Postma J., Joseph P. et al. 2020, AJ, 159, 158 Yadav J., Das M., Patra N. et al. 2021, Comparing the inner and outer star forming complexes in the nearby Spiral galaxies Ngc 628, Ngc 5457 and Ngc 6946 using UV observations, manuscript in preparation

J. Astrophys. Astr. (2021) 42:56 https://doi.org/10.1007/s12036-021-09729-z

Ó Indian Academy of Sciences Sadhana(0123456789().,-volV)FT3 ](0123456789().,-volV)

DATA PIPELINE

AstroSat/SSM data pipeline B. T. RAVISHANKAR1,* , S. VAISHALI1, D. BHATTACHARYA2, M. C. RAMADEVI1,

ABHILASH SARWADE1 and S. SEETHA3 1

Space Astronomy Group, ISITE Campus, U.R. Rao Satellite Centre, Outer Ring Road, Marathahalli, Bengaluru 560 037, India. 2 Inter-University Centre for Astronomy and Astrophysics, Ganeshkhind, Pune 411 007, India. 3 Department of Astronomy and Astrophysics, Raman Research Institute, Sadashivanagar, Bengaluru 560 080, India. *Corresponding Author. E-mail: [email protected] MS received 7 November 2020; accepted 28 January 2021 Abstract. The data pipeline at the Payload Operation Centre (POC) of the Scanning Sky Monitor (SSM) onboard AstroSat involves: (i) fetching the Level-0 data from the Indian Space Science Data Centre (ISSDC), (ii) Level-0 to Level-1 data processing followed by Level-2 data generation, and (iii) transfer of the Level-1 and Level-2 data back to ISSDC for dissemination of the re-packaged Level-2 data products. The major tasks involved in the generation of Level-1 and Level-2 data products are: (a) quality checks; time, alignment corrections, (b) temporal-HK plots generation, and, (c) image processing; light curve generation. The typical turn around time for this fully automated pipeline is about 25 min for one orbit data. In this paper, details of all the stages of this data pipeline are discussed. Keywords. Fully automated data pipeline—X-ray astronomy payload—coded mask data processing.

1. Introduction Scanning Sky Monitor (SSM) (Ramadevi et al. 2017) aboard the multi-wavelength astronomy mission AstroSat (Agrawal 2006; Singh et al. 2014) is an assembly of three coded mask cameras, monitoring the X-ray sky in 2.5–10 keV energy band (see Fig. 1). Each of the SSM cameras has one-dimensional Coded Mask optics and gas-filled proportional counter detector. The two edge cameras (SSM1 and SSM2) are canted away by 45 from the base plane of the central camera (SSM3), and further, they are also inclined in the canted plane by þ12 and -12 in order to look at different parts of the sky with some overlap in the fields of view. The whole assembly is capable of rotation on an axis near-parallel to that of the SSM3 camera as shown, and the assembly is

This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

mounted such that this is also along the ?Yaw axis of the spacecraft. The stepper motor based rotation mechanism is designed to undertake one complete rotation from 5 to 355 and reverse in stare-and-step mode in order to observe different parts of the sky, with a typical stare time of 10 min and step-angle of 10 . Among the three SSM cameras, the two edge cameras, SSM1 and SSM2 are of the same dimension and the central SSM3 camera has different dimensions (for accommodating them on the spacecraft structure). Therefore the imaging properties of the fields of view, angular resolution, etc., of SSM3 camera are also different. All these are listed in Table 1. SSM being an X-ray sky monitor the data downlinked at the Ground Station in every visible orbit is subjected to immediate processing. The highest data rate available for processing is once every  1:5 h, the orbital duration, and the data-sets are transferred from the data server at ISSDC to the POC of SSM located at the Space Astronomy Group of U. R. Rao Satellite

56

Page 2 of 10

J. Astrophys. Astr.

(2021) 42:56

The Level-2 data consist of flux estimates obtained after image reconstruction with calibrated, anode-response corrected, clean event-lists per stare per energy-band. The Level-1 data as well as the incremental Level-2 data consisting of observational updates to the sources detected, are transferred from the POC to ISSDC. At ISSDC, the incremental Level2 dataset is loaded in a database to generate the final data products via a custom built web application. Figure 1. SSM assembly flight model of the three SSM cameras with their respective electronics packages mounted on a rotating platform.

Centre (URSC). Therefore complete automation of the entire processing is essential. This paper describes the data levels and the processing undertaken to generate each level.

2. Data architecture and data pipeline Three different levels are identified for SSM data, namely, Level-0, Level-1 and Level-2. The data-flow and the interfaces for the three levels of data are depicted in the overall architecture in Fig. 2. The Level-0 data consist of the binary payload data along with the necessary auxilliary data for the same UT range. The payload HBT (high bitrate telemetry) data, including health (also termed housekeeping or HK parameters) in binary format within the Data Handling package’s envelope are generated for the three SSM cameras separately. The Level-0 datasets received at the POC are processed to generate higher level products – Level-1 data followed by Level-2 data. The Level-1 data consist of intermediate products including raw event lists and HK parameters – all UT-tagged for different energy bands and for different identified stare-durations for each of the three SSM cameras.

2.1 Data transfer and pipeline automation The data transfer operations and the pipeline processing are controlled by two sets of shell-scripts which function in tandem. The data transfer script is programmed to handle transfers of different kinds of files and in either direction between the ISSDC and the POC. The data fetching cron jobs running at the POC periodically poll the Data server at ISSDC for the availability of new Level-0 data files, and transfer the files to the POC over a dedicated Virtual Routing and Forwarding (VRF) meant for AstroSat over National Knowledge Network (NKN). Likewise separate directories are identified at the ISSDC data server to post the Level-1 and Level-2 data products. The data transfer scripts utilise Expect (which in turn is based on TCL/Tk, (Libes & O’Reilly 1995) for automating the transfer dialogue. In order to address possible link and server issues upon unsuccessful transfers, additional transfer attempts are built in the design. All Level-0 data files transferred are also verified for MD5SUM tags for possible transmission bit-errors. The cron jobs monitoring the ISSDC areas for different types of files to be transferred to the POC make sure that multiple instances are not run if by chance a previously initiated transfer has not ended owing to network delays or server issues.

Table 1. Specifications of the three SSM cameras. Parameter

SSM1/SSM2

SSM3

Mask plate width (x), mm Mask plate length (y), mm Height, mm Mask plate thickness, mm Inter-pattern gap, mm Window-support rod diameter, mm Calibration wire diameter, mm Field-of-view (FoV) (x  y) Angular resolution (x, y)

60 500 251.4 3 4 2 1 26:8  100 100 130 , 2:5

60 634 306.4 3 4 2 1 22:1  100 100 130 , 2:5

J. Astrophys. Astr.

(2021) 42:56

Page 3 of 10

56

Figure 2. Overall architecture of the SSM data pipeline with interface details between the data servers at ISSDC and the SSM POC at URSC.

The Level-0 to Level-2 pipeline is developed as a shell-script calling individual modules in two broad stages to produce Level-1 output first and use that to subsequently generate Level-2 data products. The modules are developed primarily in ANSI-C and java. For some other different stages, awk, sed, pgplot, gnuplot are also employed in the automated pipeline. The data transfer and data processing daemons interact indirectly via the inotify (Love 2005) API which monitors specified directories to trigger Level-0 data processing, and Level-1 and Level-2 data transfer, after attaching MD5SUM hashkeys. For all received Level-0 data, along with the data products a quality report is also generated based on the following criteria: (a) whether all necessary files are present, (b) possible time-lag between the instrument data and attitude data, (c) the RS-decoding parameters during the dump of the data of each of the three cameras, (d) presence of outliers among instrument house-keeping parameters, and, (e) any pipeline errors reported during the course of execution.

3. SSM Level-0 data The SSM instrument dataset is written in segments/pages each of 2048 bytes size by the onboard Processing Electronics subsystem of the respective SSM camera. Each 2k page consists of instrument house keeping parameters, time-tags of the instrument clock latched with corresponding Onboard Computer (OBC)

clock at a rate of 1.024 s, Temporal parameters (such as count-rates) and individual photon strike events recorded in the detector (with information about anode-ID and the voltages recorded at either end of the anode wire). The 2k pages are further embedded within an envelope of the Baseband Data Handling subsystem of the spacecraft which encodes the data with RS-encoding to check for transmission errors. This forms one of the main components of the Level-0 data. Along with this are packed auxiliary files including the following: (a) the SSM platform data, with information about the stare-and-step operation including the platform angle and status whether platform is rotating or is stationary, etc, at intervals of 1.024 s. (b) Attitude data, with the inertial coordinates in units of quaternions at intervals of 172 ms. (c) Time Correlation Table, correlating the SSM clock to corresponding UT, produced using the latched OBC time. (d) Make filter (MKF) file with samples at a rate  100 ms of all health and orbital parameters (like South Atlantic Anamoly (SAA) region) that will be included to obtain the good time intervals (GTIs).

4. Level-0 to Level-1 data processing The data flow diagram of the SSM Level-0 to Level1 data processing is shown in Fig. 3. The Level-0 tar file along with the corresponding trigger-file form the

56

Page 4 of 10

J. Astrophys. Astr.

(2021) 42:56

Figure 3. Data flow diagram: SSM Level-0 to Level-1 processing undertaken at the POC as part of the automated pipeline.

input. The trigger file consists of the size of tar file in bytes populated in the ISSDC data server and used at the POC for verification. After setting up the

analysis environment, the very first task is to assess the quality of the data based on the RS-decoding parameters attached to every payload 2k page. All 2k

J. Astrophys. Astr.

(2021) 42:56

Page 5 of 10

56

Figure 4. One sample plot of some of the housekeeping parameters produced using pgplot at SSM Level-1 stage; here for orbit 28198 and for SSM3 camera; The High Voltage (HV) reference is lowered during SAA passages (as also observed by the increase in Charged Particle Monitor (CPM) count-rate); Since HV is lowered during this duration SSM Veto and integrated left count rates are reduced to zero.

Figure 5. Data flow diagram: SSM Level-1 to Level-2 processing undertaken at the POC as part of the automated pipeline.

56

Page 6 of 10

J. Astrophys. Astr.

(2021) 42:56

Figure 6. Consolidated plot with some HBT temporal parameters, here for orbit 6266 and for SSM1 camera for a scheduled calibration observation with the Crab source; grey strips indicate data binned at 10 min with stare-sequence number marked in each; ILR and IRR are respectively integrated left and right count rates from all anodes; also plotted is Veto count rate, and the dip in all the count rates corresponds to the SAA region; also indicated are spans of eclipse and Earth in the FoV.

pages with uncorrectable bit-errors are dropped and all these contribute to bad-time to be accounted for in the GTI intervals. The RS-decoding parmeters are also used later to prepare a quality report to be sent back to ISSDC. The Time Correlation Table (TCT) files, the SSM platorm data with information about the stare-and-step operation and the attitude data are processed, and ASCII dumps of the same are produced. The attitude information is available in the form of quaternions and they are converted to the Right Ascension (RA) and Declination (Dec) of each of the three spacecraft body axes of Yaw, Roll and Pitch. Possible time-offsets between the instrument data and the auxilliary data for a given dump are checked. The RA and Dec of the SSM pointing axes are then computed using the attitude data of the spacecraft pointing, SSM platform angle, the status of its rotation, and the ground-measured alignment angles of the flight module namely, the Cant-angle (  45 ), the inclination angles of the two edge cameras, SSM1 and SSM2 (  12 ), and the reference position-angle for the platform rotation mechanism.

The output SSM attitude data consist of Position stream and Stare sequence, with the inertial coordinates of the three SSM cameras. The Stare sequence is determined by picking all instances when the SSM platform is in stare-mode of typically 10 min. With the available Stare sequence, the instrument event data is UT-tagged and split into individual stares, and saved as different extensions of a FITS binary table (using cfitsio library, Pence 1999). The events are also segregated into two kinds – based on whether anode-IDs at either ends match, or if one of the ends is marked 8 if recorded only at one end. Also segregated into separate FITS files – Stare-wise – are the HK parameters and temporal parameters. Plots are generated for all these parameters using the pgplot library (Fig. 4). The MKF file made available with spacecraft position, attitude parameters is augmented with some SSM specific parameters. The orbit and Low Bitrate Telemetry (LBT) files providing spacecraft parameters are also processed. The Level-1 output produced consists of the raw event-lists for each of the SSM camera, SSM attitude, augmented MKF files and SSM HK and temporal parameters, and their plots.

J. Astrophys. Astr.

(2021) 42:56

Figure 7. Data flow diagram: SSM Level-2 imaging details; corresponds to the ‘SSM camera specific imaginganalysis module’ block of the data flow diagram in Fig. 5.

5. Level-1 to Level-2 data processing The full Level-1 to Level-2 data flow diagram of SSM is as shown in Fig. 5. Level-1 data products form the input, mainly the event-lists (especially of the type in which both anode-IDs per event match), augmented MKF file, SSM attitude, SSM HK and Temporal data. After setting up the necessary output directories, plots marking individual stares with different temporal count-rates are generated using gnuplot for quick checks of the data received (Fig. 6). The next major stage is the generation of good time filter files based on the filter expressions set for individual parameters, for entire orbit data – this, for parameters common for

Page 7 of 10

56

all three SSM cameras (such as the SAA flag, CPM data), as well as those specific to individual cameras (such as Sun angle, RS decoding error count, electronic subsystem status). The good time filter files are used to consolidate and generate good time intervals. After extracting individual stare data, some more stare/dwell specific filtering is included to the GTIs (such as different voltage and temperature monitoring values). All these GTI files are applied on the raw event lists to produce clean Level-2 event lists. The consolidated GTI per camera per stare is collected to correct for the exposure time. The GTI-cleaned event lists of unambiguous anode-ID type form one of the main inputs to a camera specific imaging module. The modules leading to the imaging analysis are shown in the data flow diagram in Fig. 7. The main tasks undertaken as part of the imaging (Fig. 7) are to read energy-band details and corresponding shadow response library, mean free path values which are then populated in different data structures. Anode calibration parameters are read and applied on the GTI-filtered event-lists, producing energy-band specific Detector Plane Histogram (DPH) for every stare of each camera. The SSM catalog and the resolution elements in terms of camera coordinates of hx and hy are read. For the computed SSM attitude, the list of known sources in the FoV is determined and populated in another data structure. With the bandspecific DPHs and the list of known sources in the field for every stare, the imaging analysis is intiated. To describe here briefly, it is designed based on Bayesian Richardson-Lucy technique (see Ravishankar & Bhattacharya 2003) as well as svdfit based forward-fitting method. The energy-bands considered are: (a) 2.5–4 keV, (b) 4–6 keV, (c) 6–10 keV, and (d) 2.5–10 keV. Contributions of the known sources are fit using the response library, their response removed from the observed DPH, and in the residual, new sources are looked for. This exercise is undertaken separately for each of the four energy bands and for every camera, and, for every stare/dwell. Figure 8 shows the lightcurve for the Crab source from early 2016 for observations with SSM3 camera; here it is also compared with the MAXI (Matsuoka et al. 2009) lightcurve.1 In the inset are zoomed in plots for clarity. The data of all sources will be made public after efforts undertaken to address the dispersion in the data is applied for all observations in a data 1

The non-standard MAXI product of 2.5–10 keV light curve is generated using the MAXI on-demand Process at http://maxi. riken.jp/mxondem/.

56

Page 8 of 10

J. Astrophys. Astr.

(2021) 42:56

Figure 8. SSM3 Crab lightcurve from some set of observations across a few years (in blue). Plotted in grey is the MAXI lightcurve of Crab (the non-standard MAXI product of 2.5–10 keV light curve is generated using the MAXI on-demand Process at http://maxi.riken.jp/mxondem/) for comparison. The inset plots (a)–(d) are zoomed versions of the SSM observation slots as marked in the main panel.

regeneration campaign. An account of these efforts and details of the image processing will be provided in a future publication. The output of the imaging analysis will be propfiles (Fig. 5) which are incremental Level-2 product consisting of updates to the light curves of every detected source in four different energy bands with associated parameters of goodness of fit, earth angle, background count rate, etc. The prop-files are updated with the quality factors determined. Though the imaging is undertaken stare-wise for each energy-band, the prop-files for each SSM camera are populated detected source-wise. That is, during imaging, observations per stare/dwell, per camera, per energy-band are considered and the flux values in each energy band, goodness of fit, background rate, and other parameters like earth angle, total number of sources in the stare, etc, are noted in a data structure for every detected source. In the propfiles, however, for every detected source, all the parameters determined are populated for each stare/ dwell they are detected in, in another data structure. The prop-files are also marked separately in the data structure in three categories: (a) as incremental flux update to the light curves of known sources,

(b) alerts on outbursts in known sources, and, (c) alerts about newly detected sources. Level-2 tar packages are generated with these files. These, and the Level-1 tar-files are moved to a designated area from which the automated file-transfer daemon will pick the files and transfer them to the appropriate areas at ISSDC.

6. Turn around time The turn around time for the SSM product generation for the data corresponding to one orbit duration is about 25 min. The time it takes for the Level-0 to Level-1 processing is  5–8 min, and that for Level-1 to Level-2 processing is  12–16 min. The duration varies based on how crowded the FoV is with strong sources, and the background rate. The data transfer time itself varies based on the network bandwidth and typically takes less than a few minutes, considering the additional checks introduced as explained in Section 2.1 The POC system employed which provides this processing time involves a workstation with six core 3 GHz Intel Xeon processor and 6 GB RAM, and RHEL operating system.

J. Astrophys. Astr.

(2021) 42:56

Figure 9. SSM Data Organiser module of the automated pipeline at the ISSDC which ingests the incremental updates to the lightcurves of the sources detected into a database and manages the disseminated data products.

Page 9 of 10

56

SSMDO application is triggered. A configuration file is used to specify parameters at runtime. Each incremetal Level-2 data file is processed and the required information/observation parameters are extracted and stored in the appropriate database tables corresponding to the source observed. Using data from the tables, the Level-2 data files (FITS) for each source is updated. The source catalog file (ASCII) is also updated. The Level-2 data products – source catalog and source data files – along with metadata file are then provided to the archival system for long term archival. The output messages and error messages at every stage are logged for debugging purposes and also to provide a report on the execution of the module. Provision is made for users to view the observed source catalog of SSM and also browse through the light curves of all sources observed by SSM through a website as shown in Fig. 10. In addition, provision is made for downloading plots of light curves and hardness ratios in PNG or JPG formats, and to download the data in ASCII, FITS, and VOTable formats. All these are process validated.

8. Conclusion Figure 10. SSM Data Dissemination web portal at ISSDC provides an user interface to the SSM’s Level-2 data products.

7. Level-2 data reorganisation and dissemination SSM Data Organiser (SSMDO) is a command line based Java application to create SSM Level-2 data products (see Fig. 9). This application is a part of SSM Data Pipeline (DP)/SSM Level-1 to Level-2 data processing chain and is set up to run at ISSDC in an automated fashion. The Level-2 incremental data files need to be organised in such a way that for every source observed by SSM upto a given point of time, there is one file (FITS format) containing up-to-date observation details. SSMDO makes use of a MariaDB database to store the observation details of sources observed by SSM. One table for every source observed by SSM is maintained in the database, containing all the observation parameters pertaining to that source. Tables to store the source catalog details and alert information are also available. Once the incremental Level-2 data files are available at the designated location at ISSDC, execution of

This paper gives an account of the data pipeline implemented for the Scanning Sky Monitor on AstroSat. The data pipeline as a fully automated software, has been tested and validated.

Acknowledgements The authors are thankful to the anonymous referee for her/his very useful suggestions and comments which have helped to improve the quality of the manuscript. This research has made use of MAXI data provided by RIKEN, JAXA and the MAXI team. BTR, SV, MCR and AS thank GH of SAG, DD of PDMSA, and Director, URSC for encouragement and continuous support to carry out research and project work.

References Agrawal P. C. 2006, Adv. Space Res., 38, 2989 Libes D., O’Reilly T. 1995, Exploring Expect: A Tcl-based Toolkit for Automating Interactive Programs (O’Reilly Media Inc.) Love R. 2005, Linux J., 2005, 8

56

Page 10 of 10

Matsuoka M., Kawasaki K., Ueno S. et al. 2009, Publ. Astronom. Soc. Japan, 61, 999 Pence W. 1999, in Mehringer D. M., Plante R. L., Roberts D. A., eds, Astronomical Society of the Pacific Conference Series, Vol. 172, Astronomical Data Analysis Software and Systems VIII, 487 Ramadevi M. C., Seetha S., Bhattacharya D. et al. 2017, Exp. Astron., 44, 11

J. Astrophys. Astr.

(2021) 42:56

Ravishankar B. T., Bhattacharya D. 2003, BASI, 31, 491 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, in Takahashi T., den Herder J.-W.A., Bautz M., eds, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 9144, Space Telescopes and Instrumentation 2014: Ultraviolet to Gamma Ray, 91441S

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:25 https://doi.org/10.1007/s12036-020-09680-5

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

DATA PIPELINE

Curvit: An open-source Python package to generate light curves from UVIT data P. JOSEPH1,2,* , C. S. STALIN1, S. N. TANDON3 and S. K. GHOSH4 1

Indian Institute of Astrophysics, Bangalore 560 034, India. Department of Physics, CHRIST (Deemed to be University), Bangalore 560 029, India. 3 Inter-University Centre for Astronomy and Astrophysics, Pune 411 007, India. 4 Tata Institute of Fundamental Research, Mumbai 400 005, India. *Corresponding Author. E-mail: [email protected] 2

MS received 7 November 2020; accepted 2 December 2020 Abstract. Curvit is an open-source Python package that facilitates the creation of light curves from the data collected by the Ultra-Violet Imaging Telescope (UVIT) onboard AstroSat, India’s first multi-wavelength astronomical satellite. The input to Curvit is the calibrated events list generated by the UVIT-Payload Operation Center (UVIT-POC) and made available to the principal investigators through the Indian Space Science Data Center. The features of Curvit include: (i) automatically detecting sources and generating light curves for all the detected sources and (ii) custom generation of light curve for any particular source of interest. We present here the capabilities of Curvit and demonstrate its usability on the UVIT observations of the intermediate polar FO Aqr as an example. Curvit is publicly available on GitHub at https://github.com/ prajwel/curvit. Keyword. AstroSat—UVIT—variability.

1. Introduction The Ultra Violet Imaging Telescope (UVIT; Tandon et al. 2017a, c), consisting of two co-aligned telescopes of aperture 375 mm each, is one of the payloads onboard AstroSat. AstroSat is India’s first multiwavelength astronomical observatory, launched by the Indian Space Research Organisation (ISRO) on 28 September 2015 (Agrawal 2006). In addition to UVIT, there are three co-aligned X-ray payloads on AstroSat enabling simultaneous observation of a celestial source over a wide range of wavelengths from hard Xrays to the Ultraviolet (UV) band. UVIT, with a field of view of 28 arcminute diameter, can perform imaging and low-resolution slit-less spectroscopy. UVIT has a large number of filters and with selectable filters or gratings, simultaneous observa˚ ), neartions in far-ultraviolet (FUV, 1300–1800 A This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

˚ ), and visible (VIS, ultraviolet (NUV, 2000–3000 A ˚ 3200–5500 A) channels are possible. Of the three channels, VIS channel is used only for aspect correction, while the NUV and FUV channels are used for science observations. The detectors used in all the three channels are intensified CMOS imagers with 512  512 pixels. The telescope pointing can drift up to 30 over the night time of an orbit (duration can be up to a maximum of 1800 seconds) with a rate of 300 /second. This drift of the satellite is estimated, as a function of time, using observations obtained with the VIS channel. In the normal mode of operations, observations are carried out in the full window mode (default mode, 512  512 pixels) covering 28 diameter arcminute field resulting in a read rate of  28.7 frames/second. It is also possible to observe a partial field (that is selectable by the principal investigators (PIs) of the observing proposals) read at a higher rate. For example, the observation of a small window (100  100 pixels) will provide 640 frames per second. The NUV and FUV images are generated by

25

Page 2 of 10

combining short exposure frames with shift and add algorithm. However, for bright fields, self drift correction of NUV data with NUV and FUV data with FUV is also possible. Two modes of operation exist in UVIT: (1) photon counting mode, achievable through high electron multiplication via high voltage to the microchannel plate of the intensified imager, in the case of NUV and FUV channels, where the intensified detectors record the X and Y positions of the photons on the detectors and their arrival times and (2) integration mode, achievable through a lower electron multiplication, for VIS channel where the readout consists of image frames with a time resolution of 1 second. In the final image obtained using photon counting mode, each detector pixel is mapped to 8  8 sub-pixels by the centroiding of photon events with each sub-pixel having a plate scale of 0:41600 (Hutchings et al. 2007; Postma et al. 2011). Due to the availability of time-tagged events in photon counting mode in FUV and NUV channels, it is possible to probe the time variability of observed sources in UV and generate light curves similar to other branches of high energy astronomy such as X-rays and c-rays. UVIT has been performing as per specifications. More details and related calibration can be found in (Tandon et al. 2017b) and (Tandon et al. 2020). The UVIT Payload Operation Center (POC) at the Indian Institute of Astrophysics (IIA) runs the UVIT Level-2 pipeline (UL2P; Ghosh et al. 2021, this volume) on the Level-1 (L1) data, containing both spacecraft and observational data in FITS format, to produce science ready Level-2 (L2) data products. A description of UVIT data reduction is given in (Postma & Leahy 2017) as well as Ghosh et al. (2021). The UVIT-POC processed L1 and L2 data are made available to the PIs of the observations through the ISRO Indian Space Science Data Center (ISSDC). The L2 data contains: (i) orbit-wise calibrated events list after corrections for the drift of the spacecraft, flat field, and distortion, (ii) orbit-wise science ready images in detector coordinate and world coordinate systems and (iii) combined images that belong to a single pointing, wherein observations in a particular filter carried out over many orbits in a particular pointing are combined. L2 data from ISSDC are science ready products, and the PIs can directly carry out their photometric, spectroscopic or imaging analysis. Alternatively, PIs willing to do a custom analysis of their observations can also do so, using UL2P, along with the CALDB (that contains the calibration data files) and the CATALOG (that contains the catalogues

J. Astrophys. Astr.

(2021) 42:25

for astrometry) downloadable from ISSDC1, UVITPOC2 or the AstroSat Science Support cell3. Time variable phenomenon can be studied naturally using the ’’photon counting mode’’ of operation for the UV bands of image acquisition by UVIT. In principle, in the full window mode observations with UVIT, one can study time-varying phenomenon with a time resolution as low as 66 milliseconds in both the UV bands. Even higher time resolution is possible with smaller window observations (for example,  3 milliseconds in 100  100 pixels window). Studies of such high-resolution events will open up a new avenue of research in UV Astronomy, and for such studies, software tools are required to generate the light curves directly from the events list. The motivation is therefore to develop a software tool, that has the ability to create light curves from the events list. Here we present Curvit, an open-source Python package designed to create light curves from UVIT L2 events list. Curvit makes use of the functionalities available in other open-source Python packages such as Astropy (Astropy Collaboration et al. 2013, 2018), NumPy (Harris et al. 2020), Matplotlib (Hunter 2007), Photutils (Bradley et al. 2020, and Scipy (Virtanen et al. 2020). The availability of this tool to the PIs of UVIT will also avert the cumbersome task of first creating images of small time-bins and then doing the photometry of the target to generate light curves. gPhoton is a similar tool to create UV light curves from GALEX data (Million et al. 2016). Section 2 contains a short summary of the L2 products at ISSDC. In Sections 3 and 4, we describe the functionalities and working of the tool. In the final section, we demonstrate the usefulness of the tool by generating the light curve of intermediate polar FO Aquarii (FO Aqr), observed under the guaranteed time and now open for public after the lock-in-period.

2. UVIT data products at ISSDC The UVIT data is available at ISSDC AstroBrowse website4. Both L1 and L2 data of UVIT are available as compressed files at the archive. L2 products are organised into two categories: individual datasets (single orbit for a filter and window; see Table 1), and combined datasets over all the orbits (for a single filter 1

http://www.issdc.gov.in/. http://uvit.iiap.res.in/. 3 http://astrosat-ssc.iucaa.in/. 4 https://astrobrowse.issdc.gov.in/astro_archive/archive/Home.jsp. 2

J. Astrophys. Astr.

(2021) 42:25

Page 3 of 10

25

Table 1. Details of the data products sent to ISSDC for each orbit of observation done in a particular window-size and filter. RAS VIS Product Sky image (instrument coordinates) Sky image (astronomical coordinates) Exposure map (astronomical coordinates) Error map (astronomical coordinates) Photon events list RAS file

Description 4800  4800 sub-pixel2 FITS 4800  4800 sub-pixel2 FITS 4800  4800 sub-pixel2 FITS 4800  4800 sub-pixel2 FITS FITS binary table FITS binary table

image image image image

RAS NUV

NUV

FUV

NUV

FUV

1 1 1 1 1

1 1 1 1 1

1 1 1 1 1

1 1 1 1 1

1

1

Total 4 4 4 4 4 2



The above file structure corresponds to the ideal case when VIS, NUV, and FUV are configured by the PI. In the event of VIS not being configured, the files generated using the relative aspect series (RAS) obtained from VIS data will be missing. Similar is the case for NUV.

Table 2. Details of the combined data products sent to ISSDC. The observations carried out over the entire pointing are combined filter wise. RAS VIS Product Sky Image-A (Astronomical coordinates) Sky Image-By (Astronomical coordinates) Exposure map$ (Astronomical coordinates) Error mapy (Astronomical coordinates)

Description 4800  4800 4800  4800 4800  4800 4800  4800

sub-pixel2 sub-pixel2 sub-pixel2 sub-pixel2

FITS FITS FITS FITS

image image image image

RAS NUV

NUV

FUV

NUV

FUV

1 1 1 1

1 1 1 1

1 1 1 1

1 1 1 1

Total 4 4 4 4



Sky image-A: The astrometric accuracy of this image is limited to the accuracy of knowledge of the spacecraft aspect, which is typically around 2–3 arcmin.

y

Sky image-B: This is the final image generated after astrometry which may or may not be successful. When the astrometry is successful, the accuracy in aspect is typically 3 arcsec. When the astrometry is not successful, this image is a copy of sky image-A. The information about astrometry being successful or unsuccessful is available in the header of the FITS images.

$

They correspond to the co-ordinate system of sky image-B.

and window; see Table 2). The UVIT filters are given in Table 3.

3. Curvit workflow Curvit5 is an open-source Python package to produce light curves from UVIT data. The events list from the official UVIT L2 pipeline (version 6.3 onwards) is required as an input to the package. Curvit has two functions for light curve creation; makecurves and curve. Both the functions accept a single events list at a time which the user has to provide. We describe below each of these functions and its usage on the observation of the intermediate polar FO Aqr. 5

https://github.com/prajwel/curvit.

3.1 Function makecurves The makecurves function of Curvit automatically detects sources from the events list and create light curves for all of them. The user will have a control on the number of sources detected automatically through the use of the detection threshold parameter. Two source detection methods are available; daofind and kdtree. The user can select the preferred source detection method using the detection method parameter (the default value is daofind). The daofind method detects sources in the following manner. It first creates a 4800  4800 sub-pixel2 image from the events list and a circular mask is applied to select the central  24 arcminute region. Sources are then detected using the daofind algorithm (Stetson 1987). Mean and standard deviation values of

25

Page 4 of 10

J. Astrophys. Astr.

(2021) 42:25

Table 3. The UVIT filters in VIS, NUV and FUV channels. VIS Filter ID F1 F2 F3 F4 F5

NUV

FUV

Old filter name

New filter name

Filter ID

Old filter name

New filter name

Filter ID

Old filter name

New filter name

VIS3 VIS2 VIS1 ND1 BK7

V461W V391M V347M V435ND V420W

F1 F2 F3 F4 F5 F6 F7

Silica - 1 NUVB15 NUVB13 Grating NUVB4 NUVN2 Silica - 2

N242W N219M N245M

F1 F2 F3 F4 F5 F6 F7

CaF2 - 1 BaF2 Sapphire Grating - 1 Silica Grating - 2 CaF2 - 2

F148W F154W F169M

the background, required by daofind, are estimated by Curvit itself and the user can control the number of sources detected using the threshold parameter. Pixels in the image that have the events greater than the threshold times the standard deviation of the background will be detected. The kdtree method works as follows. A source is characterised by a cloud of events around its centroid. Therefore, to detect sources, the events are projected onto a two-dimensional Cartesian grid (with one grid cell being of 1 sub-pixel2 size), and the grid cells are sorted based on the number of events falling in each cell. Since a single source can occupy multiple grid cells, a nearest neighbour search using kdtree algorithm is performed to remove grid cells belonging to the same source (Maneewongvatana & Mount 1999). This method may not work properly on crowded fields. But in non-crowded fields, the method can detect all the sources present in the events list. However, the user may limit the number of sources to be detected using the parameter how many. Also, the aperture radius that is used to count the source events (through radius) and the size of the time bin to generate the light curves (using the parameter bwidth) can be controlled by the user. The function makecurves generates light curves for each detected sources that depend on the threshold set by the user in terms of decreasing order of brightness. The operation of the function is summarised in Fig. 1.

3.2 Function curve The function curve is similar to makecurves, with the exception that it generates light curve for a single user-defined source (through xp and yp parameters).

N263M N279N N242Wa

F172M F148Wa

Here also, the user can specify the source aperture radius and the binning time. Its operation is summarised in the Fig. 2.

4. Light curve creation From the events list FITS table, the columns Fx, Fy, MJD_L2 and EFFECTIVE_NUM_PHOTONS (hereafter ENP) are used by Curvit (see Table 4). However, the events list FITS table has more columns than the ones given in Table 4. Each row of the table characterises a single event defined by X and Y Cartesian coordinate positions (Fx and Fy) with an associated time value (MJD_L2). For the L2 data available from ISSDC, MJD_L2 provides only an approximate absolute time, good to  1 second. The time values in MJD_L2 column increment as the frame number (as denoted in the FrameCounts column of events list) changes. Therefore, MJD_L2 column can be considered as a proxy for FrameCounts column. ENP column stores the counts/second contribution from that specific event (row of the table) after including instrumental correction like flat-field across the detector (Ghosh et al. 2021, in preparation). In the methods mentioned below, each event is weighted as per the corresponding ENP value. For a given source coordinate position, it is possible to define an aperture of some radius in the detector coordinate system (x, y) and select only those events (rows of the table) which fall inside the aperture. Thus, a subset table can be created from the original table. Here, the term original table is used to refer to the input events list table. The subset table will have only those events of the original table, that is assumed to be solely coming from the source. We will refer to the time values (equivalent to frame numbers) in the

J. Astrophys. Astr.

(2021) 42:25

Page 5 of 10

25

Figure 1. Flowchart for the function makecurves in Curvit.

subset table as subset-time. Also, a separate array of unique time values is created from the original table. We will call this array as original-time. Using the resultant two arrays of time (subset-time and original-time), two separate histograms are created. The number of bins (same for both histograms) to be used is determined from the user-provided bin width (see Appendix A). For subset-time, the histogram can be interpreted as the number of events per bin coming from the source. Whereas histogram of original-time should ideally give a constant value per bin. For example, if bin width is set to 1 second, one should always get  28.7 frames per second at all the bins for 512  512 mode. However, if some frames are missing (for example, due to large telescope drift

or missing data), this will be reflected as reduced values in the original-time histogram. Therefore correction for missing frames is carried out using the original-time histogram (see Appendix B for a detailed explanation). By taking the ratio of two histograms, counts per frame (CPF) array is obtained. CPF array is then multiplied by frame-rate (  28.7 for 512  512 mode; the user may specify the frame-rate using the framecount_per_sec parameter) to obtain the counts per seconds (CPS) array. Finally, the CPS array is plotted against time as the light curve. The user is advised to look for variability in sources within the central 20 arcminute region to reduce telescope drift effects.

25

Page 6 of 10

J. Astrophys. Astr.

(2021) 42:25

Figure 2. Flowchart for the function curve in Curvit.

4.1 Background, aperture, and saturation corrections Estimation of background CPS can be obtained by manually specifying a background region (x bg and y bg) and aperture size (sky radius). It is also possible in Curvit to automatically determine background count-rate. To do this, a two-dimensional (2D) histogram of events with 16  16 sub-pixel2 bin size is created. A mask is applied on the 2D histogram to select only the central  24 arcminute region. As opposed to a source, a background region will not have the crowding of events around some centroid and have a comparatively low number of events in a 2D histogram bin containing it. Also, the values of

histogram bins containing background regions are assumed to be normally distributed. Therefore, the bins with sources are removed using sigma clipping of the 2D histogram values and locations of background histogram bins are randomly selected to estimate the mean sigma clipped background count-rate using an aperture size of sky radius. The end-user has also the option to apply aperture and saturation corrections using aperture correction and saturation correction parameters. Depending on the size of the radius used, the measured CPF will change. This difference is represented as a table of encircled energy at radii from 1.5–95 sub-pixels for both UV channels in (Tandon et al. 2020). Cubic

J. Astrophys. Astr.

(2021) 42:25

Page 7 of 10

25

Table 4. Sample events list that show only the columns used by Curvit. Frame counts .. . 3 3 4 4 4 5 5 5 6 6 .. .

Fx

Fy

ENP

MJD_L2

.. . 2461.9 3139.5 875.9 1444.0 2166.7 3355.4 3216.9 2798.5 3113.4 4367.8 .. .

.. . 2918.0 3651.2 2924.2 3605.5 3934.6 1229.8 1497.7 3836.8 2230.7 2483.6 .. .

.. . 28.5 25.6 23.8 23.7 24.8 25.6 26.3 25.7 27.5 20.7 .. .

.. . 213453019.048 213453019.048 213453019.084 213453019.084 213453019.084 213453019.120 213453019.120 213453019.120 213453019.156 213453019.156 .. .

Figure 3. Fraction of frames with no events as a function of total counts per frame.

interpolation was used to create a continuous function mapping radius to encircled energy in the stipulated range. From the encircled energy, aperturecorrected CPF can be represented as follows: Measured CPF  100 : Aperture  corrected CPF ¼ Encircled energy ð%Þ ð1Þ Thus, aperture correction is applied to CPF values in each bin. If the average photon rate per frame is not 1, the effects of saturation will make the measured CPF different from the real CPF by a value of RCORR. As long as the measured CPF is \0.6, RCORR can be estimated as below (Tandon et al. 2017b): ICPF5 ¼  lnð1  measured CPFÞ;

ð2Þ

ICORR ¼ ICPF5  measured CPF;

ð3Þ

RCORR ¼ ICORR  ð0:89  ð0:30  ICORR2 ÞÞ; ð4Þ Real CPF ¼ Measured CPF þ RCORR:

ð5Þ

Following the method above, saturation correction is applied to CPF values in each bin.

4.2 Zero event frames and centroid parity errors When the total count-rate for the observed region is small, then there will be many frames with no events; as the total count-rate go down, the fraction of zero event frames go up. This can be modelled using Poisson statistics (Tandon et al. 2017b).

Figure 4. The sources that are detected within the central 24 arcminute region (dashed circle) by makecurves.

F0total ¼ exp ðXtotal Þ;

ð6Þ

where Xtotal is the CPF for the whole field of view and F0total is the fraction of frames with no events (see Fig. 3). For Xtotal values above 4.6, the F0total is less than 1% and F0total is less than 5% for Xtotal values above 3. Since original-time histogram is used to account for the missing frames, CPF values for the source will be overestimated depending on Xtotal value. Additionally, a small fraction of the events (less than 0.01%) can be lost due to centroid parity errors. Since both zero event frames and centroid parity errors are randomly distributed, any light curves with periodicity and/or high count-rate can be taken to be having true variability.

25

Page 8 of 10

J. Astrophys. Astr.

5. Test on sample data We took the publicly available L2 data (UL2P version 6.3) of FO Aqr (observation ID: G06_084T01_ 9000000710) from the ISSDC AstroBrowse website. As an example, the Curvit software was run on L2 events lists that correspond to observations with the FUV filter F148W to create light curves. Using makecurves function with the detection threshold value of 4, six sources were automatically detected and Curvit generated light curves for all of them (Fig. 5).

(2021) 42:25

They are labelled as (a) to (f) in Fig. 4. An aperture radius of 6 sub-pixels (2.5 arcseconds) and time bin of 50 seconds was used. The observed variance (R2 ) of each of the light curves in Fig. 5 is due to contributions from intrinsic source variability and measurement uncertainty. In the event of the source being non-variable, the contribution of source variability to the observed variance is zero. Therefore variance becomes equal to the average value of squared errors (r2 ) of the light curve points. For each light curve, we calculated the variance and the mean of squared errors in the points

Figure 5. Light curves generated by the makecurves function in Curvit.

J. Astrophys. Astr.

(2021) 42:25

Page 9 of 10

Table 5. Variability measure of detected sources. Source

R2 =r2

Fvar

Fvar error

a b c d e f

11.87 0.88 1.01 1.39 0.35 0.61

0.14 – 0.04 0.38 – –

0.02 – 0.58 0.16 – –

and the ratio R ¼ R2 =r2 . For non-variable sources, this ratio R will be close to unity, and a source is considered variable if R is much larger than unity. The values of R calculated for the light curves of all the sources is given in Table 5. We also calculated the normalised excess variance, Fvar , for the light curves to test the significance of their variations following (Rani et al. 2019). For variable sources, Fvar will have a real and positive value (see Table 5). It is evident from Table 5 that only for the source (a), namely FO Aqr, (i) the ratio R is much larger than unity and (ii) Fvar is much larger than the error in Fvar . These indicate that the larger variance of the source (a) is due to the intrinsic variability of the source. Though sources (c) and (d) have real and positive Fvar values, and R greater than unity, for (c) the error in Fvar is much larger than Fvar itself and for (d) Fvar is not that significant considering its error. Thus from the light curves of sources (a) to (f), statistical tests confirm that only source (a) is variable. We note that the light curves of the sources presented here pertain to one orbit of data and such an analysis of one orbit data can only pick out short-period variables and will miss out long-period variables. Therefore, for sources where the period of variability is much longer than the one orbit data presented here, it is advisable to generate light curves for the complete observation (that can spread over many orbits), which might amount to carrying out photometry on each orbit wise images and then check for the presence of variability.

6. Conclusion The UVIT-POC at IIA processes the L1 data for UVIT received from ISSDC, generates science ready L2 products and transfers both the corrected L1 and L2 products to ISSDC for archival and dissemination to the PIs. Among the various files sent to ISSDC is the calibrated orbit-wise photon events list. The Curvit

25

software tool presented here is a standalone package in Python that can generate light curves from the events list. This overcomes the cumbersome task of first creating images at any time resolution from the events list and then doing photometry on the images to generate the light curves of any desired object in the observed field. The Curvit package has the capability to (1) generate light curves for all the sources in the observed field detected above a threshold for any time binning given by the user and (2) generate light curve for any particular source at any time binning desired by the user. We have also shown an example light curve of a target FO Aqr observed by UVIT. Curvit is publicly available on GitHub at https://github.com/ prajwel/curvit.

Acknowledgements This publication uses the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). This publication uses UVIT data processed by the payload operations centre at IIA. The UVIT is built in collaboration between IIA, IUCAA, TIFR, ISRO and CSA.

Appendices Appendix A. To estimate the number of bins To calculate the number of bins, the very first and last values of the original-time array is taken to estimate the width of the time array. Then, the time array width is divided by the bin width, and integer part of the resultant value is taken as the number of bins. original time width : ðA1Þ Number of bins ¼ bin width Appendix B. Missing frame correction Assume that an ideal non-variable source has a flux of 0.1 CPF (or  2.87 CPS for 512  512 mode). If the bin width were set to 1 second, then the subsettime histogram would be as follows: ½2:87; 2:87; 2:87; 2:87; 2:87; 2:87; . . .: However, if some of the frames vis-a-vis rows are missing from events list FITS table, we might get an array as given below: ½2:87; 2:53; 1:01; 2:84; 1:94; 2:87; . . .:

25

Page 10 of 10

A false variability can be inferred from the above array. To overcome this, the original-time histogram is used. The missing frames will be reflected as a reduced number of events per bin in original-time histogram. For the case above, the original-time histogram would be as follows: ½28:7; 25:3; 10:1; 28:4; 19:4; 28:7; . . .: By taking the ratio of two histograms, the real light curve in CPF is obtained as follows: ½0:1; 0:1; 0:1; 0:1; 0:1; 0:1; . . .: CPF is then converted to CPS by multiplying with the frame-rate (  28.7 for 512  512 mode).

References Agrawal P. 2006, A broad spectral band Indian Astronomy satellite ‘Astrosat’, Advances in Space Research, 38, 2989, Spectra and Timing of Compact X-ray Binaries Astropy Collaboration: Robitaille T. P., Tollerud E. J. et al. 2013, Astropy: A community Python package for astronomy, A&A, 558, A33 Astropy Collaboration: Price-Whelan A. M., SipHocz B. M. et al. 2018, The Astropy Project: Building an Open-science Project and Status of the v2.0 Core Package, Astron. J., 156, 123

J. Astrophys. Astr.

(2021) 42:25

Bradley L., Sip} ocz B., Robitaille T. et al. 2020, astropy/ photutils: 1.0.1, https://doi.org/10.5281/zenodo.4049061 Harris C. R., Millman K. J., van der Walt S. J. et al. 2020, Nature, 585, 357362 Hunter J. D. 2007, Matplotlib: A 2D graphics environment, Computing in Science & Engineering, 9, 90 Hutchings J. B., Postma J., Asquin D., Leahy D. 2007, PASP, 119, 1152 Maneewongvatana S., Mount D. M. 1999, in Center for Geometric Computing 4th Annual Workshop on Computational Geometry, vol. 2, 1–8 Million C., Fleming S. W., Shiao B., Seibert M., Loyd P., Tucker M., Smith M., Thompson R., White R. L. 2016, Astrophys. J., 833, 292 Postma J., Hutchings J. B., Leahy D. 2011, PASP, 123, 833 Postma J. E., Leahy D. 2017, PASP, 129, 115002 Rani P., Stalin C., Goswami, K. D. 2019, MNRAS, 484, 5113 Stetson P. B. 1987, PASP, 99, 191 Tandon S., Stalin C., Subramaniam A., Ghosh S., Hutchings, J. 2017a, Curr. Sci., 113, 583 Tandon S. N., Subramaniam A., Girish V. et al. 2017b, Astron. J., 154, 128 Tandon S. N., Hutchings J. B., Ghosh S. K. et al. 2017c, J. Astrophys. Astron., 38, 28 Tandon S. N., Postma J., Joseph P. et al. 2020, Astron. J., 159, 158 Virtanen P., Gommers R., Oliphant T. E. et al. 2020, Nature Methods, 17, 261

J. Astrophys. Astr. (2021)42:30 https://doi.org/10.1007/s12036-020-09689-w

 Indian Academy of Sciences Sadhana(0123456789().,-volV)F T3](0123456789().,-volV)

DATA PIPELINE

UVIT data reduction pipeline: A CCDLAB and UVIT tutorial JOSEPH E. POSTMA*

and DENIS LEAHY

Department of Physics and Astronomy, University of Calgary, 2500 University Dr NW, Calgary, Alberta T2N 1N4, Canada. *Corresponding Author. E-mail: [email protected]; [email protected] MS received 26 October 2020; accepted 19 December 2020 Abstract. For the AstroSat 5-Year Special Issue, we present a tutorial on the usage of the CCDLAB Pipeline for UVIT data reduction from Level 1 raw data to completed science images. The tutorial informs us of the unique data-processing requirements for reducing UVIT data, including one unique development borne out of such reduction applicable to the field of astronomy in general in the form of a novel approach to solving world coordinate solutions. Keywords. Instrumentation: detectors—methods: data anaysis—techniques: image processing.

1. Introduction The Ultra Violet Imaging Telescope (UVIT) (Kumar et al. 2012) is one of the main instruments onboard India’s first major orbital observatory AstroSat (Agrawal 2006), launched by the Indian Space Research Organization (ISRO) on September 28 in the year of 2015. UVIT is composed to two co-pointing telescopes comprising three imaging detectors covering the far ultraviolet (FUV, 120 nm to 180 nm), near ultraviolet (NUV, 200 nm to 300 nm), and visible (VIS, 320 nm to 550 nm) wavelength channels. One telescope is dedicated to the FUV detector system while the other telescope utilizes a beam splitter to separate the NUV and VIS wavelengths to orthogonally-placed detectors. The design image resolution in the ultraviolet is approximately one arcsecond. After a period of degassing with the telescope in a safe mode the system was opened for first light in December of 2015. While the aim of most familiar and common telescope hardware is to provide a stable and precise platform for integration imaging, requiring fine-guidance pointing and the like, AstroSat is commanded to oscillate its pointing on orthogonal This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’

UVIT image axes at a rate of a few arcseconds per second with an amplitude of a few arcminutes. The purpose of this oscillation is to protect the detector components from bright objects, and while such a procedure performed on a typical integrating detector would ruin the image, for a two-dimensional photon counter such as UVIT the nominal image field at instrumental resolution may be recovered by deshifting and coadding the count centroids as a function of the pointing oscillating. Within UVIT circles we call this oscillation ‘‘drift’’ and the time-sequence of the oscillation the ‘‘drift series’’, and this series is measured by tracking the positions of point sources in the VIS channel images or alternatively by tracking sources within the FUV/NUV centroids themselves. While the UVIT detectors are photon counters scanning a 512 9 512 CMOS at 28.7 Hz over a 28 9 28 arcminute field, the VIS channel is run in a special integrating configuration mode at lower voltage where photon counts are integrated on the chip for 1 s, thus allowing identification of point sources and subsequent tabulation of the drift series at that spatio-temporal cadence. Given the above consideration, viewing the first light image was not simply a matter of converting downloaded (from orbit) bits directly into some display format. Firstly, UVIT image data is composed of photon count centroids which must be stacked in order

30

Page 2 of 15

to form an image, and secondly, the centroids must be corrected for the drift series as well as have other instrumental and calibration corrections applied to form science-ready images. Naturally the drift series correction requires precise timing knowledge between the FUV/NUV detector clocks and the VIS detector clock, and those relative to the spacecraft clock, and it also requires software to perform the necessary driftseries tracking and then shift-and-stack operations, etc. Along with other data corrections such as flat field and distortion mitigation, etc., a complete software data reduction package is called a ‘‘pipeline’’. The first-light image reduced for UVIT is shown in Fig. 1, and a candid photo capturing the moment is shown in Fig. 2. Initial results and performance evaluation of UVIT may be found in Tandon and Hutchings (2017), inorbit calibration in Tandon and Subramaniam (2017), and additional calibration may be found in Tandon and Postma (2020).

2. Discussion In general, a data reduction pipeline for some mission should not be run by end-users, that is, by the scientists who are subscribing for time and the resulting

Figure 1. First-light image produced by the CCDLAB pipeline of UVIT in NUV of spiral galaxy NGC 2336. NGC 2336 is approximately 200,000 light-years in diameter, making it twice as large as the Milky Way galaxy, and is approximately 90 million light-years distant (December 18, 2015).

J. Astrophys. Astr. (2021)42:30

scientific data which they are interested in procuring. Only in the simplest of cases where the reduction is trivial is it reasonable to leave the processing of data to end users, for example if for image data the only corrections required are bias, dark field, background, and flat field. Otherwise, the peculiarity of instrumentation may often grow beyond the scale in ability, interest, and relevance of the end-user to be left to process themselves. That being said, peculiarities of instrumentation and the resulting data reduction may be made trivial by sufficiently well-written pipeline software, although for this to be achievable naturally requires well-behaved data as well increasing expenditure of software developer-hours. In the case where one is confronted with peculiar instrumentation, and complex methods required for the data reduction, and data which is not necessarily well-behaved, then data reduction for a mission should likely be left to the person or small group of people who are themselves responsible for the writing of the reduction pipeline software. Of course, this requires the software developers themselves to have some familiarity with and respect towards the handling of scientific data and its purposes, and it is helpful if such people also have some participation in the development of the instrumentation in the engineering phases. By ‘‘well-behaved data’’ we define the opposite of ‘‘poorly behaved data’’: data which is prone to novel sources of variation which are unpredictable and which render existing solutions to reduction of data from the same instrument under the same configuration non-functional. Of course, we hope that there is a limit to such poor behavior. In the end, the question of data reduction complexity goes from the trivial case of requiring no user input or knowledge of the process whatsoever, to requiring constant user-input and fulltime management of the reduction sequence from raw data input to science data output; UVIT has tended more towards the later of this range. At this time, five years post-launch, the CCDLAB UVIT pipeline (Postma & Leahy 2017) has been developed to a point where a significant fraction of the data reduction sequence may be run as automated, although options remain in the software program settings to scale the automation back and be run with user-input for each phase instead: there are scenarios of observed image fields or observational peculiarities which sometimes require user intervention. We shall refer to steps in the data reduction sequence as ‘‘phases’’. There remain several phases of the reduction procedure which unavoidably require user input given that these sequences are dependent upon the

J. Astrophys. Astr. (2021)42:30

Figure 2. Shyam Tandon (Indian instrument P.I., right) and Joseph E. Postma (UVIT technical support, left) enjoying the first-light UVIT image (December 18, 2015; Picture location: Indian Institute of Astrophysics, Bangalore; Photographer: Koshy George).

observed image field itself, and there is simply so much variation in target fields that it is not practically possible to develop software to identify and process all scenarios…they are better left to the software of the human mind. These cases will be discussed in the text ahead. What follows is a tutorial on the usage of the CCDLAB UVIT pipeline as run in maximallyautomated mode. Automated reduction scenarios which fail, and there are any number of failure scenarios, are best left to the ‘‘experts’’ to reduce with manual management of the sequence. We shall describe each phase of the reduction with discussion towards the instrumental peculiarities which necessitate each sequence. For reference, the development and data processing machine which the CCDLAB UVIT pipeline runs upon uses an Intel 7th generation chip with 4 physical cores and hyperthreading, supplying 8 processor threads running at 4.4 GHz. It is equipped with 16 GB of DDR4 RAM, and a PCIe 3.0 NVMe 2GB solid state drive. The CCDLAB UVIT pipeline and FITS image processor has been developed in Microsoft Visual Studio C?? over several releases of that product.

2.1 Phase A: Extraction, digestion, and drift series The general scenario for a given target field for some observation proposal is that it will typically be observed over multiple orbits. UVIT can only observe while in the shadow of Earth, limiting any particular

Page 3 of 15

30

observation to a maximum of around 1800 s, with typical observations in the vicinity of 1ks, whereas proposals typically request 10 ks to 100 ks observation times. Thus, the Level 1 (L1) data product from ISRO for a given proposal comprises a zip archive containing multiple individual orbit-wise data sets as FITS binary tables. We select the archive file in the usual manner of an ‘‘open file dialog’’ initiated by double-clicking the ‘‘Extract L1 zip Archive’’ item in the CCDLAB UVIT menu as shown in Fig. 3. A single-click of the zip Extraction menu will bring up the automation options for this phase, where each option is dependent upon the previous (higher) item having been completed. Thus, if the first option is deselected, then all following options are automatically deselected as well, and so on if intermediate items are deselected. Before we describe the meaning of each option, we must first understand the ‘‘digestion’’ processing of the data, the options of which are found in the next menu item ‘‘Digest L1 Fits File(s)’’, as shown in Fig. 4. In non-automated mode one would be required to double-click the menu item shown in Fig. 4, and use an ‘‘open file dialog’’ to open the FITS binary table files extracted out of the zip archive from the previous menu procedure. In both automated and nonautomated modes, there are a series of options which apply corrections and perform data fidelity checks upon the extracted L1 data centroid files: • PC Mode Apply FPN Correction: This option applies the ‘‘fixed pattern noise’’ correction to the centroid data. As discussed in the engineering development phases (Hutchings et al. 2007; Postma et al. 2011), the weighted-mean of an undersampled photon event or PSF results in a centroid which systematically tends toward the center pixel of the centroid kernel. Because we can centroid photon event PSFs to sub-pixel accuracy using a 3 9 3 pixel kernel, we gain resolution higher than the instrumental pixel scale itself, however, we must correct for the systematic bias in the centroid calculation of the undersampled PSFs. This effect was calibrated on-ground and the correction tables are part of the UVIT Calibration Database. • PC Mode Apply CPU Distortion Correction: By CPU we refer to the ‘‘camera proximity unit’’ of UVIT, i.e., the main camera and all of its components. Of course, there are three such cameras comprising the visible (VIS), near ultraviolet (NUV), and far ultraviolet (FUV)

30

Page 4 of 15

J. Astrophys. Astr. (2021)42:30

Figure 3. Level 1 data archive extraction.

Figure 4. Level 1 FITS digestion.

channels. Systematic field distortion likely arises mainly from the fiberoptic taper channeling photon pulses from the phosphor screen down to the 512 9 512 CMOS array. Field distortion was calibrated on-ground with further improvements developed from in-orbit data, and field distortions maps are part of the UVIT Calibration Database. A single-click of this item will open options for the interpolation scheme within the map: either no interpolation (CMOS pixel scale only) or bilinear interpolation, with bilinear being the default-selected option. • PC Mode Discard Duplicate Data Sets: Early in the mission it was found that L1 data sets were being provided with metadata which made the

data sets appear to be originating from unique observations; however, many data sets were simply duplicates of previous observations. The reason for this had to do with the onboard data buffer only refreshing itself after a certain amount of data collection, whereas the entire buffer is repeatedly downloaded to ground for processing into L1 format. Recent L1 data no longer suffers from this duplication issue as software mitigation has been applied at the L1 level, but early L1 data may still suffer from it. It is a relatively trivial fidelity check which consumes little computation time, and so it is good to leave this option selected. • PC Mode NUV Transform NUV to FUV Frame: The NUV camera shares a telescope with the

J. Astrophys. Astr. (2021)42:30

VIS channel, where the wavelengths are directed to orthogonally-placed cameras by means of a beam splitter differentiating the visible from near ultra violet wavelengths. This inverts the NUV field relative to the FUV and VIS fields, and additionally due to mounting orientation the NUV field is rotated by approximately 32 degrees relative to those channels. Thus, this option transforms the NUV centroids through an inversion and rotation matrix such that the NUV centroids nominally share the same field orientation as the FUV and VIS fields. • INT Mode Skip: This option would typically only be used when a user is investigating problems with PC mode data. The VIS images require the most time for extraction given that typically *104 images are available to be processed for a given target campaign, comprising many Gigabytes of data, whereas the FUV and NUV centroid lists are typically less than 100 MB each. • INT Mode Degradient Images: The VIS drifttracking data are full-frame reads of the CMOS array with an integration time of 1s, and the line-scans generate a relatively uniform horizontal gradient of several hundred ADU’s on top of a bias of approximately 1400 ADU’s, and so the gradient is significant. Sufficient uniformity in the horizontal gradient allows for its correction by the subtraction of the median of each column of the CMOS array from each column as such. Correcting this gradient assists in the identification and selection of sources in the images to use for drift tracking, and assists in the reliability of the subsequent sourcetracking routine. • INT Mode Clean Images: The VIS channel detector system frequently develops artefacts due to the various effects of in-orbit space radiation upon the detector system hardware. These artefacts manifest as multiple well-separated bright lines running horizontally across half of the image field, although the artefacts are inconsistent in placement, and degree, for any particular appearance of them. This menu item has options for thresholds to detect these artefacts with default values supplied. The algorithm is such that, as the VIS images are extracted out of the FITS binary tables into individual images, they are scanned row-by-row for a given number of bright pixels above the given threshold; if a bright line artefact is

Page 5 of 15

•

•

•

•

30

detected as such, then the offending pixels in the row are replaced by the average of the pixels above and below the given line. These artefacts can render VIS image data unusable for their purpose of drift-tracking, given that sources can drift over these line artefacts and the artefacts are typically much brighter than the sources. The solution described here mitigates this problem. Discard Data Sets Less Than: This item opens a drop-down list to select a value specifying the number of minutes under which an exposure should be discarded from the reduction. There are frequent 30 s observations which are performed simply for brightness-safety checks for the detectors before a full exposure is commanded, but such short exposures have too-low of SN to be accurately combined in registration with other science exposures. The default value is 2 min. Filter Correction: Early in the mission there was difficulty in aligning the time-position of the filter wheels with the time of the centroid image data, because these two systems use different electronics hardware. The detector system electronics unit has its own internal clocks, whereas the filter wheel system is a completely separate piece of hardware with its own clocks. So-called ‘‘housekeeping’’ metadata files are supplied in the L1 archives which serve to correctly align which filter was being used for a particular observation. This option generally is not required any longer at this point as the correction now occurs at the L1 creation level, but for older data it is still sometimes required. TBC: This refers to ‘‘Time Bit Correction’’ which mitigates a stuck 20th bit in the clock on one of the UVIT channels. The mitigation for this problem is now corrected at the L1 level, and so is not required. Delete Files After Digestion: This option deletes the FITS binary table files after the FUV/NUV centroid and VIS image data has been extracted and digested. It is somewhat redundant because CCDLAB will ask the user if they wish to delete all intermediate processing files at the very last step later when the science images are finalized, although it helps to reduce disk space usage at intermediate processing steps given that the intermediate-processing data files can grow to order

30

Page 6 of 15

102 of Gigabytes. The original zip archive L1 file will never be deleted by CCDLAB. When CCDLAB is installed all optimal default options will be preselected. We may now return to the ‘‘Extract L1 zip Archive’’ menu item of Fig. 3 which will initiate the extraction, digestion, and much of the processing of the data upon double-click in automated mode when all sub-menu items are selected. We shall describe the effect of each item: • Auto Run to VIS Background: This option will have the CCDLAB pipeline run through the extraction of the zip file, collating all extracted orbit-wise data into individual directories and sub-directories parented by the directory of the selected zip file, perform all data-fidelity checks and instrumental corrections as explained under the ‘‘Digest L1 Fits File(s)’’ menu item description, and will automatically determine the VISchannel background image to use for subtraction from the VIS image drift-tracking data. Typically this sequence of the phase requires several minutes. • Auto Proceed with VIS Background: This option will have the pipeline automatically proceed with the subtraction of the VIS background from all VIS image data. Typically there are order 104 of VIS images which require background correction, and the process requires several minutes. • Auto Proceed with VIS Tracking: This option will have the pipeline process each orbit-wise directory of VIS tracking images, automatically determining sources in the initial image and then tracking their centroids through the sequence of images following for each orbit. This is the first non-trivial area where it is possible for the automated mode to fail: failure would be found in the VIS image fields simply not having any good sources to use for automated tracking, in which case the user would be required to manually select very faint sources and observe when and where failure in tracking occurs and then remember to not use those sources in following attempts; this may occur in a few percent of observations. This phase displays the tracking sequence graphically on the CCDLAB image window, showing the paths for each source being tracked as in Figure 5. The automated drift-tracking algorithm will attempt to track as many sources as it can, and it will discard poor tracks and subsequentlyuntrackable sources ‘‘on the fly’’ as it runs

J. Astrophys. Astr. (2021)42:30

through the orbit-wise image sequences. The drift series are determined as differentials from the initial position for each source, and thus the multiple source tracks can be merged at the end of the process as a mean. • Auto Apply VIS Drift: This option will apply the drift series from all orbits determined in the previous step to all orbits of FUV/NUV data. A drift series typically takes on the forms for the x- and y-axes as shown in the plots of Fig. 6. In Fig. 6 there occurs an apparent deltafunction in the drift series just after the halfway mark on the x-axis; this originates from one of the several individual sources contributing to the series being skewed by some noise at that instant, and such variations are mitigated by taking a ‘‘robust mean’’ of the multiple drift series from the multiple sources tracked. By ‘‘robust mean’’ we mean an iterative average where the values of a sequence which exceed 3 standard deviations of the sequence relative to the sequence’s mean are replaced by the median of the sequence, until no values of the sequence exceed 3 sigma from the new mean. The exposure map for the image of the centroid list is created at the stage of the application of the drift series for the given centroid list. The exposure map is created by following-along the pointing of the telescope as measured by the drift series, where the active field-of-view of the detector, measured on ground from flat-field calibration, is moved about the padded field of view as unity pixel values summed into the exposure map for each frame read of the observation. At the completion of the application of the drift series we have reached the limit of full-automization reasonable to develop for the pipeline. At this intermediate stage there will be drift-corrected images along with their exposure maps and other intermediate data files in each orbit-wise subdirectory for each channel and filter, and the images will be displayed as an image-set to the user in the main CCDLAB window. If only a single orbit was ever observed for a target, then only a single image will exist and the user can skip ahead to Phase D. Otherwise, multiple orbits must be registered together to align the fields. A typical proposal and its L1 data would have required ten minutes to reach this point. Given that images have been created at the end of this sequence, we should mention the image-

J. Astrophys. Astr. (2021)42:30

Figure 5. CCDLAB displays the paths as each source is tracked in the drift series.

creation options available in the menu item ‘‘Convert Event List to Image’’, as shown in Fig. 7. • Filter Cosmic Ray Frame: The option to filter cosmic rays will remove frames from the centroid list which exceed a certain specified number of counts in the frame, given that a nominal frame should only contain one or two dozen of photon events from real sources, whereas a cosmic ray event in the frame will generate many tens to thousands of events as a ‘‘splash’’. This menu item presents options to set a cosmic ray frame detection threshold either by total count within frame, or by the number of standard deviations above which the number of counts in a frame should qualify a frame as

Page 7 of 15

30

likely containing a cosmic ray; the latter is likely the safer option to use since this accounts for variation in background. This cosmic ray filtering option is never used and the frequency of cosmic ray ‘‘splashes’’ in the centroid list simply forms a part of the nominal background. However, the option could be used for improving the signal to noise ratio for faint sources; for example, the background in the M87 region reduces by approximately 15% (from 2 9 10-4 c/pix/s) in NUV when using a 4-sigma threshold, and likewise and significantly for FUV the background reduces by 65% (from 2 9 10-5 c/pix/s). If this option is used and cosmic ray ‘‘splash’’ frames are removed from the centroid list, the final integration time for the image is appropriately adjusted given the number of frames removed. • Apply Max–Min Threshold: Each centroid comes with a ‘‘diagnostic’’ of the maximumcorner pixel minus the minimum-corner pixel of the 5 9 5 pixel kernel centered on the peak pixel of the photon event. By limiting this range as a threshold this option will ignore centroids which have potential ‘‘contamination’’ from coincident events. This option is never used for science purposes. • Centroid Image Padding: This option pads the nominal field of view so that drift correction can move centroids into regions which may have exceeded the nominal instrumental field-ofview. A default option of 44 pixels around the 512 9 512 CMOS array is selected. The exposure map uses the same padding.

Figure 6. An example of the drift series in x- and y-axes. The drift series for each source are tabulated as a differential from their initial positions, and the plots here represent several source drift series plotted together in overlay.

30

Page 8 of 15

J. Astrophys. Astr. (2021)42:30

Figure 7. Menu for converting the centroid lists to images.

• Apply Exposure Array Weighting: This option scales each centroid by its location within the exposure map such as to normalize the image to a uniform exposure time. The exposure map is applied here much like a flat-field is applied, although nominally most of the field is uniformly observed and only the periphery of the field where the drift moved the sky in and out of the field-of-view receives a non-unitary scaling correction. The exposure map is included with the final science image and so the correction may be removed (by multiplying-in the exposure map) in order to get observed total counts, etc. The exposure map originates with the active field of view of the detector and thus it also captures any ‘‘bad pixels’’ as determined in ground calibration. • Apply Flat Field Weighting: This option provides each centroid a weight, nominally of unity value, but with small variations given the flat field for each detector and filter and the original location of the centroid on the image. • Pixel Resolution: Science images are finalized at 1/8th CMOS pixel resolution, although the intermediate phase of orbit-wise image registration typically uses 1/4th pixel resolution which is set in the registration menu item to be discussed in Phase B. Final images are in total counts, and the final exposure time in seconds is given by the header keyword ‘‘RDCDTIME’’.

2.2 Phase B: Registration of orbit-wise images Generally, we will have a series of nominally-driftcorrected orbit-wise images, and due to drift these images are almost always at some unique translational offset at the scale of a few arcminutes relative to each other. Additionally, rotation of the field may enter between (but not within) orbits, thus requiring a rotational transformation as well as the translational one in order to align the fields to a common frame. Naturally a transformation matrix must be applied to the centroids in order to thus align their image fields, and therefore requires a precise determination of the translation and rotation parameters for each field. This task is accomplished with user interaction and although it is perhaps theoretically possible to code an automated routine here, the complexity of such an algorithm has seemed to the developer to exceed the simplicity of the interactive approach. Registration is accessed through the ‘‘Registration, Rotation, Transformation’’ menu item under the main UVIT menu, and is initiated by double-clicking the ‘‘General Registration’’ submenu item as shown in Fig. 8. The option to ‘‘Masterize Singles’’ is a simple housekeeping option that will format filenames and structure subdirectories as ‘‘master files’’ when there is a single orbit-observation for a given channel and filter combination. The option to ‘‘Folder Browse Scan for Most Recent XYInts Lists’’ will provide a folder browser dialog to select the channel and filter directories which one wishes to register, otherwise the user

J. Astrophys. Astr. (2021)42:30

Page 9 of 15

30

Figure 8. Menu for orbit-wise registration of images.

must use an open file dialog to then manually search for the specific centroid files wish they wish to register. The numeric dropdown item is nominally set to 4, indicating  pixel resolution at which to register the images, and this can be set to 2 (‘ pixel) or 1 (full pixel) in scenarios where the increased signal from coarser binning is required in order to identify sources. We must note that at this point the data files are structured in subdirectories such that the original location of the L1 zip archive is the parent directory, and under this parent are FUV, NUV, and VIS subdirectories. The VIS directory contains subdirectories of all orbit-wise image sets, whereas the FUV and NUV directories contain subdirectories first for a given filter, and then within each filter directory are orbit-wise subdirectories containing the centroid list data sets and their drift-corrected images. Thus, there are several options for the manner in which the registration of all orbit-wise images may be approached. All available images will currently be loaded into CCDLAB as an image set and the user may blink through them making note of corresponding sources across the image set (either mentally, or with CCDLAB by marking the sources via a right-click on the image window), where each image will have some translational and possible rotational offset relative to the others. The results of this visual scan of the images are the following possibilities: (i) There are obvious corresponding sources across all orbit-wise channel-filter images. In this case, the parent directory for all files can be selected with the aforementioned folder browser dialog, and all images can be registered as a common set in this way. (ii) There are not obvious corresponding sources across all orbit-wise channel-filter image combinations, due to astrophysical and instrumental

differentiation of the brightness of sources detected. In this case, one must typically select and process only the FUV and NUV directories separately as there always is enough correspondence for a given channel across all of its filters to be able to identify common sources for registration. In this case a final registration to a common frame for all channel-filters will occur later with the orbit-merged centroid images in Phase C. When the registration procedure is started, the user will receive simple instructions from CCDLAB to use the cursor to select sources in the initial image field. Registration is an iterative process and the procedure may be repeated as needed. For example, if there is no field rotation between the orbits then it is sufficient to select only a single source to track the translational shifts between orbits. If a user selects two sources then the registration will compute both translational and rotational shifts, and if three or more sources are selected then the rotation will compute a full 2D transformation matrix in order to effect the rotation and translation between orbits. It should be noted here that while a rotation and translation transformation should be sufficient for a given channel-filter, small residuals in the accuracy of the distortion maps as well filters which generate unique distortions and scale offsets (particularly for NUVB15 which has unique residual distortions relative to the other channel-filters on the scale of *2 arcseconds) benefit in the application of a full 2D transformation in order to align all channel-filters to the same scale and field rotation. If the first registration iteration uses only a single point, and upon the translation transformation it is then witnessed in the CCDLAB image window when blinking through the images that some rotation exists, then the user may iterate the process again and select more points for transformation. The first point that the

30

Page 10 of 15

J. Astrophys. Astr. (2021)42:30

user selects will become the ‘‘anchor point’’, and then the secondary points may be ‘‘grabbed’’ by the cursor in order to rotate them about the anchor into the position aligned with their sources in subsequent images after the first image has been used for the point selection. If all images across all channel-filter combinations could have been registered in this way then registration will be finished at this point, however, if the orbitwise NUV and FUV centroid images could only be registered separately at this point then registration will be iterated a final time after the merging of orbit-wise fields in Phase C.

After the merge one will have a ‘‘master’’ data set for each channel-filter observed for the proposal. If all fields were able to be registered previously then the registration will remain and the user may proceed to Phase D. Otherwise, if the FUV and NUV fields were not able to be registered previously due to large differentials in source detection, then with the increased signal-to-noise of the merged files, having typically ten-times the total exposure compared to a single orbit-observation, one will now be able to identify common sources such as to effect the final registration for all master channel-filter fields, and such registration may again be performed iteratively if required.

2.3 Phase C: Merging channel-filter orbit-wise data

2.4 Phase D: Optimizing the PSF

At this phase, we assume that any given filter-channel orbit-wise data will be aligned via registration. Thus, we simply merge the orbit-wise data into a ‘‘master’’ data file where all centroid data of contributing orbits are merged into a master list, from which a master image may then be produced. The exposure maps must also be correctly merged in this process so that the total exposure time may be correctly evaluated for regions of the periphery of the final images, and this is handled internally. Merging is initiated by doubleclicking the ‘‘Merge Centroid Lists’’ menu item as shown in Figure 9, under the ‘‘Registration, Rotation, Transformation’’ menu, and the option to ‘‘Delete Contributing Directories’’ cleans up the previous intermediate orbit-wise data subfolders from each channel-filter directory during the merge, and the ‘‘Folder Browse Scan for Most Recent XYInts Lists’’ option allows the user to have to only select the parent directory of the FUV and NUV folders instead of searching for the data files manually in order to begin the merge.

Due to the drift-series tracking occurring at a sampling rate of 1 Hz, and the occasional occurrence of the movement of the Scanning Sky Monitor (SSM) camera on the spacecraft inducing high-frequency differentials in the drift track, it is not possible for the VIS drift series to be a perfect representation of the telescope’s pointing oscillations. Additionally, we have found a thermal stick-slip inducing a small differential in the pointing of the FUV telescope relative to that of the VIS telescope at the CMOS pixel scale which is much larger than the instrumental resolution of 1/3rd pixel; the stick-slip is shown in Fig. 10. The effect upon the point sources when this phenomenon manifests is to extend a point source into a small ‘‘streak’’. To correct this residual in the drift series, the user may optimize the PSF of the images via the menu item as shown in Fig. 11. This procedure may be run for images which do not have a noticeable problem with the PSF and such images will also benefit in significant improvement of the profiles. At this point the merged data set images will all be loaded into

Figure 9. Merging orbit-wise centroid lists.

J. Astrophys. Astr. (2021)42:30

Page 11 of 15

30

Figure 10. Plot of thermal stick-slip which causes a differential in the pointing between the VIS and FIV telescopes.

Figure 11. Optimizing the point source spread function.

CCDLAB for viewing, and with the cursor the user may move the region-of-interest sub window over the brightest sources in the image and right-click the mouse to mark these source coordinates. If multiple sources are selected, then double-clicking the ‘‘Optimize Point Source ROI’’ button as shown in Fig. 11 will have CCDLAB automatically determine the best solution to optimize the PSF for all sources. If only a single source is selected, then the user must specify the ‘‘Stack Time’’ via the drop-down submenu, and then initiate the procedure; the stack time specifies how many seconds to use for sampling the source centroids for averaging their position within the subwindow. The region of interest subwindow size should be set such that it just contains the PSF of the brightest source…typically 11 9 11 science-image pixels. It is important to make a note here about the optimization of PSFs. If the metric for optimal PSF is either the narrowest FWHM, or a maximized peak value, then the best result is simply found in correcting all centroids from a given source PSF to fall within a single 1/8th pixel bin, i.e., in correcting the

PSF spread of centroids for a given source into a delta function. But then what is the effect of doing this to all other centroids and sources to which those corrections would be applied across the image? The instrumental PSF is produced by ostensibly random effects, and thus, correcting a single source into a delta-function induces a convolution of its PSF into all other sources, thus degrading the resolution for the rest of the image. That is, a single source would be artificially narrowed into a delta-function to produce its ‘‘best PSF’’, but the effect of the corrections required to perform this operation on the single source applied to the rest of the centroids of the image would be to convolve the PSF of the single source into all other centroids, thus degrading the resolution for the rest of the image. This is why we cannot automatically apply the corrections for the ‘‘best PSF’’ of a single source across the entire centroid list, and why multiple sources should be used to determine systematic effects of residual drift across those sources. A stack-time for a single source of ‘‘20’’ (seconds) would thus be a good choice for cases where only a single point source is available.

30

Page 12 of 15

2.5 Phase E: World coordinate solution and derotation to sky coordinates Given that there are inter-orbit variations in the field rotation aspect of the telescope, likewise the final merged master images will be rotated at some random value relative to sky coordinates. We also wish to have a world coordinate solution (WCS) in any case so that locations in the image may be mapped to catalogue and sky coordinates. Given that we are handling centroid data, we may derotate the image field via the photon event centroid list in order to align the image axes to sky coordinates as necessary once we have a determination of the field rotation from the WCS. CCDLAB implements the trigonometric algorithm as described in Postma and Leahy (2020), and this is accessed through the WCS menu item on the main CCDLAB menu bar as shown in Fig. 12. The trigonometric algorithm is an entirely novel solution to the problem of solving World Coordinate Systems, and is borne specifically out of the problem of solving WCS for UVIT images in the FUV; the algorithm

Figure 12. CCDLAB implements an automatic world coordinate solver.

J. Astrophys. Astr. (2021)42:30

however is generally applicable to any image in astronomy and can determine solutions instantaneously in most cases. CCDLAB has also implemented ‘‘astroquery’’ (Ginsburg et al., 2019) via Python script kindly supplied by Dr. Eric Rosolowski of the University of Alberta, and this menu item is shown in Figure 13. The user may simply double-click the ‘‘AstroQuery’’ menu item in order to download the GaiaDR2 catalogue file relevant to the region of the image. The catalogue region to download for UVIT images is given by the header keywords ‘‘RA_PNT’’ and ‘‘DEC_PNT’’, the field radius is specified as 17 arcminutes, and the user may specify either a circular or square region for the catalogue query (as shown in Fig. 13); these specifications may be modified within the menu item for other images with different header keywords and field sizes, etc. The RA and Dec key values may be in either numeric degree format or as textual sexagesimal format. The trigonometric algorithm is not an absolute blind solver (at this point) and so it does require a catalogue region specification of coordinates within *‘ of the field center with respect to the field width, although more accurate specifications improve performance of the trigonometric algorithm.

Figure 13. Astroquery is used to download the GaiaDR2 catalogue information for an image region.

J. Astrophys. Astr. (2021)42:30

Page 13 of 15

30

Figure 14. The images may be de-rotated to align the axes with sky coordinates.

Once astroquery is finished, the user may then click the ‘‘Solve’’ menu item and the WCS will be determined by the algorithm. Thus, at this phase, CCDLAB should have all channel-filter master images loaded for viewing and they should all be aligned from registration, and the user may then use the WCS menu to solve the WCS solution for the set of aligned fields. Please note that the WCS solution will be computed for the current image being viewed only, and given that we are querying the GaiaDR2, the image which the user should choose to determine the solution for should be the filter which is closest to visible wavelengths.

As a final correction to the data, we may derotate the image field based on the initial solution of the WCS. This task is initiated by selecting the menu item as shown in Fig. 14, ‘‘De-Rotate Loaded Images vis WCS’’. This procedure will use the WCS solved for the currently-viewed image, and use the solved field rotation therein to de-rotate all of the fields such as to align the axes to sky coordinates, i.e., vertical is increasing declination and leftward is increasing right ascension. A new WCS for the derotated field will automatically be computed and this WCS will be copied into all of the other headers of the final images.

2.6 Phase F: Finalize science products

Figure 15. Menu item to finalize science image products.

The last step is to package the final science image files for distribution to end-users, and to clean up all intermediate processing folders and files from the computer system. If the user clicks the ‘‘Finalize Science Products’’ menu item as shown in Fig. 15, then CCDLAB will create a zip file in the parent directory containing the final science images along with their exposure maps, with appropriate naming of the files such as to distinguish them. The units of the science image will be in total counts, its exposure time is given by header keyword RDCDTIME, and the exposure map is normalized to RDCDTIME so that most of the exposure map is unit value. The user will be asked if all intermediate processing files should be deleted, and if so, the original L1 zip archive will still

30

Page 14 of 15

be allowed to remain so that the source data is not lost and could be re-processed without having to download it again in the future if the need arises. The ‘‘intermediate files’’ are the centroid lists which include lists of the centroids, their solar-system barycentric Julian dates, flat-field and exposure weighting, and their detector frame numbers and frame times. These may be of interest if one wishes to examine temporal variations within the span of a single orbit, but otherwise the orbit-wise images of approximate tento-twenty-minute integrations and their mean time of observation may be taken as data points for light curves, etc. The FITS image headers list ‘‘BJD0’’ and ‘‘MEANBJD’’ as the Barycentric time of the start of the image and the mean Barycentric time of the image, in Julian Days. This is done by converting the detector clock times of the centroids to UTC via Level1 ‘‘housekeeping’’ files provided inside the L1 archives, and then converting UTC to geocentric Julian Day and then correcting to Barycentric time. The algorithm is listed in Appendix A. 3. Conclusion We have presented a tutorial for the processing of Level 1 data into final science image products. A video-tutorial may also be watched at this YouTube address: https:// www.youtube.com/watch?v=4_48yRcN3nc. Also, CCDLAB may be installed on Windows by downloading from this address: https://github.com/ wer29A/CCDLAB/releases. The UVIT Calibration Database may likewise be downloaded from: https://drive.google.com/file/d/ 1dD4R7qvsW7Eny93AqgD0IE1weWTaHj0_/view? usp=sharing. If any user desires more information, source code, or assistance with CCDLAB, please contact the authors.

Acknowledgements This publication uses the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). This publication uses UVIT data processed by the payload operations centre at IIA. The UVIT is built in collaboration between IIA, IUCAA, TIFR, ISRO and CSA. We thank the referee(s) for their review.

J. Astrophys. Astr. (2021)42:30

Appendix A: MATLAB algorithm for determining barycentric Julian Date function result = HJDC(JD,lat,long,ra,dec,varargin) % Usage: [nx3]answer = HJDC(JD,lat,long,ra,dec); % Result is returned in arbitrary ‘‘answer’’ variable as a nx3 row-major % matrix, where n is number of JD input values. 1st column is Barycentric % Julian Day corrections, 2nd column is Barycentric radial velocity % corrections, 3rd column is airmass. % % % % % %

Input: JD = full Geocentric Julian Day day.day lat = latitude of observation deg.deg long = west longitude of observation deg.deg RA = right ascension of target hours.hr dec = declination of target deg.deg

%DEFINITIONS vrot_eq = 465.1; % earth equatorial rotational linear velocity in m/s, based on spherical % earth using quadratic-mean (polar-equatorial) radius; can be improved to % take into account non-sphericity and geographical elevations, but these % are 2nd order corrections at best. au = 1.49597870e11; % astronomical unit (m) cs = 173.14463348; % speed of light (au/d) % BEGINNING OF COMPUTATIONS % west longitude of observatory in hours: L = long/15; % Greenwhich Mean Sidereal Time at JD: GMST = rem(18.697374558 ? 24.06570982441 908*(JD - 2451545.0),24); % Local Sidereal Time at JD and longitude: LST = GMST - L; %local hour angle of target: ha = LST - ra; ha = ha*pi/12; %latitude in radians: lat = lat*pi/180; %right ascension in radians: ra = ra*pi/12; %declination in radians: dec = dec*pi/180; %altitude:

J. Astrophys. Astr. (2021)42:30

alt = asin(sin(lat).*sin(dec)?cos(lat).*cos(dec).* cos(ha)); % true zenith angle; don’t care about stuff below horizon: zt = (pi/2 - alt); zt(zt [ pi/2-pi/2/50) = NaN; A = ( 1.002432*cos(zt).^2 ? 0.148386*cos(zt) ? 0.0096467 ) ./ (cos(zt).^3 ? ... 0.149864*cos(zt).^2 ? 0.0102963*cos(zt) ? 0.000303978); % Airmass of target: Young, A. T. 1994. Air mass and refraction. Applied % Optics. 33:1108–1110. % exact decimal day number from J2000.0 UT 12hr: n = JD-2451545.0; % mean anomaly, in radians, at day number n: g = rem((357.528 ? .9856003.*n).*pi/180,2.*pi); % mean longitude, in radians, at n: L = rem((280.46 ? .9856474.*n).*pi/180,2.*pi); % ecliptic longitude, in radians, at n: lam = L ? 1.915.*pi/180.*sin(g) ? .020.*pi/ 180.*sin(2.*g); % ecliptic obliquity, in radians, at n: eps = 23.439.*pi/180 - .0000004.*pi/180.*n; % distance of earth from sun in au’s at JD: R = 1.00014 - 0.01671.*cos(g) - 0.00014.*cos(2.*g); % rectangular coordinates of earth wrt solar system barycenter referred to % equinox and equator of J2000.0, in au’s: X = -R.*cos(lam); Y = -R.*cos(eps).*sin(lam); Z = -R.*sin(eps).*sin(lam); % first deriv’s of XYZ above, wrt time in days (au/d). Note: deriv’s of % XYZ w d/dt eps *0: Xdot = .0172.*sin(lam); Ydot = -.0158.*cos(lam); Zdot = -.0068.*cos(lam); % rv in direction of target due to earth’s rotational motion, ?ve away, m/s rv_rot = cos(lat).*cos(dec).*sin(ha).*vrot_eq; % rv due to earth’s orbital motion, ?ve away au/d: rv_orb = -Xdot.*cos(ra).*cos(dec) - Ydot.*sin(ra).*cos(dec) - ... Zdot.*sin(dec);

Page 15 of 15

30

rv_orb = rv_orb.*au/86400; %convert to m/s RVC = rv_rot ? rv_orb; RVC = round(RVC.*100)/100; %rounded to 2 decimal place = 0.01 m/s. % Radial Velocity due to earth’s rotation and orbital motion referred to % barycenter, in direction of target, ?ve away. BJDC= 1/cs.*(X.*cos(ra).*cos(dec) ? Y.*sin(ra).*cos(dec) ? Z.*sin(dec)); % Barycentric Julian Day Correction. ADD this to geocentric input JD % Otherwise known as Heliocentric JD (HJD) but there is ambiguity here. % Difference between HJD and BJD \\ BJDC (*1s/*5min) in almost all cases, % so only important for very-high precision timing if *isempty(varargin) if varargin{1} == ’X’ result = A(:); end if varargin{1} == ’H’ result = BJDC(:); end else result = [BJDC(:) RVC(:) alt(:)]; end

References Agrawal P. C. 2006, Adv Space Res 38, 2989 Ginsburg A. et al. 2019, AJ 157, 98G Hutchings J. B., Postma J., Asquin D., Leahy D. 2007, PASP 119, 1152 Kumar A., Ghosh S. K., Hutchings J. et al. 2012, Proc SPIE 8443, 84431N Postma J., Hutchings J. B., Leahy D. 2011, PASP 123, 833 Postma J. E., Leahy D. 2017, PASP 129, 115002 Postma J. E., Leahy D. 2020, PASP 132, 054503 Tandon S. N., Hutchings J. B. et al. 2017, J. Astrophys. Astr., 38, 28 Tandon S. N., Postma J. et al. 2020, AJ 159, 158 Tandon S. N., Subramaniam A. et al. 2017, AJ 154, 128

J. Astrophys. Astr. (2021) 42:29 https://doi.org/10.1007/s12036-020-09686-z

Ó Indian Academy of Sciences Sadhana(0123456789().,-volV)FT3 ](0123456789().,-volV)

DATA PIPELINE

Performance of the UVIT Level-2 pipeline S. K. GHOSH1,5,*, S. N. TANDON2,3, P. JOSEPH3, A. DEVARAJ3, D. S. SHELAT4,5

and C. S. STALIN3 1

Tata Institute of Fundamental Research, Mumbai 400 005, India. Inter-University Centre for Astronomy and Astrophysics, Pune 411 007, India. 3 Indian Institute of Astrophysics, Bangalore 560 034, India. 4 Space Application Centre (ISRO), Ahmedabad 380 015, India. 5 National Centre for Radio-Astrophysics (NCRA-TIFR), Pune 411 007, India. *Corresponding Author. E-mail: [email protected] 2

MS received 31 October 2020; accepted 10 December 2020 Abstract. Performance of the Level-2 pipeline, which translates the UVIT data created by the ISRO’s ground segment processing systems (Level-1) into astronomer ready scientific data products, is described. This pipeline has evolved significantly from experiences during the in orbit mission. With time, the detector modules of UVIT developed certain defects which led to occasional corruption of imaging and timing data. This article will describe the improvements and mitigation plans incorporated in the pipeline and report its efficacy and quantify the performance. Keywords. Telescopes: UVIT—instrumentation: pipeline.

1. Introduction The Ultra-Violet Imaging Telescope (UVIT) on board AstroSat carries out simultaneous imaging of the sky in three bands, viz., Far-UV (FUV: 130–180 nm), Near-UV (NUV: 200–300 nm) and visible (VIS: 320– 550 nm). Using a selectable filter for each band, radiation in limited ranges of wavelength are allowed to reach the respective detectors. Gratings for UV bands also allow low spectral resolution slit-less spectroscopy. The detector for each of the 3 bands consist of an image intensifier assembly consisting of a photo-cathode deposited on the window which proximity focuses photo-electrons onto a pair of Micro-Channel Plate assemblies biased to suitable high voltages to multiply them. These secondary electrons are further accelerated by electric field to hit a phosphor (acting as anode) to generate corresponding optical light pulses. This light is coupled through a This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

de magnifying fibre-optic taper onto a CMOS imager (512  512 pixels) which covers the entire circular active area of the detector. The full field of view of UVIT is 28 arc-min (diameter) for each band and it is read at a rate  28.7 frames/s, but a centrally positioned square window of selectable size allows for imaging in smaller fields (smallest: 5:50  5:50 corresponding to 100  100 pixels) with proportionately faster sampling in photon counting mode imaging operation of the UV detectors. The Far-UV and NearUV bands are operated in photon-counting mode, with a high gain of the Micro-Channel Plate assembly, while the VIS band is operated in an integration mode, with a low gain of the Micro-Channel Plate assembly and a much slower image read out rate which aid determination of drift of the spacecraft with high accuracy. Each band is configured for sky observations by effecting several user selectable parameters, the major ones being filter/grating and window size (coupled image frame rate). Further details about UVIT can be found elsewhere (Kumar et al. 2012a; Tandon et al. 2017a, b).

29

Page 2 of 7

The data originating from UVIT (detector modules/electronics, filter drive units) as well as the spacecraft systems (time, aspect, etc) are sorted in time and collated on ground by various processing stages at the centres of ISRO (ISTRAC, URSC, ISSDC). The resulting data products, called Level-1 (L1), are provided to Payload Operation Centre, POC, for UVIT (at IIA, Bangalore). At POC further processing are carried out using the UVIT’s Level-2 pipeline along with some auxiliary programs, which result in final astronomer ready L2 products for dissemination and archiving. The most significant role of the Level-2 pipeline is to first extract spacecraft drift parameters with time (from stars detected in sky images in VIS) and apply corresponding corrections, along with various systematic effects inherent within UVIT, to combine the series of short exposure UV data into integrated sky images. The development of the UVIT’s Level-2 pipeline, hereafter ‘‘pipeline’’ was initiated well before the launch of AstroSat/UVIT, but significant changes of strategy/algorithms/coding were needed to be carried out after real L1 data corresponding to in orbit operations became available. This article chronicles the evolution of the pipeline over the initial years after the launch in September 2015. The performance of the pipeline in terms of achieved angular resolution (characteristics of Point Spread Function) and efficiency with which input L1 data could be translated into L2 products have been quantified.

2. Functionalities of the pipeline Typical observation of any astronomical target is organized as a sequence of exposures in selected Filter-Window size configurations of Far-UV and Near-UV bands as per scientific requirements of the observer. The VIS band is configured with largest window (512  512; covering the full 28 arc-min sky field) and a ‘safe’ filter such that the brightest star in the field would not cross the nominal safety threshold set to trigger the on-board autonomous Bright Object Detect, BOD, logic of the detector. Identical detector safety concerns are also addressed while selecting acceptable filters for the UV bands. The UVIT can observe only during the dark part of any orbit due to large background from scattered solar and earth radiation in the bright part. This allows a maximum of  2000 second uninterrupted exposure at one instance (this could become even shorter in practice, due to passage of spacecraft through South Atlantic

J. Astrophys. Astr.

(2021) 42:29

Anomaly, SAA, or constraints like minimum Sun angle, etc). Hence, observation needing longer total integration time is spread over multiple orbits. Each uninterrupted imaging operation with a fixed FilterWindow configuration is called an ‘‘Episode’’. The dark part of one orbit can also accommodate more than one Episode (e.g. shorter exposures with different Filter/Window). Generally, observations of a specific target are scheduled in successive orbits sandwiched between two spacecraft slews—one for pointing to the target and the next away from it (typically pointing to another target) called a ‘‘Pointing’’. Accordingly, UVIT observations from each Pointing would generally contain multiple Episodes. The complete Level-1 data from a particular Pointing are finally combined by the ISRO ground station software systems into a single ‘‘Merged L1’’ data bundle. Initially, ISRO promptly provides Level-1 data from individual orbit dumps (say, ‘‘Orbit L1’’). While Merged L1 would mandatorily hold data from multi-Episode observations for each band of UVIT, the Orbit L1 would most often hold single Episode per band. The L1 data bundle for UVIT consists of Science (Imaging) data from all 3 detectors and Auxilliary (Aux) data containing inputs regarding UVIT Filters and various information from Spacecraft Systems, e.g. time calibration, attitude of satellite reference axes, position in orbit, gyro sensors’ output and house keeping information. The pipeline has been designed to automatically handle an entire input L1 data bundle at a time and carry out a series of tasks sequentially to generate all output products from its single run. It uses the UVIT’s Calibration Database (CALDB) and the user selected various parameters and switches through a Parameter Interface Library (PIL). As mentioned in the introduction, the detectors for all 3 bands of UVIT are constructed identical intensified CMOS-imaging systems, which can be operated either with a high gain effecting Photon Counting mode (PC) or at a low gain effectively functioning in Integration mode (INT). The detectors for Far-UV and Near-UV bands are operated in PC mode and frames are read out at a high speed (  28.7 fps for full field; max  640 fps for smallest field) while for VIS band INT mode is used at a low frame rate (  1 fps). Successive frames from the CMOS imager are read out continuously during active imaging (Kumar et al. 2012b; Tandon et al. 2017c, 2020). The on-board processing of these frames depends on the mode. For INT mode, pixel values for the image within the selected Window constitute the detector data. On the

J. Astrophys. Astr.

(2021) 42:29

other hand, for PC mode, for each frame the part within the selected Window is processed to identify photon events from the light distribution and compute their centroid coordinates along 2 axes of the detector. The details of individual detected photon events constitute the imaging data for PC mode. In view of the above mentioned differences in the data emanating from the detector operated in INT/PC modes, the processing schemes employed in the pipeline are separately dedicated for each mode. The key functionality of the pipeline is to translate UVIT measurements to astronomer ready sky images by recovering absolute aspect and the angular resolution in presence of various systematic effects and random perturbations of the spacecraft pointing. While the many processes involved to achieve it are rather complex, they are divided into two main themes: (a) Extraction of the drift and disturbances to the optical axes of UVIT/spacecraft reference axes, and use these to combine the frames using shift and add algorithm. (b) Generation of sky images of intensity and corresponding exposure arrays and uncertainty arrays, by applying all applicable corrections for instrumental effects, e.g. flat-field corrections and corrections for distortion as quantified in the calibration database. Exposure arrays are required because exposure over the field varies due to drift of the pointing, and uncertainty arrays are required because the actual number of photons depends on flat field corrections in addition to the exposure and the intensity. Each of these have been implemented as stand alone processing chain handling UVIT single band data from one Episode of observation, named Relative Aspect (RA) and Level-2 imaging (L2) respectively. Given that two very different modes of operation of UVIT, viz., INT and PC, each chain gets further diversified into two, making in total four chains. The relevant chains are operated on individual Episode data sequentially since the L2 chain needs results from RA chain run for corresponding time range. Most common instance of processing of data from a specific Episode, involves execution of the RA chain on VIS band data to extract drift series followed by two separate executions of the L2 chain on NUV and FUV band data using the corresponding drift series. In the rare situation of absence of VIS data for the particular Episode, drift series is extracted using the NUV data. This sequence completes processing for one Episode generating a complete set of resulting products. The most important ones are: UV sky images of intensity, exposure and

Page 3 of 7

29

statistical uncertainty arrays (4800  4800; pixel  0.4 arc-sec) in both the detector as well as equatorial coordinate systems (along with astrometric corrections when successful), final corrected UV event table with details for timing studies. As stated earlier, in general, the total exposure for a target with a specific Filter-Window configuration for an UV band would constitute multiple Episodes. Hence, additional processing are involved in combining the results from individual Episodes. The ‘‘combining’’ operation of multiple Episodes involves determination of relative shifts and rotations between UV images from individual Episodes and applying these corrections to align and then stacking. This is carried out in the following steps: identification of a ‘master’ Episode (with largest exposure); tabulation of stars and their centroid coordinates detected in the UV intensity image from each Episode; correlation of stars between ‘master’ and every other Episodes and determine shifts and rotations relative to the ‘master’; application of shifts and rotations to products of all non-master Episodes aligning them with the ‘master’ Episode; accumulation of corresponding products from all Episodes generating the final ‘‘combined’’ products. One important detail regarding generation of combined UV intensity image—the offsets for each Episode are applied to the centroids of individual photons first and finally gridding them on to a common array, thereby retaining the precision/angular resolution achieved in individual Episodes, even in the most general case for offset including a rotation in addition to shifts along the two axes. On the other hand, all shift and rotation operations for exposure and uncertainty arrays are performed on the array elements (following first sub-division to 9600  9600 then shift and rotation operations on sub pixels, re-gridding followed by binning back to 4800  4800 to minimize the loss of precision due to finite pixel size). The success of this scheme for combining critically depends on availability of bright enough UV stars in the field. Accordingly, the success rate for the FUV band is lower than that for the NUV band. The astrometric refinements are carried out on the combined multi-Episode UV images also. All the usual L2 products (except events list) are generated for the multi-Episode cases.

3. Evolution of the pipeline during early in orbit operations of UVIT The development of the pipeline had begun well before the AstroSat/UVIT launch when the instrument details were frozen after engineering model was

29

Page 4 of 7

realized successfully. The architectural and structural detail of the pipeline were finalized early on. Given the similarities of certain processes used in multiple chains, these were developed as modules. The chains called several such modules in appropriate sequences supplemented by unique processes dedicated for the chain. Once the laboratory data from UVIT detectors became available, many functionalities of the modules could be tested and validated. However, the real inputs for the pipeline, the L1 data bundles, became available only after UVIT was turned on in orbit. For the first time concurrent data originating from the spacecraft sub systems along with UVIT could enable testing of modules dependent on such interconnected dataset. Initially there were some surprises leading to discovery of minor oversights in the pipeline realization. This resulted in many additional features to be incorporated. Here we briefly describe a few of them. It may be recalled that the data from the UVIT Filter drive units are collected directly by the spacecraft subsystem appearing in the Aux data in L1 as opposed to UVIT detector data from UVIT electronics as Science

J. Astrophys. Astr.

(2021) 42:29

data stream. The L1 product introduces the Filter information in to detector data using time correlation between Aux and Science data. However, it was noticed that the reported identity of the Filter was incorrect at times. Since no credible reason could be attributed to this anomaly, UVIT POC devised a scheme to correct for this and generating a L10 data bundle, which was sent to ISSDC for archiving and dissemination. As time progressed, the UVIT detector electronics was found vulnerable to charged particle hits (Single Event Effects/Latchups). This manifested as peculiar anomalies in the Science data stream rendering many logics of the pipeline to fail. A few examples are described here. The raw images of VIS band showed bright (pixel with higher ADU counts) stripes along one axis of the CMOS imager, which affected the success of star finding algorithm. Since the affected pixels appeared in a pattern and also constituted a negligible fraction, appropriate logic was developed and incorporated in the pipeline to address them without any loss of functionality (introducing a new block named ‘‘Artifact Handler’’). As time progressed,

Figure 1. Example of stripes (along horizontal axis) in Quick Look image generated from VIS band images from one Episode. Such artifacts are handled successfully by the pipeline while extracting the drifts. A few curved tracks correspond to detected stars and their movement due to spacecraft drift during the observation Episode.

J. Astrophys. Astr.

(2021) 42:29

Page 5 of 7

29

Figure 2. Quick Look image in NUV band generated by plotting centroid coordinates of all photons detected during one observation Episode. The broad patches and narrow stripes (along horizontal axis) are unexpected artifacts.

the number of such stripes increased (see Fig. 1.), which required further tweaking of the fix. Later, the positions of the stripes too started jumping around, which required completely new kind of logic to by pass them. Still, the RA_INT chain could successfully extract the drift series without any degradation of accuracy. Since mid-2017, no new kind of artifacts have been observed in the VIS band images and the current Artifact Handler continues to mitigate such effects successfully. At times the Science data from successive Episodes for the NUV band showed extremely unusual patches rendering these dataset to be completely unusable (see Fig. 2). Fortunately, a power RESET could recover the NUV band from these artifacts. The L1 dataset was expected to correlate UVIT’s internal clocks (one per band) to well calibrated Universal Time Clock (UTC) based on periodic simultaneous time sampling by the spacecraft. Hence, the pipeline was originally designed using UTC as the primary reference for time. However, often this time correlation was found to be unreliable. Accordingly an additional functionality was introduced in the pipeline which allows use of UVIT’s Master Clock for timing (UVIT is configured selecting an unique band, VIS, as Master for all the 3 bands ensuring inter-band time synchronization). This timing scheme has since been used successfully. Even, while by passing the UTC for

primary processes in the pipeline, a parallel scheme has been introduced to provide approximate (good to  1 s) absolute time (MJD_UT) for every frame, by identifying selected patches of time where UTC correlates well with the UVIT Master Clock. This allows timing studies using the pipeline product providing photon event table with all details including MJD_UT, even in UTC by passed mode. In the L1 Science data for UV bands operated in PC mode, various anomalies were noticed. For example: jumps (spike) in Frame Number or Frame Time, completely discontinuous Frame Number and Frame Time inserted within normal good data sequence; data from a particular frame repeated elsewhere in the data from the same Episode; abrupt discontinuity in Frame Number which violated monotonicity. These artifacts were discovered gradually during early in orbit phase. Appropriate remedial logic were designed and incorporated within the DataIngest block of the pipeline.

4. Performance of the pipeline The pipeline is routinely operated at the UVIT Payload Operation Centre (POC) at the Indian Institute of Astrophysics, Bangalore. The L1 data are regularly received by the POC from ISSDC/ISRO. The POC sends L10 data (modified) and L2 product bundles

29

Page 6 of 7

J. Astrophys. Astr.

(2021) 42:29

Figure 3. Distribution of the quality values (from achieved PSF size and amount of pedestal) for FUV and NUV band images.

Figure 4. Distribution of the yield (fraction of input Level-1 data translated to Level-2 image products, for FUV and NUV bands) achieved by the pipeline.

back to ISSDC for dissemination to the Proposers and archiving. An installer for the pipeline, Calibration Database and relevant instructions are publicly available from the sites: https://uvit.iiap.res.in/Downloads; http://astrosat-ssc.iucaa.in/?q=uvitData; https://www. tifr.res.in/*uvit/. Though the pipeline has undergone major evolution during the initial year or so, it has since stabilized and has been performing satisfactorily. The data for the early phase are also being re-processed using the last stable version at the POC. The astrometric accuracy achieved over the central 240 diameter of the field is better than 0:300 (rms), but it is not so good for larger diameters and the errors could be up to 0:800 for parts of the outer annular region with radius between 120 and 140 . The photometric error, as determined from multiple observations of SMC fields, is found to be \6% (rms) (Tandon et al. 2020). Two key parameters to quantify the success of the pipeline are: (a) size of the PSF and (b) the amount of L1 data incorporated in the final L2 sky products. The former checks not only the quality of drift tracking but also application of all corrections for systematic effects. Accordingly, the POC has devised a quality factor which grades the intensity image products from the FWHM size of the core of the PSF as well as the fraction of intensity in the pedestal, for a few point like sources spread across the field. The best value of the quality factor is ‘10’ gradually reducing with degradation to lower values. Figure 3 shows the distribution of the quality factor for a large sample of pipeline products. The quality report for every data set is always included in the standard L2 product bundle. For a sizable fraction of the observations in both NUV and FUV bands, a score of 9 or above has been

achieved. This implies the FWHM of the PSF to be \ 1.8 arc-sec and the pedestal \20%. The second aspect relates to the ‘‘yield’’ of the pipeline. The sum of exposure time of frames contributing to the final sky images (T_L2) is compared with the corresponding frames available in the L1 data set (T_L1), as a fraction, yield = (T_L2/T_L1). The distribution of yield among a large sample of datasets is presented in Fig. 4.

5. Future plans The UVIT’s Level-2 pipeline has been performing quite satisfactorily as evidenced in the earlier section. However, at times some issues have hampered the generation of Merged L1 data sets. Since the L1 datasets corresponding to individual dump orbits are available, a scheme to stitch them in to pseudo merged L1 data has been developed at the POC. It has been planned to generate the missing merged L1 data using this scheme. Another plan for future augmentation of the pipeline is to utilize optical stars detected in VIS frames in the logic for combining multi-Episode data (instead of UV stars used currently). This would also help improving the success rate of astrometry block.

Acknowledgements The UVIT project is a result of collaboration between IIA, Bangalore, IUCAA, Pune, TIFR, Mumbai, many centers of the Indian Space Research Organization (ISRO), and the Canadian Space Agency. We thank these organizations for their support. We gratefully

J. Astrophys. Astr.

(2021) 42:29

thank members of the Ground Segment software teams of ISRO for their support. We also thank members of the AstroSat Project and the AstroSat Science Working Group for their feedback. References Kumar A., Ghosh S. K., Hutchings J. et al. 2012a, SPIE, 8443, 84431N

Page 7 of 7

29

Kumar A., Ghosh S. K., Kamath P.U. et al. 2012b, SPIE, 8443, 84434R Tandon S. N., Ghosh S. K., Hutchings J. B. et al. 2017a, Curr. Sci., 113, 583 Tandon S. N., Hutchings J. B., Ghosh S. K. et al. 2017b, J. Astrophys. Astr., 38, 28 Tandon S. N., Postma J., Joseph P. et al. 2020, AJ, 159, 158 Tandon S. N., Subramaniam A., Girish V. et al. 2017c, AJ, 154, 128

 Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:37 https://doi.org/10.1007/s12036-021-09716-4

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

DATA PIPELINE

A generalized event selection algorithm for AstroSat CZT imager data A. RATHEESH1,2,3,* , A. R. RAO1,4, N. P. S. MITHUN5, S. V. VADAWALE5,

A. VIBHUTE4, D. BHATTACHARYA4, P. PRADEEP6, S. SREEKUMAR6 and V. BHALERAO7 1

Tata Institute of Fundamental Research, Homi Bhabha Road, Colaba, Mumbai 400 005, India. Department of Physics, Tor Vergata University of Rome, Via della Ricerca Scientifica 1, 00133 Rome, Italy. 3 INAF - IAPS, Via Fosso del Cavaliere 100, 00133 Rome, Italy. 4 Inter University Centre for Astronomy and Astrophysics, Post Bag 4, Ganeshkhind, Pune 411 007, India. 5 Physical Research Laboratory, Navrangpura, Ahmedabad 380 009, India. 6 Vikram Sarabhai Space Centre, Kochuveli, Thiruvananthapuram 695 022, India. 7 Indian Institute of Technology Bombay, Mumbai 400 076, India. *Corresponding author. E-mail: [email protected]; [email protected] 2

MS received 7 November 2020; accepted 24 January 2021 Abstract. The Cadmium–Zinc–Telluride (CZT) Imager on board AstroSat is a hard X-ray imaging spectrometer operating in the energy range of 20–100 keV. It also acts as an open hard X-ray monitor above 100 keV capable of detecting transient events like the Gamma-ray Bursts (GRBs). Additionally, the instrument has the sensitivity to measure hard X-ray polarization in the energy range of 100–400 keV for bright on-axis sources like Crab and Cygnus X-1 and bright GRBs. As hard X-ray instruments like CZTI are sensitive to cosmic rays in addition to X-rays, it is required to identify and remove particle induced or other noise events and select events for scientific analysis of the data. The present CZTI data analysis pipeline includes algorithms for such event selection, but they have certain limitations. They were primarily designed for the analysis of data from persistent X-ray sources where the source flux is much less than the background and thus are not best suited for sources like GRBs. Here, we re-examine the characteristics of noise events in CZTI and present a generalized event selection method that caters to the analysis of data for all types of sources. The efficacy of the new method is reviewed by examining the Poissonian behavior of the selected events and the signal to noise ratio for GRBs. Keywords. AstroSat—CZT Imager—cosmic rays—detectors: X-rays—detectors: noise.

1. Introduction CZT Imager (hereafter CZTI) is one of the four coaligned instruments used for pointed observations in the Indian multi-wavelength astronomical satellite AstroSat (Singh et al. 2016). CZTI is a hard X-ray instrument operating above 20 keV (Bhalerao et al. 2017). It uses passive collimation and coded mask imaging to make spectral and timing measurements in This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

the 20–100 keV region. Above these energies the collimators and shield become increasingly transparent and the instrument can be used as an all sky monitor to detect transient events like Gamma-Ray Bursts (GRBs) and measure their spectral, timing and, most importantly, polarisation properties (Rao et al. 2016; Chattopadhyay et al. 2019). Though the basic CZT detectors are sensitive to only up to 200 keV, the Compton scattered events can be used to measure the total energy up to 380 keV. Further, some individual detector elements, called pixels, happened to be of low gain and hence can be used to extend the energy

37

Page 2 of 13

range of the instrument to 700 keV (see Chattopadhyay et al. 2021; Abhay Kumar et al. 2021). Non-focussing hard X-ray instruments are generally background dominated and in CZTI the coded mask technique effectively measures the background and the source simultaneously within its primary field of view. At higher energies, where CZTI has some attractive additional scientific features like the spectro-polarimetric study of GRBs, the background can be highly variable and unpredictable, limiting the sensitivity of measurements. There are, however, several design features in CZTI, though used earlier in several hard X-ray instruments but rarely simultaneously in any instrument, which can be very useful in reducing and eliminating background. AstroSat is in a low inclination (6 ) low Earth orbit, thus making the satellite only skim the surface of the high background South Atlantic Anomaly (SAA) region in most of the satellite orbits. This drastically reduces the proton induced radioactivity. Among the currently operating hard X-ray instruments only the NuSTAR satellite is in such an equatorial orbit. Secondly, CZTI uses a limited amount of shielding because it is used for imaging only up to 100 keV. This results in a much lower amount of high-Z materials thus reducing the effective induced background. The most important feature of CZTI, however, is the continuous availability of the time-tagged information for each of the registered ionising event: energy of the event, its location in the detector plane, and the time of arrival (correct to 20 ls). Since hard X-ray detectors are single photon counting devices, the effect of background can be understood and perhaps eliminated by examining the data in multiple dimensions (spatial, temporal and energy) and demanding a strict adherence to the Poisson nature of random events. Thus a pixelated detector with information of time, energy and position of interaction can distinguish between particles, particle induced noises, electronic pixel noises and genuine X-rays. This is due to the fact that the interaction process of the photons are different from particles, and their statistical properties in space, time and energy differ. A pixelated detector like CZTI is an ideal detector to search for short transients (\1 s) like short Gamma Ray Bursts (sGRBs), counterparts to gravitational wave events and fast radio bursts (FRBs) (Rao 2018). The standard CZTI pipeline primarily provide the analysis of data for on-axis sources and the search for such transient events is currently done by fine-tuning the pipeline parameters. CZTI reports the detection of around 80 GRBs per year. However, most of these detections are based on trigger times obtained from other instruments like Gamma Ray Burst monitor onboard Fermi Gamma-ray Space Telescope (Fermi-

J. Astrophys. Astr. (2021)42:37

GBM), Burst alert monitor onboard Neil Gehrels Swift Observatory (Swift-BAT) and so on. These detections are also prone to manual fine tuning of some parameters to get the maximum signal to noise ratio for that GRB. Some efforts have been made to search for transients in CZTI data, however such untriggered searches for transients require a substantial reduction of noise, without which such efforts will result in a large fraction of false alarms. Other than the electronic noises, the main source of noise in a semi conductor detector in space is cosmic particle induced noises. Particles get scattered within the detector crystal resulting in a large amount of ionisation. Apart from triggering pixels at time of arrival of a particle, pixels become ‘noisy’ for a further time period due to large deposition of charge. This time period is of the order of few tens of milli seconds, which affects the search for transients especially at short time scales. In this work, we examine the properties of ‘events’ in CZTI to segregate noise events from science analysis to to develop a source independent algorithm for reducing the cosmic ray induced noises and to improve the performance of the instrument. Primarily we study the characteristics of the cosmic particle induced noises in addition to a better handling of the electronic noises. In a companion paper (Paul et al. 2021) we presented the results of the study of the data for spatial clustering and found signatures for particle track interactions that can mimic short transients at time scales of a few hundred milli-seconds, as seen in the INTEGRAL PICsIT instrument (Segreto et al. 2003). Here we take a holistic view and examine the data in all the available dimensions (spatial, temporal, and energetic) and develop an improved algorithm to segregate the noise from genuine X-ray ‘events’. The details of the event selection methods employed in the current CZTI pipeline and its inadequacies are discussed in the next section. Temporal, spatial, and spectral characteristics of noise events in CZTI are presented in Section 3 and the details of a new algorithm for removal of these events is provided in Section 4. In Section 5, the efficacy of this algorithm is demonstrated and in the last section, we conclude discussing its advantages.

2. An overview of CZTI data and event selection in the data analysis pipeline CZTI consists of 4 independent quadrants each having an array of 16 pixellated CZT detector modules (Bhalerao et al. 2017). Each CZT detector module is

J. Astrophys. Astr. (2021)42:37

of size  40 mm  40 mm and composed of 256 pixels with 2.5 mm pitch. All quadrants of CZTI also include veto detectors, made of CsI(TI), placed underneath the CZT detector plane, which helps in identifying and removing coincidentally detected particle events and high energy gamma rays. CZTI operates in different modes, depending on the storage space available, charge particle background and various other external and internal factors. The normal mode (M0) is the default mode of operation in which it records the time tagged information for all the events that are registered. For each event, the recorded data contains the information regarding the detected energy of incident photons (PHA), pixel number, module identification (id), time stamp correct to 20 ls, a flag indicating if the event is from on-board calibration source and the energy deposited in the veto detector if it is a veto-tagged event. This list of events constitute the basic data from CZTI for all subsequent analysis. Apart from genuine X-ray photons from the source and the X-ray background, charged particles also contribute to the ‘events’ recorded by X-ray detectors like CZTI. Particles interacting with the detector lose energy continuously and produce tracks in the detector plane. Charged particles interacting with the instrument or spacecraft body can produce secondary particles and X-ray photons, which can also deposit energy in the detectors. For the scientific use of data from instruments, such particle induced events are to be identified and removed by the data processing chain. As CZTI is composed of pixellated detectors, each pixel acts as an independent X-ray or particle detector. One charged particle, however, can produce events in many pixels of CZTI at the same time by the interaction of the same particle as well as its secondaries. Due to the finite time required for polling and readout of event details from each pixel, it is possible that these events are recorded with different time stamps, although they occurred at nearly the same time. Hence, such particle interactions will be recorded as successive events having time stamps that differ by zero or by the time resolution element of CZTI: 20 ls (The 20 ls time bin is digitally generated; the on board electronics is capable of recording events at a time scale of  1 ls, and the time required for polling and readout of an event is a few ls, much shorter than the CZTI time resolution element). We call these train of events as ‘bunches’, as they are bunched temporally. As the detectors in CZTI are designed for very high count rate applications, they have very short (  1

Page 3 of 13

37

ls) charge readout time. Hence, when a high energy particle deposits significant amount of charge in a pixel, it can trigger multiple events that are recorded individually. Thus, bunches also include multiple events from one pixel. It may be noted that as two events that are recorded within 20 ls can be Compton scattered events that carry information about the polarization of the source (Vadawale et al. 2018; Chattopadhyay et al. 2019), bunches are defined as three or more events clustered in time. Since such bunched series of events are not from the source under observation, an on-board algorithm identifies them and removes them from the event records written in the on-board storage for transmission to ground. For each bunch of events discarded this way, a summary including the number of events in the bunch, total duration, full information for three events (first, second, and the last), detector numbers for additional four events, are recorded instead. For the rest of the events, complete information as mentioned earlier are transmitted to the ground and all further selection of events are carried out by the data analysis pipeline on ground. In most cases the pruned data was found to be an order of magnitude lower in volume than the noisy data, thus providing a huge advantage in the data transmission time. Analysis of in-flight data has shown that often there are residual additional events in time scales of the order of a milli-second after the end of bunched events. These are understood as due to electronic noise induced by the ‘bunch’. The low energy threshold of the instrument is kept just above the electronic noise (  20 keV) to facilitate the collection of a large number of Compton scattered events. This has the effect of making the pixels prone to some false triggering. Some pixels become quite noisy and they are suppressed by ground commands. Further, some pixels become noisy after the cosmic ray charge deposition and hence some lingering noise is also seen after the bunches induced by cosmic rays. In the CZTI Level-1 to Level-2 data processing chain, ‘cztbunchclean’ task takes care of this postbunch event cleaning. Bunches are defined as series of events which have time stamps of successive events differing by 20 ls (the time resolution of CZTI) or less. Events that are part of bunches are removed from the event list by ‘cztbunchclean’. Further, events for certain duration after each bunch are also removed by this task. As it is observed that bunches with less number of events are generally localized in the detector plane, post bunch events removal is done differently for short bunches and long bunches. For

37

Page 4 of 13

bunches with length less than 20, all events within 1 ms (T2) duration after the bunch, registered in the same detector where the bunch occurred are discarded. For bunches with length more than 20, all events within 1 ms (T3) are discarded irrespective of the detector. The algorithm also includes a provision to ignore events after the bunch irrespective of the bunch length for certain duration (T1); however, this is not used in the default processing and is not recommended. It may be noted that there is provision to vary the threshold bunch length and time scales from the default values. Although all the events removed in this manner are expected to be particle induced events, a very small fraction of genuine X-ray events coinciding with the bunch duration or post-bunch duration also get removed. Hence, the live time of the instrument needs to be corrected for this. The ‘cztbunchclean’ task also computes the live time losses incurred by the removal of particle shower duration, which is used in subsequent stages to correct the exposure times. The reduction in exposure is of the order of  1–2%. It is found that approximately 90% of the recorded events are due to events discarded as bunches and they cause this 1–2% reduction in exposure. The typical background rate, per quadrant, is 150 to 200 s1 . Apart from these particle induced train of events, other spurious events occur due to certain pixels having higher noise. During the ground testing and in-flight commissioning phases of the instrument, pixels that have extremely high noise have been disabled. Similar exercise is carried out from time to time during the mission operation. However, during each observation, it is likely that some additional pixels misbehave and generate spurious events, but not high enough to decide to disable that pixel. In such cases, events from these ‘hot pixels’ also need to be removed from further analysis. It is to be noted that near room temperature pixelated semiconductor detectors like CZT are prone to increased noise in some pixels and these noisy pixels are isolated and removed by ground software (see, for example, Segreto et al. 2010, for a description of the noise reduction method for Swift/BAT.). In CZTI data analysis software, the task named ‘cztpixclean’ identifies and discards events from these noisy pixels. This is carried out in two steps. In the first step, pixel-wise histograms for each quadrant for the entire observation is computed. For a total of 4096 pixels per quadrant, and with an average count rate of  150 counts/s, none of the pixels are expected to deviate beyond 5 r from the mean due to the Poisson statistics that a photon counting instrument should

J. Astrophys. Astr. (2021)42:37

follow. Hence, pixels that show a deviation more than 5r from the total average in a Detector Plane Histogram (DPH) are considered as ‘‘noisy’’ pixels and are removed from any further analysis. At least a data set with at least 2000 s are required to get enough statistics per pixel to allow such an analysis. This process is done iteratively and the iteration is continued until all remaining pixels have counts within the five sigma limit. In this process, the number of the events removed as noisy pixel events depends on the number of noisy pixels and counts per noisy pixel in that observation. In general around one percent of the pixels are flagged as ‘‘noisy pixels’’. In the next step, flickering detectors and pixels that have noisy events only within certain duration of the observation are identified and events from them are discarded only for that duration. Such flickering pixels and detectors are identified as those having count rates higher than a certain limit, considering the maximum deviations expected from a Poisson distribution with the mean count rates. As the average count rates observed by CZTI is dominated by background, it is expected to remain constant in case of observations of persistent X-ray sources and thus the default threshold rates (hereafter cztpipeline_lowthresh_run) for pixels and detectors required for the algorithm are estimated assuming the typical background rates. Thus, the algorithm discards events from pixels for one second time bins that register more than 2 counts/s and from detectors that register higher than 25 counts/s. However, this assumption is not valid in the case of transients like GRBs, where the count rates increase significantly above the background rates. In such cases, this part of the algorithm in ‘cztpixclean’ with the default parameters will tend to remove significant fraction of source events in addition to the noise events. Thus, for purposes like search for transients and analysis of GRBs with CZTI observations, this part of the event selection in the data analysis pipeline is bypassed by providing high values for the detector and pixel count thresholds (hereafter cztpipeline_highthresh_run). While this ensures that no source events are rejected, some of the spurious events remain in the final list of selected events. Various values based on the flux level of the GRBs are manually set by trial and error to get the maximum signal to noise ratio and to get a stable background. Although it is possible to optimize the threshold parameters based on the count rate of an already identified GRB, it will not be possible to do that in case of requirements like blind search for transients.

J. Astrophys. Astr. (2021)42:37

The event selection algorithms employed in CZTI data analysis pipeline as described here is found to be effective to identify and remove most of the noise events for observations of persistent X-ray sources. With the low threshold configuration in ‘cztpixclean’, it is estimated that the contribution from the residual noise events in the clean event files is less than 10% of the statistical error due to the background (see Section 5). For the analysis of bright transients like GRBs, however, the software parameters had to be tweaked depending on the brightness of the GRBs so that the source counts are not suppressed as noise. Considering these limitations in the event selection algorithm in the current data analysis pipeline, here we re-examine the characteristics of the noise events in CZTI and propose improvements to these methods that overcome these limitations.

3. Noise characteristics at millisecond time scales We have investigated the events at milli-second time scales to study the characteristics of the noise events. The main motivation to explore milli-second time scales is the observation that the lingering effects of cosmic particles can extend up to a few tens of milliseconds, even though the primary interactions and energy deposition are in nano-seconds. To perform this study, we use the output event file obtained from ‘cztbunchclean’ with no post bunch cleaning (T1 ¼ 0, T2 ¼ 0 and T3 ¼ 0). Though the new algorithm is tested on several data, for a systematic study that is presented here we have selected two data sets for a detailed analysis: AstroSat observation IDs 9000000618 (hereafter obsID1) and 9000000276 (hereafter obsID2). The data sets are of sufficiently long duration so that they cover the observed diurnal variation in the background. We further flag the ‘noisy’ pixels from this analysis, as their source is already identified as electronic noises. Since the cosmic rays interact locally at the module level, the noise triggered by it should also be locally clustered at the place of interaction of the particle. Hence we probe the time scale of lingering noise at the module level. We generated lightvcurves at 0.5 ms, 1.0 ms, 2.0 ms, 4.0 ms, 8.0 ms and 16.0 ms for each module. From the mean count-rate we estimate the expected counts from a Poisson distribution and compare it with the observed counts. The estimated Poissonian counts and observed counts histogram for each module is then added to get statistics at the quadrant level. Since the background of CZTI shows a slow orbital variation,

Page 5 of 13

37

we divide the data into chunks of 100 s intervals. The observed and expected counts from all these chunks are summed up to get the final histogram. An example of such a histogram for 8.0 ms binning is shown in Fig. 1. A consistent behaviour is seen for both the obsIDs as well as for all quadrants of CZTI. It can be seen that there are a large number of occasions when counts in excess of 10 are found, while the expected number from Poisson distribution is practically zero. We note here that any deviation in the Poisson distribution by slow variation in the counts due to orbital background variation is explicitly taken care of by calculating the expected Poisson distribution for the each 100 s chunk of data. Further, by taking the counts for each module we essentially reduce the average expected rate and thus enhancing the contrast due to noise because the noise variations are expected to be localised to each module. We can use these histograms to estimate the residual noise counts by integrating the observed histogram and subtracting the expected Poisson counts. Thus for the 8.0 ms binned data, the estimated noise count rates in all the quadrants are 23.53 counts/s and 17.9 counts/s in obsID1 and ObsID2, respectively, for the 8.0 ms binned data. We use this technique to get an estimate of the clustered noise events at various time scales. The estimated noise event rates at 0.5 ms, 1.0 ms, 2.0 ms, 4.0 ms, 8.0 ms and 16.0 ms timescales are plotted in Fig. 2. Again a consistent trend is seen in the two obsIds and across quadrants: though there is about 20% variation among the different sets of data. It demonstrates that the noise events have a time scale of several ms and most of the noise events are captured when we examine the data at 8 ms time scale. Hence we chose 8 ms for further estimating the amount of noise events. Since the expected Poissonian count histogram falls to zero beyond 10 counts per 8 ms per module (see Fig. 1), all the counts detected beyond it have to be noise counts. We used these counts to study their distribution in energy across the detector plane to further investigate their characteristics. When we made a plot of the energy spectra of these events, it showed multiple peaks at different energies (top panel of Fig. 3). On a closer inspection of the data we realised that these peaks are module dependent and hence we scaled the energy scale of the plot to the Lower Level Discriminator (LLD) value of each module (we note here that in CZTI there are no pixel wise LLD, rather there is only a module wise LLD: each module having only one analog to digital converter). Different modules have their LLDs in the range 15 keV to 65

37

Page 6 of 13

J. Astrophys. Astr. (2021)42:37

Figure 1. Histogram of counts per 8 ms per detector module for obsID1 (top) and obsID2 (bottom). Columns from the left shows quadrant 0, 1, 2 and 3, respectively. The blue line shows the expected Poissonian distribution and the red line shows the observed count distribution.

Figure 2. The noise counts at different time scales for all the quadrants. The noise counts are estimated from the deviation from Poissonian. Top and bottom plots are for obsID1 and obsID2. Different quadrants are labelled within the plot.

keV. In the distribution of the normalized energy (taking the energy scale from the LLD of each module), we see that there is only a single peak near the LLD, thus signifying that the noise events are clustered around the electronic LLD of the modules. Hence we keep a threshold of 10 keV from the LLD of each module to identify the clustered noise events if they satisfy further conditions as outlined in the next section. 10 keV is an adequate value to cover the entire range of post bunch noises and the genuine

events from being flagged based on trial and error on different data sets. We also examined the detector plane histogram (DPH) of the above mentioned noise events. The counts are uniform across the quadrant except for a few remaining noisy or flickering pixels as seen in the detector plane histogram (DPH) of these events (Fig. 4). This indicates that these are electronic noises, however they are uniform and hence not associated to just a few noisy pixels. We conclude that these extra noise events are caused by the lingering effects of cosmic ray induced bunches in all active pixels and they appear uniform in the DPH. Now we examine the lingering effects after a particle interaction in the detector plane. When a particle interacts and triggers multiple pixels within 20 ls, then they are registered as a bunch as mentioned above. We find excess counts for few tens of milliseconds just after the interaction of many ‘bunches’. However these are mainly found for ‘bunches’ with more than 15 events. Fig. 5 shows the post bunch noise associated with the bunch, the light curve showing excess counts after the bunch, and energy distribution which shows excess counts near the LLD. The time interval for the energy distribution and the DPH are indicated as green shaded part in the light curve. The time of the ‘bunch’ is indicated by the vertical red line. From the DPH the post-bunch noise along the track left by the particle can be seen clearly. In most of the cases the time scale of post bunch effects is around a few tens of milli-seconds, however in some extreme cases, the time scale can go as large as 250 ms. The plotted data corresponds to quadrant 0 of obsID2. Similar properties in energy and temporal

J. Astrophys. Astr. (2021)42:37

Page 7 of 13

37

4. A new event selection algorithm

Figure 3. Top: The spectral energy distribution of events selected from 8 ms bins of module wise data when the observed counts go beyond 10 counts per bin (deemed to be noise events, see text). Multiple peaks are seen in the distribution. Bottom: The same distribution when the energy scale is re-normalised for each module by taking the start point as the LLD of that module. A clear peak is seen near the LLD and the red vertical line shows the energy threshold used for further selection criteria. Different quadrants are indicated by the labels within the plots. The data corresponds to obsID2.

Figure 4. The detector plane histogram (DPH) of the noise events selected according to the criterion that the integrated counts over 8 ms bins goes beyond 10 counts. The plotted data correspond to obsID2.

regimes for post bunch noises in comparison to the non-Poissonian noise identified above indicate that the post bunch noises are the main source of the nonPoissonian noise.

Based on all the above findings, we have formulated a strategy for removing the noise events. This essentially replaces the routines ‘cztbunchclean’ and ‘cztpixclean’ in the present analysis pipeline. The algorithm is divided into four different subsequent tasks. At first we remove the gross noisy pixels from the detector plane histogram (DPH) of a quadrant, as it is the predominant component of noise in the detector. This is done in an iterative manner. All the pixels in a quadrant which register counts above 5 sigma from the mean are classified as ‘noisy’ pixels and are removed from the analysis. Since normal pixels and ‘low gain’ pixels have a difference in the count rate, the flagging is done separately for them. Since the edge pixels in a module have less geometric area, all the counts in DPH are normalised by the geometric area of the particular pixel. The mean and sigma are recalculated after the first iteration to catch and exclude fainter ‘noisy’ pixels. The iteration is repeated until no further pixels are caught as ‘noisy’. The deviation from the mean (5r) is parametrised as noisypixsigmathresh. This step is similar to the algorithm employed in ‘cztpixclean’ of the CZTI pipeline, but with some refinements such as handling the low gain pixels separately. This same method of flagging pixels is additionally used in the final event file for identifying noises in the ‘neighbouring double’ events used for polarimetry. Two events occurring within in the time resolution of the instrument are termed as ‘double’ events, and such events in neighbouring pixels are termed as ‘neighbouring double’ events. Since the corner pixels angles register less events, we flag the outliers separately for the corner and side double events. We further split the flagging process for low gain pixels and normal pixels as the count rate observed in them are different. If a pixel is found noisy for any of the segregated DPHs, then no further Compton events are used from that pixel for polarimetry. In the second stage of event selection, we identify and remove post-bunch noise events. This task is carried out in two steps based on the number of events in the bunch. As bunches with more than 100 events span more than one module, the module identification becomes tricky as they are identified from 7 module numbers among all the bunch events. When such a bunch occurs it may trigger pixels across at least 2 modules and hence can disturb a large number of

37

Page 8 of 13

J. Astrophys. Astr. (2021)42:37

Figure 5. Left: The light curve binned at 0.01 s showing the post bunch noise after a bunch. The time of the bunch is indicated as red vertical line. The green shaded region shows time selection for the energy distribution and the DPH. Middle and right: The energy distribution and DPH for the post bunch noise.

pixels. However, these bunches can be identified from the total number of events in a bunch. At present we keep a threshold of 100 events (parametrised as superbunchsize) to distinguish such bunches. These bunches are classified as ‘super bunches’. After identifying these bunches we perform the DPHclean method outlined in Paul et al. (2021) to check if there is a DPHstructure or spatial clustering in the DPH for the next 100 ms (parametrised as DPHtime) from the start time of the bunch. If a clustering is found, that time interval is excluded for that quadrant. The fractional live-time during this time are updated accordingly. In the next part of the post bunch clean, we include bunches with events greater than 15 events (parametrised as heavybunchsize) as well. These bunches are called as ‘heavy bunches’, and the rest of the smaller bunches are termed as ‘small bunches’. These particle tracks that trigger less number of bunch events, can be locally identified in a module or two. For such a bunch, 4 further criteria determine if these post bunch events need to be excluded. These events are excluded if all the four criteria are met at the same time • the event is within 250 ms (parametrised as heavybunchtime) after the bunch, • the energy of the event is less than LLD ? 10 keV. The LLD here is the LLD of the pixel in which the event is found, • the event is in the same module as the bunch, or in the neighbouring 4 columns and 4 rows of pixels adjacent to that module, • the event is clustered in 8 ms with another event that also follows the above three criteria. Furthermore, all events from the same module or neighbouring pixels are excluded for 5 ms (parametrised as heavybunchtimedet) or 1 ms (parametrised as smallbunchtimedet) for ‘heavy bunches’ or ‘small bunches’. The pixel exposure values of each pixels are corrected accordingly.

The final task of the algorithm is to identify and remove events from flickering pixels. Since pixels can flicker from very short time scales to long time scales, we apply two strategies to find and exclude events from flickering pixels. The first strategy is to find pixels that show non-Poissonian behaviour with respect to time and thus identified from the deviations in the individual pixel light curves. The light curves are binned at 100 seconds for each pixels and the deviations from the mean are checked to find the flickering pixels. A bin size of 100 s is appropriate to find the flickering pixels as it gives enough statistics per bin. The significance of deviation (parametrised as flickersigma) and the number of allowed times for the deviation (parametrised as flickernumtimes) are decided by the total exposure of the observation. Since the overall background of CZTI follows a trend, the pixel wise light curves are normalized by the total light curve for the quadrant. This will essentially avoid flagging short time increase in count-rate due to genuine GRBs detection, as flickering episodes because such increase will be more or less uniformly distributed across the detector plane. Further, the count rate in each bin is corrected for the fractional exposures calculated from the good time intervals (GTIs) of each quadrants. Pixels caught as flickering will be excluded for the entire observation period. The second strategy to find flickering pixels is looking for non-Poissonian behaviour in the DPH at shorter time scales, i.e. 1 s and 10 s. Depending upon the mean count rate in that time interval, the pixel counts which are not probable to occur at least once are calculated, starting from 2 counts, and all the pixels that register counts above that are removed for that particular interval. Pixel exposures are updated accordingly after this step. Table 1 summarize all the parameters and the respective default values used in the new algorithm. The flow chart given in Fig. 6, outlines the entire algorithm. The abbreviations used in the flow chart for visual representation are given in the Table 1.

J. Astrophys. Astr. (2021)42:37

Page 9 of 13

Table 1. Parameters of the event selection algorithm. flickersigma and flickernumtimes are based on the exposure of an observation. The range of values used are given in the table. Parameter

Default value

noisypixsigmathresh (Nr ) superbunchsize (SBlen ) DPHtime (tDPH ) heavybunchsize (HBlen ) heavybunchtime (tHB ) heavybunchtimedet (tHBM ) smallbunchtimedet (tSBM ) flickersigma (FNr ) flickernumtimes (Ntflick )

5 100 100 ms 15 250 ms 5 ms 1 ms 5–8 1–3

5. Results We now employ the algorithm described in the previous section to obtain clean events for CZTI observations. The efficacy of the new event selection strategy is evaluated by examining any residual clustering in temporal, spatial, and spectral domains for the cleaned events. Finally, we use data from detected gamma ray bursts to quantify the improvement in signal to noise ratio with this method.

5.1 Clustering at millisecond time scales We re-examine the non-Poissonian behaviour at 8 ms time scale in the cleaned event files to quantify the reduction in temporally clustered noise. We find that the noise events calculated from the difference in the Raw event list excluding bunches

Compute mean (μ) and std deviation (σ) of counts per pixel

Find noisy pixels with count> μ + Nσ × σ

observed and the expected count rates decreased by an order of magnitude. The noise count rate after the cleaning is 3.33 counts/s and 3.72 counts/s in obsID1 and ObsID2 in comparison to 23.53 counts/s and 17.9 counts/s before the cleaning. We also compare these noise count rates with the results of the both the cztpipeline run configurations (as outlined in Section 2) and for both the observations. For the cztpipeline_lowthresh_run the noise count rate is 5.16 count/s and 4.39 count/s in obsID1 and ObsID2, and in cztpipeline_highthresh_run is 20.43 count/s and 14.96 count/s in obsID1 and ObsID2. Figure 7 shows the expected Poissonian counts in quadrant 0, the observed counts before and after noise clean, and also for the cztpipeline_lowthresh_run and cztpipeline_highthresh_run of the cztipipeline. The other three quadrants also show similar behaviour as in Fig. 7. From Fig. 7 and from the noise count rates, it is evident that the current algorithm decreases the number of noise events better than the cztpipeline_lowthresh_run of the cztpipeline, and at least by a factor of 5 with respect to the cztpipeline_highthresh_run.

5.2 Spatial clustering and DphStructures Using the DPHclean algorithm discussed in Paul et al. (2021) we examine the level of spacial clustering within a bin size of 100 ms at different steps event selection algorithm. This DPHclean algorithm is previously used within the event selection algorithm to exclude time intervals after a bunch of size 100 events (see Section 4), but here we use it on the entire data set to quantify any spatial clustering in the

Input event list for post-bunch filtering

Input event list for flickpixclean

Loop over each bunch

Compute μi and expected σi for normalized rate in each pixel for 100s time bin

Discard post-bunch events within tHB if they meet energy, pixel, and clustering conditions

Length > HBlen ?

Yes

No No

Any noisy?

Discard all events within tHBM in the same module

Length > SBlen ? Yes Check for DPHStructure in DPH of events within tDP H

Discard all events within tSBM in the same module

No

Present?

Yes Discard events from noisy pixels

37

No

Last bunch?

Yes Discard all events within tDP H

Find number of time bins (N ti ) with rates higher than μi + F N σ ∗ σi Discard all events from pixel i, if N ti > N tf lick Generate normalized pixel lc for 10s and 1s time bins Discard events from pixel for 1 or 10s time bins if counts exceeds threshold

Yes Event list removing noisy pixels

Event list after post-bunch filtering

Final clean event list

Figure 6. Flow chart outlining the new event selection algorithm sequentially. The abbrevations used are given in the bracket for different parameters in Table 1.

37

Page 10 of 13

J. Astrophys. Astr. (2021)42:37

Figure 7. Histogram of counts per 8 ms per detector module for obsID1 (left) and obsID2 (right) after the new event selection algorithm (green), as well as low_thresh_config (purple) and cztpipeline_highthresh_run (magenta) products. The plotted data corresponds to quadrant 0. The blue line shows the expected Poissonian counts and the red line shows the observed counts without any post bunch clean.

remaining selected events. A bin size of 100 ms was used since this time scale is identical to the time scale of post bunch noises. DPH for every 100 ms starting from the start time to the stop time is generated and checked for spatial clustering using the above mentioned algorithm. Those DPHs where spatial clustering was found are termed as DphStructures. In this section we try to quantify the rate of DphStructures in event files at different steps of the noise clean algorithm. Since post bunch noises are the prominent form of spatially clustered noises, the rate of DphStructures indicates the amount of residual post bunch noise remaining in the data. We find that the the rate of DphStructures after the ‘noisy’ pixel clean is 0.93 per second and 0.95 per second in obsID1 and obsID2. After the postbunchclean1 it reduced to 0.42 per second and 0.48 per second for obsID1 and obsID2. Further, after postbunchclean2 it reduces to 0.22 per second and 0.19 per second for obsID1 and obsID2 respectively. After flickpixclean it again reduced to 0.08 per second and 0.07 per second for obsID1 and obsID2. This indicates that the noise due to DphStructures reduces by an order of magnitude after the noise clean algorithm.

5.3 Spectra of clean events Now, we examine the spectra of cleaned events. Figure 8 shows the energy spectrum for clean single pixel events for ObsID2, along with spectrum of the scattered events of the double pixel Compton events that are used for polarimetry. The single event spectrum shows the expected Tantalum K  a and K  b

Figure 8. The spectra of single and Compton double events showing peak around 40 keV as also seen in the ‘bunches’ indicating the particle origin of these events.

lines and an additional line feature around 40 keV. This line feature at 40 keV is also seen in Compton event spectrum and is known to be a prominent feature in the events that constitute the bunches, as shown by the bunch event spectrum in the figure. Thus, a small fraction of events having similar spectral characteristics as the bunch events remain present in the clean event files. To estimate the contribution of such events, the spectra around the line were fitted with a powerlaw continuum and a Gaussian line and fraction of events within the line were computed. It is seen that about 1–1.2% of the total clean single and Compton events are arising from this line at 40 keV and the line fraction is about 40% for bunch events. Assuming that the spurious events left in the clean event files have spectrum similar to the bunch events, we estimate the fraction of residual noise events to be  2.5–3%.

J. Astrophys. Astr. (2021)42:37

Page 11 of 13

37

Table 2. Signal-to-noise ratio (S/N) and 3r outliers per quadrant calculated for 11 GRBs using the cztpipeline_lowthresh_run, cztpipeline_highthresh_run and new event selection algorithm. For extremely bright GRB160821A cztpipeline_lowthresh_run shows a negative S/N since the counts during the GRB interval was killed and instead of a peak it resulted in a drop in the light curve. GRB name GRB191225B GRB190315A GRB180703B GRB180504B GRB180120A GRB180411C GRB171126A GRB171223A GRB200916B GRB200920B GRB160821A

cztpipeline_lowthresh_run

cztpipeline_highthresh_run

S/N

3r outliers

S/N

3r outliers

S/N

3r outliers

2.76 2.52 1.51 1.69 4.48 3.06 3.76 2.02 2.88 0.98 1.76

2.01 5.3 6.64 4.15 11.57 5.09 4.15 5.75 5.54 2.95 35.67

5.35 4.01 3.06 3.48 3.45 5.92 2.99 5.49 2.09 3.73 2.88

2.3 5.54 6.97 4.22 11.99 5.45 4.34 6.23 6.43 3.25 38.2

1.52 0.85 1.08 0.94 0.92 2.08 1.25 0.91 1.19 1.21 1.41

2.07 4.24 3.36 3.89 8.76 5.01 4.13 3.9 3.97 3.41 0:51

Although the effect of these residual events are visible in the energy spectrum, they do not show any clustering in time or detector position, which is why they are not removed even after the improved event filtering techniques. As these residual events are random in time and uniform over the detector plane, they act as additional background events and thus get removed in background subtraction. Thus, this small fraction of residual events do not contribute to any additional systematic errors.

5.4 Signal-to-noise of GRBs We calculated the signal to noise ratio in 11 gamma ray bursts (GRBs), previously detected by CZTI to show the reduction in the background noise without compromising on the source counts for the new algorithm. We compare our results with the cztpipeline_lowthresh_run and cztpipeline_highthresh_run of the cztpipeline. The light curve of the short and long GRBs were binned at 1 s. Bins with exposure times less than 0.3 are excluded from the analysis. The trigger time (Ttrig ) are obtained from the AstroSat CZTI GRB catalog (http://astrosat.iucaa.in/ czti/?q=grb). The source region in the light curve was selected manually. 500 seconds before and after the GRB region was selected as background. However the background pre- and post-GRB background was taken 5 seconds away from the GRB to avoid contamination from the source counts in

New algorithm

modelling the background. Since the background of CZTI follows a trend we fit a quadratic function to detrend the pre- and post-GRB backgrounds for all the quadrants. The fitted function was then subtracted from the light curve to obtain a background subtracted light curve. The signal-to-noise ratio of the GRB is estimated by M/S, where M is the mean during the background subtracted GRB time interval, and S is the standard deviation of the de-trended background. The sample consists of 5 short GRBs (T90 \2 s) and 6 long GRBs (T90 [ 2 s). Table 2 shows the signal-to-noise ratio of all 11 known GRBs in comparison with the cztpipeline_highthresh_run and cztpipeline_lowthresh_run. We also calculated the amount of 3r peaks per quadrant above the poisson limit. This is calculated by subtracting the estimated amount of peaks beyond 3r level from the Poissonian equation from the observed peaks in the light curve beyond the 3r level at the quadrant level. Table 2 shows that while the signal have been lost in some cases for the cztpipeline_lowthresh_run, 3r peaks per quadrant increase in case of the cztpipeline_highthresh_run. However the new algorithm retains the GRB signal without compromising on the amount of particle induced 3r peaks. It is also seen that there is a slight improvement of the signal-to-noise ratio, as well as a reduction in the particle peaks in comparison to both the methods from cztpipeline. Figure 9 shows the lightcurves of the GRB from the cztpipeline_highthresh_run of cztpipeline and after noise clean algorithm. The peaks that disappear after the new event selection algorithms are the particle induced noises. It

37

Page 12 of 13

J. Astrophys. Astr. (2021)42:37

Figure 9. CZTI light curves of different GRBs at quadrant level from cztpipeline_highthresh_run (left) and the new event selection algorithm (right). The vertical blue dotted line indicates the GRB trigger time. The 3r outliers are marked by red dots and are seen more in the cztpipeline_highthresh_run.

is evident from the light curves, that outliers in light curves in second and sub-second time scales has reduced significantly, and the search for astrophysical transients will not be influenced by particle induced and electronic pixel noises. Hence the false alarm rates due to particle induce noises will be drastically reduced for an automatic transient search.

6. Conclusions In this paper, we have outlined a method of event selection in the CZTI data based on their clustering properties at spatial, spectral and temporal dimensions. It is found that the cosmic ray induced noise in the CZTI data has been reduced by a significant

J. Astrophys. Astr. (2021)42:37

amount without source flux dependent manual configuration of the parameters. The same algorithm provides cleaned data for both highly variable short duration bright transient sources as well as for the background dominated on-axis source observations, with the same configuration of the parameters. We also found that the cosmic ray tracks which can mimick short GRBs are also reduced significantly. Searches for transients, especially short transients, like short GRBs, counterparts to gravitational wave events, counterparts to fast radio bursts (FRBs), will benefit by using this algorithm as the false trigger rate will be reduced significantly. The reduction in the particle tracks not only improves the transient and GRB searches but also the science products like spectrum, polarisation and localisation for them. This is because the particle tracks are not uniformly distributed across time for short timescales like a few hundreds of seconds, which makes the background subtraction in energy and DPH improper. Since the new algorithm is robust at divergent count levels, it will be very useful for combining data obtained for long duration at different flux levels like that needed for the off axis pulsar searches, especially for the fainter milli-second pulsars. Further co-adding data to search for sources at lower flux levels is easier with this new algorithm because the parameters can be set uniformly across data sets and an automatic analysis procedure can be established. With additional features in flagging ‘noisy’ and ‘flickering’ pixels, this algorithm has a better handling of the pixelated noises in temporal and spatial regimes in comparison to the cztipipeline. Currently a beta version of the algorithm is available. This will be used in multiple data sets for different uses. The algorithm is organised in a structured way such that the default parameters can be tweaked, the order and sequence of analysis can be experimented such that a robust understanding of the various aspects of noise could be arrived at. A new version of cztipipeline is currently being developed to include improved aspects of energy calibration and imaging and we plan to incorporate this noise clean algorithm in the new pipeline.

Page 13 of 13

37

Acknowledgements The data used in this work is from the AstroSat mission of the Indian Space Research Organization (ISRO), archived at the Indian Space Science Data Centre (ISSDC). We acknowledge the AstroSat CZTI Payload operation centre at IUCAA for their help in developing and testing this algorithm. The CZT Imager instrument was built by a TIFR-led consortium of institutes across India, including VSSC, ISAC, IUCAA, SAC, and PRL. The Indian Space Research Organisation funded, managed and facilitated the project. We thank Mayuri Shinde for her contributions in developing the algorithm into a code in C?? programming language. We also thank Avinash Aher for his contributions in data analysis.

References Abhay K., Chattopadhyay T., Vadawale S. V. et al. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036021-09711-9 Bhalerao V., Bhattacharya D., Vibhute A. et al. 2017, J. Astrophys. Astr., 38, 31 Chattopadhyay T., Vadawale S. V., Aarthy E. et al. 2019, ApJ, 884, 123 Chattopadhyay T., Gupta S., Sharma V. et al. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036021-09718-20 Paul D., Rao A. R. et al. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036-021-09750-2 Rao A. R. 2018, J. Astrophys. Astr., 39, 2 Rao A. R., Chand V., Hingar M. K. et al. 2016, ApJ, 833, 86 Segreto A., Cusumano G., Ferrigno C. et al. 2010, A&A, 510, A47 Segreto A., Labanti C., Bazzano A. et al. 2003, A&A, 411, L215 Singh K. P., Stewart G. C., Chandra S. et al. 2016, in Proc. SPIE, Vol. 9905, Space Telescopes and Instrumentation 2016: Ultraviolet to Gamma Ray, 99051E Vadawale S. V., Chattopadhyay, T., Mithun, N. P. S. et al. 2018, Nature Astron., 2, 50

 Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:31 https://doi.org/10.1007/s12036-021-09690-x

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

Dorado and its member galaxies II: A UVIT picture of the NGC 1533 substructure R. RAMPAZZO1,2,* , P. MAZZEI2, A. MARINO2, L. BIANCHI3, S. CIROI4,

E. V. HELD2, E. IODICE5, J. POSTMA6, E. RYAN-WEBER7, M. SPAVONE5 and M. USLENGHI8 1

INAF Osservatorio Astrofisico di Asiago, Via Osservatorio 8, 36012 Asiago, Italy. INAF Osservatorio Astronomico di Padova, Vicolo dell’Osservatorio 5, 35122 Padua, Italy. 3 Department of Physics and Astronomy, The Johns Hopkins University, 3400 N. Charles St., Baltimore, MD 21218, USA. 4 Department of Physics and Astronomy, University of Padova, Vicolo dell’Osservatorio 3, 35122 Padua, Italy. 5 INAF-Osservatorio Astronomico di Capodimonte, Salita Moiariello 16, 80131 Naples, Italy. 6 University of Calgary, 2500 University Drive NW, Calgary, Alberta, Canada. 7 Centre for Astrophysics and Supercomputing, Swinburne University of Technology, Hawthorn, VIC 3122, Australia. 8 INAF-IASF, Via A. Curti, 12, 20133 Milan, Italy. *Corresponding Author. E-mail: [email protected] 2

MS received 30 October 2020; accepted 17 December 2020 Abstract. Dorado is a nearby (17.69 Mpc) strongly evolving galaxy group in the Southern Hemisphere. We are investigating the star formation in this group. This paper provides a FUV imaging of NGC 1533, IC 2038 and IC 2039, which form a substructure, south west of the Dorado group barycentre. FUV CaF2-1 UVIT-Astrosat images enrich our knowledge of the system provided by GALEX. In conjunction with deep optical wide-field, narrow-band Ha and 21-cm radio images we search for signatures of the interaction mechanisms looking in the FUV morphologies and derive the star formation rate. The shape of the FUV luminosity profile suggests the presence of a disk in all three galaxies. FUV emission is detected out to the optical size for IC 2038, and in compact structures corresponding to Ha and H II bright features in NGC 1533. A faint FUV emission, without an optical counterpart, reminiscent of the H I structure that surrounds the outskirts of NGC 1533 and extends up to IC 2038/2039, is revealed above the local background noise. Keywords. Ultraviolet: galaxies—galaxies: elliptical and lenticular—cD—galaxies: spiral—galaxies: interaction—galaxies: evolution..

1. Introduction One of the breakthrough provided by the Galaxy Evolution Explorer (GALEX hereafter) (Martin et al. 2005; Morrissey et al. 2007) is the direct evidence of galaxy transformation in groups. UV-optical colour magnitude diagrams (CMDs hereafter) highlighted an This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

intermediate region, the green valley, populated by transforming galaxies (see e.g. Salim et al. 2007; Schawinski et al. 2007) located between the sequence of red galaxies, mostly evolved early-type (Es?S0s = ETGs hereafter), and the blue cloud composed of late-type (LTGs hereafter), star forming galaxies. Ranking groups according to their blue vs. red sequence and green valley galaxy populations, UVoptical CMDs contribute to correlate the group structure, kinematics and dynamics to its members

31

Page 2 of 13

evolutionary phase. Marino et al. (2010, Marino et al. 2013) investigated loose groups, rich of LTGs, analogs of our Local Group. This kind of groups lack a well defined red sequence. At odd, less dispersed groups with an increasing fraction of ETGs start to develop a galaxy population inhabiting both the red sequence and the green valley. Very rich groups, in an advanced stage of virialization, like NGC 5486 group, the third richest association in the nearby universe after Virgo and Fornax, show a well developed red sequence and a still rich green valley (Marino et al. 2016; Mazzei et al. 2014b, 2019 and references therein). The enrichment of the red sequence and of the green valley traces the transition from loose yet un-virialized groups to rich, more compact and virialized ones (see e.g. Rampazzo et al. 2018). Driven by the gravitational force, that collapses groups and makes galaxies to interact, mechanisms leading to a galaxy morphological metamorphosis can either quench or re-ignite the star formation (SF). Mechanisms involved are of several types and seem to depend on the environment density (see e.g. Boselli & Gavazzi 2006, 2014). Mergers can transform spirals into ellipticals and S0s (see e.g. Toomre & Toomre 1972; Barnes 2002; Mazzei et al. 2014a and references therein) and quench SF by ejecting the interstellar medium via supernovæ, AGN or shock-driven winds (see e.g. Di Matteo 2015 and references therein). Ram-pressure, that may strip the gas reservoirs is supposed to work mostly in rich environments (Boselli & Gavazzi 2014; Ramatsoku et al. 2019), but there are evidences (e.g. in H I ) that it also works for groups (Bureau & Carignan 2002; Kantharia et al. 2005). There are examples of mass-transfer between gas-rich and gas-poor companion galaxies, e.g. physical pairs composed of a LTG and an ETGs, that may re-fuel SF (see e.g. Domingue et al. 2003; Keel 2004; Chung et al. 2006; Plana et al. 2017a, b). In this paper, we focus on the nearby Dorado group in the Southern Hemisphere (RA ¼ 64:3718 [deg], Dec ¼ 55:7811 [deg]) as defined by Kourkchi and Tully (2017) (see also, Firth et al. 2006 and references therein). Throughout the paper, we adopt 17.69 Mpc as the distance of all Dorado candidate members. Dorado CMD is described in Fig. 1 of Cattapan et al. (2019). The red sequence of the group includes several ETGs. A set of intermediate luminosity member galaxies is still crossing the green valley. Basically only NGC 1566, a bright grand design spiral, is still located in the blue cloud. With respect to evolved groups (e.g. NGC 5486, see Marino et al. 2016) Dorado seems to be in an earlier and active

J. Astrophys. Astr. (2021)42:31

evolutionary phase. Several indications support this view like the group clumpy structure. A compact group, SGC 0414-5559 (Iovino 2002) is located at the barycentre of the group as defined by Kourkchi & Tully (2017). The compact group is dominated by two ETGs, NGC 1549 an E and NGC 1553 an S0, both showing a wide shell structure. The Dorado group members morphology shows indeed the nearly ubiquitous presence of interaction signatures such as shells, asymmetries, tails (see e.g. Malin & Carter 1983; Cattapan et al. 2019). Star-forming rings have been also revealed in several ETGs (Rampazzo et al. 2020). Dorado is an atomic gas rich group. Nearly half of the entire H I content of the Dorado group, 3:5  1010 M , is located in the spiral member NGC 1566, although H I has been also detected in several other members, independently from their morphological classification (Kilborn et al. 2005, 2009). Rampazzo et al. (2020), presented their Ha?[N II] study aiming at investigating the star formation rate (SFR hereafter) of the Dorado backbone galaxies. They found that SFR in LTGs is fading while in ETGs is not yet shut-down. Rampazzo et al. (2020) suggested that mechanisms such as gas stripping and gas accretion, through galaxy–galaxy interaction, seem relevant in the evolutionary phase of Dorado.

Figure 1. NGC 1533 sub-structure. VST deep image in the SDSS g-band of the IC 2038, IC 2039 (north west side) and NGC 1533 (south east side). The image size is 140  14:70 . Astrosat-UVIT observations (see Table 1), cover the entire sub-structure.

J. Astrophys. Astr. (2021)42:31

This work complements the Rampazzo et al. (2020) ˚ ) broad filter study, using FUV CaF2-1 (1300–1800 A (similar to GALEX FUV; see Tandon et al. 2017) observations obtained with Astrosat-UVIT of the Dorado members. The UVIT targets partly cover the galaxy set observed in Ha?[N II] by Rampazzo et al. (2020). In this paper, we investigate the sub-structure, SW of the Dorado barycentre, formed by NGC 1533, IC 2038 and IC 2039. Our goal is to analyze the FUV morphological structure of galaxies in this sub-structure and the relation between Ha regions and the FUV emission. Ha emission has been found not only in the Scd galaxy IC 2038 but also in the E-S0 galaxy NGC 1533 (Rampazzo et al. 2020). The paper plan is the following. In Section 2, we summarize our knowledge about the NGC 1533 substructure. UVIT observations and the reduction are presented in Section 3. In Section 4, the photometric analysis and the comparison with GALEX observations are presented. The results are given in Section 5 and discussion is presented in Section 6. Section 7 contains summary and conclusion.

2. NGC 1533 substructure and the Dorado group According to Kourkchi and Tully (2017), Dorado has an average redshift of 1230  89 km s1 and a velocity dispersion of 242 km s1 . NGC 1533 and the pair IC 2038/39 form a sub-structure of the group with an average redshift of 764  43 km s1 (see Table 1). Figure 1 shows a deep image of the NGC 1533 substructure in the SDSS g-band (lg  30 mag arcsec2 ) obtained within the VST Early-type GAlaxy Survey (VEGAS1) (see e.g. Capaccioli et al. 2015). This substructure is located at the south west periphery of Dorado group barycentre formed by the SCG 0414-5559 compact group, well separated, both in radial velocity and in projection, from other member candidates. NGC 1533 has been investigated with GALEX by Marino et al. (2011a, b). Rampazzo et al. (2017) investigated this galaxy with Swift. All these studies evidenced an outer FUV bright ring. In correspondence to the FUV ring, Rampazzo et al. (2020) found some Ha complexes. Moreover they found that while the SFR is enhanced in NGC 1533 it is depressed in IC 2038, if compared with the general sample of ETGs (Gavazzi et al. 2018) and of Spirals (James et al. 2004), respectively. Rampazzo et al. (2020) found 1

Visit the website http://www.na.astro.it/vegas/VEGAS/Welcome. html.

Page 3 of 13

31

that in IC 2038 H II regions are distributed along the galaxy body, which appears slightly elongated towards the companion galaxy, IC 2039, in the SE direction. More recently, deep optical observations, in g and r bands, of NGC 1533 and IC 2038/39 have been analysed by Cattapan et al. (2019) using the VLT Survey Telescope (VST) at the European Southern Observatory, Chile. They evidenced a large disk around NGC 1533 and several tails suggesting that NGC 1533 and the nearby pair are evolving together (Cattapan et al. 2019). The NGC 1533 Dorado substructure has been mapped in H I by Ryan-Weber et al. (2004) (see also Kilborn et al. 2005, 2009) showing and extended H I structure that extends from 1533 up to IC 2038/39 pair. Werk et al. (2008, 2010) reported that there are some star forming regions well outside NGC 1533 with the same radial velocity as the H I gas revealed by Ryan-Weber et al. (2004; see their Fig. 7 for J0409-56). The narrow-band Ha?[N II] study by Rampazzo et al. (2020) detected three Ha regions in correspondence of the regions detected by Werk et al. (2010) confirming that such outer H II regions belong to NGC 1533 sub-structure. In this context, the study of the NGC 1533 substructure is relevant both for the general understanding of the evolution of gas-rich merging events and for the study of local SF mechanisms.

3. Observations and reduction Astrosat is a multi-wavelength satellite that has been launched by the Indian Space Research Organization on September 28, 2015. The ultraviolet-optical telescope on board is the Ultra-Violet Imaging Telescope facility UVIT (Tandon et al. 2017). It is composed of two Ritchey-Chretien telescopes with 37.5-cm aperture, a circular field-of-view of 280 diameter, observ˚ ) and ing simultaneously one in FUV (1300–1800 A ˚ ) and optical the other both in NUV (2000–3000 A ˚ band, VIS (3200–5500 A), by means of a beamsplitter directing NUV and VIS to individual cameras. Since the NUV channel was not operative during our runs, observations have been performed with the FUV channel only. We used the full field-of-view, in photon counting mode with the Filter F148W CaF2 ˚ Photons are counted on (kmean ¼ 1481, Dk ¼ 500 A). a planar CMOS array at approximately 28 Hz and stacked on the ground with shift and add algorithms (for details, see Kumar et al. 2012; Postma et al.

31

Page 4 of 13

J. Astrophys. Astr. (2021)42:31

Table 1. Relevant properties of members of Dorado in the UVIT FoV from the literature. Field centre A B

ID source

RA (J2000)

Dec. (J2000)

FUV (mag)

Vhel (km s1 )

Morpho. type

IC 2038 IC 2039 NGC 1533

04 08 53.76 04 09 02.37 04 09 51.84

-55 59 22.2 -56 00 42.1 -56 07 06.6

17:25  0:02 19:98  0:07 16:90  0:02y

712 817 764

7.0 -3.1 -2.5

Column 1: UVIT field. Column 2: Source identification. Columns 3 and 4 provide right ascension and declination of the sources. Column 5 lists the FUV total integrated magnitudes from GALEX archive (indicated with an *) and reported in NED (indicated with y ). Columns 6 and 7 report the heliocentric radial velocity and the galaxy morphological type, respectively, from Kourkchi and Tully (2017).

2011; Postma & Leahy 2017; Tandon et al. 2017) with the astrometric world coordinate solution solved automatically by a trigonometric algorithm (Postma & Leahy 2020). Table 2 reports the relevant characteristics of our UVIT observations. Astrosat-UVIT observations are the result of the proposal A07_010 (PI R. Rampazzo) and cover the south west part of Dorado, in particular Astrosat-UVIT fields contain IC 2038 and IC 2039 (Field A) and NGC 1533 (Field B).

4. Data analysis and comparison with the literature In this section, we present the data analysis performed and the comparison with the current literature mainly from GALEX study. Original images with 0:00 416 per sub-pixel have been rebinned 4  4 providing a final scale of 1:00 664 px1 .

4.1 Integrated FUV magnitudes Column 3 of Table 3 provides the FUCaF2-1, UVIT integrated magnitudes. In Fig. 2, we compared our results with integrated magnitudes reported in Table 1 from GALEX observations. The figure shows the good agreement between our values and those in the current literature.

4.2 Surface photometry The surface photometry is obtained using the IRAF task ELLIPSE (Jedrzejewski 1987). In obtaining the surface brightness profile, ELLIPSE accounts for the geometrical information contained in the isophotes and allows the variation of the ellipticity,  ¼ ð1 

b=aÞ and of the position angle, (PA), along the ellipse major axis (a). In addition, ELLIPSE provides measure of the isophotal shape parameter, the socalled a4 parameter from the fourth cosine component of the Fourier analysis of the fitted ellipse, allowing to distinguish between boxy (a4 \0) and discy (a4 [ 0) isophotes (Bender et al. 1988). This approach, widely used with optical images of ETGs, has been adopted to investigate GALEX FUV data by Jeong et al. (2009) and Marino et al. (2011b). FUV geometric profiles have not been provided by the above papers. In fitting isophotes we allowed , and PA to variate with the galacto-centric distance, in order to obtain a good description of the FUV luminosity distribution. Our FUV images either show irregular peculiar features (NGC 1533), clumpy and spiral features (IC 2038) or have a low signal-to-noise (IC 2039). This is the reason for which we provide in Table 3 only the average ellipticity, hi and position angle, hPAi, (columns 4 and 5 respectively) of the galaxy. In presence of irregular, peculiar features, sudden variations in both  and PA are expected and, in particular, the isophotal shape parameter, a4 , loses its physical meaning. The pair IC 2038/IC 2039 has not been previously investigated with GALEX. Marino et al. (2011b) obtained the surface brightness profiles NGC 1533. Figure 3 shows the comparison between the Marino et al. (2011b) and our luminosity profile. The luminosity profiles compare quite well outside 3000 , suggesting that there is a very small, if any, zero point effect, while the central region differs significantly. The reason of such a difference in the central part is not entirely clear, although in FUV the GALEX PSF FWHM is 4.200 with respect to 1.500 of UVIT (see Morrissey et al. 2007; Tandon et al. 2017, 2020). Substructures on spatial scales (projected on the sky) .0.4 kpc are washed out in the GALEX profile, but are picked up in the UVIT data. In particular, the FUV

J. Astrophys. Astr. (2021)42:31

Page 5 of 13

31

Table 2. UVIT observations. Field ID A B

Obs ID

Observing date (s)

Exp. time

Target ID

A07_010T04_9000003220 A07_010T05_9000003236

October 5, 2019 October 16, 2019

6481.278 6628.734

IC 2038 NGC 1533

Field identification in column 2 refers to the proposal A07 010 (PI. R. Rampazzo). Column 3 and column 4 report the observing date and the total reduced exposure time. In column 5, the central target is quoted. The nominal zero point magnitude of the FUV channel and the physical plate scale are 18.08 mag and 3:00 33 respectively, as reported in Tandon et al. (2017). Table 3. Relevant FUV parameters of galaxies in the NGC 1553 sub-structure. Field

ID source

FUV (mag)

hi

hPAi ( )

n

LFUV (erg s1 Hz1 )

SFR (M yr1 )

A

IC 2038 IC 2039 NGC 1533

17:19  0:07 20:03  0:11 16:74  0:15

0:72  0:02 0:19  0:05 ...

154:4  4:6 104:1  8:7 ...

0:81  0:09 1:45  0:18 2:64  0:06

1:81  1027 1:33  1025 2:74  1026

0:025  0:002 0:002  0:0002 0:038  0:006

B

Column 1: UVIT field. Column 2: Source identification; column 3: FUV integrated magnitude corrected for galactic extinction accounting for AFUV ¼ 7:9  EðB  VÞ; EðB  VÞ is 0.01 mag both for IC 2038 and IC 2039, 0.015 mag for NGC 1533 from NED. Luminosities are computed accounting for the same distance, 17.69 Mpc, for all our targets. In columns 4 and 5, the average ellipticity and position angles are provided: for NGC 1533, see Section 5. In column 6, the Se´rsic index from the best-fit of the entire profile is reported. In columns 7 and 8, the FUV luminosity and the SFR computed according to the Lee et al. (2009) provided in Section 6.3, are reported.

knot that peaks at 2500 (indicated as A in Fig. 5) is clearly smoothed out in Marino et al. (2011b), suggesting a significant role played by the PSF in driving the GALEX surface photometry.

Figure 2. Comparison between UVIT FUV CaF2 -1 integrated magnitudes of Dorado members with those from GALEX available in the literature. Magnitudes in the plot are not corrected for galactic extinction.

Figures 4 and 5 show the FUV image and the luminosity profile of the pair members IC 2038/39 and of NGC 1533, respectively. The bottom panel of these figures shows a color composite RGB image of the galaxies obtained from the present FUV image, the continuum and Ha?[N II] images from Rampazzo et al. (2020).

Figure 3. Comparison of the present FUV luminosity profile of NGC 1533 with Marino et al. (2011b) from GALEX pointed observations.

31

Page 6 of 13

J. Astrophys. Astr. (2021)42:31

Figure 4. (Top: left panel) FUV image of IC 2038 (Sbc) and of IC 2039 (E), its physical companion, in the south east. The image size is 40  40 . North on the top, east to the left. (Top: mid and right panels) FUV surface brightness profile of IC 2038 and of IC 2039 (blue dots). Single Se´rsic law fit of the luminosity profiles, discussed in Section 5 are also shown as cyan dots. The values of the Se´rsic index are n ¼ 0:81  0:09 and n ¼ 1:45  0:18, for IC 2038 and IC 2039, respectively. (Bottom panel) Color-composite RGB image using the red and the green channels for Ha and the nearby continuum (Rampazzo et al. 2020) and the blue channel for the FUV image. Both the Ha and the continuum PSF have been re-scaled to the PSF of the FUV image.

4.3 Luminosity profile fitting We fit the shape of the FUV luminosity profile with a Se´rsic law (Se´rsic 1963). Even a crude representation of the light profile, as a simple Se´rsic law fit, provides useful information, and sometimes is the only decomposition that can be compared with the literature. The Se´rsic law l / r1=n , where l is the surface brightness and r the radius, is a generalization of the de Vaucouleurs (1953) r1=4 and of the Freeman (1970) exponential laws. The variation of the Se´rsic index, n, describes the manifold of the shapes of luminosity profiles of ETGs. The watershed can be considered the value n ¼ 4 representing a

‘classic’ elliptical. A classic exponential disc (Freeman 1970) in S0s has an index n ¼ 1. FUV luminosity profiles of ETGs may reach large values of n (see e.g. Marino et al. 2011b). Rampazzo et al. (2017) suggested that in luminosity profiles with n\3 the presence of a disc starts to emerge. The Se´rsic law fit is shown in the right panel/s of Figs. 4 and 5 for IC 2038/IC 2039 and NGC 1533, respectively, superposed to the luminosity profile. The fit is extended to the entire profile without masking FUV bright sub-structures such as the ring (B) and the knot (A) in NGC 1533. The Se´rsic fit accounts for the UVIT-PSF so it is not necessary to avoid the inner part of the luminosity profile certainly ‘contaminated’ by the instrument PSF.

J. Astrophys. Astr. (2021)42:31

Page 7 of 13

31

Figure 5. (Top: left panel) FUV image of NGC 1533. The image size is 70 70 . North on the top, east to the left. (Top: right panel) Surface brightness profile of NGC 1533 (blue dots) and single Se´rsic law fit (cyan dots). The value of the Se´rsic index is n ¼ 2:64  0:06. (Bottom panel) As in Fig. 4, for NGC 1533. Labels A and B indicate H II regions in Rampazzo et al. (2020).

5. Results In the following sections, we discuss the shape of the surface brightness profile and the morphology of the FUV emission.

5.1 Members surface photometry IC 2038. In the top panels of Fig. 4, the FUV image and the luminosity profiles of IC 2038 and IC 2039 are shown.

IC 2038, classified as Sbc in optical by HyperLeda, does not show a bulge (see e.g. Rampazzo et al. 2020, their Fig. 3). The FUV emission appears clumpy in the centre of the galaxy, where H II regions are detected by Rampazzo et al. (2020), while an arm-like structure appears out the both sides of the galaxy body. The FUV surface brightness profile is flat in the central 2000 . The best-fit to the entire profile has a Se´rsic index of n ¼ 0:81  0:09 suggesting the galaxy is dominated by the disk. In the central 2500 there is an excess of luminosity with respect to the Se´rsic law best fit, in correspondence to FUV clumps.

31

Page 8 of 13

The FUV emission extends out to the optical size of the galaxy as shown by Fig. 4 and Fig. 1: there is no evidence of a XUV disk (Thilker 2008). The average ellipticity, hi ¼ 0:72  0:02 and the average position angle hPAi ¼ 154:4  4:6 , compare well with values from optical bands, 0:74  0:04 and 155.2 respectively, from the HyperLeda catalog. IC 2039. In FUV, IC 2039 is the faintest galaxy of the sub-structure (Fig. 2). We best-fit luminosity profile of IC 2039 (top right panel of Fig. 4) with a single Se´rsic law with an index n ¼ 1:45  0:18 indicating that it is basically composed of a disc. The average ellipticity is hi ¼ 0:19  0:05 and the position angle hPAi ¼ 104:1  8:7 to compare with optical values of 0:15  0:07 and 124:3 from HyperLeda. NGC 1533. The galaxy, classified as E-S0 (HyperLeda) is known to have an outer ring and a inner bar, so Comero´n et al. (2014) classified it as (RL)SB00 . However, in FUV this galaxy shows no signature of the bar. The presence of a FUV-bright outer ring was evidenced using GALEX by Marino et al. (2011b) both in NUV and FUV. UVIT clearly reveals a bright FUV spot at 2500 and the ring in the range 3900 –8100 , in agreement with Marino et al. (2011a) (their Table 1). Their presence causes a sudden jump of both  and PA. ELLIPSE provides  ¼ 0:56  0:04 and PA ¼ 63 within 3900 and   0.17  0:06 and PA¼164 outside the region of the ring up galaxy outskirts. The value of the ellipticity and PA provided by HyperLeda from optical bands are 0.36 ± 0.09 and 141.5 . The surface brightness profile (top right panel of Fig. 5) is best-fitted by a Se´rsic law with index n ¼ 2:64  0:06 suggesting the presence of a disc. This value agrees with the UV Se´rsic index of n ¼ 2:76  0:10 obtained from the

J. Astrophys. Astr. (2021)42:31 ˚ by luminosity profile in the Swift W2 filter (kc ¼ 2030 A) Rampazzo et al. (2017).

5.2 FUV regions outside galaxies main body To enhance the signal-to-noide ratio (S/N) in the galaxy outskirts and bring out any possible faint structures in the UV emission, we adopted the procedure outlined by Ebeling et al. (2006), called ASMOOTH. The only parameter required by by the procedure is the desired minimal S/N, smin . For each individual pixel, the algorithm increases the smoothing scale until the S/N within the kernel reaches a specified input value. ASMOOTH suppresses very efficiently the noise while the signal, locally significant at the selected S/N level, is preserved on all scales. In the right and left panels of Fig. 6, the FUV image has been treated with ASMOOTH selecting a S/N above the back-ground of smin ¼ 1:5 and smin ¼ 2:0, respectively. The possible physical causes of the features that emerged using this procedure are discussed below. Cattapan et al. (2019) (see their Fig. 7) superposed the emission of NGC 1533 in the W2 filter derived by Rampazzo et al. (2017) from Swift-UVOT observations to their wide field, deep g-band image obtained at VST (see Fig. 1). They found that the FUV ring of NGC 1533 is superposed to spiral-like residuals obtained after a model of the optical luminosity profile has been subtracted. The right panel of Fig. 6, the FUV image treated with ASMOOTH shows an arm-

Figure 6. (Left panel) The H I contour levels from Ryan-Weber et al. (2004) at column densities 2:5  1020 , 2:8  1020 , 3:1  1020 , 3:5  1020 , 3:9  1020 and 4:2  1020 atoms cm2 are superposed to the UVIT FUV image of the sub-structure after an adaptive smoothing (ASMOOTH) with a minimum S/N smin ¼ 1:5 (Ebeling et al. 2006) has been applied. The image size corresponds to Fig. 1. (Right panel) Zoom-in on NGC 1533 (field-of-view 70 70 ) showing the complex structure of the ring and an arm/like structure on the west side (ASMOOTH with a minimum S/N smin ¼ 2).

J. Astrophys. Astr. (2021)42:31

Figure 7. The south-east region of NGC 1533 observed in FUV by UVIT. Three boxes of 6000  6000 in the FUV image enclose the regions identified as J0409-56 E1 ? E2, E3, E4 by Werk et al. (2010) on GALEX in which they found Ha sources. The two areas, showed as sub-panels on the right of the FUV image, mark the position of H II regions, at the same redshift of the galaxy, detected by Rampazzo et al. (2020) via Ha ? [N II] observations. The centres of the regions in Rampazzo et al. (2020) correspond to the following coordinates 04 10 13.5–56 11 36 (J2000) and 04 10 10.66–56 07 27.99 (J2000) and overlap with region E1 ? E2 and E4 in Werk et al. (2010), respectively.

like structure is emerging also in FUV outside the ring in the west side. In the left panel of Fig. 6, we overplot the H I emission isophotes from Ryan-Weber et al. (2004). While the center of NGC 1533, in particular the ring, is devoid of H I the large scale structure of H I seems to corresponds to the faint extended FUV emission derived using ASMOOTH. The ATCA H I observations by Ryan-Weber et al. (2004) extended over an area at least 50 to west (partly covered by Fig. 6). They do not detect any H I to the west within their detection limit. Some FUV emission regions above smin ¼ 1:5 are still visible in that area. The presence of this FUV emission would be consistent with the N-body SPH numerical simulations by Ryan-Weber et al. (2003) describing the H I ring around NGC 1533 as the remnant of a tidally destroyed galaxy (see Section 6.2). Figure 7 shows the outer FUV emission regions of NGC 1533, investigated by Werk et al. (2008, 2010) in the first systematic search for outlying H II regions, as part of a sample of 96 emission-line point sources

Page 9 of 13

31

(referred to as ELdots-emission-line dots) derived from the NOAO Survey for Ionization in Neutral Gas Galaxies (SINGG). Such regions, selected from GALEX FUV, are indicated in the above paper as J0409-56 E1, E2, E3 and E4. Rampazzo et al. (2020) inspected also the corresponding areas looking for H II regions in their frames. Those found are shown right panels of Fig. 7 and indicated as E1?E2 and E4. Rampazzo et al. did not detect H II regions in the zone indicated as E3. Indeed, Werk et al. (2010) found that their targets span over a large range in Ha luminosities which correspond to a few O stars in most of the nearby cases. Werk et al. (2010) emphasized that often FUV sources are mixed to unresolved dwarf satellite companions and background galaxies. Outer H II regions of NGC 1533 may be linked to the strong interacting phase suggested by Cattapan et al. (2019) and well imaged in their Fig. 6. H II regions detected and their embedded young stars are definitely correlated with H I as their velocities are the same (see Section 6.2).

6. Discussion The NGC 1533 sub-structure in the Dorado group has been recently studied by Cattapan et al. (2019) in g and r SDSS-bands and by Rampazzo et al. (2020) in Ha?[N II]. This substructure with a heliocentric velocity Vhel  764 km s1 is still evolving separately from the Dorado core, the compact group SCG 04145559 (Iovino 2002), with Vhel  1230 km s1 . Both IC 2038/IC 2039 and NGC 1533 show several interaction signatures as described in Cattapan et al. (2019) (see their Fig. 6). Their environment is H I rich (Ryan-Weber et al. 2003; Kilborn et al. 2009). The above studies suggest a common evolutionary picture of the galaxies members of the NGC 1533 substructure.

6.1 FUV and evolution of member galaxies Mazzei et al. (2014a, 2019) and references therein, using smoothed particle hydrodynamic simulations with chemo-photometric implementation (SPH-CPI), investigated the evolutionary path of NGC 1533, the dominant member of the sub-structure. From a large grid of simulations of galaxy encounters and mergers, starting from triaxial halos of gas and dark matter, a simulation matching the global

31

Page 10 of 13

properties of this SB0 galaxy (absolute B mag, SED, luminosity profile, morphology at different wavelengths and kinematics properties) has been single out. According to this study NGC 1533 is the result of a major merging occurred at z ¼ 2:3. 40% of its current mass is assembled before z ¼ 1. The FUV bright ring has been one of the morphological features used to single out the simulation from the SPH-CPI grid. The ring is, indeed, well reproduced by the selected simulation as the result of a resonance that appeared when the galaxy is 8 Gyr old and is maintained up to now. The simulation showed the path of NGC 1533 in the (NUV-r) vs. Mr colormagnitude diagram. NGC 1533 is 13.7 Gyr old, that it spends as follows. It lies on the blue cloud for 7.2 Gyr, from there it takes about 0.9 Gyr to reach the green valley that will cross reaching the red sequence in 1.6 Gyr, finally it gets its current position on the red sequence after additional 4 Gyr. The evolution of the pair IC 2038/IC 2039 has not been studied yet. The pair badly lack a detailed kinematic study in order to constrain the SPH-CPI grid of simulations. As relevant examples in this context, we mention the SPH-CPI studies of the pair NGC 454 by and of the false pair NGC 3447/NGC 3447A by Plana et al. (2017a, b), Mazzei et al. (2018). 6.2 FUV and SF regions The FUV emission is a short scale (107 years) SF indicator. Therefore it can be associated to a measure of the SF obtained from Ha (106 years) (see Kennicutt & Evans 2009). Next sections will focus on SF regions and SFR estimates from Ha and FUV. The FUV emission of IC 2038 is more extended than the area where H II regions are found. Indeed, the FUV emission covers the entire galaxy seen in the continuum (optical) image (bottom panel of Fig. 4). The FUV emission is also present in the inner regions of IC 2039 (Fig. 4). However, no H II regions have been detected in this E-S0 by Rampazzo et al. (2020). At odds, in NGC 1533 H II regions are found in two small complexes by Rampazzo et al. (2020) and labelled as A and B in Fig. 5. In this galaxy the FUV emission concentrates in the ring and a bright FUV spot, that includes the above H II regions. We revealed in this paper a faint FUV emission (Fig. 6) associated to a complex H I structures composed of two major arcs one north west and a second south east (Ryan-Weber et al. 2004). No optical counterparts are connected with these arcs. Faint H II

J. Astrophys. Astr. (2021)42:31

Figure 8. Comparison between the SFR computed from Ha and FUV emission without accounting for internal extinction. The blue solid line show the relation found by Lee et al. 2009 (see text).

regions are found south east of NGC 1533 (Werk et al. 2008, 2010; Rampazzo et al. 2020) and are correlated to the H I envelope. In particular regions indicated in Fig. 7 as E1–E2 have a recession velocity of 831 and 846 km s1 (another region indicated E5 by Ryan-Weber et al. 2004, see their Fig. 2) has a recession velocity of 901 km s1 ). With these recession velocities H II regions are compatible with being associated to NGC 1553 sub-structure (see Table 1) and with the H I structure as well. Concerning the origin of the H I clouds around NGC 1533, Ryan-Weber et al. (2003) suggested that it could be the merger remnant of a tidally destroyed galaxy. Ryan-Weber et al. (2004) noticed that the H I gas in the south east cloud has velocity dispersion up to 30 km s1 and velocity gradient in the range 7–50 km s1 kpc1 . These conditions makes this site unlikely for SF since the latter usually requires the gas to have a low velocity dispersion in order to collapse. We conclude that the H I arcs detected by RyanWeber et al. (2004), the correspondent faint FUV structure revealed in this paper in addition to the optical evidence of faint tails and arcs shown by the (Cattapan et al. 2019) deep surface photometry are indication that galaxy–galaxy encounters, leading to galaxy tidal disruption (Ryan-Weber et al. 2003) and merging events (Mazzei et al. 2019) are the drivers of

J. Astrophys. Astr. (2021)42:31

Page 11 of 13

the complex evolution within the NGC 1533 substructure. 6.3 SFR from FUV integrated galaxy luminosity In this section, we compare the SFR for our three targets as derived by Ha emission (Rampazzo et al. 2020) with that by the integrated FUV emission in this paper. We follow the recipes of Lee et al. (2009, their equation (3)) to compute the SFR for FUV: SFR ðM yr1 Þ ¼ 1:4  1028 LFUV ðergs1 Hz1 Þ: ð1Þ Lee et al. (2009) found the following relation when the effects of internal dust attenuation are not included: logðSFRðFUVÞÞ ¼ 0:79 logðSFRðHaÞÞ  0:20:

ð2Þ

This is highlighted by a solid blue line in Fig. 8. Lee et al. (2009) noticed that Ha and FUV SFRs agree to within factor of 2 for all galaxies with SFR 0:01M yr1 . The SFR of IC 2038, the only LTG in the substructure, does not deviate from the Lee et al. (2009) relation. However, according to Rampazzo et al. (2020) the SFR of IC 2038 estimated from Ha is below the average for its morphological class estimated by James et al. (2004). The SFR of the two ETGs, IC 2039 and NGC 1533 seems to deviate from the Lee et al. (2009) relation, suggesting a higher SFR from Ha than from UV. This point is quite surprising given that UV stars trace SFR with a time scale longer than Ha and the activity of SF is residual for ETGs. Figure 2 and Table 2 in Lee et al. (2009) showed the trend of the ratio log[SFR(Ha)/ SFR(FUV)] as a function of the B-band galaxy absolute magnitude. At the distance of Dorado, the absolute B magnitude of NGC 1533 and IC 2039 is MB ¼ 19:52 and 16:3, respectively. At these magnitudes, Lee et al. (2009) reported the average values of log½SFRðHaÞ=SFRðFUVÞ  0:10  0:36 (1r) and log½SFRðHaÞ=SFRðFUVÞ  0:12  0:18. Our measured values are log½SFRðHaÞ=SFRðFUVÞ ¼ 1:02 and log½SFRðHaÞ=SFRðFUVÞ ¼ 0:6 for NGC 1533 and IC 2039, respectively. However, the bottom panel of Fig. 2 in Lee et al. (2009) showed discrepant cases of galaxies, with values of log½SFRðHaÞ=SFRðFUVÞ similar to ours. Investigating

31

the SF properties in the Local Volume of Galaxies with Ha and FUV fluxes, Karachentsev and Kaisina (2013) provided a log½SFRðHaÞ=SFRðFUVÞ ¼ 1 for NGC 1533, in well agreement with our value. We plan to further explore the comparison of SFR from Ha and FUV using the entire UVIT data-set of the Dorado backbone, considering extinction effects in more detail than in Lee et al. (2009) assumptions (Rampazzo et al. in preparation). The SPH chemophotometric simulation of NGC 1533 by Mazzei et al. (2019) provided an estimate of the internal galaxy extinction. We will use also such estimates.

7. Summary and conclusion We performed with UVIT a FUV photometric study of a substructure of the Dorado group of galaxies that includes three galaxies: the mixed morphology pair IC 2038 (Sbc)/2039 (E-S0) and NGC 1533. We derived their luminosity profiles and discussed their FUV morphologies. We found the following results: • The shape of the FUV luminosity profile indicates the presence of a disc in all three galaxies. The presence of disc suggests that dissipative mechanisms have been at work. • In IC 2038, the FUV emission is detected out to the optical size for IC 2038, further out than the H II regions system detected by Rampazzo et al. (2020). There is no evidence of a XUV disk (Thilker 2008). • In IC 2039, the FUV emission is detected in the inner regions where no H II regions have been detected by Rampazzo et al. (2020). • In NGC 1533, the FUV emission is more extended than the system of H II regions detected by Rampazzo et al. (2020). The extended FUV emitting regions likely correspond to outer arm-like structures detected at different wavelengths (Marino et al. 2011a; Rampazzo et al. 2017; Cattapan et al. 2019). • We reveal a faint FUV emission, just above the local background noise, reminiscent of the wide H I structure detected by Ryan-Weber et al. (2004). In the east and south east regions of this FUV emission lurk few H II regions highlighted by Werk et al. (2008, 2010) and Rampazzo et al. (2020) with the same redshift as the H I structure (Ryan-Weber et al. 2004) and of the NGC 1533 sub-structure as well.

31

Page 12 of 13

• We derive the SFR from the FUV luminosity and we compare the results with SFR, following the Lee et al. (2009) recipe, with SFR from Ha by Rampazzo et al. (2020). The SFR for IC 2038 only agrees with Lee et al. (2009) expected ratio between FUV and Ha derived values. At odds, our measure of the SFR Ha/ FUV for NGC 1533 well agrees with that found by Karachentsev and Kaisina (2013). Lee et al. (2009) SFR(FUV)–SFR (Ha) relation does not account for internal dust effects. Such relation will be investigated further in the analysis of our UVIT FUV observations of the whole galaxy group. In a forthcoming paper, we will analyze the central part of Dorado, the compact group SGC 0414-5559 Iovino (2002), composed of NGC 1553, NGC 1549, NGC 1546 and IC 2058 plus the dwarf galaxy PGC 75125. The study of the SBc galaxy NGC 1536 and of the mixed pair NGC 1596/NGC 1602 (S0?Irr), part of the A07 program but non yet observed, will complete our study of the Dorado backbone with UVIT. Acknowledgements This publication uses the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). This publication uses UVIT data processed by the payload operations centre at IIA. The UVIT is built in collaboration between IIA, IUCAA, TIFR, ISRO and CSA. References Barnes J. E. 2002, MNRAS, 333, 481 Bender R., Do¨bereiner S., Mo¨llenhoff C. 1988, A&AS, 74, 385 Boselli A., Gavazzi G. 2006, PASP, 118, 517 Boselli A., Gavazzi G. 2014, A&A Review, 22, 74 Bureau M., Carignan C. 2002, AJ, 123, 1316 Capaccioli M., Spavone M., Grado A. et al. 2015, A&A, 581, A10 Cattapan A., Spavone M., Iodice E. et al. 2019, ApJ, 874, 130 Chung A., Koribalski B., Bureau M., van Gorkom J. H. 2006, MNRAS, 370, 1565 Comero´n S., Salo H., Laurikainen E. et al. 2014, A&A, 562, A121 de Vaucouleurs G. 1953, MNRAS, 113, 134. Di Matteo T. 2015, in: IAU General Assembly, vol. 29, 2257908

J. Astrophys. Astr. (2021)42:31 Domingue D. L., Sulentic J. W., Xu C. et al. 2003, AJ, 125, 555 Ebeling H., White D. A., Rangarajan F. V. N. 2006, MNRAS, 368, 65 Firth P., Evstigneeva E. A., Jones J. B. et al. 2006, MNRAS, 372, 1865 Freeman K. C. 1970, ApJ, 160, 811 Gavazzi G., Consolandi G., Pedraglio S. et al. 2018, A&A, 611, A28 Iovino A. 2002, AJ, 124, 2471 Jedrzejewski R. I. 1987, MNRAS, 226, 747 James P. A., Shane N. S., Beckman J. E. et al. 2004, A&A, 414, 23 Jeong H., Yi S. K., Bureau M. et al. 2009, MNRAS, 398, 2028 Karachentsev I. D., Kaisina E. I. 2013, AJ, 146, 46. Kantharia N. G., Ananthakrishnan S., Nityananda R., Hota A. 2005, A&A, 435, 483 Keel W. C. 2004, AJ, 127, 1325 Kennicutt R. C., Evans N. J. 2009, Ann. Rev. A&A, 50, 531 Kilborn V. A., Koribalski B. S., Forbes D. A. et al. 2005, MNRAS, 356, 77 Kilborn V. A., Forbes D. A., Barnes D. G. et al. 2009, MNRAS, 400, 1962 Kourkchi E., Tully R. B. 2017, ApJ, 843, 16 Kumar A., Ghosh S. K., Hutchings J. et al. 2012, Proc. SPIE, 8443, 84431N Lee J. C., Gil de Paz A., Tremonti C. et al. 2009, ApJ, 706, 599 Malin D. F., Carter D. 1983, ApJ, 274, 534 Marino A., Bianchi L., Rampazzo R. et al. 2010, A&A, 511, A29 Marino A., Bianchi L., Rampazzo R. et al. 2011a, ApJ, 736, 154 Marino A., Rampazzo R., Bianchi L. et al. 2011b, MNRAS, 411, 311 Marino A., Plana H., Rampazzo R. et al. 2013, MNRAS, 428, 476 Marino A., Mazzei P., Rampazzo R., Bianchi L. 2016, MNRAS, 459, 2212 Martin D. C., Fanson J., Schiminovich D. et al. 2005, ApJ, 619, L1 Mazzei P., Marino A., Rampazzo R., Galletta G., Bettoni D. 2014a, Adv. Space Res., 53, 950 Mazzei P., Marino A., Rampazzo R. 2014b, ApJ, 782, 53 Mazzei P., Marino A., Rampazzo R. et al. 2018, A&A, 610, A8 Mazzei P., Rampazzo R., Marino A., Trinchieri G., Uslenghi M., Wolter A. 2019, ApJ, 885, 165 Morrissey P., Conrow T., Barlow T. A. et al. 2007, ApJS, 173, 682 Plana H., Rampazzo R., Mazzei P. 2017, MNRAS, 472, 3074. Plana H, Rampazzo R., Mazzei P. et al. 2017, MNRAS, 472, 3074 Postma J., Hutchings J. B., Leahy D. 2011, PASP, 123, 833

J. Astrophys. Astr. (2021)42:31 Postma J., Leahy D. 2017, PASP, 129, 981 Postma J., Leahy D. 2020, PASP, 132, 1011 Ramatsoku M., Serra P., Poggianti B. M. et al. 2019, MNRAS, 487, 4580 Rampazzo R., Mazzei P., Marino A. et al. 2017 A&A, 602, A97 Rampazzo R., Mazzei P., Marino A. et al. 2018, Ap&SS, 363, 80 Rampazzo R., Ciroi S., Mazzei P. et al. 2020, A&A, 643, A176 Ryan-Weber E. V., Meurer G. R., Freeman K. C. et al. 2004, AJ, 127, 143 Ryan-Weber E., Webster R., Bekki K. 2003, The IGM/ Galaxy Connection: The Distribution of Baryons at z ¼ 0, ASSL Conference Proceedings, vol. 281, edited by Jessica L. Rosenberg and Mary E. Putman, ISBN: 1-4020-1289-6, Kluwer Academic Publishers, Dordrecht, p. 223

Page 13 of 13

31

Salim S., Rich R. M., Charlot S. et al. 2007, ApJS, 173, 267 Schawinski K., Kaviraj S., Khochfar S. et al. 2007, The Astrophys. J. Suppl. Series, 173, 512 Se´rsic J. L. 1963, Jan, Boletin de la Asociacion Argentina de Astronomia La Plata Argentina, 6, 41 Tandon S. N., Subramaniam A., Girish V. et al. 2017, AJ, 154, 128 Tandon S. N., Postma J., Joseph P. et al. 2020, AJ, 159, 158 Thilker D. 2008, Astrophys. Space Sci. Proc., Galaxies in the Local Volume, edited by Koribalski B. S., Jerjen H., 109 Toomre A., Toomre J. 1972, ApJ, 178, 623 Werk J. K., Putman M. E., Meurer G. R. et al. 2008,ApJ, 678, 888 Werk J. K., Putman M. E., Meurer G. R. et al. 2010, AJ, 139, 279

 Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:34 https://doi.org/10.1007/s12036-021-09708-4

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

Imaging and photometric studies of NGC 1316 (Fornax A) using Astrosat/UVIT NILKANTH D. VAGSHETTE1,*, SACHINDRA NAIK2, NEERAJ KUMARI2

and MADHAV K. PATIL3 1

Department of Physics and Electronics, Maharashtra Udayagiri Mahavidyalaya, Udgir 413 517, India. Astronomy and Astrophysics Division, Physical Research Laboratory, Navrangpura, Ahmedabad 380 009, India. 3 School of Physical Sciences, Swami Ramanand Teerth Marathwada University, Nanded 431 606, India. *Corresponding Author. E-mail: [email protected] 2

MS received 28 October 2020; accepted 13 January 2021 Abstract. We present imaging and photometric studies of the radio galaxy NGC 1316 (Fornax A) using high spatial resolution near-ultraviolet (NUV) and far-ultraviolet (FUV) imaging telescopes of the first Indian multi-wavelength space observatory AstroSat. The residual maps of UV emission obtained from the subtraction of smooth models witness peculiar features within the central few kpc (1–2 kpc) region. The spatial correspondence between the radio emission maps and FUV imaging study reveal that the UV emitting sources are displaced away from the centre by the AGN outburst (radio jet). The presence of rims and clumpy structures in the outskirt of this galaxy delineate that the galaxy has acquired a large fraction of gas through merger-like events and is still in the process of settling. The estimates of the star formation rates (SFR) using FUV and NUV luminosities are found to be 0.15 M yr1 and 0.36 M yr1 , respectively, and provide the lower limit due to the screen effect. The estimated lower rates of SFR in this galaxy probably represent its quenching due to the AGN driven outflows emanating from the central engine of NGC 1316. Keywords. Galaxies: formation—ultraviolet: galaxies—galaxies: evolution—galaxies: star formation.

1. Introduction NGC 1316 (Fornax A) is a nearby (z ¼ 0:00587) giant peculiar S0 galaxy hosting numerous tidal tails, shells, unusual dust patches all embedded within a much larger outer envelop of stars and a prominent dust lane oriented along its optical minor axis (Malin & Carter 1983; Deshmukh et al. 2013). In addition to the intricate dust patches and shells, NGC 1316 also hosts filamentary, nebular emission features, ripples, arcs and several complex filamentary loops of other phases of inter stellar medium (ISM) (Schweizer & Seitzer 1988). The faint tidal tails – wisps and shells of stars evident around the galaxy suggest that they have been torn from their original locations and flung This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’

into the intergalactic space through complex gravitational effects. Schweizer (1980) reported the presence of a compact disk of gas near its center that has different orientation and much faster rotation relative to the stars. All of these signs of complex dust features, shells, loops and other sub-structures evident in NGC 1316 collectively point to its violent past builtup through the merger of several smaller dust rich galaxies in distant past (Terlevich & Forbes 2002). Schweizer (1980) based on the expansion time of the outer stellar loops and Goudfrooij et al. (2001) on the spread in the ages of globular clusters demonstrate that the mergers might have happened between 1 to 3 Gyr ago and then NGC 1316 has continued accretion of smaller satellite galaxies (Iodice et al. 2017). NGC 1316 is one of the nearest Central Dominant (CD) giant radio galaxy, at a luminosity distance of  25 Mpc (Wright 2006) that exhibit filamentary low-

34

Page 2 of 8

ionization nuclear emission, an unresolved bright nucleus in UV bands, and interestingly hosts a strong radio core with steep spectrum and dual jets. The double-lobe radio continuum source from this galaxy extend well beyond its telescopic field-of-view, spanning over several degrees on the sky (Ekers et al. 1983; Drinkwater et al. 2001), and exhibit sharp edges in its outer part (Ekers et al. 1983). The S-shaped nuclear radio jet in the central region of NGC 1316 appears to bend at south of NW shell and north of SE blob due to the interaction between the radio jet and the ISM (Morokuma-Matsui et al. 2019). In the photographic plates. NGC 1316 appears to interact with a close companion spiral galaxy NGC 1317 on its North, however, this companion is not dominant enough to distort NGC 1316 at the observed scales. Though formation of NGC 1316 through merging episodes is unquestionable, the composition and characteristics of the multi-phase ISM in the galaxy is not yet fully understood. Relative to its very large stellar content 6  2  1011 M , the dust content of NGC 1316 is very low 2  106 M (Draine et al. 2007; Deshmukh et al. 2013), and is probably due to a 10 : 1 merger between a dominant, dust-poor earlytype galaxy and a smaller, gas-rich spiral (Lanz et al. 2010). Our earlier study (Deshmukh et al. 2013) of wavelength-dependent nature of dust extinction over the range near-infrared (NIR) to UV revealed that the dust grains in NGC 1316 posses identical physical and chemical characteristics as that of the canonical grains in the Milky Way and was confirmed through the parallel extinction curve. However, the smaller values of Rk indicate that the dust grains in the environment of NGC 1316 are relatively smaller than the canonical grains (Patil et al. 2007). This study also quantifies the dust mass assuming screening effect of dust (Patil et al. 2007; Vagshette et al. 2012), to be  2:12  105 M , while that estimated from the MIPS far-infrared flux was found to be 3:21  107 M . As ultraviolet (UV) and Far-Infrared (FIR) emission originate from the young stars, therefore, the UV and IR luminosities are often used in literature for investigating star formation histories of external galaxies (Buat et al. 2010; Kennicutt & Evans 2012). Particu˚ provides with larly, the UV continuum at k\2000 A the more reliable estimates of star formation rates (Salim et al. 2007). As a result, several attempts have been made in the past to better understand the star formation in external galaxies using UV observations.

J. Astrophys. Astr. (2021)42:34

However, the Earth’s atmosphere acts as a main hindrance in acquiring UV data on such galaxies, as a result, we need to rely on space-based observatories like GALEX and/or AstroSat for observing such galaxies in UV bands. Therefore, the UV observations of NGC 1316 acquired using AstroSat observatory are crucial for understanding the star formation and hence evolution of the galaxy. The merger episodes in NGC 1316 fuel the central supermassive black hole and transform it into a strong radio source, an Active Galactic Nucleus (Croton et al. 2006; Oogi & Habe 2013; Oogi et al. 2016; Rodriguez-Gomez et al. 2016; Vagshette et al. 2016, 2017, 2019; Pandge et al. 2012; Sonkamble et al. 2015). The mergers drive the external gas inflow into the central part of the galaxy and hence also activate the star formation in central region (Hopkins et al. 2008). But the ignited AGN develop powerful jets, which then heat and blow up the surrounding ISM and hence suppress the star formation (Croton et al. 2006; Fabian 2012). As a result, the detailed study of the star formation and interaction of radio jets with the surrounding environment in merger remnants is interesting. Therefore, NGC 1316 is one of the suitable candidates of recent mergers to investigate star formation and interplay between the AGN and ISM. Using recent high resolution radio observations with MeerKAT, Serra et al. (2019) have demonstrated that NGC 1316 hosts a significant fraction of H I emission distributed in a variety of scales i.e. all the way from its center till the large-scale environment. Serra et al. (2019) found that the H I emission detected in the central  5 kpc as additional component to the complex interstellar medium in this region: including dust (Carlqvist 2010; Deshmukh et al. 2013), dust emission (Lanz et al. 2010), molecular gas (Horellou et al. 2001; Morokuma-Matsui et al. 2019), ionised gas (Schweizer 1980; Morokuma-Matsui et al. 2019), and X-ray emitting gas (this paper). Serra et al. (2019) have also revealed the extension of H I emission up to 70–150 kpc and oriented along both the tails of NGC 1316 with a total mass content of 7  108 M , roughly 14 times larger than the past detection. Interestingly, both these H I tails of NGC 1316 show a spatial association with the complex optical tidal features confirming their common origin through the merging of a gas-rich progenitor system. They further propose that the merger was so significant that the tidal forces pulled a large fraction of gas and stars out to larger radii apparent in the form of optical tails.

J. Astrophys. Astr. (2021)42:34

Further, the low frequency observation between 154 and 1510 MHz shows that both the radio lobes exhibits a steep spectrum (Mancuso et al. 2017). These steep spectrum lobes are basically characterised by optically thin synchrotron radio emission from jets and they are associated with very massive black holes in the central early type galaxies (Mancuso et al. 2017). In this paper, we present a detailed study of NGC 1316 galaxy using the high spatial resolution AstroSat ultra-violet imaging telescope (UVIT) data. The paper is structured as follows: Section 2 reports the observation and data analysis methods, whereas Section 3 describes the results obtained from the UV imaging analysis. Section 4 discusses the star-formation and AGN outflow in NGC 1316 and lastly, Section 5 summarises the results obtained from the present study.

2. Observations and analysis The first Indian multi-wavelength astronomical satellite AstroSat was launched by Indian Space Research Organization (ISRO) on 28 September 2015 (Agrawal 2006, Singh et al. 2014). The observatory is sensitive to photons from visible, UV and X-ray bands simultaneously through five sets of instruments such as Ultraviolet Imaging Telescope (UVIT; Tandon et al. 2017a, b), Soft X-ray Telescope (SXT; Singh et al. 2017), Large Area X-ray Proportional Counters (LAXPCs; Agrawal et al. 2017), Cadmium–Zinc– Telluride Imager (CZTI; Rao et al. 2017), and a Scanning Sky Monitor (SSM; Ramadevi et al. 2018). The UVIT consists of two co-aligned telescopes with diameters of 38 cm, in Ritchey-Chretien configuration. Out of the two, one telescope is fully dedicated for FUV observations, whereas the other one observes in NUV and visible channels. The UV imaging telescope covers  280 circular field-of-view with an angular resolution of 1.200 for the NUV and \1.400 , for FUV channels. The details of UV imaging telescope and instrumentation can be found in Tandon et al. (2017a, b). The UVIT on-board AstroSat was used to observe our target galaxy, NGC 1316 at a redshift located at a luminosity distance1  25 Mpc (Wright 2006). The angular scale obtained by angular size distance of 100 corresponds to 0.122 kpc. The observations of the galaxy NGC 1316 were carried out in four UVIT 1

http://www.astro.ucla.edu/%7Ewright/CosmoCalc.html.

Page 3 of 8

34

filters. The characteristics of the filters, their mean wavelength and band width, exposure time, zero point and unit conversion factors are tabulated in Table 1. Photometric calibrations for FUV and NUV filters were carried out by following the procedure described in Tandon et al. (2017a, b).

3. Imaging analysis The high resolution UVIT data on NGC 1316 acquired using AstroSat was reduced following standard routine described in Tandon et al. (2017a). The level 2 image files were obtained from level 1 data by applying the UVIT Level-2 Pipeline (UL2P) task. This consists of the preliminary reduction, masking of bad pixels and flagging of multi-photon event blocks, detection of cosmic-ray events that affect the science data and finally applying the drift correction to the event centroids. Then, the multiple orbital image files were combined after applying astrometric corrections delivered by the UVIT_DriverModule task. The final images thus produced were the intensity maps in units of counts per second. This paper uses data sets acquired using two UV bands, namely NUV (N219M) and FUV (F148W) while data in other bands were rejected due to the poor signal. The flux and AB magnitude measurements of NGC 1316 were obtained by multiplying the count rates in a particular filter by its corresponding conversion factor as given by Tandon et al. (2017b). The source and background counts were extracted using imcnts task within xspatial package of Image Reduction and Analysis Facility (IRAF). The background counts were extracted by selecting multiple source free regions in image files. Then the counts in the science image were corrected for background emission and normalized considering the integration time. The astrometry corrected and 1.5r Gaussian smoothed FUV and NUV emission maps, derived from the analysis of the AstroSat data are shown in the left panels of Fig. 1, while those in the right panels compare them with that of the data from GALEX. The small scale features apparent within few arcsec (500 ) in AstroSat FUV data are not detected by GALEX. Similar small scale structures are also evident in the AstroSat NUV images. The imaging comparison of the data from AstroSat and GALEX observatories find the extent of FUV emission up to  2000 against  7000 in GALEX, while that in NUV bands is found to be  3000 against  20 in GALEX. To enhance the visibility of the UV emission features in NGC 1316,

34

Page 4 of 8

J. Astrophys. Astr. (2021)42:34

Table 1. Details of the observation of NGC 1316. Name

Mean (k) ˚ (in A)

Dk ˚ (in A)

Zero-point

FUV channel F148W (F1) F154W (F2) F169M (F3) F172M (F5)

CaF2 -1 BaF2 Sapphire Silica

1481 1541 1608 1717

500 380 290 125

18.016 0.01 17.778 0.01 17.455 0.01 16.342 0.02

NUV channel N219M (F2)

NUVB15

2196

270

16.50 0.01

Filter(slot)

Unit conversion (UC) ˚ (1015 erg cm2 s1 A) 3.09 3.55 4.392 10.74

± ± ± ±

Exp. time (in second)

0.029 0.04 0.037 0.16

830 300 1280 874

3.50 ± 0.035

3747

Figure 1. The upper panel represents the FUV images of NGC 1316 using ASTROSAT observation (left panel) and GALEX observation (right panel). While the lower panel visualize the NUV images of NGC 1316 using ASTROSAT observation (left panel) and GALEX observation (right panel).

the residual maps presented in Fig. 2 are derived after subtracting the smooth models in FUV and NUV bands of AstroSat data. Here, smooth models of the FUV and NUV emission were obtained by fitting ellipses using task ellipse within IRAF 2.16 to the surface brightness distribution in the respective bands.

This was achieved by obtaining its smooth models after fitting ellipses to the isophotes of the galaxy image using ellipse task within IRAF (for details, see Patil et al. 2007; Vagshette et al. 2012). Smooth 2D models thus derived which were then subtracted from the cleaned science frames of NGC 1316. One of

J. Astrophys. Astr. (2021)42:34

Page 5 of 8

34

features like knots, multiple rim-like structures along the NE and SW directions and a depression in the surface brightness near its nucleus (within central 1–2 kpc) and collectively point towards the complex morphological structures of NGC 1316. With an objective to investigate the association of the apparent UV extinction, we have made use of the optical B and V band data sets acquired using CTIO 1.5 m telescope and have generated the (B–V) colour index map of NGC 1316. The (B–V) colour index map reveals very complex dust extinction regions (darker shades) oriented along the optical minor axis of NGC 1316 in the form of a lane, which then appears in the form of a complex arc shape on either side of the lane. Dust lane, apparent in the central region of (B–V) colour index map, is orthogonal to its UV emission. However, the UV emission appears to be absorbed at the locations of dust in the central region of NGC 1316. We also generated a map of hot X-ray emitting gas distribution within NGC 1316 from the analysis of high resolution 0.3 to 6 keV Chandra data and is shown in the lower panel of Fig. 2. X-ray emission from this galaxy also exhibit a complex structure, however, by and large the diffuse X-ray emission appears to be oriented along dust lane extended to even larger extent (  2 arcmin).

4. Discussion

Figure 2. Top panel shows the AstroSat/UVIT FUV(F148W) residual image of NGC 1316, created by subtracting the model image from the original background subtracted raw image. Arrow marks in the image point to the excess emission from the central region. The FUV emission directly represents the location of young massive stars in the nebula. Middle panel represents the BVcolour map and bottom panel shows the distribution of hot gas.

the residual maps of NGC 1316 in FUV is shown in Fig. 2, which confirms the presence of the hidden

The rate at which the galaxies form stars is only one of the several fundamental astrophysical processes in galaxy evolution. Massive young O, B or A-type stars are hot and emit a quantitative amount of ultraviolet radiation and thus exhibit an ideal tracer to identify and measure the star formation activity of the galaxy. The space-based telescopes such as AstroSat and GALEX provide direct estimate of the current star-formation rate (SFR) by measuring the UV radiation. The star formation rate of NGC 1316 was calculated by measuring the luminosity (or flux) in NUV and FUV bands. The total measured in FUV and NUV are flux within 2000 12 erg cm2 s1 and ð38:8 ð24:0  8:0Þ  10 12 2 1 erg cm s , respectively. This cor8:5Þ  10 responds to the luminosity of ð1:85  0:62Þ  1042 erg s1 and ð2:99  0:66Þ  1042 erg s1 , respectively. The SFR in the galaxy were estimated from its UV luminosity and employing the relation given by Iglesias-Pa´ramo et al. (2006) which can be expressed as

34

Page 6 of 8

J. Astrophys. Astr. (2021)42:34

Figure 3. The FUV (F148W) images are Gaussian smoothed with 1.500 . The left panel shows the 1.45 GHz radio contours overlaid on the FUV image, whereas the right panel shows the soft X-ray contours overlaid on the FUV image.

LðFUVÞ ðerg s1 Þ  109:51 ; 3:83  1033 LðNUVÞ ðerg s1 Þ  109:33 : SFR ðNUVÞ ðM yr1 Þ ¼ 3:83  1033 SFR ðFUVÞ ðM yr1 Þ ¼

Using the estimated FUV and NUV luminosities from the UVIT observation, the star formation rates of NGC 1316 are, respectively, found to be 0:15  0:05 M yr1 and 0:36  0:07 M yr1 and provide lower limit due to the foreground screening effect. But the apparent significant and complex dust extinction region imply that the UV emission from this galaxy is extinguished and therefore the SFR estimated using the UV fluxes alone represent a substantial underestimate. As a result, other possible tracers need to be explored for better representation. The UV photons absorbed by the dust grain produces re-emission at longer wavelengths like IR or FIR and hence IR emission analysis is called for this system. Duah et al. (2016) using 22 lm luminosity derived the star-formation rate of  0.7 M yr1 , whereas the IRAS FIR luminosities yielded the SFR as 0.82 M yr1 (Terrazas et al. 2017). NGC 1316 is a well-known merger type galaxy and merger induces star formation at the centre (100–1000 pc) of the galaxy (Bournaud 2011). The nuclear activity in merger-induced starburst galaxies have long been given more importance in the study of merger and star formation connection (Sanders et al. 1988; Duc et al. 1997; Lawrence et al. 1989). Though, NGC 1316 satisfies characteristics of merger remnant, gas-rich system, the SFRs in this galaxy are significantly less. Understanding the cause of quenching of star formation in NGC 1316 is,

therefore, a challenge for studies of evolution of the galaxy. A different theoretical explanation has been proposed for this challenge though a general agreement is yet to reach. The most acceptable explanation is the Active Galactic Nucleus (AGN) outflows can remove the substantial amount of gas from the host galaxy, thereby shutting off the star formation. On the other hand, these outflows compress the surrounding gas and trigger star formation in compressed outer region. To investigate the outflow activity, we used radio image of NGC 1316, observed on 31 May 2002 (ObsID: AH787) available in VLA archive. The left panel of Fig. 3 shows the FUV image of NGC 1316 along with the 1.45 GHz VLA radio contours. The direction of the radio jet can be seen along the FUV surface brightness depression region implying that the radio jet pushes (remove) the gas in outward direction from the central region of NGC 1316. From this map it is also apparent that some amount of gas gets compressed along the NW and SE direction and emits radiation in the ultraviolet region. Thus, the AGN driven outflows remove gas from the nuclear region and suppresses star formation in the central region. The NUV and FUV emission maps derived from the analysis of ASTROSAT data exhibit multiple rimlike structures along the NE and SW directions and a significant depression in the central region of surface brightness from NGC 1316. The presence of such peculiar structures in NGC 1316 is likely to be due to the tidal forces that pulled gas and stars form core after the merger episode or due to the AGN outburst. The radio jets originating from central engine pushes away the gas in outward direction during outbursts (O’Sullivan et al. 2011; David et al. 2009; McNamara

J. Astrophys. Astr. (2021)42:34

et al. 2000; Sanders & Fabian 2007; Vagshette et al. 2019). However, the recent high resolution radio observations performed using MeerKAT suggests its origin to be due to a  10:1 merger between a dominant early-type galaxy and a smaller, gas-rich progenitor (Morokuma-Matsui et al. 2019). Spatial correspondence between the FUV, 0.5–3 keV X-ray and 1.45 GHz radio emission from NGC 1316 are shown in Fig. 3. X-ray emission map was obtained using high resolution Chandra 30 ks archival data (Obs. ID 2022) with the ACIS-S3 as the target. The left panel of Fig. 3 shows 1.4 GHz radio contours overlaid on the FUV image of NGC 1316, while in the right panel, the 0.5-3 keV X-ray contours are overlaid on the FUV image. The figure clearly shows the morphological similarity in the FUV and X-ray emission. Both these indicate deficient regions along the SW and SE directions, whereas excess emission is apparent in both the bands along the SE and SW directions pointing towards their common origin. It is likely that the observed deficiencies are carved by the central AGN, radio jets of which pushes the UV and X-ray emitting gas from the central region of the galaxy.

5. Conclusion We have observed the nearby merger remnant galaxy NGC 1316 in search of the connection between nuclear activity and star formation in its central region using high resolution NUV and FUV imaging telescopes on board the first Indian dedicated astronomical satellite AstroSat. The unsharp masked image as well as the surface brightness profiles from the analysis of these data sets confirm that the UV emission from this galaxy is not smooth but exhibits perturbations and are in agreement with the results presented by several other researchers. Some of the important results from this study are: • The residual image confirms the presence of peculiar features in the nuclear (1–2 kpc) region of NGC 1316. • The hidden structures like rim, clumps and their strong spatial correspondence with imagery at other wavelengths confirm that the origin of gas and dust in this system is due to the merger like episodes. • The estimates of star formation rates in NGC 1316 based on FUV and NUV luminosities are 0.15 M yr1 and 0.36 M yr1 , respectively.

Page 7 of 8

34

• A comparison of 1.4 GHz emission contours and FUV and emission map revealed that the UV emitting sources are displaced away by the radio jets emanating from the central engine, thereby confirming that the AGN driven outflows are responsible for the quenching of the star formation in the gas-rich merger remnant galaxy NGC 1316.

Acknowledgements We thank the reviewer for encouraging comments and suggestions on the paper. This work uses the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). This publication uses UVIT data processed by the payload operations centre at IIA. The UVIT is built in collaboration between IIA, IUCAA, TIFR, ISRO and CSA. Data from VLA and Chandra archive, Extragalactic Database (NED) and NASA’s Astrophysics Data System (ADS) are also used in this paper. NDV thanks Science and Engineering Research Board (SERB), India for providing a research fund (Ref. No.: YSS/2015/001413). NDV also thanks IUCAA, Pune for the use of library facility. References Bournaud F. 2011, in Charbonnel C., Montmerle T., eds, EAS Publications Series, vol. 51, 107–131 Buat V., Giovannoli E., Burgarella D. et al. 2010, MNRAS, 409, L1 Carlqvist P. 2010, Ap&SS, 327, 267 Croton D. J., Springel V., White S. D. M. et al. 2006, MNRAS, 365, 11 David L. P., Jones C., Forman W. et al. 2009, ApJ, 705, 624 Deshmukh S. P., Tate B. T., Vagshette N. D., Pandey S. K., Patil M. K. 2013, Res. Astron. Astrophys., 13, 885 Draine B. T., Dale D. A., Bendo G. et al. 2007, ApJ, 663, 866 Drinkwater M. J., Gregg M. D., Holman B. A., Brown M. J. I. 2001, MNRAS, 326, 1076 Duah Asabere B., Horellou C., Jarrett T. H., Winkler H. 2016, A&A, 592, A20 Duc P. A., Mirabel I. F., Maza J. 1997, A&AS, 124, 533 Ekers R. D., Goss W. M., Wellington K. J. et al. 1983, A&A, 127, 361 Fabian A. C. 2012, ARA&A, 50, 455 Goudfrooij P., Alonso M. V., Maraston C., Minniti D. 2001, MNRAS, 328, 237

34

Page 8 of 8

Hopkins A. M., McClure-Griffiths N. M., Gaensler B. M. 2008, ApJ, 682, L13 Horellou C., Black J. H., van Gorkom J. H. et al. 2001, A&A, 376, 837 Iglesias-Pa´ramo J., Buat V., Takeuchi T. T. et al. 2006, ApJS, 164, 38 Iodice E., Spavone M., Capaccioli M. et al. 2017, ApJ, 839, 21 Kennicutt R. C., Evans N. J. 2012, ARA&A, 50, 531 Lanz L., Jones C., Forman W. R. et al. 2010, ApJ, 721, 1702 Lawrence A., Rowan-Robinson M., Leech K., Jones D. H. P., Wall J. V. 1989, MNRAS, 240, 329 Malin D. F., Carter D. 1983, ApJ, 274, 534 Mancuso C., Lapi A., Prandoni I. et al. 2017, ApJ, 842, 95 McKinley B., Yang R., Lo´pez-Caniego M. et al. 2015, MNRAS, 446, 3478 McNamara B. R., Wise M., Nulsen P. E. J. et al. 2000, ApJ, 534, L135 Morokuma-Matsui K., Serra P., Maccagni F. M. et al. 2019, PASJ, 71, 85 Oogi T., Habe A. 2013, MNRAS, 428, 641 Oogi T., Habe A., Ishiyama T. 2016, MNRAS, 456, 300 O’Sullivan E., Giacintucci S., David L. P. et al. 2011, ApJ, 735, 11 Pandge M. B., Vagshette N. D., David L. P., Patil M. K. 2012, MNRAS, 421, 808 Patil M. K., Pandey S. K., Sahu D. K., Kembhavi A. 2007, A&A, 461, 103

J. Astrophys. Astr. (2021)42:34 Rodriguez-Gomez V., Pillepich A., Sales L. V. et al. 2016, MNRAS, 458, 2371 Salim S., Rich R. M., Charlot S. et al. 2007, ApJS, 173, 267 Sanders D. B., Soifer B. T., Elias J. H. et al. 1988, ApJ, 325, 74 Sanders J. S., Fabian A. C. 2007, MNRAS, 381, 1381 Schweizer F. 1980, ApJ, 237, 303 Schweizer F., Seitzer P. 1988, ApJ, 328, 88 Serra P., Maccagni F. M., Kleiner D. et al. 2019, A&A, 628, A122 Sonkamble S. S., Vagshette N. D., Pawar P. K., Patil M. K. 2015, Ap&SS, 359, 21 Tandon S. N., Subramaniam A., Girish V. et al. 2017a, AJ, 154, 128 Tandon S. N., Hutchings J. B., Ghosh S. K. et al. 2017b, J. Astrophys. Astr., 38, 28 Terlevich A. I., Forbes D. A. 2002, MNRAS, 330, 547 Terrazas B. A., Bell E. F., Woo J., Henriques B. M. B. 2017, ApJ, 844, 170 Vagshette N. D., Naik S., Patil M. K. 2019, MNRAS, 485, 1981 Vagshette N. D., Naik S., Patil M. K., Sonkamble S. S. 2017, MNRAS, 466, 2054 Vagshette N. D., Pandge M. B., Pandey S. K., Patil M. K. 2012, New A, 17, 524 Vagshette N. D., Sonkamble S. S., Naik S., Patil M. K. 2016, MNRAS, 461, 1885 Wright E. L. 2006, PASP, 118, 1711

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:42 https://doi.org/10.1007/s12036-020-09684-1

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

Study of Galactic structure using UVIT/AstroSat star counts RANJAN KUMAR1, ANANTA C. PRADHAN1,*, DEVENDRA K. OJHA2,

SONIKA PIRIDI1, TAPAS BAUG3 and S. K. GHOSH2 1

Department of Physics and Astronomy, National Institute of Technology, Rourkela 769 008, India. Tata Institute of Fundamental Research, Homi Bhabha Road, Mumbai 400 005, India. 3 Kavli Institute for Astronomy and Astrophysics, Peking University, 5 Yiheyuan Road, Haidian District, Beijing 100871, China. *Corresponding Author. E-mail: [email protected] 2

MS received 4 November 2020; accepted 16 December 2020 Abstract. The structure of our Galaxy has been studied from ultraviolet (UV) star counts obtained with the Ultra-Violet Imaging Telescope (UVIT) on board the AstroSat satellite, in Far-UV (FUV) and Near-UV (NUV) bands. The F154W (BaF2) and N263M (NUVB4) filters were used in the FUV and NUV bands, respectively. The point sources are separated from the extra-galactic sources of UVIT observations using infrared (IR) color cut method. The observed UVIT star counts match well with the simulations obtained from the Besanc¸on model of stellar population synthesis towards several Galactic directions. We also estimated the scale length and scale height of the thick disc and the scale height of the thin disc using the space density function and the exponential density law for the stars of intermediate galactic latitudes. The scale length of the thick disc ranges from 3.11 to 5.40 kpc whereas the scale height ranges from 530  32 pc to 630  29 pc. The scale height of the thin disc comes out to be in the range of 230  20 pc to 330  11 pc. Keywords. Stars: distances—ultraviolet: stars—galaxy—disc.

1. Introduction A major objective of modern astrophysics is to understand when and how the galaxies are formed, and how they have evolved since. Our own Galaxy, the Milky Way, provides a unique opportunity to study a galaxy in exquisite detail, by measuring and analyzing the properties of large samples of individual stars. The stellar population synthesis models based on star counts methods in conjunction with large area sky survey observations have considerably helped in predicting the different structural parameters of the Galaxy such as stellar densities, scale length and scale height (Gilmore & Reid 1983; Robin et al. 2003; Girardi et al. 2005; Juric´ et al. 2008; Ivezic´ et al. 2012; Chen et al. 2017). The Galaxy models are This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

mostly based on the photometric surveys from infrared (IR) to ultraviolet (UV) bands of the electromagnetic spectrum such as massive data sets of the Sloan Digital Sky Survey (SDSS) (York et al. 2000), the Two-Micron All-Sky Survey (2MASS) (Skrutskie et al. 2006), Galaxy Evolution Explorer (GALEX) (Martin et al. 2005), Global Astrometric Interferometer for Astrophysics (GAIA) (Gaia Collaboration et al., 2018) including millions to billions of stars. Star counts studies in UV have made a stride after the advent of GALEX, which provided a wide sky coverage in UV allowing for a new analysis of the UV sky (Bianchi et al. 2011; Pradhan et al. 2014). Bianchi et al. (2011) have used the TRILEGAL model of stellar population synthesis (Girardi et al. 2005) to produce UV star counts in GALEX bands and found that their model gives the closest prediction to the observed star counts. Pradhan et al. (2014) have upgraded the Besanc¸on model of stellar population synthesis to include the UV bands of GALEX and

42

Page 2 of 7

UVIT to produce UV star counts towards different parts of the sky. They found that the model predicted star counts match well with the observed star counts in FUV and NUV bands of GALEX. Although they developed the model to predict UV star counts in FUV and NUV filters of UVIT incorporating respective filter responses, no comparison could be carried out with observations at that time (UVIT was launched later on). Based on the model UV colors they could separate out white dwarfs (WDs) and blue horizontal branch stars (BHBs) which are evasive at other wavelength bands due to their high temperature and low luminosities. The Besanc¸on model of stellar population synthesis is extensively described in Robin et al. (2003, 2012) which produces star counts of different evolutionary stages contained in Galactic thin disc, thick disc, halo, bar and bulge. The thick disc density law is generally approximated by a double exponential which is a function of both scale length and scale height. The thick disc has larger scale length and scale height than the thin disc (Ojha et al. 1996; Robin et al. 1996; Chen et al. 2001; Chang et al. 2011; Chen et al. 2017) although the values of these parameters are still debatable. A few studies have also revealed that the scale length of the thin disc is larger than the thick disc which contradicts to the earlier consensus (Bensby et al. 2011; Cheng et al. 2012). Even with the availability of much improved data collections, a convergence in the values of structure parameters has not yet obtained (BlandHawthorn & Gerhard 2016).

2. Data reduction and analysis UVIT consists of 38-cm twin telescopes; one for ˚ ) and the other for NUV (2000– FUV (1300–1800 A ˚ ˚ ). The light from 3000 A) and visible (3200–5500 A the latter telescope is split into NUV and visible bands using a dichroic mirror. The FUV and NUV telescopes comprise of five filters each and are operated in photon counting mode whereas the visible filters are operated in integration mode and are mostly used for tracking purpose. The field of view of both the FUV and NUV telescopes is 280 and the resolution is \1.500 . The details about the UVIT telescopes and their calibration are reported in Tandon et al. (2017, 2020). We have observed five fields towards different Galactic directions such as Galactic Center (GC), Galactic Anti-Center (GAC), and South Galactic Pole ˚ ) and one (SGP) using NUVB4 filter (keff ¼ 2632 A

J. Astrophys. Astr. (2021)42:42

field (GC47-42, towards GC) in F154W /BaF2 filter ˚ ) of UVIT. The observation details of (keff ¼ 1541 A the observed fields are given in Table 1. The data reduction of the UVIT observations was performed with a customized software package CCDLAB (Postma & Leahy 2017). After performing all the corrections, we aligned and co-added all the orbits to obtain the final science image to perform photometry. We applied astrometry from GAIA data release 2 (Gaia DR2, Gaia Collaboration et al., 2018) catalog using IRAF ccmap package. We were able to achieve an overall astrometric precision of 0:100 for our images. We performed point spread function (PSF) photometry on the reduced final science images using DAOPHOT package (Stetson 1987) in IRAF.1 We selected 30–40 isolated sources for PSF modelling of the images. We got an average FWHM of 1.200 for the PSF model sources on the observed images. Once the model was developed, we did an aperture photometry on NUV and FUV images. We used ALLSTAR routine to obtain the relative magnitudes of the sources in the crowded field over aperture photometry of the sources at FWHM of the PSF model stars. We performed a curve of growth analysis on PSF modeled sources to find an aperture correction value for the relative magnitudes provided by ALLSTAR routine. The aperture correction was applied to the relative magnitudes of the detected sources. Finally, we generated a catalog of UVIT observed sources by applying various selection criteria such as magnitude error cut, sharpness and chi-fit of the profile of stars. The magnitudes of the sources were corrected for extinction using EðB  VÞ values from Schlafly & Finkbeiner (2011) and then employing the extinction law of Cardelli et al. (1989). GALEX exposure times vary from observation to observation around the nominal exposures of 100s for all sky imaging survey (AIS). For this exposure time, the typical 5r detection limits for the FUV and NUV filters of GALEX AIS are  20 and  21 ABmag, respectively, and for medium imaging survey (MIS) for an exposure time of 1500s the depth is  22.7 ABmag in both FUV and NUV filters (Bianchi et al. 2017). The typical 5r detection limits for an exposure time of 200s for the UVIT BaF2 (FUV) and NUVB4 (NUV) wavebands are 20.0 and 21.2 ABmag, respectively. So, the typical depths reached in both GALEX and UVIT are almost similar. The magnitude error plots for UVIT FUV and NUV observed sources 1

https://iraf.net/.

J. Astrophys. Astr. (2021)42:42

Page 3 of 7

42

Table 1. The details of the UVIT observations of various Galactic fields. Galactic positions

Fields GAC146-46 GC47-42 GC47-42 SGP30-90 GC15?60 GAC175?60

RA (J2000)

DEC (J2000)

29:4583 326:6767 326:6767 12:8583 222:3558 160:8379

13:0000 8:6110 8:6110 27:1283 14:9447 41:9471

0.35

Filters

Exposure time in seconds

NUVB4 BaF2 NUVB4 NUVB4 NUVB4 NUVB4

4,665 4,989 5,521 4,638 5,718 5,694

0.25

0.30 0.20 Magnitude Error

Magnitude Error

0.25 0.20 0.15 0.10

0.15

0.10

0.05 0.05 0.00

0.00 14

15

16 17 18 19 20 21 22 NUVB4 Magnitude (AB system)

23

24

14

15

16 17 18 19 20 21 22 23 24 FUV BaF2 Magnitude (AB system)

25

Figure 1. NUVB4 (left) and BaF2 (right) magnitudes of the observed point sources are plotted against their corresponding errors. We have retained sources with errors less than 0.2 ABmag for our analysis.

are shown in Fig. 1. We have deep observations with exposure times more than 4.5 ks and we see that the sources up to 23.5 ABmag have the magnitude errors of less than 0.2 ABmag in both the filters. We have retained sources with errors less than 0.2 ABmag in both the NUVB4 and BaF2 filters. The Two-Micron All-Sky Survey (2MASS) and Wide-Field Infrared Survey Explorer (WISE) observations cover almost the entire sky and our UVIT observation overlaps with WISE þ 2MASS survey. We cross-matched observed UVIT sources with WISE þ 2MASS catalog within a search radius of 300 using CDS X-match service available in TOPCAT software package2 (Taylor 2005). We applied the IR color cut (J – W1 [ 1.2 mag, where J is 2MASS band at 1.24 lm and W1 is a WISE band at 3.4 lm) method to exclude the extra-galactic sources from our catalog (see Pradhan et al. 2014). Separation of the point sources from the extra-galactic sources is clearly visible in the color-magnitude diagram as shown in Fig. 2.

2

http://www.star.bris.ac.uk/*mbt/topcat/.

3. Comparison of model star counts Besanc¸on model is a population synthesis model developed using Galactic evolutionary scenarios and dynamics of different components of the Milky Way such as discs, bulge, halo, etc. (Robin et al. 2003, 2012). It uses a set of evolutionary tracks, a star formation rate and an initial mass function to produce stars of different populations. Pradhan et al. (2014) have extended this model to UV passbands by including the GALEX and UVIT filters. The model has already been validated with UV star counts of GALEX catalog and also improved to generate the star counts in UVIT filters. Here, we verify the model star counts with our UVIT observations. We generated model simulations for the BaF2 and NUVB4 filters of UVIT in different Galactic directions. The Galactic directions were chosen in such a manner that the observation should cover regions towards GC and GAC. We have also considered the south Galactic pole direction to check the star count variation near Galactic poles. We retained all the point sources with errors less than 0.2 ABmag in both the filters. In Fig. 3, we have

42

Page 4 of 7

J. Astrophys. Astr. (2021)42:42 14

14 UVIT BaF2

15

16

16

17

17

18

18 BaF2

NUV B4

UVIT NUVB4

15

19

19

20

20

21

21

22

22 23

23 24 −0.5

24 0.0

0.5

1.0 1.5 J-W1

2.0

2.5

3.0

−0.5

0.0

0.5

1.0 J-W1

1.5

2.0

2.5

Figure 2. Color-magnitude diagrams, J – W1 vs. NUVB4 (left) and J – W1 vs. BaF2 (right). A vertical dashed line is drawn at the color J – W1 = 1.2 mag in the both panels. Stellar sources are well separated from the extra-galactic sources with J – W1 \ 1.2 mag.

compared the model simulated star counts (solid line) with observations (solid circles) in UVIT NUVB4 and BaF2 filters at different Galactic latitudes. The observed and model simulated star counts were binned in a magnitude interval of 1.0 ABmag. We see that the observations in both the filters are matching with the model simulations up to their completeness limits.

ratio between fields at the same Galactic latitude towards GC and GAC directions is given by AGC =AGAC ¼ expðþ2 jR  R0 j=hR Þ;

where jR  R0 j ¼ d cosðbÞ is the distance of stars from the Sun on the Galactic plane. Hence, the scale length of the discs obtained from the above formula is hR ¼

4. Scale length and scale height of the thick disc Since the first evidence for the existence of the thick disc provided by Gilmore and Reid (1983) from the star counts analysis, it has been established as a chemically and dynamically distinct component of the Galaxy which is substantiated by the analysis of large samples of data (Yong et al. 2005; Bensby et al. 2011; Jacobson et al. 2011; Beraldo e Silva et al. 2020, and references therein). In order to derive the structural parameters of the Galaxy (i.e., the scale length and scale height of the thin and thick discs) using star counts method, we use the density law which is approximated by a double exponential:     R  R0 jzj exp  ; ð1Þ qðR; zÞ ¼ qðR0 Þ exp  hz hR where qðR0 Þ is the normalized stellar density at the solar neighborhood, R0 ¼ 8:33  0:35 kpc (Gillessen et al. 2009) is distance of the Sun from the Galactic center, z ¼ d sinðbÞ is the height above the Galactic plane where b is the Galactic latitude, R is the Galactocentric distance projected on the Galactic plane, and hR and hz are the scale length and scale height of the disc, respectively. We have used star counts ratio in two galactic directions (i.e., GC and GAC) to derive the disc parameters. The star counts

ð2Þ

2d cosðbÞ : logðAGC =AGAC Þ

ð3Þ

While estimating the scale length of the thick disc using Equation (3), we have assumed that the stellar population is homogeneous in both the Galactic directions. However, this is not exactly the case when large distances are being probed: stellar population in the inner disc, for instance, are more metal rich, hence, they could produce UV-bright sources with a different efficiency than the metal poorer populations in the outer disc. This is the kind of effect that requires a more detailed investigation, and it would be beyond the scope of this paper. We obtained distance, d of the observed UVIT sources using their parallax values from Gaia DR2 catalog (Gaia Collaboration et al., 2018) and then calculated the scale length of the thick disc using the UVIT observed star counts in a magnitude interval of 18.0 to 20.0 AB magnitude at GC and GAC directions at similar latitude (see Table 2). We calculated the space densities of the observed stars at two Galactic latitudes, 60 and 42 in northern and southern Galactic hemispheres using the following equation: qðr1 ; r2 Þ ¼ N1;2 =DV1;2 ;

ð4Þ

where r1 , r2 are the limiting distances, N1;2 is the total number of stars within distances r1 and r2 , and partial

J. Astrophys. Astr. (2021)42:42 350

Page 5 of 7

l = 15o , b = 60o Model simulation UVIT NUVB4

300

42

l = 175o , b = 60o Model simulation UVIT NUVB4

140 120 Starcounts /deg 2

Starcounts /deg 2

250 200 150 100 50

80 60 40 20

0

0 14

650

15 16 17 18 19 20 21 22 NUVB4 Magnitude (AB system)

23

24

14

l = 47o , b = −42o Model simulation UVIT NUVB4

600 550

15 16 17 18 19 20 21 22 NUVB4 Magnitude (AB system)

23

24

23

24

l = 146o , b = −46o Model simulation UVIT NUVB4

175

500

150

450

Starcounts /deg 2

Starcounts /deg 2

100

400 350 300 250 200

125 100 75 50

150 100

25

50 0

0 14

120

100

15 16 17 18 19 20 21 22 NUVB4 Magnitude (AB system)

23

24

14

l = 47o , b = −42o Model simulation UVIT FUV(BaF2 )

15 16 17 18 19 20 21 22 NUVB4 Magnitude (AB system)

l = 30o , b = −90o Model simulation UVIT NUVB4

140

Starcounts /deg 2

Starcounts /deg 2

120 80

60

40

100 80 60 40

20 20 0

0 14 15 16 17 18 19 20 21 22 23 24 25 FUV(BaF2 ) Magnitude (AB system)

14

15 16 17 18 19 20 21 22 23 NUVB4 Magnitude (AB system)

24

Figure 3. Model predicted star counts are compared with the observed star counts for the NUVB4 and BaF2 filters of UVIT towards various Galactic directions. The solid circles represent the observed UV star counts along with the error bars due to Poisson noise. The solid lines represent the model generated star counts. The star counts are binned in a magnitude interval of 1.0 ABmag. Table 2. Scale length and scale height of the thin and thick discs derived from UVIT star counts. Galactic positions b 60 -42

Thick disc

l

Scale length (hR , in kpc)

Scale height (hz , in pc)

Thin disc Scale height (hz , in pc)

15 175 47 146

3.11 3.11 5.40 5.40

570  54 650  49 530  32 630  32

320  17 280  05 330  11 230  20

42

Page 6 of 7

J. Astrophys. Astr. (2021)42:42 10−1

10−1 Northern Galactic b ≈ 60o

GAC-fit GAC-UVIT

−2

Southern Galactic b ≈ −42o

GAC-fit GAC-UVIT

−2

10

10

10−3

10−3

10−4

10−4

10−5

10−5

10−6

10−6

Figure 4. Space density (counts per cubic parsec) vs. height above the Galactic plane (z, in kpc) for GAC fields at intermediate latitudes towards the northern (left panel) and southern (right panel) Galactic directions. Thin disc stars were fitted with an exponential density law for z\1:2 kpc, while thick disc stars were fitted from z ¼ 1:2 to 2.8 kpc.

volume DV1;2 ¼ ðp=180Þ2 ðh=3Þðr23  r13 Þ with h being the field size in square degrees. We then fitted the analytical density law function (solid line) to the space density (solid circles) incorporating all the associated parameters (Fig. 4). The space densities of the UVIT observed point sources towards GC and GAC directions at two latitudes 60 and 42 are calculated using Equation (4). We assumed a local density ratio for thin and thick disc stars to be 100:5 as suggested in Ojha et al. (1996) along the radial directions. We find a turnover at z  1:2 kpc in the space density of stars, which shows that there are two physically distinct components: a thin disc with z\1:2 kpc and a thick disc with z [ 1:2 kpc. We fit the exponential density laws (Equation (1)) for thin disc stars for z\1:2 kpc and for thick disc stars from z  1:2 to 2.8 kpc. The scale length of the thick disc and the scale heights of the thin disc and thick disc are calculated at four Galactic directions which are given in Table 2. A wide range of values of scale length of the thick disc and scale height of the thick disc and thin disc has been published in literature using multi-wavelength photometric surveys which demonstrates the persistence of uncertainty in these structure parameters (Bland-Hawthorn & Gerhard 2016). Our study of UV star counts gives a scale length of the thick disc between 3.11 kpc and 5.40 kpc, which is in close agreement with the literature values of 3.00 to 5 kpc (Chen et al. 2001; Ojha 2001; Siegel et al. 2002; Juric´ et al. 2008; Yaz & Karaali 2010; Chang et al. 2011; Chen et al. 2017). Similarly, we estimated the scale height of thick disc to be from 530  32 pc to 650  49 pc with several intermediate values. Our estimation matches with the estimation of the previous works; 550–720 pc (Bilir et al. 2008), 490–580 pc (Ak et al. 2007) and 580–720 (Chen et al. 2001). Our measured

scale height of the thin disc ranges from 230  20 pc to 330  11 pc which matches with results of the recent work by Juric´ et al. (2008), Yaz & Karaali (2010), Chang et al. (2011), Polido et al. (2013) and Lo´pez-Corredoira & Molgo´ (2014).

5. Conclusion We present the preliminary results of the UV star counts analysis performed using the observations obtained from UVIT on board AstroSat satellite. The Besanc¸on model of stellar population synthesis upgraded to produce simulations for UVIT filters is validated for two of its filters. The scale height ranges of thick disc and thin disc obtained from the UV star counts analysis are from 530  32 pc to 630  29 pc and 230  20 pc to 330  11 pc, respectively. The scale length of the thick disc varies from 3.11 to 5.40 kpc. The values of these parameters are in well agreement with the already reported literature values. Here, we have limited our analysis using observations through one FUV and one NUV filter of UVIT mostly covering GC and GAC regions. In the future, we will present our analysis by comparing the model predictions in other filters of UVIT towards various possible Galactic directions. This will provide us an opportunity to filter out the hot sources such as white dwarfs and blue horizontal branch stars from the sample.

Acknowledgements We would like to thank the referee for giving useful suggestions to improve the manuscript. We would like to thank Dr. A. C. Robin for letting us use their model of stellar population synthesis and for giving her

J. Astrophys. Astr. (2021)42:42

useful inputs on the Besanc¸on model. RK would like to acknowledge CSIR Research Fellowship (JRF) Grant No. 09/983(0034)/2019-EMR-1 for the financial support. ACP would like to acknowledge the support by Indian Space Research Organization, Department of Space, Government of India (ISRO RESPOND project No. ISRO/RES/2/409/17-18). ACP also thanks Inter University centre for Astronomy and Astrophysics (IUCAA), Pune, India for providing facilities to carry out his work. DKO and SKG acknowledge the support of the Department of Atomic Energy, Government of India, under Project Identification No. RTI 4002. TB acknowledges the support from the National Key Research and Development Program of China (2017YFA0402702, 2019YFA0405100). This publication uses the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Center (ISSDC). The UVIT data used here was processed by the Payload Operations Centre at IIA. The UVIT is built in collaboration between IIA, IUCAA, TIFR, ISRO and CSA.

References Ak S., Bilir S., Karaali S., Buser R. 2007, Astronomische Nachrichten, 328, 169 Bensby T., Alves-Brito A., Oey M. S., Yong D., Mele´ndez J. 2011, ApJL, 735, L46 Beraldo e Silva L., Debattista V. P., Nidever D., Amarante J., Garver, B. 2020. arXiv:2009.03346 Bianchi L., Herald J., Efremova B. et al. 2011, Astrophys. Space. Sci., 335, 161 Bianchi L., Shiao B., Thilker D. 2017, ApJS, 230, 24 Bilir S., Cabrera-Lavers A., Karaali S. et al. 2008, Proc. Astron. Soc. Aust., 25, 69 Bland-Hawthorn J., Gerhard O. 2016, ARA&A, 54, 529 Cardelli J. A., Clayton G. C., Mathis J. S. 1989, ApJ, 345, 245 Chang C.-K., Ko C.-M., Peng T.-H. 2011, ApJ, 740, 34 Chen B., Stoughton C., Smith J. A. et al. 2001, ApJ, 553, 184 Chen B. Q., Liu X. W., Yuan H. B. et al. 2017, MNRAS, 464, 2545 Cheng J. Y., Rockosi C. M., Morrison H. L. et al. 2012, ApJ, 752, 51

Page 7 of 7

42

Gaia Collaboration, Brown A. G. A., Vallenari A. et al. 2018, A&A, 616, A1 Gillessen S., Eisenhauer F., Trippe S. et al. 2009, ApJ, 692, 1075 Gilmore G., Reid N. 1983, MNRAS, 202, 1025 Girardi L., Groenewegen M. A. T., Hatziminaoglou E., da Costa L. 2005, A&A, 436, 895 Ivezic´ Zˇ., Beers T. C., Juric´ M. 2012, ARA&A, 50, 251 Jacobson H. R., Friel E. D., Pilachowski C. A. 2011, AJ, 141, 58 Juric´ M., Ivezic´ Zˇ., Brooks A. et al. 2008, ApJ, 673, 864 Lo´pez-Corredoira M., Molgo´ J. 2014, A&A, 567, A106 Martin D. C., Fanson J., Schiminovich D. et al. 2005, ApJL, 619, L1 Ojha D. K. 2001, MNRAS, 322, 426 Ojha D. K., Bienayme O., Robin A. C., Creze M., Mohan V. 1996, A&A, 311, 456 Polido P., Jablonski F., Le´pine J. R. D. 2013, ApJ, 778, 32 Postma J. E., Leahy D. 2017, PASP, 129, 115002 Pradhan A. C., Ojha D. K., Robin A. C., Ghosh S. K., Vickers J. J. 2014, A&A, 565, A33 Robin A. C., Haywood M., Creze M., Ojha D. K., Bienayme O. 1996, A&A, 305, 125 Robin A. C., Marshall D. J., Schultheis M., Reyle´ C. 2012, A&A, 538, A106 Robin A. C., Reyle´ C., Derrie`re S., Picaud S. 2003, A&A, 409, 523 Schlafly E. F., Finkbeiner D. P. 2011, ApJ, 737, 103 Siegel M. H., Majewski S. R., Reid I. N., Thompson I. B. 2002, ApJ, 578, 151 Skrutskie M. F., Cutri R. M., Stiening R. et al. 2006, AJ, 131, 1163 Stetson P. B. 1987, PASP, 99, 191 Tandon S. N., Subramaniam A., Girish V. et al 2017, AJ, 154, 128 Tandon S. N., Postma J., Joseph P. et al. 2020, AJ, 159, 158 Taylor M. B. 2005, in Shopbell P., Britton M., Ebert R., eds, Astronomical Society of the Pacific Conference Series, Vol. 347, Astronomical Data Analysis Software and Systems XIV, 29 Yaz E., Karaali S. 2010, New Astronomy, 15, 234 Yong D., Carney B. W., Teixera de Almeida M. L. 2005, AJ, 130, 597 York D. G., Adelman J., Anderson John E., J. et al. 2000, AJ, 120, 1579

 Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:59 https://doi.org/10.1007/s12036-021-09715-5

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

The central region of the enigmatic Malin 1 KANAK SAHA1,* , SURAJ DHIWAR2,3, SUDHANSHU BARWAY4,

CHAITRA NARAYAN5 and SHYAM TANDON1 1

Inter-University Centre for Astronomy and Astrophysics, Pune 411 007, India. Department of Physics, Savitribai Phule Pune University, Pune 411 007, India. 3 Dayanand Science College, Barshi Road, Latur 413 512, India. 4 Indian Institute of Astrophysics (IIA), II Block, Koramangala, Bengaluru 560 034, India. 5 National Centre for Radio Astrophysics-TIFR, Pune 411 007, India. *Corresponding Author. E-mail: [email protected] 2

MS received 8 November 2020; accepted 17 January 2021 Abstract. Malin 1, being a class of giant low surface galaxies, continues to surprise us even today. The HST/F814W observation has shown that the central region of Malin 1 is more like a normal SB0/a galaxy, while the rest of the disk has the characteristic of a low surface brightness system. The AstroSat/UVIT observations suggest scattered recent star formation activity all over the disk, especially along the spiral arms. The central 900 (  14 kpc) region, similar to the size of the Milky Way’s stellar disk, has a number of far-UV clumps—indicating recent star-formation activity. The high resolution UVIT/F154W image reveals far-UV emission within the bar region (  4 kpc)—suggesting the presence of hot, young stars in the bar. These young stars from the bar region are perhaps responsible for producing the strong emission lines such as Ha, [OII] seen in the SDSS spectra. Malin 1B, a dwarf early-type galaxy, is interacting with the central region and probably responsible for inducing the recent star-formation activity in this galaxy. Keywords. Galaxies: structure—galaxies: evolution—galaxies: interaction.

1. Introduction Since its discovery (Bothun et al. 1987), Malin 1 continues to surprise us. Malin 1 belongs to the class of giant low-surface brightness (GLSB) galaxies and probably, is one of the largest disk galaxies with spiral arms extending up to 130 kpc in radius at the distance of Malin 1 (Moore & Parker 2006; Boissier et al. 2016). Most of the stars in the outer disk of Malin 1 follows an exponentially declining surface brightness profile. The outer disk is so faint, that when extrapolated to the centre, it has a central surface brightness of 25.5 mag arcsec2 in the v-band (Impey & Bothun 1989). Using the deep images in the six photometric bands from the NGVS and GUViCS surveys, Boissier et al. (2016) have shown that the extended lowThis article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

surface brightness disk has a long and low star-formation history. However, the central region has a complex structure—with a bulge and a bar as revealed by the Hubble Space Telescope (HST) observation (Barth 2007). In other words, the morphology of the central region is similar to the high surface brightness (HSB) galaxies with SB0/a morphology. This, in turn, makes Malin 1 a classic example of a hybrid galaxy with central HSB-like and outer LSB-like envelope. The central HSB-like structure is also manifested in terms of the inner steeply rising rotation curve (Lelli et al. 2010). Several questions arise here—is Malin 1 a galaxy in transition from LSB to HSB? What drives such a transformation? In addition to stars, the galaxy contains plenty of neutral hydrogen gas (Pickering et al. 1997) as in typical late-type spirals. Even then, it appears that the galaxy may not be forming stars efficiently as it lacks CO molecules (Braine et al. 2000), although, the non-detection of CO may also be

59

Page 2 of 14

attributed to the low temperature of molecular gas compounded with sub-solar metallicity (Braine et al. 2000). Nevertheless, the central region appears to have recent star formation activity—based on the strong emission lines such as Ha, Hb, [OII], [OIII] from the SDSS central 300 fibre spectra and much recently from a detailed spectroscopic study by Junais and Epinat (2020). Such an activity is probably induced by an external perturber. Indeed, Malin 1 is interacting with another galaxy called Malin 1B which an early type galaxy at a distance of only 900 (14 kpc) from the centre. Recently, Reshetnikov et al. (2010) suggest that Malin 1 is also probably interacting with another galaxy SDSSJ123708.91?142253.2 at a distance of about 358 kpc away. Such an interaction can lead to the formation of the central bar (Miwa & Noguchi 1998; Lang et al. 2014; Łokas et al. 2014). Accretion of smaller galaxies like Malin 1B or giant clumps could enhance star-formation, increase the central concentration—in other words, grow the central bulge (Aguerri et al. 2001; Eliche-Moral et al. 2018). One might expect to see the induced star-formation activity in the UV emission and indeed, this is evident from the far-UV (FUV) and near-UV (NUV) observation of this galaxy by GALEX. However, due to the large 500 PSF, the central 900 region remains practically unresolved in both FUV and NUV filters. To achieve a detailed understanding of the recent star-formation activity and the nature of the central disk, we observed Malin 1 with the Ultra-Violet Imaging Telescope (UVIT) on-board AstroSat, simultaneously in farultraviolet (FUV) and near ultraviolet (NUV) bands. Our analyses show that there is clumpy star-formation in the central disk hosting a bar and S0 morphology. We also find evidence of prominent emission in the FUV band from the bar region of the galaxy. The paper is organized as follows Section 2 describes UVIT observation and data reduction. The treatment we use to correct for dust extinction is briefly described in Section 2.3. In Section 2.4, we compare the UVIT observations with that of GALEX and CFHT. We discuss the color composition of the central bulge and bar region in Section 3 and a detailed decomposition of the optical surface brightness profile in Section 4. In Section 5, we cover the UV surface brightness as well as the HI emission. We discuss in detail on the central bar and the central star formation activity in Section 6. Finally, in Section 7 we summarise and conclude.

J. Astrophys. Astr.

(2021) 42:59

2. UVIT observation and data analysis The giant low-surface brightness galaxy Malin 1 (at a redshift of z ¼ 0:0827) was observed simultaneously ˚ ) and near-UV in the far-UV (F154W; 1250–1750 A ˚ ) filters by the UVIT during (N263M; 2461–2845 A February 2017 (PI: Kanak Saha; proposal id: G06016). The observations were carried out in the photon counting mode (frames taken every 34 millisecond during each orbit) resulting in about  45000 frames accumulation in a typical good dump orbit. The orbitwise L1 dataset was processed using the official L2 pipeline. The pipeline throws away the cosmic-ray affected frames during data reduction process and these frames were excluded in the final science-ready images in FUV and NUV bands and the subsequent calculation of the photometry. The final science-ready images had a total exposure time of texp ¼ 9850 s in ˚ ) and texp ¼ 9600 s in F154W (bandwidth ¼ 380 A ˚ N263M (bandwidth ¼ 275 A). Astrometric correction was performed using the GALEX and CFHT u band images as the reference images. We have used an IDL program which takes an input set of matched xpixel/ypixel from UVIT images (F154W and N263M) and RA/Dec from GALEX and CFHT u band images. We then perform a TANGENTPlane astrometric plate solution similar to ccmap task of IRAF (Tody 1993). The astrometric accuracy in NUV was found to be  0:1500 while for FUV, the RMS was found to be  0:2200 , approximately half a (sub) pixel size. Note that the absolute astrometric accuracy for GALEX based on QSOs data is about 0:500 (Morrissey et al. 2007). It is also known that there is systematic offset between SDSS and GALEX  0:1–0:300 . The photometric calibration is performed with a white dwarf star Hz4; the photometric zero-points are 17.78 and 18.18 for F154W and N263M respectively (Tandon et al. 2017). Once photometric calibration and astrometric correction are successfully applied, we extract an image of size 60000  60000 as shown in Fig. 1. The morphology and color of the galaxy are discussed in Section 3 We run SExtractor (Bertin & Arnouts 1996) on this cutout image of size 60000 and extract the 3r sources. After removing the 3r sources, we place a large number of random apertures of 7  7 pixel boxes avoiding the location of the extracted sources. This was repeated with boxes of size 11 pixels. We use the mean of the resulting histogram (nearly symmetric) as the background. For F154W, the background is Bf ¼ 9:5  106 cts1 pix1 with a

(2021) 42:59

Page 3 of 14

a

S1

14.340

14.380

14.300

Dec (J2000)

14.340

14.380

F154W

b

14.260

14.260

Dec (J2000)

N263M

S1

59

14.300

J. Astrophys. Astr.

189.320

189.280

189.240

189.200

cc

189.320

189.280

189.240

189.200

60"

RA (J2000)

RA (J2000)

9"

lin

ma

1B

malin 1A

Figure 1. (a, b) AstroSat/UVIT observation of Malin 1. S1 (at a distance of 230:600 ¼ 358 kpc) is believed to be interacting with Malin 1. (c) An RGB colour image of the region marked by red circle—using CHFT g-band (red), AstroSat N263M (green) and F154W (blue). Both N263M and CFHT g-band images are convolved with F154W PSF and resampled to F154W pixel scale.

rf ¼ 3:94  105 cts1 pix1 . In N263M filter, the sky background is Bn ¼ 7:24  105 cts1 pix1 with a rn ¼ 6:48  105 cts1 pix1 . Note that the sextractor background from the FUV image is even lower and it is B ¼ 1:15  108 cts1 pix1 with an rms of 1:9  105 cts1 pix1 . We did not consider this in our subsequent calculation.

for NUV modelling. The SBP reveals long wing outside the core in FUV (see the star’s image in the upper panels of Fig. 2). The FUV star profile is fitted with two components: a Moffat profile for the core, and an exponentially declining profile for the wing as follows:

2.1 PSF modeling

where Im0 and Iw0 denote the central intensity values of the inner Moffat and outer exponential profiles respectively. For the Moffat, the FWHM ¼ pffiffiffiffiffiffiffiffiffiffiffiffi ffi 1 2a 2b  1; where b and a are the free parameters, b determining the spread of the Moffat function. Note the Gaussian function is a limiting case of Moffat function (b ! 1). The best fit model was obtained with two components (Moffat and exponential). With the Moffat fit, the estimated PSF obtained is 1:6700 with b ¼ 1:4. We discuss the wing of the PSF separately below.

FUV modelled PSF is shown in Fig. 2. To model the PSF and estimate the FWHM, we have chosen a bright star (RA: 189.2595, Dec:14.2518) right at the bottom of Malin 1 (see Fig. 1). We used IRAF (Tody 1993) to obtain the surface brightness profiles (SBP) of the star in FUV band. The star in NUV is heavily saturated, so we have avoided a discussion of the NUV PSF here. Instead, we have considered a circular symmetric PSF from the recent calibration paper (Tandon et al. 2020)

" IPSF ðrÞ ¼ Im0

 2 #b r r 1þ þIw0 e h0 ; a

ð1Þ

59

Page 4 of 14

J. Astrophys. Astr.

(2021) 42:59

line of sight. The mean EðB  VÞ ¼ 0:038 along the direction of Malin 1. The values of foreground extinction parameter for the FUV and NUV bands are then derived using Ak ¼ kk EðB  VÞ; where kk is obtained following Calzetti extinction law (Calzetti et al. 2000). For F154W, AF154W ¼ 0:39 mag with kF154W ¼ 10:18 while for N263M, AN263M ¼ 0:29 mag with kN263M ¼ 7:57. The nebular colour excess can be estimated using the method of Balmer decrement (Osterbrock and Ferland 2006) assuming case B recombination, temperature T ¼ 104 K and electron density ne ¼ 100 cm3 as   ðHa=HbÞobs ; ð2Þ EðB  VÞ ¼ 1:97 log 2:86

Figure 2. Top panel: Image of the star in F154W filter. Bottom panel: PSF modeled with a Moffat and an exponential profile for the wing.

2.2 PSF wing The wing of the PSF in F154W follows an exponential fall-off. This is also true for the N263M band where the exponential profile extends up to  3500 or  84 pixels (not shown here). In F154W, the wing has been modelled till  1400 or  34 pixels. In F154W, the scale-length of the exponential profile is h0 ¼ 3:500 or 8.4 pixels. Accurate estimation of the faint emission from the galaxy will require a thorough knowledge of the PSF wing which reflects amount of the scattered light in the same wavelength. We found that the wing of the PSF contains about 19% of the total flux while 81% is in the core in accordance with recent UVIT calibration (Tandon et al. 2020).

2.3 Dust correction Both FUV and NUV fluxes are susceptible to both foreground and internal dust extinction. Schlegel et al. (1998) full-sky 100 lm map provides us with the foreground dust reddening EðB  VÞ along a given

where Ha and Hb are the observed line fluxes. Based on the SDSS spectra of the central 300 region, the observed Ha and Hb fluxes are 224:5  5:8  1017 and 67  4:6  1017 erg s1 cm2 respectively. The internal color excess EðB  VÞnebular ¼ 0:1323. In the subsequent calculation, we use the Calzetti relation (Calzetti et al. 2000) for stellar continuum color excess EðB  VÞstar ¼ 0:44  EðB  VÞnebular ¼ 0:058—suggesting low extinction in the central region of the galaxy (Junais & Epinat 2020). This is in-sync with general trend found in LSB galaxies which are often devoid of dust and molecular gas compared to their HSB counterparts (Rahman et al. 2007; Das et al. 2006). In the rest of the calculation, we use these values of extinction for the full galaxy.

2.4 Comparison with GALEX observation The FUV image from GALEX has an exposure time of 3154 s while in UVIT/F154W band, it is 9850 s. Compared to GALEX NUV, our NUV band N263M is ˚ bump in the extincnarrower and avoids the 2175 A tion curve. In Fig. 3, we show a one-to-one comparison between the GALEX and UVIT observation of Malin 1. Rectangular regions marked 1 and 2 along the spiral arms of Malin 1 in the GALEX images are resolved into clumps in the UVIT images. With the UVIT FUV image, it is thus possible to study the young star-forming clumps along the spiral arms. The region 2 in GALEX FUV image resolves into four clear clumps (marked by a, b, c, and d) in UVIT F154W image. We have measured their aperture magnitudes and the corresponding signal-to-noise

J. Astrophys. Astr.

(2021) 42:59

Page 5 of 14

59

Figure 3. A comparison between the GALEX FUV and AstroSat/UVIT FUV image (upper panels). The bottom panels show the NUV images. The radius of the red circle centred on Malin 1 is 6000 and that of the inner green one is 900 . On the UVIT FUV image (top right panel), Malin 1B (RA: 12:36:58.83, Dec: ?14:19:44.11) is marked by the magenta circle ð200 radius). In the region 2 (top right panel), we have marked four clumps a, b, c and d. Boxes are of size 2400  1600 .

ratios. The faintest of the four far-UV clumps ‘b’, has a background subtracted magnitude of 24.27 AB mag with a S/N ¼ 4:2. While the brightest among them is the clump ‘c’ having a magnitude of 23.3 AB mag and S/N ¼ 7:3. One of the clumps marked by cyan circle in region 1 (bottom right panel of Fig. 3) in the N263M band has been detected at S/N ¼ 7:2 and has a magnitude of 23.65 AB mag within a circular aperture of radius 1:600 . Due to the improved image resolution and sensitivity of UVIT, we are able to measure the far-UV flux from Malin 1B , marked by magenta circle, located at a distance of 900 from the centre (see Figures 3 and 4). Within an aperture of radius 200 , Malin 1B has a magnitude of 24.9 AB mag (and 24.5 AB mag after foreground correction) with S/N ¼ 3:3. In comparison, Malin 1B is completely blended in the GALEX FUV as well as in the NUV images (see the left panels of Fig. 3). The inner regions (denoted by green circle with radius ¼ 900 ) in the GALEX images are also not resolved. Whereas in their UVIT counterpart, it is

possible to locate a few clumps in the FUV F154W band and N263M band—indicating recent star-forming activity. A closer inspection of the central region of the F154W image shows that there is elongated structure in FUV; when zoomed into this, we found two clumps aligned to give an impression of a bar-like structure. We discuss this structure in detail in the following section. Interestingly, the bar-like structure is also seen in the N263M (relatively dust-free); although it appears like a elliptical blob. In Fig. 4, we compare the UVIT/FUV morphology with the CFHT u-band image. The better resolution and depth of the CFHT observation (Boissier et al. 2016) makes it possible to compare both regions 1 and 2 in considerable detail. The star-forming far-UV clumps (a, b, c, d) marked in Fig. 3 are clearly visible in the CFHT u, g and i bands. The observation of these clumps in the optical bands, in particular the i-band, alongside the far-UV, indicates that they are older than 100 Myr. In other words, the spiral arms in Malin 1 although appear bluer, are not as young as a few 100

59

Page 6 of 14

J. Astrophys. Astr.

(2021) 42:59

Figure 4. A comparison between the CFHT u-band and AstroSat/UVIT FUV image. The radius of the blue circle centred on Malin 1 is 6000 and that of the inner green one is 900 . The magenta circle marks the location of Malin 1B.

Myr. The combination of optical and far-UV emission suggests that there is a mixture of both young and old stellar populations along the spirals arms, e.g., see regions 1 and 2 as well as the color-map in Fig. 5.

3. Morphology and colormap The bottom panel of Fig. 1 shows a color composite of Malin 1 in CFHT g-band and the two UVIT filters. The spiral arms extend to large radii—it is possible to trace it clearly up to a radius of 6000 ¼ 93 kpc, perhaps even larger radii Boissier et al. (2016). Interestingly, the spiral arms are not wide open, like in NGC 1566

Figure 5. CFHT g  i color map of Malin 1 and its central 900 region. The unit of the color bar is in magnitude. Inset figure highlights the central disk with F154W contour of the bar region overlaid on the g  i color map. A part of the FUV emission appears misaligned with respect to the bar.

(Gouliermis et al. 2017). The CFHT g-band image reveals more of a tightly wound type, although it is believed to be interacting with an external galaxy SDSS J123708.91?142253.2 (Reshetnikov et al. 2010) located at a distance of about 358 kpc away. Apart from the spiral arms, the disk in the outer part is nearly invisible—in other words, even with the deepest CFHT image (Ferrarese et al. 2012), we see just the spiral arms. These spiral arms are asymmetric (Moore & Parker 2006), both in radii and azimuths; the asymmetry is prominent along the North-East and South-West direction (Fig. 1). One can easily trace the F154W light (in blue) along the same direction. However, as one moves inward, the galaxy morphology changes, rather dramatically. There is clearly a brighter component in the central 900 ¼ 14 kpc region. Interestingly, this central region is nearly the same size as that of the stellar disk of the Milky Way. As revealed in the HST observation, this central region harbours a bar (Barth 2007). In Fig. 5, we show the 2D g - i color map of the galaxy using the CFHT deep observation and the FUV–NUV color from AstroSat/UVIT. Globally, the galaxy has a clear color gradient with bluer outskirts, see Boissier et al. (2016). By comparing with Fig. 3, one can easily trace the regions 1 and 2 with bluer g  i color. However, the central 900 region is comparatively redder with g  i  11:5, except the central bulge. The color map reveals the central bar very clearly with g  i ’ 1:29. The central bulge is red with g  i  2—suggesting old stellar population. The comparative bluer g  i color of the bar in Malin 1 suggest triggering of star formation in the bar region given the Malin 1 seems to be interacting with the neighbour Malin 1B (Barway & Saha 2020). The evidence of star-formation activity in the bar region is also clear from the FUV contours

J. Astrophys. Astr.

(2021) 42:59

shown on the optical g  i color map. The FUV–NUV color map of Malin 1 is noisy. There are several pixels with zero count/s in the FUV observation. On a visual inspection, it appears that the overall FUV–NUV ’ 0, although the very central part is redder. In the subsequent analysis, we do not use this FUV–NUV color map to draw any inference.

4. Multi-component decomposition In the following, we obtain the surface brightness profile (SBP) using the IRAF ellipse task on the CFHT g-band image (Fig. 6). We model this SBP using a combination of the standard Sersic profile (Sersic 1968) for the bulge and exponential profile (Freeman 1970) for the stellar disk. We also use the Sersic profile to model the central bar as in Pahwa & Saha (2018). The Sersic ? exponential profiles are convolved with a Gaussian PSF of FWHM 0:700 (in CFHT g-band) and the resulting model is fitted to the observed g-band SBP using profiler (Ciambur 2016).

4.1 Faint exponential envelope The 1D profile decomposition (Fig. 6) shows that there is a very faint low-surface brightness exponential disk whose central surface brightness is l0;g ¼ 26:7 mag arcsec2 and a large scale length h0 ¼ 30:200  47 kpc. The origin of such a faint and diffuse stellar disk is elusive. Could it be due to the interaction with the SDSS galaxy (Reshetnikov et al. 2010) or due to the accretion of Malin 1B, a dwarf early-type

Page 7 of 14

59

galaxy which in-spiraled towards the central region as a result of the dynamical friction? The outer region beyond 2000 has spiral arms as can be seen by two kinks on the radial light profile—one at  2200 and another at  3500 . From the nature of the spiral arms (more or less like a tightly wound), it might be possible to infer that Malin 1B might have excited it during its inward in-spiraling movement. Interestingly, the spiral arms do not extend all the way to the central region; it stops around 8  1000 from the center and this is apparently the high surface brightness region of the galaxy with a morphology of SB0/a. It might be possible that this hotter inner region is acting like a Q-barrier for sustaining the spiral structure in the outer disk (Saha & Elmegreen 2016).

4.2 Intermediate exponential disk The slope of the SBP changes significantly at radii below 2000 . We fit the intermediate region (from about 6–2000 ) with an exponential profile motivated by the nature of the profile in this region. The central surface brightness of this disk is 21:79 mag arcsec2 in g-band and has a scale length of 2:5100 ¼ 3:9 kpc. Considering the B-band l0 ¼ 22:5 mag arcsec2 and the same in r-band as 21 mag arcsec2 as the definition of LSB disk (Pahwa & Saha 2018), we obtain l0 ¼ 21:9 mag arcsec2 in the gband for the definition of LSB disk. According to this and the comparatively larger scale-length, the intermediate disk is also an LSB disk or a marginal LSB disk. The best-fit parameter reveals a compact pseudobulge at the central region with an effective radius re ¼ 1:0800 ¼ 1:6 kpc and Sersic index n ¼ 0:66.

Figure 6. CFHT g band surface brightness profile fitted with a sersic function for the bulge (dashed red line), and two exponential components (dashed purple and blue lines) and spiral arms and non-uniformities fitted by Gaussians (dashed cyan lines). The black solid line shows the best-fit model.

59

Page 8 of 14

However, the central region is better modelled with the higher resolution images from HST as Malin 1 has been observed in the HST/WFPC2/F300W and F814W bands (see Barth 2007). Before we discuss the central region and the bar in detail, it is useful to understand the global UV surface brightness profile (especially the central region) and cold neutral hydrogen gas distribution in the galaxy.

5. UV surface brightness profile and HI Boissier et al. (2016) have performed a detailed modelling of the FUV and NUV surface brightness profiles using GALEX observation. Although they exclude the central 2000 region from their analysis, measurements of FUV and NUV surface brightness levels are shown at 1000 and they are about 27 mag arcsec2 . Whereas at a radius of 2000 from the centre the FUV and NUV surface brightness levels are 28.7 and 28:5 mag arcsec2 respectively. In Fig. 7, we show the surface brightness profile (SBP) of Malin 1 in F154W and N263M bands. Since the galaxy is extremely diffuse in FUV and NUV, we have not run the IRAF ellipse task to derive its SBPs. Instead, the SBPs are derived by placing concentric circular apertures on the galaxy and reading the counts within each aperture. Both the F154W and N263M SBPs are corrected for the foreground dust extinction as mentioned in Section 2.3. It is useful to compare our UV SBPs with that from GALEX. At 10 from the centre, the F154W surface brightness level is

J. Astrophys. Astr.

(2021) 42:59

 27:5 mag arcsec2 and in N263M band, it is about  28:4 mag arcsec2 . At a radius of 2000 , the N263M SB level is  29:5 mag arcsec2 while the FUV SBP stays flat at  28 mag arcsec2 except at the location of the spiral arms. When comparing these with the previous studies (Boissier the central region is better. 2016), we find that the UVIT probes deeper in surface brightness level at 1000 radius while at 2000 the N263M filter goes 1 mag arcsec2 deeper than GALEX NUV. At smaller radii, i.e., within 1000 radius, the surface brightness level rises steeply both in FUV and NUV and this has become possible because of the better image resolution in UVIT. From the UV surface brightness profiles, we find that the FUV  NUV ’ 0 in the central region consistent with previous result (Boissier et al. 2016). However, at radii [ 1000 , the FUV SBP flattens out and FUV  NUV ’ 1— indicating young, bluer stellar population in the outskirts. Since Malin 1 is interacting with two other galaxies—one is in-spiraling, i.e., Malin 1B and another, SDSS J123708.91?142253.2 (Reshetnikov et al. 2010), currently at a larger distance, might be responsible for the excitation of star-formation activity in the overall galaxy and might have elevated the FUV flux in the whole galaxy. This is not so unlikely as the galaxy Malin 1 has a huge reservoir of neutral hydrogen gas. The most validated 21-cm emission from VLA observations of Malin 1 by Pickering et al. (1997) shows an extended HI disk up to  6000 radius (see Fig. 8). The total HI mass of this huge disk was estimated by them to be 6:8  1010 M corresponding to

Figure 7. Surface brightness profiles in near and far-UV bands of AstroSat/UVIT. The galaxy has extremely low surface brightness level outside the central region.

J. Astrophys. Astr.

(2021) 42:59

an integrated flux of 2.5 Jy km/s (with a H0 ¼ 75 km s1 Mpc1 ). They also found a slowly rising rotation curve that flattened to 210 km s1 at  4000 . However, Lelli et al. (2010) note that the slow rise in the central region could well be an effect of beam smearing (which affects only the central parts). On a re-analysis of Pickering et al. (1997) data with a technique that reduces beam smearing, they find that the circular velocity touches 237 km s1 at 800 itself. This implies that Malin 1 has a steeply rising rotation curve, which is a unique feature of an HSB disk. Based on this, Lelli et al. (2010) argue that Malin 1 has an inner early-type HSB disk followed by an extended LSB disk. As mentioned earlier, the outcome from the rotation curve analysis is in accordance with findings from HST observation which describes the central region of Malin 1 as a normal SB0/a galaxy surrounded by a huge LSB disk (Barth 2007). As evident in Fig. 8, the apparent distribution in the extended HI disk is far from centrally concentrated. Pickering et al. (1997) conclude that the HI disk is very strongly warped from the twisted contours seen in their individual channel maps. A strong northern warp (towards us) coupled with the inclination of 38 may well explain the observed non-axisymmetry in HI distribution. Yet, it is unclear whether the warp alone

Figure 8. HI 21 cm map of Malin 1 (reproduced here with permission from T. Pickering). The HI extends up to 6000 (white circle). The blue circle corresponds to the size of Malin 1’s central 900 region. A rough estimate of the HI mass within the blue circle could be MHI ¼ 2:6  109 M . Pickering et al. (1997) find evidence for a strong warp, which may explain all or part of the apparent nonaxisymmetric HI distribution seen here.

Page 9 of 14

59

is sufficient to explain it or the disk is also lopsided to an extent which could be a result of the ongoing tidalinteraction (Yozin & Bekki 2014; van Eymeren et al. 2011). From Fig. 8, it also clear that the central 900 region which harbours a bar has a lopsided HI distribution. This is interesting as well as exciting given the morphology of the central region. Normally, galaxies with S0 morphology are gas-poor (Chamaraux et al. 1986; Haynes et al. 1990; Masters et al. 2012) and if there is a bar, one expects to find HI holes (van Driel & van Woerden 1991; Athanassoula et al. 2013; Newnham et al. 2020). In the following section, we discuss the implication and possible connection between the bar, far-UV emission and the HI observation. 6. Central disk and the bar The high resolution HST image has revealed a central bar (Barth 2007) in the heart of Malin 1 within 900 radius. Here for the sake of completeness, we model the surface brightness profile of the central 900 region observed in HST/F814W band (see Fig. 9). The central region has three components—a bulge, a bar and an exponential disk. We have used profiler (Ciambur 2016) to model the inner surface brightness profile taking into account the HST PSF. The bulge is a compact pseudobulge with sersic index n ¼ 1:4 and re ¼ 0:5600 ¼ 0:87 kpc in the F814W band. The exponential disk has a central surface brightness of l0 ¼ 17:64 mag arcsec2 and a scale-length of 4.8 kpc (in F814W). This disk is clearly a high surface brightness disk like our own Galaxy and our analysis confirms previously findings in this regard (Barth 2007; Lelli et al. 2010; Boissier et al. 2016). The lower panels of Fig. 9 show the radial variation of the ellipticity and position angle (PA) of the central region. The ellipticity has a peak at around 2:600 ð¼ 4 kpc) which is considered here as the radius of the bar (Erwin 2005). Beyond this region, i.e., outside the bar (  300  500 ), the PA changes abruptly and there is a hint of faint spiral arm (on the East-ward direction, see the CFHT u-band image in Fig. 4) which might have been driven by the bar itself (Saha et al. 2010; Saha & Elmegreen 2016). Otherwise, the inner part resembles S0 galaxies with a bar and pseudo-bulge (Vaghmare et al. 2015, 2018). If we consider the optical color g  r ¼ 1:16 for this 900 central region, we find the mass-to-light ratio is 9.1 following Bell & de Jong (2001) and a stellar mass of M ¼ 8:9  1011 M —this is about 17 times more

59

Page 10 of 14

Figure 9. The top panel shows the HST F814W surface brightness profile fitted with two Sersic components: one for the bulge (dashed orange line), and the other for the bar (dashed green line). The inner exponential profile is indicated by the dashed blue line. Black solid line is the best fit model. The middle and lower panels show the change in ellipticity and the position angle respectively.

massive than the Milky Way’s stellar mass, although the size is similar. Using our rough estimate of the HI mass within this region, we find MHI =M ¼ 0:003— pushing the central region of Malin 1 close to one of the most gas-poor galaxies (Haynes et al. 1990; Masters et al. 2012). Note that the galaxy contains a significant amount of cold HI gas but because of the huge stellar mass (  1:8  1012 M ), the overall gas fraction is only about  4%.

J. Astrophys. Astr.

(2021) 42:59

Figure 10 shows the central morphology of Malin 1 in HST/F814W, CFHT g-band, and UVIT/F154W bands. The bar is most prominent in the HST F814W but not so prominent in the CFHT optical images. However, as we saw earlier, the CFHT g  i color map reveals the bar mostly clearly (see Fig. 5). In Fig. 10, we also show the far-UV emission morphology in the central region alongside other bands. The central region has a few clumps in far-UV suggesting young, recent star-formation. But most of the FUV emission is originating from the bar region. The optical contours and the position angle show that the bar is oriented along the south-east to north-west direction with a position angle of 63 for the optical bar. We also see emission from the bar region in the near-UV N263M band (see the bottom right panel of Fig. 3). The background subtracted farUV flux within the bar region is S ¼ 0:008775 ct/s corresponding to 22.92 AB mag. After the aperture and foreground dust extinction correction using Schlegel et al. (1998), the bar region has 22.64 AB mag in F154W band. In the N263M band, the background subtracted flux within the bar region (same aperture) is 0.018219 ct/s corresponding to 22.53 AB mag which after aperture and foreground correction becomes 22.28 AB mag. Within the same aperture (of 2:600 radius), we have computed the bar magnitudes in CFHT u, g, and i bands and they are 19.65, 17.98 and 16.69 AB mag respectively (all mags are foreground dust corrected). The FUV–NUV (F154W–N263M) color of the bar region is 0.36—suggesting extremely blue and the presence of hot, young O, OB stars. While the g  i ¼ 1:29 color of the bar suggest red and older stellar population. The same is reflected in other UV-optical colours such as FUV  i ¼ 5:7, NUV  g ¼ 4:27— indicating recent star formation activity (Schawinski et al. 2009) in the bar region. The broad-band colours and the central strong emission lines such as Ha and [OII] suggest mixed stellar population with very young stars. Contrary to quenching due to the bar (George et al. 2019), we see rejuvenation of the bar (Barway & Saha 2020) probably induced by the companion galaxies but this remains to be investigated with more detailed IFU-like observation. Overall, it appears that the gigantic Malin 1 has a massive Milky Way size stellar disk at its heart.

6.1 Central star formation activity The central region of Malin 1 shows both absorption and emission lines. The strong absorption lines are

J. Astrophys. Astr.

(2021) 42:59

Page 11 of 14

59

Figure 10. Central morphology of Malin 1—multi-wavelength view from HST I-band to AstroSat/UVIT far-UV. A bar of size 2:600 is visible in the HST/F814W band. The radius of the green circle is 900 (which contains the inner exponential disk). F814W contours are overlayed on the rest of the images. There is FUV emission from the bar region. The brightest pixel in F154W within the bar region is 0.00051 ct/s.

due to metals in the stellar atmospheres of mostly lowluminosity stellar population. It has several absorption lines e.g., Mg, Na, g-band, Ca(H/K)—suggestive of metal-rich environment or old evolved stars. We use the O3N2 relation (Pettini & Pagel 2004) to estimate the gas-phase oxygen abundance. we find the metallicity of the galaxy to be 8.69 which is close to solar metallicity. The emission lines such Ha, [OII], [OIII] suggest recent star formation activity. In fact, there are scattered recent star-formation activity in the entire disk all the way up to about 6000 as revealed by our UVIT image in F154W band, especially along the spiral arms. This is evident upon close inspection of the CFHT g-band and the F154W image—where one can trace spiral arms and FUV clumps; also clear from the g  i color map (Fig. 5). We estimate the central star formation rate using the SDSS spectroscopic data which provides line fluxes for the central 300 region. We use the Kennicutt (1998) relation SFRðM yr1 Þ ¼ 7:93  1042 LHa ðerg s1 Þ;

ð3Þ

to estimate the SFR within central 300 region (which covers the middle part of the bar) from the Balmer corrected Ha flux 336:7  8:7  1017 erg s1 cm2 (see Section 2.3). Then the dust corrected Ha luminosity is 5:74  0:14  1040 erg s1 . The estimated central SFR is found to be 0:45  0:01M yr1 .

6.2 Far-UV SFR Central SFR within 2.6 arcsec is calculated using the FUV magnitude. We correct the observed FUV magnitude of the bar region (see Section 6) for the

internal dust extinction using Balmer decrement method and following Calzetti et al. (2000) dust attenuation law ˚ The dust corrected (both internal at kmean ¼ 1541 A. and foreground) FUV magnitude of the bar region is 22.4 AB mag. The corresponding flux from the ˚ 1 bar region is 5:0  1:6  1017 erg s1 cm2 A (3:9  1:2  1029 erg s1 cm2 Hz1 ) and luminosity ¼ 6:74  2:1  1026 erg s1 Hz1 . Further we estimate the far-UV SFR using the following relation (Kennicutt 1998): SFRðM yr1 Þ ¼ 1:4  1028 LFUV ðerg s1 Hz1 Þ: ð4Þ The estimated FUV SFR within the central 2:600 (i.e., the bar region) is 0:094  0:03M yr1 . In deriving the above relation, Kennicutt (1998) uses the Salpeter Initial Mass Function (IMF) with mass limits of 0.1 to 100M and stellar population models with solar abundance. We also estimate the SFR within 2.6 arcsec using the recent empirical relation by Karachentsev and Kaisina (2013): log SFRðM yr1 Þ ¼ 2:78  0:4  mFUV þ 2 logðDÞ; ð5Þ where D is in Mpc and mFUV refers to the FUV magnitude. The estimated FUV SFR is 0:094  0:008M yr1 in good agreement with that obtained using (Kennicutt 1998). The specific star formation rate sSFR ¼ 1:00  1013 yr1 —placing the central region of the galaxy in the quenched category marked by log sSFR ¼ 11:8 (Barway & Saha 2020). Even if the star formation rate jumps by a factor of 10, the central region of Malin 1 would be classified as

59

Page 12 of 14

quenched. Nevertheless, the detailed process of quenching remains unclear (van den Bosch et al. 2008; Martig et al. 2009; Peng et al. 2015). The strong emission lines from the central region (Junais & Epinat 2020) and our UV observation suggest that there is recent star-formation activity and the presence of hot blue stars. Together, we seem to arrive at a complex scenario of the central region of Malin 1. Whatever might the case, it would be interesting to investigate in detail the central region of this galaxy.

7. Discussion and conclusions Malin 1 is a classic example of a giant low-surface brightness galaxy (Impey & Bothun 1989; Moore & Parker 2006; Boissier et al. 2016) which is currently undergoing tidal interaction with two external galaxies (Reshetnikov et al. 2010). The central region (i.e., within 14 kpc) of Malin 1 has a HSB stellar disk with a compact, central pseudobulge, and a bar. The bar has a radius of 4. kpc. This central region is equivalent to  17 Milky Way size stellar disks put together within this 900 region. The central disk has a few FUV clumps ( [ S=N ¼ 3)—indicating hot, young stars but only a hint of faint spiral structure. So the emerging picture of Malin 1 is that at the heart it, has a star-forming HSB Milky Way size stellar disk surrounded by an intermediate LSB exponential disk extending upto about 30 kpc. Beyond this lies the extremely low-surface brightness stellar envelope. The apparently simpler giant LSB galaxy has a complex layered structure— might hold important clues to the true complexity of galaxy formation yet to be unfolded (Cole et al. 2000; Martin et al. 2019; Kulier et al. 2020). The star-forming bar in Malin 1 is surprising because of the central S0-like morphology. Our UVIT observation shows FUV emission from the bar region - indicating ongoing star formation. In addition, the SDSS central 300 fibre shows strong emission lines such as Ha and [OII] indicating again the most recent (maybe 10 Myr old) star formation activity in the bar; a similar scenario has been observed in NGC 2903 (George et al. 2020). The bar appears to play a dual role in the star-formation activity in galaxies. It can funnel gas inwards (Combes 2004) and ignite starformation activity in the central region of a galaxy and lead to the formation of a pseudobulge (Athanassoula 1992; Kormendy & Kennicutt 2004; Jogee et al. 2005; Lin et al. 2017). On the other hand, a bar can also lead to the star-formation quenching (James et al. 2009;

J. Astrophys. Astr.

(2021) 42:59

James & Percival 2016; Khoperskov et al. 2018; George et al. 2019, 2020). The bar in Malin 1 has induced star-formation, probably due to the ongoing interaction; especially due to the in-spiraling Malin 1B. In addition, the strong lopsidedness of the atomic hydrogen gas might also lead to inward movement of gas due to outward angular momentum transport (Saha & Jog 2014). Our primary conclusions from this work are: • Using the high resolution UVIT observation, we show that there is scattered star formation all over the disk up to about 93 kpc. The FUV emission is clearly seen to follow the large scale spiral arms of Malin 1. • We perform multi-component decomposition of the light profile of Malin 1 in - HST/F814W and CFHT g-band. The CFHT deep image in the gband reveals that Malin 1 has a very faint lowsurface brightness outer disk, and an intermediate LSB disk. Both disks follow exponential profile— the outer one has a scale-length of 47 kpc while the intermediate one has a scale-length of 3.9 kpc. • The central 14 kpc region is modelled using F814W bands. Our decomposition shows that the central region has three components namely a compact pseudobulge, a bar and a HSB exponential stellar disk. • The high resolution UVIT imaging data shows FUV emission from the bar region - indicating ongoing star-formation activity. The FUV-NUV colour of the bar is 0.36—indicating the presence of hot, young stars. Based on the extinction corrected FUV flux of the bar region, we estimate an SFR ¼ 0:09M yr1 . Whereas the SFR based on the central Ha line flux is ¼ 0:45M yr1 . • The better PSF and sensitivity of UVIT have allowed us to detect FUV fluxes within the Malin 1B. The galaxy Malin 1B has a magnitude of 24:9  0:32 AB mag. • The surface brightness profiles in FUV and NUV up to about 77 kpc shows that FUV is flatter than NUV in the outer region—suggesting bluer population and scattered star-formation till the end of the disk.

Acknowledgements The AstroSat/UVIT observation of Malin 1 and the preliminary data analysis was done as part of the

J. Astrophys. Astr.

(2021) 42:59

collaborative program under bilateral grant DST/INT/ South Africa/P-03/2016 of the Indo-South Africa Flagship Program in Astronomy. A good part of this work was initiated during the visit of KS to South African Astronomical Observatory, Cape Town. KS and SB acknowledge the generous financial support under the same. The UVIT project is a collaboration between IIA, IUCAA, TIFR, ISRO from India and CSA from Canada. This publication uses the UVIT data obtained from Indian Space Science Data Centre (ISSDC)of ISRO, where the data for AstroSat mission is archived. We thank the anonymous referee for an insightful comments which improved the quality of the paper. We thank T. Pickering for kindly making the HI data available to us. SD gratefully acknowledges the support from Department of Science and Technology (DST), New Delhi under the INSPIRE faculty Scheme (sanctioned No: DST/INSPIRE/04/ 2015/000108).

References Aguerri J. A. L., Balcells M., Peletier R. F. 2001, Astron. Astrophys., 367, 428 Athanassoula E. 1992, Mon. Not. R. Astron. Soc., 259, 345 Athanassoula E., Machado R. E. G., Rodionov S. A. 2013, Mon. Not. R. Astron. Soc., 429, 1949 Barth A. J. 2007, Astron. J., 133, 1085 Barway S., Saha K. 2020, Mon. Not. R. Astron. Soc., 495, 4548 Bell E. F., de Jong R. S. 2001, Astrophys. J., 550, 212 Bertin E., Arnouts S. 1996, A&AS, 117, 393 Boissier S., Boselli A., Ferrarese L. et al. 2016, Astron. Astrophys., 593, A126 Bothun G. D., Impey C. D., Malin D. F., Mould J. R. 1987, Astron. J., 94, 23 Braine J., Herpin F., Radford S. J. E. 2000, Astron. Astrophys., 358, 494 Calzetti D., Armus L., Bohlin R. C. et al. 2000, Astrophys. J., 533, 682 Chamaraux P., Balkowski C., Fontanelli P. 1986, Astron. Astrophys., 165, 15 Ciambur B. C. 2016, Pub. Astron. Soc. Aus., 33, e062 Cole S., Lacey C. G., Baugh C. M., Frenk C. S. 2000, Mon. Not. R. Astron. Soc., 319, 168 Combes F. 2004, IAUS, 222, 383 Das M., O’Neil K., Vogel S. N., McGaugh S. 2006, Astrophys. J., 651, 853 Eliche-Moral M. C., Rodrı´guez-Pe´rez C., Borlaff A., Querejeta M., Tapia T. 2018, Astron. Astrophys., 617, A113 Erwin P. 2005, Mon. Not. R. Astron. Soc., 364, 283 Ferrarese L., Coˆte´ P., Cuillandre J.-C. et al. 2012, Astrophys. J. Suppl., 200, 4

Page 13 of 14

59

Freeman K. C. 1970, Astrophys. J., 160, 811 George K., Joseph P., Mondal C. et al. 2019, Astron. Astrophys., 621, L4 George K., Joseph P., Mondal C. et al. 2020, Astron. Astrophys., 644A, 79 Gouliermis D. A., Elmegreen B. G., Elmegreen D. M. et al. 2017, Mon. Not. R. Astron. Soc., 468, 509 Haynes M. P., Herter T., Barton A. S., Benensohn J. S. 1990, Astron. J., 99, 1740 Impey C., Bothun G. 1989, Astrophys. J., 341, 89 James P. A., Bretherton C. F., Knapen J. H. 2009, Astron. Astrophys., 501, 207 James P. A., Percival S. M. 2016, Mon. Not. R. Astron. Soc., 457, 917 Jogee S., Scoville N., Kenney J. D. P. 2005, Astrophys. J., 630, 837 Junais, Boissier S., Epinat B. et al. 2020, Astron. Astrophys., 637, A21 Karachentsev I. D., Kaisina E. I. 2013, Astron. J., 146, 46 Kennicutt R. C. Jr. 1998, Astrophys. J., 498, 541 Khoperskov S., Haywood M., Di Matteo P., Lehnert M. D., Combes F. 2018, Astron. Astrophys., 609, A60 Kormendy J., Kennicutt R C. Jr 2004, Ann. Rev. A&A, 42(1), 603 Kulier A., Galaz G., Padilla N. D., Trayford J. W. 2020, Mon. Not. R. Astron. Soc., 496, 3996 Lang M., Holley-Bockelmann K., Sinha M. 2014, Astrophys. J. Lett., 790, L33 Lelli F., Fraternali F., Sancisi R. 2010, Astron. Astrophys., 516, A11 Lin L., Li C., He Y., Xiao T., Wang E. 2017, Astrophys. J., 838, 105 Łokas E. L., Athanassoula E., Debattista V. P. et al. 2014, Mon. Not. R. Astron. Soc., 445, 1339 Martig M., Bournaud F., Teyssier R., Dekel A. 2009, Astrophys. J., 707, 250 Martin G., Kaviraj S., Laigle C. et al.. 2019, Mon. Not. R. Astron. Soc., 485, 796 Masters K. L., Nichol R. C., Haynes M. P. et al. 2012, Mon. Not. R. Astron. Soc., 424, 2180 Miwa T., Noguchi M. 1998, Astrophys. J., 499, 149 Moore L., Parker Q. A. 2006, Pub. Astron. Soc. Aus., 23, 165 Morrissey P., Conrow T., Barlow T. A. et al. 2007, Astrophys. J. Suppl., 173, 682 Newnham L., Hess K. M., Masters K. L., Kruk S., Penny S. J., Lingard T., Smethurst R. J. 2020, MNRAS, 492(4), 19 Osterbrock D. E., Ferland G. J. 2006, Astrophysics of gaseous nebulae and active galactic nuclei Pahwa I., Saha K. 2018, Mon. Not. R. Astron. Soc., 478, 4657 Peng Y., Maiolino R., Cochrane R. 2015, Nature, 521, 192 Pettini M., Pagel B. E. J. 2004, Mon. Not. R. Astron. Soc., 348, L59 Pickering T. E., Impey C. D., van Gorkom J. H., Bothun G. D. 1997, Astron. J., 114, 1858

59

Page 14 of 14

Rahman N., Howell J. H., Helou G., Mazzarella J. M., Buckalew B. 2007, Astrophys. J., 663, 908 Reshetnikov V. P., Moiseev A. V., Sotnikova N. Y. 2010, Mon. Not. R. Astron. Soc., 406, L90 Saha K., Elmegreen B. 2016, Astrophys. J. Lett., 826, L21 Saha K., Jog C. J. 2014, Mon. Not. R. Astron. Soc., 444, 352 Saha K., Tseng Y.-H., Taam R. E. 2010, Astrophys. J., 721, 1878 Schawinski K., Lintott C., Thomas D. et al. 2009, Mon. Not. R. Astron. Soc., 396, 818 Schlegel D. J., Finkbeiner D. P., Davis M. 1998, Astrophys. J., 500, 525 Sersic J. L. 1968, Atlas de Galaxias Australes Tandon S. N., Subramaniam A., Girish V. et al. 2017, Astron. J., 154, 128

J. Astrophys. Astr.

(2021) 42:59

Tandon S. N., Postma J., Joseph P. et al. 2020, Astron. J., 159, 158 Tody D. 1993, in Hanisch R. J., Brissenden R. J. V., Barnes J., eds, Astronomical Society of the Pacific Conference Series, Vol. 52, Astronomical Data Analysis Software and Systems II, 173 Vaghmare K., Barway S., Mathur S., Kembhavi A. K. 2015, Mon. Not. R. Astron. Soc., 450, 873 Vaghmare K., Barway S., Va¨isa¨nen P. et al. 2018, Mon. Not. R. Astron. Soc., 480, 4931 van den Bosch F. C., Aquino D., Yang X. et al. 2008, Mon. Not. R. Astron. Soc., 387, 79 van Driel W., van Woerden H. 1991, Astron. Astrophys., 243, 71 van Eymeren J., Ju¨tte E., Jog C. J., Stein Y., Dettmar R. J. 2011, Astron. Astrophys., 530, A30 Yozin C., Bekki K. 2014, Mon. Not. R. Astron. Soc., 439, 1948

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:71 https://doi.org/10.1007/s12036-021-09698-3

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

Planetary nebulae with UVIT: A progress report N. KAMESWARA RAO1,* , F. SUTARIA1, J. MURTHY1, A. RAY2,3 and G. PANDEY1 1

Indian Institute of Astrophysics, Bangalore 560 034, India. Tata Institute of Fundamental Research, Colaba, Mumbai 400 005, India. 3 Homi Bhabha Centre for Science Education (TIFR), Mumbai 400 088, India. *Corresponding Author. E-mail: [email protected] 2

MS received 8 November 2020; accepted 7 December 2020 ˚ contains important spectral lines to understand the Abstract. The spectral region between 1250–3000 A morphological structures and evolution of planetary nebulae. This is the region sampled by UVIT through various filter bands both in the continuum and in emission lines (e.g.. [C IV], [He I], [Mg II] etc.). We have mapped several planetary nebulae with different characteristics, ranging in morphology from bipolar to wide and diffuse, and in various states of ionization, comparing the UV with the X-ray morphologies wherever the X-ray images were also available. The major unanticipated discovery with UVIT has been the detection of previously undetected, cold, fluorescent, H2 gas surrounding some planetary nebulae. This may be a possible solution to the missing mass problem. Here we present a review of our studies so far done (both published and on going) with UVIT. Keywords. Stars: AGB and post-AGB—stars: winds, outflows—planetary nebulae: ISM: planetary nebulae: general—planetary nebulae: individual: NGC 6302.

1. Introduction Planetary nebulae (PNs) are splendid remnants of extraordinary deaths of ordinary stars in the mass range of 1–8M . They disburse the nucleosynthetically processed stellar material like carbon and s-process elements into the interstellar medium, thus enriching the matter which forms the next generation of stars. The extensive, slow, stellar wind, moving at speeds of 10 to 15 km s1 , with a mass-loss rate of  107 M yr1 , that starts on the thermally pulsing asymptotic giant branch (AGB) – double shell (He and H) burning sources – transforms into a heavy super-wind with mass-loss rates of  104 M yr1 (Delfosse et al. 1997) as the star evolves to the tip of AGB in the H-R diagram. In a relatively short time most of the mass is lost through a super-wind till the envelope mass falls below 103 –104 M , when a structural change occurs to the star as a degenerate CO oxygen core (which This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

ultimately becomes a white dwarf) develops. The photospheric radius shrinks and the effective temperature Teff starts to increase keeping the luminosity almost constant. Consequently, the mass-loss rate of stellar wind decreases to about 108 M yr1 and the wind speed picks up to 200 to 2000 km s1 . This fast stellar wind plows into the material that was earlier lost through super-wind generating a shock at the interface, while the stellar radiation heats and ionizes the ejecta. The circumstellar material starts to glow as the planetary nebula, and ‘‘illuminates the pages of the book that tells the star’s story’’ (Bianchi 2012). In the interacting stellar wind model (Kwok et al. 1978), it is the interaction of the the high speed stellar wind and the slowly expanding super-wind material that shapes the planetary nebulae. Presence of a companion and or magnetic fields may further alter the morphology of the PN. In general, PNe show very many shapes ranging from spherical to bipolar to multipolar, with some even having chaotic geometries. Morphological studies of these objects reveal their past history of mass ejections, their time scales, kinematics, properties of

71

Page 2 of 20

the ionizing source, wind interactions as well as interactions with interstellar medium etc. The UV region is important for the study of both the central stars (CSPNs) as well as the nebula, because the most important lines of the most abundant elements and ˚ , [C IV] 1550 A ˚, their ionization states like [C II] 1335 A ˚ , [N III] 1760 A ˚ , [C II] 2326 A ˚ etc., fall in [He II] 1640 A this region. These lines are important for modeling the ionization structure, shocked regions, chemical composition etc., and for the estimation of the Teff of the hot CSPNs. Moreover, the interstellar extinction ˚ bump can be studied only in the UV through 2179 A band. The Ultraviolet Imaging Telescope(s) (UVIT) on AstroSat (Singh et al. 2014), with broad and narrow band filters which cover important spectral lines and continuum with an angular resolution of about 100 .5, over a 280 field-of-view, are well-suited for the study of PNs. Details of UVIT are provided in Kumar et al. (2012) and its in-orbit performance is described in Tandon et al. (2017a) and Tandon (2020). UVIT is one of the five payloads on the multi-wavelength Indian astronomical satellite AstroSat that was launched on 2015 September 28. It consists of two 38-cm aperture telescopes, one of which is optimized for FUV, while the other has a dichroic beam splitter that reflects NUV and transmits the optical. Each UV channel can be studied in five broad and narrow band filters, as well as by low resolution transmission gratings. The Visual channel (VIS channel), which operates only in the integration mode, is used for tracking. Our project uses UVIT imaging of X-ray bright and X-ray faint planetary nebulae of different morphologies in various UV emission lines, particu˚ , [C II] 2326 A ˚ , [O II] 2470 A ˚, larly [C IV] 1550 A ˚ , [Mg II] 2800 A ˚ , [He II] 1640 A ˚ etc., [Si IV] 1400 A using various filters of the UVIT–FUV and UVIT– NUV channels. We aim to study the UV morphologies, shocked regions and correspondence of UV and X-ray emissions in PNs, and to that end, several PNs of varied morphological types in both near (NUV) and far (FUV) UV ranges have been observed. Unfortunately, the NUV channel became dysfunctional after 2017. In this paper, we discuss our observations conducted so far of the selected PNs (Table 1), as well as some of the results and surprises that emerged. Detailed studies of individual objects would be presented else where but some salient observational features particularly brought out by UV studies are dealt in the current presentation. Detailed discussion of NGC 40 and NGC 6302 have been presented in Kameswara et al. (2018a, b).

J. Astrophys. Astr.

(2021) 42:71

Table 1 shows a broad morphological classification of the nebulae we observed with UVIT so far which range from compact bipolar nebulae (B) to large elliptical (E) and round (R) nebulae. Some are irregular. Figure 1 illustrates typical nebular emission lines that are enclosed by UVIT filters that were used for our PN studies.

2. Results and discussion 2.1 Bipolar nebulae—FUV halos and arcs Most of the nebulae we have observed so far belong to the group of PNe with bipolar morphology. Our sample includes NGC 40 (the ‘‘bow tie nebula’’), NGC 650, NGC 2440, OH231.8?4.2 (the ‘‘Calabash nebula’’), NGC 2818, Mz 3, NGC 6302 and NGC 7027. One of the reasons we observed them is to look for systematic features (aspects) in the UV that would characterise the group – e.g. the cold circumnebular H2 gas. Detail UVIT imaging studies of NGC 40 and NGC 6302 are to be found in Kameswara et al. (2018a, b), while for NGC 2818 in Kameswara Rao et al., A&A (submitted). Although observations of NGC 650, OH231.8?4.2, Mz3 (and IC 4997) have been done the data is not yet available from ISSDC. The compact low excitation planetary nebula, NGC 40, was the first object we studied with a view to look for correspondence of high excitation UV line regions with Chandra X-ray images. It has been imaged in the far-ultraviolet filters F169M (UVIT/FUV-F3 with ˚ and F172M (UVIT/FUV-F5 with keff keff ¼ 1608 AÞ ˚ of 1717 A), as well as in the near-ultraviolet (UVIT/ NUV) filters N245M (UVIT/NUV-B13) and N279N ˚ ). The filters (UVIT/NUV-N2 with keff of 2792 A ˚ selected would allow imaging in [C IV] 1550 A ˚ (F169M) and [C II] 2326 A (N245M) emission lines, as well as in the continuum (F172M) and (N263). Morphological studies in optical and infrared (IR) show that NGC 40 has ionized high density central core surrounded by faint filamentary halo with circumnebular rings that are seen only in Ha but not in ˚ emission [O III]. UVIT studies show that [C II] 2326 A is confined mostly to the core and shows similar morphology as low excitation lines in optical. How˚ emission is present in the ever, strong [C IV] 1550 A core and shows similar morphology and extent as that of X-ray (0.3–8 keV) emission observed by Chandra, suggesting interaction of the high-speed wind from WC8 central star (CS) with the nebula. An unexpected

00 01 04 07

07 07 07 08 09

09 11 12 13 16

17

20 21 21 22

NGC 40 NGC 650 NGC 1514 A 21

NGC 2440 OH231.8?4.2 JrEr1 A 30 NGC 2818

EGB6 NGC 3587 Lo Tr5 MyCn18 Mz 3

NGC 6302

A 70 NGC 7027 Hu 1-2 NGC 7293

31 07 33 29

13

52 14 55 39 17

41 42 57 46 16

13 42 09 29

m

33.2 01.8 08.3 38.5

44.4

59.0 52.8 33.8 35.1 15.0

54.9 16.9 51.6 53.5 01.5

01.0 19.7 17.0 02.7

s

-07 ?42 ?39 -20

-37

?13 ?55 ?25 -67 -51

-18 -14 -53 ?17 -36

?72 ?51 ?30 ?13

°

05 14 38 50

06

44 02 53 22 59

12 42 25 52 37

31 34 46 14

0

18.0 10.0 09.5 13.6

11.0

34.9 00.0 30.6 51.7 42.2

29.7 50.2 16.9 46.8 37.4

19.1 31.5 33.5 48.4

00

E B Ea E

B

B

R Rsm E

B, M B Es Rs B

B Br R Es

Typea

9 9 9 9

9 9 9 9 9

9 9 9 9

0.16 3.5 0.18 0.21

0.55 0.83 6.32 0.13 0.67

0.61 1.97 1.67 10.25

0.8 0.5 0.67 13.4

9 9 9 9

0.65 0.4 0.3 13.4

1.0 9 4.85

0.16 3.5 0.18 0.21

0.55 0.5 6.3 0.29 0.67

0.61 3.2 1.67 10.25

Size (arcmin)

G06_071 A05_103T06 A07_134T03 A05_149T01 A07_059T05 G05_187 G07_31 A07_134T01

G06_065 A07_059To1 G06_066 G05_178T01 A07_134T02 G07_034 A05_103 G06_061 G06_068 A05_149 A09_047 G08_029 A05_149T03 G05_182 G09_061T02 A05_103

Proposal ID 2016.12.09 2019.10.04 2016.12.26 2016.09.30 2019.11.04 2017.04.04 2019.03.03 2018.01.22 2016.12.26 2018.12.21 2020.06.09 2018.04.04 2018.12.22 2016.06.04 2020.06.06 2019.07.26 2019.06.07 2017.03.18 2019.07.24 2020.05.14 2018.10.05 2019.10.24 2016.09.30 2017.07.24 2019.10.25

Date

F2, F3, F2, F3, F2, F3, F2, F3, F2, F3, F3, F5 F3, F5 F3, F5

F3, F5 F2, F3, F2, F3, F3, F5 F2, F3, F3, F5 F3, F5 F2, F3, F2, F3, F2, F3, F1, F2, F1, F2, F2, F3, F3,F5 F2, F3, F2, F3, F5 F5 F5 F5 F5

F5 F5

F5 F5 F5, Gr F3, F5 F3, F5 F5

F5

F5 F5

FUV filter

B4, B13

B15, Gr, N2

B13, B4

B4, B13, N2, B15

B15, N2, Gr

B4, B13

B4, B13, N2 B4, B13

B4, N2, Gr

NUV filter

n.d. Helix Position 2, n.d. Position 3, n.d.

A&A n.d. n.d.

00 :5  00 :5, n.d. n.d., hour glass n.d.

n.d., ppn Calabash PNG164.8?31.1 Hyd.def A&A(p) (d.n.r)

A&A n.d.

Comments1

The types are taken from IAC Morphological Catalog of Northern Galactic Planetary Nebulae (Manchado et al. 1996b). 1 n.d. data is not available, Level 2 data not received from ISSDC. 2 Galex FUV size is bigger 70 .

a

(2021) 42:71

The morphological classification is as follows: compact bipolar nebulae (B), large elliptical (E) and round (R). Subscripts ‘s’, ‘sm’ and ‘a’ refer to inner structure, multiple shells and ansae respectively. Objects that have been proposed but not observed to date are noted in the last column as ‘n.d.’.

h

Nebula

a, d(2000)

Table 1. Target characteristics, and observation log of the PNs selected for our program.

J. Astrophys. Astr. Page 3 of 20 71

71

Page 4 of 20

J. Astrophys. Astr.

(2021) 42:71

NGC 6302 -Nebula SWP 05211 SWP 30986

C IV

SWP 33379 N IV]

He II

N III] C III]

O IV]

[Ne IV]

O III]

0 1400 NGC LWR LWP LWP

6302 -Nebula 09064 13129 13120

1600

1800

2000

[Ne IV]

O III

Mg II

Figure 1. IUE low resolution nebular spectra of NGC6302 are shown to illustrate the wavelength range of UVIT filters. FUV is plotted on top and NUV at the bottom. Relative effective areas of various UVIT filters used for PN studies and typical nebular emission lines they include are shown.

UVIT discovery is the presence of faint large emission halo in FUV F169M surrounding the central core (Fig. 2, top). This FUV halo is absent in the other filters. This emission halo is unlikely to be due to ˚ emission, or due to dust scattering. [C IV] 1550 A Instead, it most likely is due to UV fluorescence emission from Lyman bands of H2 molecules since a few vib-rotational lines have already been detected in the IR from Spitzer spectra. The FUV halo in NGC 40 highlights the extensive existence of cold H2 molecules in the regions even beyond the optical and IR halos. Thus UV studies are important to estimate the amount of H2 , which is probably the most dominant molecule and significant for mass-loss studies. Central star and the nebular core occur in the north-west edge of the FUV halo in the direction of the star’s proper motion vector suggesting a possible interaction with the surrounding interstellar medium (ISM). Presence of much bigger and more extensive FUV halo was discovered around the famous high excitation PN, NGC 6302 the butterfly nebula (Kameswara et al. 2018b). It has been imaged in F169M and F172M, as well as in N279N and in N219M (UVIT/ ˚ ). NUV-B15 with keff of 2196 A Very detailed Hubble Space Telescope (HST) images have been discussed by Szyszka et al. (2009) who also identified the elusive central star. The optical narrow band images show two main lobes with complicated clumpy small scale structure in the East–West direction separated by a dark lane of a very dense disc

of gas (neutral and molecular) and dust, stretching to North–South. It formed into a toroid, that obscures the central star with visual extinction of about 8 magnitudes (Matsuura et al. 2005; Peretto et al. 2007; Szyszka et al. 2009; Wright et al. 2011). Meaburn et al. (2008) determined the distance to the nebula as 1:17  0:14 kpc from expansion parallax using proper motions of features in the North–West lobe. This estimate seems to be consistent with measurements of proper motions from Hubble images of the eastern lobe (Szyszka et al. 2011). From 3D photoionization modelling of the nebula, Wright et al. (2011) derived the properties of the central star as hydrogen deficient with Teff of 220000 K, log g of 7, L of 14300L and mass of 0.73–0.82M . They also estimated the initial mass to be around 5:5M . Extensive studies of the circumstellar torus from infrared to radio wavelengths (Dinerstein & Lester 1984; Kemper et al. 2002; Matsuura et al. 2005; Peretto et al. 2007; Santander-Garcı´a et al. 2017) suggest the structure is that of a broken disc containing 2.2M of dust and molecular gas expanding at 8 km s1 , presumably ejected from the star some 5000 years ago, over a duration of  2000 years. The torus also obscures both the star and an ionized gas disc (detected in 6-cm free–free continuum) around the star. Kinematical studies of the east and west lobes seem to suggest that an explosive event initiated a kind of Hubble flow (i.e. a flow in which the velocity increases outward in proportion to its distance from

J. Astrophys. Astr.

(2021) 42:71

Page 5 of 20

71

Figure 2. Top: FUV images of NGC 40 in F172M (left) and in F169M (right). The faint FUV halo in F169M, extending beyond the bright central region, is absent in the F172M image (Kameswara et al. 2018a). Bottom: The extensive FUV lobes and jets in NGC 6302 are shown in the F169M image which extends much beyond the optical image. The F172M image (not shown) does not show these lobes and jets.

the star) in both lobes about 2200 years back (Meaburn et al. 2008; Szyszka et al. 2009). The formation and flow of matter probably was directed by the torus into East–West lobes. Our F169M image of this nebula shows faint emission lobes that extend to about 50 on either side of the central source. Faint orthogonal jets are also present on either side of the FUV lobes through the central source (Fig. 2, bottom). These lobes and jets are not present in either of the two NUV filters or in FUV F172M. Optical and IR images of NGC 6302 show brightly emitting bipolar lobes in the East–West direction with a

massive torus of molecular gas. Dust is seen as a dark lane in the North–South direction. FUV lobes are much more extended and oriented at a position angle of 113 . The FUV lobes and jets might be remnants of earlier (binary star) evolution, prior to the dramatic explosive event that triggered the Hubble type bipolar flows about 2200 years back. The source of the FUV lobe and jet emission is not known, but most likely is due to fluorescent emission from H2 molecules. The cause of the difference in orientation of optical and FUV lobes is also indeterminate, although we speculate that it could be related to the binary interactions.

71

Page 6 of 20

2.1.1 NGC 2440. A different kind of FUV halo is seen in the bipolar (multi-polar) PN NGC 2440 in our UVIT observations. The two prominent lobes of bipolar structure prominently seen in the optical images (e.g. HST heritage image) are resolved in the various nebular line filter images in to at least two interlocking, differently oriented, bipolar structures (Lopez et al. 1998). They are oriented at position angles (PA) of 35 and 85 , with a third one at 60 (Lopez et al. 1998). These multipolar structures suggest changes in the direction of sporadic mass outflows from the central object (Lopez et al. 1995; Manchado et al. 1996a). Many molecular emissions and outflows have been mapped. Mapping of H2 (v ¼ 1  0Sð1Þ) rotational transition shows a spiky spherical structure of  7300 diameters (Muthumariappan et al. 2007—personal communication; Wang et al. 2008) with spokes emanating from the centre. CO (3-2) emission closely follows the PA 35 bipolar axis in three clumps extending to about a radius of  3600 from the central clump (Wang et al. 2008). HCN, HCOþ emission is also detected within the nebular diameter of  7100 (Schmidt & Ziury 2016). Cuesta and Phillips (2000a) analysed and modeled the optical nebular line filter images. Ramos-Larios and Phillips (2009) show the Spitzer images at 3.6, 4.8, and 8l infra-red emission extending to  8000 diameter centered on the central source where two bright nebular knots (NK and SK) separated by  6.200 occur and define another axis. Thus the, ionized, molecular gas and dust are all confined within  8000 diameter centered around the centre. Lago and Costa (2016) modeled the morpho-kinematical structure with two bipolar components with PA 35 and 85 . The nebular abundances and the central star properties have been studied recently by Miller et al. (2019). Our UVIT images are obtained in the FUV filters F169M and F172M as well as in the NUV filters N219M, N245M and N263M (Fig. 3). F169M ˚ , [He II] 1640 A ˚ includes the lines of [C IV] 1550 A close to the star whereas F172M mostly displays the ˚ emission continuum and a weakly, the [N III] 1760 A ˚, feature. The images of N245M includes [C II] 2326 A N263M [Mg II] and the continuum. Comparison of UV images with optical nebular lines ground based (Lopez et al. 1998; Cuesta & Phillips 2000b) and HST show that N245M and N263M are very similar to ˚ whereas F172M image is similar to that [N II] 6584 A, of continuum emission (Cuesta & Phillips 2000b) both in size as well as in the presence of the features consistent with low excitation line contribution.

J. Astrophys. Astr.

(2021) 42:71

However, the images obtained with F169M differ from those of the other UV filters and the optical lines. The bright central part of the image shows broadly ˚ image and shows the two similar to the [O III] 5007 A  bipolar systems at PA 35 and 85 (Fig. 4). The orientation of the central knots also is similar. The most intriguing features of the F169M image are the faint halo extending beyond the central bright nebula in the North–East (NE), and to a lesser extent, on the South–West (SW). The faint halo extends beyond the  8000 nebular diameter estimated from dust and molecular emission. It extends to  3800 beyond the bright nebula on the NE side and  1800 on the SW. The axis of this halo seem to coincide with PA 35 nebular axis. In addition, the most interesting is a thin jet that extends to  4400 to the South–West beyond the nebula, parallel to the PA 35 axis. (The bright condensation at the southern end of the jet is a field star. It is also present in the F172M image.) The FUV halo in NGC 2440 is similar in nature to the other two systems NGC 40 and NGC 6302 and most likely caused by the fluorescent emission from cold H2 molecules, excited by the diffuse UV radiation of the hot central star. This cold H2 might be a product of much earlier mass loss from the system when it (or possibly, the primary) was on early AGB phase. The FUV jet might also be an earlier ejection from the binary system. 2.1.2 NGC 2818. The bipolar PN NGC 2818 presents a different kind of FUV emission. Instead of a FUV halo around the PN, NGC 2818 shows FUV emitting circum-nebular arcs at a distance from the nebula, possible remnants of much earlier mass ejections and mass-loss. The PN NGC 2818 is one of the few known PNs that are members of galactic clusters which makes it possible to estimate a lower limit to the original main sequence mass from the cluster turnoff. In the cse of NGC 2818 it is around 2.0 to 2.2M . NGC 2818 is a high excitation nebula with lines of [He II], [C IV], [N V] same time strong lines of low excitation as well as vibrational-rotational lines of H2 in the near and mid-IR. The Spitzer images in mid-IR wavelengths show dust emission extending beyond the optical nebula, particularly on the western lobe (Hora et al. 2006). From the optical spectral analysis the Teff of the central star is estimated to be  169000 K (Mata et al. 2016). Very detailed Hubble Space telescope (HST) images (Hubble Heritage image collection) have been discussed by Vazquez (2012) along with its

J. Astrophys. Astr.

(2021) 42:71

Figure 3. NUV N219M image (left) is compared with UVIT image in F169M (F3) of NGC 2440 (right).

kinematical structure. The kinematical age of the wide lobes is estimated as  8400±3400 years. The optical narrow band images show bipolar lobes in the East– West direction with complicated small scale structure and a pinched, hourglass type narrow equatorial waist in the middle stretching to north-south. The semimajor axis is estimated to extend to 7500 through

Page 7 of 20

71

optical nebular lobes in East–West and a minor axis extending to 5500 North–South with 1400 diameter central region, which is potentially the remnant of an equatorial enhancement A number of cometary knots are seen in images of low excitation lines e.g. [N II], that are preferentially located inside a radius of 2000 around the central star (Vazquez 2012). The bipolar planetary nebula, NGC 2818 and the open cluster have been imaged in three far-ultraviolet (FUV) filters , F154W: keff of 1541, F169M: keff of 1608 and F172M: keff of 1717 with UVIT. The F154W image shows faint emission of a partial nebular ring and couple of nebular arcs (shell) that surround the central nebula at a distance of 37000 and 17000 from the central star (Fig. 5). F169M image also shows traces of these features but not as prominently as in F154W image. But the images in F172M filter, NUV from GALEX and optical filters do not show any trace of these emission features. The FUV emission from partial ring at a distance of 6.4 pc from the star suggests an ejection that took place about 60,000 years back (or more) from the central star. The observed expansion velocity of 105 km s1 of the polar lobes and the distance of 3.56 kpc determined from Gaia parallaxes for both the cluster and the

˚ obtained with Figure 4. Top left panel: Images of NGC 2440 in the filter F169M (left) is shown along with [O III] 5007 A HST. The FUV halo and jet are indicated in UVIT image. Right panel: The same image is shown with the three bipolar orientations (PA 35 (a) and 85 (b)). Bottom left panel: The UVIT image of NGC 2440 in F172M is shown along with ˚ . Bottom right panel: The UVIT image in N265M compared with the HST image in [N II] HST image in [N II] 6584 A ˚ 6584 A is shown.

71

Page 8 of 20

nebula suggests such an age. This is by far the most distant and oldest relic of mass ejection observed for a planetary nebula. The FUV emission in these nebular features is most likely due to UV fluorescent H2 molecules. From the Teff and luminosity of the star it appears that enough stellar UV radiation reaches the nebular arc to produce sufficient H2 fluorescence. The original formation of the shell (or a ring) might have involved shocks or high temperature and high velocity gas, but in due course that gas has recombined and cooled to the present cold molecular gas. The FUV images of the central bipolar nebula show bright emission region dominated by He II k1640 and to lesser extent [C IV] k1550 emission, around the star. Another prominent morphological aspect of FUV emission, particularly seen in F169M image is the presence of radial filaments (Fig. 6) diverging from the central star in almost all directions. These filaments are spread more in the direction of eastern lobe than towards western lobe. They extend to about 4800 into eastern lobe. These filaments are much more prominent in F169M image than in [O III] or F172M images, suggesting that they represent [He II] (and [C IV]) line emitting regions. The filaments have a width of about 0.065 pc at the distance of the nebula.

Figure 5. UVIT/FUV F154W image of NGC 2818 shows a faint, narrow, nebular, partial ring-like feature about 37000 East of the nebula and also two nebular arcs at about 17000 North-West of the nebula (top right and bottom right). These features are absent in F172M (top left) and GALEX NUV (bottom left) images.

J. Astrophys. Astr.

(2021) 42:71

This structure of the filaments is very similar to the radial rays surrounding the main ring in helix nebula as shown by O’Dell et al. (2004) . These radial filaments seem to provide channels for the hot stellar wind to flow. It is amazing that FUV studies could bring out relics of 60000 years past mass-loss from the pre-CSPN star of NGC 2818. 2.1.3 NGC 7027. Although our analysis of this PN is on going, we illustrate here F154W, F169M and F172M filters image (Fig. 7) from our observations and point out some interesting features. NGC 7027 (PN G084.9-03.4) – also known as the ‘‘magic carpet’’ nebula or ‘‘pink pillow’’ nebula – is located at 1 kpc (Zijlstra et al. 2008) and has a kinematical age of just 600 years (based on its radio flux – Masson 1989). It is a compact and young PN, one of the brightest nebulae in the sky and the most extensively studied one. NGC 7027 is a carbon-rich nebula with a very highexcitation spectrum showing lines of [O IV], [Mg V] etc. It hosts one of the hottest central stars known to date, with a Teff  200000 K (Latter et al. 2000). A small, essentially ellipsoidal, expanding ionized shell surrounds the central star (Masson 1989). Further outwards, a thin shell indicates the presence of H2 a photo-dissociation region (PDR) and shows signs of recent interaction with collimated outflows (Cox et al. 2002). The nebula also shows many molecular emission lines, e.g. lines of CO, CHþ , H2 O, and even HeHþ ion have been detected from beyond the PDR (Wesson et al. 2010; Santander-Garcı´a et al. 2017; Gusten et al. 2019). The PN was discovered to be an X-ray source by Kastner et al. (2001), who attributed the X-ray emission to shock heating by a fast wind from the central star impacting the slow wind which the progenitor star ejected while on the asymptotic giant branch (AGB). Zhang et al. (2005) detected in the spectrum of this bright young PN of Raman-scattered [O VI] features at ˚ and 7088 A ˚ pointing to the existence of 6830 A abundant neutral hydrogen around the ionized region FUV is very sensitive to the internal and external extinction. NGC 7027 has high and variable extinction across the nebula that has been mapped by Walton et al. (1988). Montez and Kastner (2018) link the Xray emission to the distribution of extinction across the nebula. UVIT images in FUV would reflect such extinction variations (Fig. 8) that would be explored later. UVIT images do not show presence of FUV halo or arcs as in other bipolar PNs listed earlier although

J. Astrophys. Astr.

(2021) 42:71

Figure 6. UVIT/FUV F169M image of NGC 2818 is ˚ ground-based, image (inset: compared with [O III] 5007 A Vazquez 2012). While several features are in common, a major difference is the prominence of radial streamers (or emission filaments) in the F169M image is marked by green arrows. The bandpass of F169M filter covers mainly [He II] ˚ and [C IV] k1550 A ˚ emission lines. Presence of k1640 A these streamers suggest that they might have been swept up by strong stellar wind from the central star.

Figure 7. FUV F154W, F169M and F172M images of NGC 7027. Note that the South-East part is fainter and affected by extinction. FUV is very sensitive to dust extinction. The differential extinction can be studied from UV images.

deep images in optical (HST) do show faint circular rings around the main nebula. One possible reason is that because of high internal extinction no UV photons from the central star reach to the outer H2 region.

2.2 Round nebulae—FUV halos The members of this group that were observed by UVIT are NGC 1514, A 30 , EGB 6, and NGC 3587 2.2.1 NGC 1514: Shining fluid! Observing NGC 1514 on 1790 November 13, William Herschel called it ‘a most singular phenomenon!’ This PN is unusual because of its very bright central star and large diameter low surface brightness nebulosity. The observation by Herschel is termed as ‘an important

Page 9 of 20

71

event in the history of astronomy’ (Seaton 1980) because of his realisation that ‘we therefore either have a central body which is not a star or have a star which is involved in a shining fluid, of a nature totally unknown to us’ (Seaton 1980). It turns out that there is not one star at the centre but two, a sdO ? A0III. Recent observations show that the binary system has a period of 3306 days and eccentricity of 0.46. The estimates of the mass for the cooler secondary is about 2.3M and the hot primary is of 0.9M (Jones et al. 2017). Ressler et al. (2010) described the morphological structure of NGC 1514 (see their Fig. 1) as a nebula with two shells, inner and outer with diameters of 13200 and 18300 . The inner shell has numerous bubble structures at its edge pushing into the outer shell. The monochromatic optical images of NGC 1514 show that the amorphous appearance of the nebula contains very little nebular emission within about 3000 of the nucleus (Balick 1987). This is also confirmed by the absence of [C III] k1909 emission which is normally the strongest nebular feature in IUE spectra of PNe (Ressler et al. 2010). A pair of infrared bright axisymmetric rings that surround the visible nebula were discovered by Ressler et al. (2010), particularly dominant in 22 lm. Such a structure is not suggested in any of the visible wavelength images which is probably resulted from binary interaction. NGC 1514 is also a X-ray source (Tarafdar & Apparao 1988; Montez et al. 2015). We have obtained UVIT images in F154W, F169M, F172M as well as in N245M, N263M, and N279N. Although our analysis is ongoing and not complete ,we present few images and show the comparison with the optical image (a combination of B, V, R þ I – Ressler et al. 2010) in Fig. 7. The UV emission seem to be mostly to the inner shell. There is UV (nebular) emission within 3000 . The nebular extent in FUV is less than that in near UV and optical. There seems to be streams connecting the central region to the inner shell particularly in F245M. The structure in NUV F245M is very similar to the optical except the bubbles look sharper. 2.2.2 Born-again planetary nebula A30. Abell 30 (PNG208.5?33.2, A30) is archetypal born-again PN of about 2 arc minute diameter. The central star of A30 is believed to have undergone a very late thermal pulse (VLTP) that caused the ejection of hydrogen deficient material, prominently seen in the light of [O III] lines, about 850 years back. The inner parts of the nebula are filled with this material whereas the

71

Page 10 of 20

Figure 8. FUV F154W image (left) is shown along with the X-ray (Kastner et al. 2001) and HST optical image (right). The faint circular rings seen in the optical image are absent in FUV image.

outer rim of the nebula is of H-normal composition and of about 12500 years of age. A30 is also an X-ray source showing both a diffuse source covering the inner few arc seconds covering the hydrogen deficient knots and a point source located on the central star. Born-again planetary nebulae (PNe) are believed to have experienced a VLTP (Iben et al. 1983) while the star was descending the white dwarf cooling track. During this event, the remaining stellar helium envelope reaches the critical mass required to ignite its fusion into carbon and oxygen (e.g., Herwig 2005; Miller Bertolami & Althaus 2006; Lawlor & MacDonald 2006); the sudden increase of pressure leads to the ejection of the newly processed material and, as the stellar envelope expands, its temperature decreases and the star returns in the Hertzsprung– Russell (HR) diagram to the locus of the asymptotic giant branch (AGB) stars. Soon after, the contraction of the envelope will increase the stellar effective temperature, boosting the UV flux, and giving rise to a new fast stellar wind. So far, the only bonafide bornagain PNe are Abell 30 (A 30), Abell 58 (A 58, Nova Aql 1919), Abell 78 (A 78), and V4334 Sgr (the Sakurais object). Among them, A30 and A78 are the more evolved ones, with large limb-brightened, H-rich outer nebulae surrounding the H-poor, irregularshaped structures that harbor the cometary knots in the innermost regions (Jacoby 1979; Hazard et al. 1980; Meaburn & Lo´pez 1996; Meaburn et al. 1998). HST images in the [O III] emission line of the central regions have revealed equatorial rings (ERs) and compact polar outflows (POs) in the central regions of both PNe (Borkowski et al. 1993, 1995). The dynamics are revealing: while the outer nebulae show

J. Astrophys. Astr.

(2021) 42:71

shell-like structures expanding at velocities of 30 to 40 km1 , the H-poor clumps present complex structures, with velocity spikes of 200 km1 (Meaburn & Lo´pez 1996; Chu et al. 1997; Meaburn et al. 1998). The morphology and kinematics of the H-poor knots unveil rich dynamical processes in the nebulae. The material photo-evaporated from the knots by the stellar radiation is swept up downstream by the fast stellar wind, which is otherwise mass loaded and slowed down (Pittard et al. 2005). The interactions are complex, resulting in sophisticated velocity structures, as well as X-ray emitting hot gas (Chu & Ho 1995; Guerrero et al. 2012; Toala´ et al. 2015). To study the correspondence of UV emission with the X-ray emission as well as the hydrogen deficient ejecta, we imaged A30 with UVIT in 3 FUV and 2 NUV filters. Two FUV filters, F154M (F2) and ˚ ) transmit the high excitation F169M (F3; keff 1608 A lines of [He II], [C IV] etc. as shown in Fig. 1). The ˚ ) allows other FUV filter F172M (F5 with keff 1717 A mostly the nebular continuum. The NUV filters ˚ ) and N279N2 (N2 with N219M (B15 with keff 2196 A ˚ keff 2796 A) allow mostly low excitation lines or continuum. The images of the nebula in these filters are shown in Fig. 9. The nebula is most intense in F169M and F154W where [He II] line emission dominates. The UVIT provides a spatial resolution of  1.00 3. In the present work, we contrast the UV images with both X-ray contours as well as ground based [O III] and H-alpha images (Arturo Manchado – personal communication). In the FUV F2, F3 images, the hydrogen deficient nebular knots are not as conspicuous as in the [O III] image (Fig. 8). The FUV F2, F3 show radial streamers, which are quite prominent almost extending from central region (Fig. 10) to the edge of the nebula. They provide the channels for the material photo-evaporated from the knots by the stellar radiation is swept up downstream by the fast stellar wind, which is otherwise mass loaded and slowed down (Pittard et al. 2005). At the edge of the channel where it interacts with the nebular boundary, arch type structures are seen suggesting that the boundary is being pushed out by the flow of stellar wind swept material. X-ray emission has been detected within A30 (e.g., Guerrero et al. 2012; Kastner et al. 2012; Montez et al. 2015). This born-again PN has been studied with ROSAT (PSPC and HRI), Chandra, and XMM-Newton X-ray satellites (Chu & Ho 1995; Chu et al. 1997; Guerrero et al. 2012). Its X-ray emission originates from the CSPN, but there is also diffuse emission spatially coincident with the

J. Astrophys. Astr.

(2021) 42:71

Page 11 of 20

71

Figure 9. Top: PN A30 in various UVIT FUV and NUV filters is shown. Bottom: The UVIT/FUV F169M image of PN A30 (left) is compared with ground-based [O III] image (right). The knots and cometary tail-like filaments in the centre of the [O III] image appear to have counterparts in the F169M image. The [O III] image was supplied by Arturo Manchado (personal commun.).

clover leaf-shaped H-poor structure detected in [O III]. The X-ray emission from both the CSPN and the diffuse extended emission is extremely soft. The XMM/Newton X-ray continuum contours (Chu et al. 1997) are shown in Fig. 11 superposed on F169M image. They are confined mostly to the inner nebular region where the [He II] dominated gas is present. The Chandra X-ray region is displayed in Fig. 11 as an inset displaying the HST image of the inner 1000 radius of the nebula. The X-ray region contains both diffuse emission and knots. 2.2.3 FUV halo. The most surprising result of our UV imaging of A 30 is the presence of a FUV halo in the F154M and F169M filters, extending beyond the known optical and NUV nebular size (Fig. 9). The halo is not present in the F172M image, nor in any of the NUV images or even in the optical images. This situation is similar to that in NGC 40, NGC 6302 and NGC 2440. The FUV emission is very likely, the result of H2 molecular fluorescent emission from the AGB ejecta, from molecules excited by the diffuse UV radiation from the CSPN seeping through to the cold molecular region. In spite of the presence of a very hot central star, with Teff of 115000 K (Toala´ et al. 2015), and an earlier excursion to hot PN stage (born-again), the nebula seems to still possess some unionized molecular gas. Is this gas a survivor of 12000 years of the PN evolution? The FUV halo seem to be distributed asymetrically only on one side of the nebula. Earlier, Dinerstein and Lester (1984)

discovered an infra-red disk inside the nebula. A possible connection of this dust disk to the FUV halo needs to be explored. 2.2.4 NGC 3587—the owl nebula. This is a well studied PN with an angular size of  30 and of symmetrical morphology consisting of triple shell structure with a round double shell which forms the main bright nebula, and a faint outer halo (Chu et al. 1987; Chu et al. 1993; Guerrero et al. 2003). The bright inner shell is about 18200  16800 and resembles the face of the owl with each eye being of  3500 . The outer shell is almost circular with a diameter of  20800 . The outer shell is about 25% larger than innershell. The surface brightness of the outer shell decreases outward in Ha and [O III]. It shows a limb brightening along the PA 15 to 13 and in PA 180 to 230 . The halo is prominent in [O III] images than in Ha or [N II] but is present in all the optical images (Hajian et al. 1997). This behavior is also common to other PN halos and is most likely an effect of the hardening of ionizing radiation (Guerrero & Manchado 1999). At the faintest level the halo is circular with a radius of 35000 (Guerrero et al. 2003) although overall morphology is asymmetrical (Fig. 10). The halo is kinematically independent of the main nebula (Chu et al. 1998). The relative line strength in the halo are also different from the main nebula. The [N II]/Ha ratio of the halo is a factor of 4 higher than the other parts of the main nebula which indicates that the halo is photo-

71

Page 12 of 20

Figure 10. Top left: A30 image in F169M showing radial streamers from CSPN showing channels of streaming stellar wind (green arrows). The pink arrows point to the arches where the channeled flow hit the outer boundary. Right: Ground-based [O III] image superposed by an inset showing the region of Chandra X-ray emission (Guerrero et al. 2012). Bottom: Image of A30 F169M superposed by the XMM Newton X-ray continuum counters.

ionized. The morphology of the halo suggests an interaction with the surrounding interstellar medium and the gas in the halos is ionized by stellar UV radiation leaking through the material of the main nebula (Guerrero & Manchado 1999; Guerrero et al. 2003). IUE spectra obtained 10 .5 away from the central star, almost into the halo, still show weak [He II] 1640 ˚ and [C III] 1909 A. ˚ The halo gas is clearly ionized A but of low density. The hot CSPN with Teff ¼ 104000 K and log g ¼ 7:0 shows weak stellar wind (Werner et al. 2019; Garcia-Diazet al. 2018). A detailed spatio-kinematical study was presented by Garcia-Diaz et al. (2018), who treat the owl kind of nebulae as a seperate class. One of the puzzles presented by the owl is the existance of cavities (the eyes of the owl) and how are they maintained. At a Gaia distance of 880 pc, the halo extends to 0.4 pc from the central star.

J. Astrophys. Astr.

(2021) 42:71

Figure 11. Top: FUV UVIT F169M (left), N245M and N263M (right) images of NGC 1514. The nebula is brighter in NUV N245M than in FUV F169M. Bottom: The UVIT N245M image of NGC 1514 is compared with optical image (bottom right) (Ressler et al. 2010). The UV emission is confined to the inner ring and centered around the central star.

Halo expansion velocity is assessed as about 10 km1 (Guerrero et al. 2003) which suggests an age [40,000 years. The time scales indicate that the main nebula consists of super-wind from AGB phase and the faint halo from an early AGB wind. Our observation with UVIT consists of FUV images in F154W, F169M, and F172M. We coupled these with the NUV image from GALEX (GI6-015012PK148p57d1pp-nd-int), and compare them with image in the optical narrowband H-b and [N II] filters. Images in all filters show the same general features as in the optical – i.e. the two shells, and halo. However, the cavities (the eyes) are not as prominent in the FUV as in optical. Moreover, the NUV image shows a sharp wiggly boundary on the North–East showing the interaction with ISM in the direction of motion. The south western side is more diffuse and faint but more extended than NE. The FUV F169M image possibly dominated by [He II] emission, shows a more diffuse but brighter halo that extends all around the outer shell similar to NUV. The asymmetrical distribution of the halo does suggest interaction with ISM. One of the special features seen in the F169M image is a jet in the North–West cavity (see Fig. 10)

J. Astrophys. Astr.

(2021) 42:71

Page 13 of 20

71

Figure 12. Images of A30 in F154W, F169M and F172M showing the FUV halo around F154 and F169M images and the absence of it around F172M image.

Figure 13. Top: Images of NGC 3587 (Owl Nebula) in F169M and Galex NUV showing the halos around the deep images (marked as (a)). The nebula in F169M and Galex NUV with two shells shown are marked as (b). Ground-based Ha, [N II] and [O III] images (from Garcia-Diaz et al. 2018) are shown for comparison with UV images. The features of F169M and Galex NUV are similar to [O III] and Ha respectively. Bottom: Image of NGC 3587 in F154W (left) and F169M (right). F169M image shows a jet-like feature (shown by the arrow) that is absent in F154W image.

that is not seen either in F154W or F172M or even NUV images. This could be a hot [He II] emitting jet, a real sign of activity in a docile nebula. 2.2.5 EGB6. EGB 6 (PN G221.6?46.4) is a large (130  110 ) and very low-surface-brightness planetary nebula, serendipitously discovered by Bond on POSS prints in 1978. The central star, PG 0950?139 is a very hot DAOZ white dwarf with Teff ¼ 105000  5000 K, logg ¼ 7:4  0:4 (Werner et al. 2018). The

CSPN has an apparent cool dwarf companion shrouded in dust at a separation of 0.00 16, which was initially detected through near infrared excesses (Bond et al. 2016). Initial spectroscopic observations showed the central star has strong [O III] emission. Later Liebert et al. (1989) showed that the strong [O III] nebular lines arise from a compact emission knot (CEK), which is unresolved and appears to coincide with the PNN in ground-based images. However, recent HST observations (Bond et al. 2016) showed

71

Page 14 of 20

J. Astrophys. Astr.

(2021) 42:71

Figure 14. Top: UVIT image of EGB 6 in F148W filter showing the fainter outer nebula and smaller central nebula (shown by the arrow) along with an optical image of EGB 6 in Ha and [N II] (Jacoby & Van De Steene 1995) and an optical colour image of the outer nebula from Don Goldman (astrodonimaging.com). Note that the inner smaller nebula is not present in the two optical images. Bottom: Galex FUV image of EGB 6 (left) is compared with UVIT image in F148W. The inner small nebula is shown by the arrow (right). Note the absence of inner small nebula in the GALEX image.

that even the emission knot is associated with a companion at 0.00 166 away from the CSPN. This corresponds to a projected linear separation of  118 AU, for a nominal distance of about 725 pc. The electron density of the CEK is remarkably high, about 2:2  106 cm3 , according to an emission-line analysis by Dopita and Liebert (1989). Thus, EGB 6 raises several astrophysical puzzles, including how to explain the existence and survival of a compact dense [O III] emission nebula apparently associated with a cool M dwarf, located at least 118 AU from the source of ionizing radiation (Bond et al. 2016). It is to be noted that very weak [O III] k5007 emission attributed to the large PN has been detected serendipitously in the SDSS spectra of two faint galaxies that happen to lie behind EGB 6 (Yuan & Liu 2013). Ackers et al. (1992) listed the relative inten˚ sities of Ha, Hb, and 5007 A. Bond et al. (2016) suggested a scenario in which the EGB 6 nucleus is descended from a wide binary similar to the Mira system, in which a portion of the wind from an AGB star was captured into an accretion disk around a companion star; a remnant of this disk has survived to the present time and is surrounded by gas photoionized by UV radiation from the WD.

Our UVIT observation include imaging EGB 6 in F148W, F154W, F169M and F172M filters. The low surface brightness PN is present in all the filters with about same dimensions and appearance as the Ha and [N II] images (Jacoby & Van De Steene 1995). The south-west part of the nebula is the brightest and even the F148 image shows wavy appearance which coincides with proper motion direction of the star. The most interesting and unique structure that is only present in FUV filters F148W, F154W and F169M is a smaller nebula close to the CSPN extending to about 2.0 4 away from it. It has a bow-like appearance away from the central star (Figures 12 and 13) in the general direction of proper motion of the star. This central nebula is not even present in GALEX image. The existence of this small nebula adds a new puzzle to the already listed ones by Bond et al. (2016).

2.3 Elliptical nebulae The members of this group from Table 1 are A 21, Jr Er 1, LoTr5, A70, Hu 1-2 and NGC 7293. Reduced Level 2 data of LoTr5, Jr Er1, A 70, Hu 1-2 is not

J. Astrophys. Astr.

(2021) 42:71

Figure 15. Top: UVIT image of EGB 6 in F154W filter showing the faint outer nebula and smaller central nebula along with F148W image of the outer nebula. Note that an inner smaller nebula is present in the two FUV images. Bottom: FUV image of EGB 6 in F148W. Note that the bow of the inner small nebula is towards South-West direction.

available. We present the results on thee PNs A21, and NGC 7293 in the following sections. 2.3.1 Abell 21, medusa nebula, A21. A21 (PNG205.1?14.2) is a very wide (68500  53000 ) evolved PN with a very hot white dwarf central star WD0726?133 also known as YM29. The Teff and log g of the CSPN are estimated to be 140000 K and 6.5 respectively. The WD central star appears as a point source superposed on diffuse emission in the MIPS 24 lm image (Chu et al. 2009). The flux density in the MIPS 24 lm band is almost three orders of magnitude higher than the expected photospheric emission. However, WD 0726?133 remains as a puzzle since no companion has been detected (Ciardullo et al. 1999). The IR excess is attributed to a dust disk. The orgin of the dust is unclear whether it is accreted material or remnants of the dusty AGB phase. Whether the star is a binary is also a possibility (Clayton et al. 2014). A21 shares this property with JrEr 1 and NGC 7293. Short exposure UVIT observations have been obtained in F169M and F172M in FUV and N263M and N245 in NUV filters. The central star as expected is conspicuous. The images are very similar to the

Page 15 of 20

71

Figure 16. Top: UVIT image of NGC 7293 in F169M filter (red) showing the outer nebula and the [He II] filament (marked by arrow) connecting the central region (and the star). The top right panel shows the absence of [He II] filament in the N245M image. Bottom: Inner 40 regions of NGC 7293 in F169M (left) and F172M (right) superposed with intensity contours. The [He II] filament is seen in the F169M image (marked by arrow) whereas it is absent in the F172M image.

optical ones. The NUV images show great similarity to Ha image with a large number of filaments (Fig. 14). However the FUV images look more diffuse similar to [O III] k5007 image (Manchado et al. ˚ and 1996a). The F169M (includes [He II] 1640 A ˚ [C IV] 1540 A) image shows a faint filament connecting the central star. However better observations are required. 2.3.2 NGC 7293, helix nebula. Helix nebula (PNG036.1-57.1) is one of the nearest (Gaia distance of 201 pc) and well-studied PN in almost all wavelengths (O’Dell et al. 2004, 2005, 2007; Meaburn et al. 2005, 2008; Hora et al. 2006; Montez et al. 2015; Van De Steene et al. 2015). It is also the largest PN in the sky with a diameter of about 13.0 5. A wide variety of phenomena has been studied in this PN ranging from exotic molecules, to X-rays, number of intriguing structures, from small cometary knots to large-scale arcs, to bipolar outflows, dusty disks, shock fronts etc. The inner helical structure is composed of thousands of cometary

71

Page 16 of 20

J. Astrophys. Astr.

(2021) 42:71

Figure 17. A surface and contour study of NGC 7293 in UVIT/FUV F169 and F172M bands. Top: Surface contours of the central source in F169 (left) and F172M (right) showing distored contours for F169M indicating the possible presence of [He II] bright clouds or clumps around the CSPN. In F172M, the contours are more circular and suggests the presence of CSPN alone. Bottom: The figure shows the volume contours of the central source in F169M ([He II] emission line) and F172M (continuum) filters. The CSPN is seen to be accompanied by three [He II] clumps in F169M whereas the profile in F172M shows CSPN alone.

knots of lowly ionized and molecular gas (O’Dell et al. 2007; Etxaluze et al. 2014). The white dwarf central star WD 2226-210 with a surface temperature of 103600  5500 K (Napiwotzki 1999) ionizes the AGB nebula. Su et al. (2007) also showed the presence of a 35–150 AU diameter debris disk around this central star. The main nebula consists of two rings of highly ionized gas and a faint outer filament. The three-dimensional (3D) structure of the main nebula has been investigated by Zeigler et al. (2013) who noted that the structure of the helix projects as if it were a thick walled barrel composed of red and blue-shifted halves in a bipolar geometry. The barrel axis of the helix is tilted about 10 East and 6 South relative to the line of sight . There are two intriguing aspects that were revealed by infrared studies. Strong emission lines of [O IV] 25.9 lm and [Ne V] 24.3 lm have been detected in the Spitzer spectrum. This source of emission is identified with a point soure centred on the CSPN. The excitation of [O IV] requires photons of 54.9 eV. In such a case, [He II] lines which also needs 54.4 eV for ionization (and recombination) are expected to produce a strong point source centered on the star. Images of [He II] k4686 have not shown such a point source. Secondly it has been known that Chandra X-ray imaging showed a hard X-ray point source, even at subarc second resolution, centered on the CSPN. The source of X-rays is unknown. The surface temperature of CSPN is not hot enough to generate hard X-rays

Figure 18. NUV image of NGC7293 in N245M (left) compared with GALEX NUV image. Image in N263M (right) is again compared with NUV GALEX image of NGC 7293. Note the higher spatial resolution of UVIT images and the radial filaments.

from its photosphere. FUV region contains [He II] ˚ line, which is expected to be about 16 times 1640 A stronger (recombination) than [He II] k4686, occurs in the F169M filter of UVIT. With a view to map the hot gas at higher spatial resolution (1.00 3) we observed the central regions of helix nebula. We observed the central region of helix in the first instance in two FUV filters F169M and F172M and

J. Astrophys. Astr.

(2021) 42:71

Page 17 of 20

71

Figure 19. Top left panel: Images of A21 in Ha (Manchado et al. 1996a) and in N263M showing the similarity of the filamentary structure. Right panel: The FUV images in F169M and F172M. The arrow points to the faint filament that connects to the CSPN. Bottom: UVIT composite image of A21 in three colours: F169M (blue), N263M (green) and N245M (red).

Figure 20. Details of UVIT image of NGC 7293 in the N263M filter. There are some new structures that are not seen in optical images.

two NUV filters N245M and N263M. F172M images expected to provide the continuum emission in contrast to F169M. NUV filters might include mostly the nebular continuum in addition to some low excitation emissions. IUE spectra up to 2 arc minutes away from ˚ line provides stronCSPN show that [He II] 1640 A gest emission in the inner regions. Figure 15 shows the relative comparison of the spatial resolution of UVIT with respect to GALEX both in FUV and NUV. Several nebular knots (cometary) seen in HST optical images (O’Dell et al. 2007) are also present in UV (Figures 16 and 17). F169 images of the central region of helix showed mainly two surprising features in contrast to F172M (and N245M and N263M). The region around CSPN has nebulous clumps surrounding the point source

Figure 21. Details of UVIT image of NGC 7293 in the N263M filter. The fine optical knots detected in HST image (left) (O’Dell et al. 2005) are seen in the UV image in F263M (right) obtained with UVIT. The same knots are shown in both images. Some of the cometary knots are also present in UV images.

71

Page 18 of 20

(Fig. 15). The intensity contours around the central source show few arc seconds extended regions suggesting that the central source is surrounded by clouds of [He II] (Fig. 18). Secondly, there is a nebulous streamer connecting the central region to the outer ring (Fig. 19). The [He II] streamers could possibly be providing mass flow to and from the central source. Could this be the source of accretion to the central WD? Although NGC 7293 is of great astrophysical interest, it is not an easy object to observe because of its low surface brightness and its large angular extent in the sky. The field-of-view of UVIT is adequate to cover major portions of the nebula at any given pointing. We have observed two more locations in NGC 7293 covering the whole nebula including the shocked regions. Some of the data is not yet available. We hope to present a detailed paper later (Figures 20, 21).

3. Concluding remarks In this paper, we have presented UVIT observations of 11 of the 19 proposed objects in our program. This compendium forms an ‘‘atlas’’ of sorts, of deep UV imaging of thise objects, with a spatial resolution of 100 .5. The data for the 8 objects is not yet available. The main theme that has been developed here is existence of the FUV halos. Halos around PN are well known and well studied (Ramos-Larios & Phillips 2009). For example, in NGC 3587, the halo is seen in all wavelengths from UV (Section 2.2.3) to the optical, and consists mainly of ionized gases (Guerrero & Manchado 1999). Extended ionized halo has been found around 60% of the PNs for which proper imaging has been done (Corradi et al. 2003). The halos are thought to be a result of mass-loss at the end of the AGB phase, their edges being the signature of the last thermal pulse (Steffen & Schnberner 2003). In contrast, the UVIT discovered FUV halos, and jets around bipolar nebulae and A 30 are only seen in wavelengths shortward of k1650 not in longer wavelength images. Warner and Lyman bands of H2 start appearing shortward of k1650. Ultraviolet fluoresence spectra of H2 as modeled by France et al. (2005) with IC 63, show strong emission peak at k1608 (keff of F169M filter) and no emission shortward of k1650. UVIT studies have brought out a totally new aspect to the hidden mass of the planetary nebulae namely existence of FUV halos, jets and arcs, mostly due to very cold H2 gas around young, bipolar and even

J. Astrophys. Astr.

(2021) 42:71

some old PNs like born-again PNs as well. Such cold gas could only be seen through UV fluorescence of H2 molecule. Big FUV lobes, and jets much bigger than optical nebule have been detected through FUV studies by UVIT. UV imaging in each case of PNs we studied, revealed a new aspect observationally which reiterates the importance of UV studies. ‘Planetary nebulae are like a box of chocolates, you never know what you are going to get’ – Monic Yourcg.

Acknowledgements We would like to express our sincere thanks to several people who have helped us in carrying out this study of PNs, especially K. Sriram, Arturo Manchado for supplying ground based images of some the PN in nebular lines, Don Goldman for the colour image of EGB 6, and Annapurni Subramaniam for the contour images of NGC 7293. This publication uses the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). The publication uses UVIT data processed by the payload operations centre at IIA. The UVIT is built in collaboration between IIA, IUCAA, TIFR, ISRO and CSA. UVIT and AstroSat observatory development took about two decades before launch. Several people from several agencies were involved in this effort. We would like to thank them all collectively. NKR would like thank Department of Science and Technology for their support through grant SERB/F/2143/2016-17 ‘Aspects in Stellar and Galactic Evolution’. Some of the data presented in this paper (eg. GALEX, IUE, HST) were obtained from the Mikulski Archive for Space Telescopes (MAST). STScI is operated by the Association of Universities for Research in Astronomy Inc., under NASA contract NAS5-26555. Support for MAST for non-HST data is provided by the NASA Office of Space Science via grant NNX09AF08G and by other grants and contracts. AKR acknowledges the support of the Raja Ramanna Fellowship of the Department of Atomic Energy prior to the submission of this work.

References Ackers A., Ochsenbein F., Stenholm B. et al. 1992, Strasbourg-ESO Catalogue of Galactic Planetary Nebulae

J. Astrophys. Astr.

(2021) 42:71

Balick B. 1987, AJ, 94, 671 Balick B., Frank A. 2002, ARAA, 40, 439 Balick B., Frank A., Liu B., Huart-Espinosa H. 2017, ApJ, 843, 208 Bianchi L. 2012, in Manchado A., Stanghellini L., Schonberner D., eds, Planetary Nebulae: An Eye to the Future, Proc. IAU Symp., 283, 45 Bond H. E., Ciardullo R., Esplin T. L., Hawley S. A., Libert J., Munari U. 2016, ApJ, 826, 139 Borkowski K. J., Harrington J. P., Tsvetanov Z. 1995, ApJ, 449, L143 Borkowski K. J., Harrington J. P., Tsvetanov Z., Clegg R. E. S. 1993, ApJ, 415, 47 Clayton G. C., De Marco O., Nordhaus J. et al. 2014, AJ, 147, 142 Chu Y.-H., Chang T. H., Conway G. M. 1997, ApJ, 482, 891 Chu Y.-H., Ho C.-H. 1995, ApJ, 448, 127 Chu Y.-H., Grundl R. A., Conway G. M. 1987, AJ, 116, 1882 Chu Y.-H., Kwitter K. B., Kaler J. B. 1993, AJ, 106, 650 Chu Y.-H., Gruendi R. A., Conway G. M. 1998, AJ, 116, 1882 Chu Y.-H., Su K. Y. L., Bilikov J. et al. 2011, AJ, 142, 75 Chu Y.-H., Gruendl R. A., Guerrero M. A. et al. 2009, AJ, 138, 691 Ciardullo R., Bond H. E., Sipior M. S., Fullton L. K., Zhang C.-Y., Schaefer K. G. 1999, AJ, 118, 488 Corradi R. L. M., Schonberner D., Steffen M., Perinotto M. 2003, MNRAS, 340, 417 Cox P., Huggins P. J., Maillard J.-P. et al. 2002, A&A, 384, 603 Cuesta L., Phillips J. P. 2000a, ApJ, 543, 754 Cuesta L., Phillips J. P. 2000b, AJ, 120, 2661 Delfosse X., Kahane C., Forveille T. 1997, A&A, 320, 249 De Marco O. 2009, PASP, 121, 316 Dinerstein H. L., Lester D. F. 1984, ApJ, 281, 702 Dopita M. A., Liebert J. 1989, ApJ, 347, 910 Etxaluze M., Cernicharo J., Goicoechea J. R. et al. 2014, A&A, 566, 78 Fang X., Guerrero M. A., Marquez-Lugo R. A., Toala J. A. et al. 2014, ApJ, 797, 100 Feibelman W. A. 2001, ApJ, 550, 785 France K., Andersson B.-G., McCandliss S. R., Feldman P. D. 2005, ApJ, 628, 750 Garca-Segura G., Villaver E., Langer N., Yoon S.-C., Manchado A. 2014, ApJ, 783, 74 Garcia-Diaz M. T., Steffen W., Henny W. J. et al. 2018, MNRAS, 479, 3909 Gusten R., Wiesemeyer H., Neufeld D. 2019, Nature, 568, 357 Guerrero M. A., Manchado A. 1999, ApJ, 522, 378 Guerrero M. A., Ruiz N., Hamann W.-R. et al. 2012, ApJ, 755,129 Guerrero M. A., Chu Y.-H., Manchado A., Kwitter K. B. 2003, AJ, 125, 321

Page 19 of 20

71

Hazard C., Terlevich R., Morton D. C., Sargent W. L. W., Ferland G. 1980, Nature, 285, 463 Hajian A. R., Frank A., Balick B., Terzian Y. 1997, ApJ, 477, 226 Herald J. E., Bianchi L. 2011, MNRAS, 417, 2440 Herwig F. 2005, ARA&A, 43, 435 Hora J. L. Latter W. B., Smith H. A., Marengo M. 2006, 652, 426 Hsia C.-H., Chau W., Zhang Y., Kwok S. 2014, 787, 25 Hua C. T., Dopita M. A., Martinis J. 1998, A&AS, 133, 361 Iben I. Jr., Kaler J. B., Truran J. W., Renzini A. 1983, ApJ, 264, 605 Jacoby G. H. 1979, PASP, 91, 754 Jacoby G. H., Van De Steene G. 1995, AJ, 110, 1285 Jones D., Van Winckel H., Aller A., Exter K., De Marco O. 2017, A&A, 600, 9 Kameswara Rao N., Sutaria F., Murthy J., Krishna S., Mohan R., Ray A. 2018, A&A, 609, L1 Kameswara Rao N., De Marco O., Krishna S., Murthy J., Ray A., Sutaria F., Mohan R. 2018, A&A, 620, 138 Kastner J. H., Vrtilek S. D., Soker N. 2001, ApJ, 550, 189 Kastner J. H., Montez R. Jr., Balick B., et al. 2012, AJ, 144, 58 Kemper F., Molster F. J., Jager C., Waters L. B. F. M. 2002, A&A, 394, 679 Kumar A., Ghosh S. K., Hutchings J. et al. 2012, in Proc. SPIE., vol. 8443 Kwitter K. B., Chu Y. H., Downes R. A. 1993, IAUS, 155, 209 Kwok S., Purton C. R., Fitzgerald P. M. 1978, ApJ, 219, L125 Lang D., Hogg D. W., Mierle K., Blanton M., Roweis S., 2010, AJ, 139, 1762 Latter W. B., Kelly D. M., Hora J. L., Deutsch L. K. 1995, ApJS, 100, 159 Lawlor T. M., MacDonald J. 2006, MNRAS, 371, 263 Lester D. F., Dinerstein H. L. 1984, ApJ, 281, L67 Liebert J., Green R., Bond H. E., Holberg J. B. et al. 1989, ApJ, 346, 251 Lopez J. A., Vazquez R., Rodriguez L. F. 1995, ApJ, 455, L63 Lopez J. A., Meaburn J., Bryce M., Holloway A. J. 1998, ApJ, 493, 803 Lago P. J. A., Costa R. D. D. 2016, RmxAA, 52, 329 Latter W. B., Dayal A., Bieging J. H., Meakin C. et al. 2000, ApJ, 539, 783 Manchado A., Stanghellini L., Guerrero M. A. 1996a, ApJ, 466, L95 Manchado A., Guerrero M. A., Stanghellini L., Serra-Ricart M. 1996b, The IAC Morphological Catalog of Northern Galactic Planetary Nebulae, IAC, La Laguna Mata H., Ramos-Larios G., Guerrero M. A. et al. 2016, MNRAS, 459, 841 Matsuura M., Zijlstra A. A., Molster F. J., Waters L. B. F. M., Nomura H., Sahai R., Hoare M. G. 2005, MNRAS, 359, 383

71

Page 20 of 20

Martin C., Hurwitz M., Bowyer S. 1990, 354, 220 Masson C. R. 1989, ApJ, 336, 294 Meaburn J., Lo´pez J. A. 1996, ApJL, 472, L45 Meaburn J., Lopez J. A., Steffen W., Graham M. F., Holloway A. J. 2005, AJ, 2005, 130, 2303 Meaburn J., Lopez J. A., Bryce M., Redman M. P. 1998, A&A, 334, 670 Meaburn J., Lloyd M., Vaytet N. M. H., Lopez J. A. 2008, MNRAS, 385, 269 Miller T. R., Henry R. B. C., Balick B. et al. 2019, MNRAS, 482, 278 Miller Bertolami M. M., Althaus L. G. 2006, RMxAC, 26, 48 Montez R. Jr., Kastner H., Balick B. et al. 2015, ApJ, 800, 8 Montez R., Kastner J. H. 2018, ApJ, 861, 45 Murthy J., Rahna P., Sutaria F. et al. 2017, A&C, 20, 120 Napiwotzki R., 1999, A&A, 350, 101 O’Dell C.R., McCullough P. R., Meixner M. 2004, AJ, 128, 2339 O’Dell C. R., Henney W. J., Ferland G. J. 2005, AJ, 130, 1720 O’Dell C. R., Henney W. J., Ferland G. J. 2007, AJ, 133, 2343 Peretto N., Fuller G. A., Zijlstra A. A., Patel N. A. 2007, A&A, 437, 207 Pittard J. M., Dyson J. E., Falle S. A. E. G., Hartquist T. W. 2005, MNRAS, 361, 1077 Rahana P. T., Murthy J., Safonova M. et al. 2017, MNRAS, 471, 3028 Ramos-Larios G., Phillips J. P. 2009, MNRAS, 400, 575 Ressler M. E., Cohen M., Wachter S. et al. 2010, AJ, 140, 1882 Santander-Garcı´a M., Bujarrabal V., Alcolea J., CastroCarrizo A., Sa´nchez Contreras C., Quintana-Lacaci G., Corradi R. L. M., Neri R. 2017, A&A, 597, 27 Schmidt D. R., Ziury L. M. 2016, ApJ, 817, 175 Seaton M. J. 1980, QJRAS, 21, 229 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, Proc. SPIE, 9144, 91441S

J. Astrophys. Astr.

(2021) 42:71

Steffen M., Schnberner D. 2003, IAUS, 209, 439 Su K. Y. L., Chu Y.-H., Rieke G. H., Huggins P. J. et al. 2007, ApJ, 657L, 41 Szyszka C., Walsh J. R., Zijlstra A. A., Tsamis Y. G. 2009, ApJ, 707, L32 Szyszka C., Zijlstra A. A., Walsh J. R. 2011, MNRAS, 416, 715 Tandon S. N., Hutchings J. B., Gosh S. K. et al. 2017a, JApA, 38, 28 Tandon S. N., Subramaniam A., Girish V. et al. 2017b, AJ, 154, 128 Tandon S. N. et al. 2020, AJ, 159, 158 Tarafdar S. P., Apparao K. M. V. 1988, ApJ, 327, 342 Toala´ J. A., Guerrero M. A., Todt H. et al. 2015, ApJ, 799, 67 Van De Steene G. C., Van Hoof P. A. M., Exter K. M., Barlow M. et al. 2015, A&A, 574, 134 Vazquez R. 2012, ApJ, 751, 116 Walton N. A., Pottasch S. R., Reay N. K. 1988, A&A, 200, 21 Wang M.-Y., Muthumariappan C., Kwok S. 2006, IAUS, 234, 537 Wang M.-Y., Hasegawa T. I., Kwok S. 2008, ApJ, 673, 264 Werner K., Rauch T., Kruk J. W. 2018, A&A, 616, 73 Werner K., Rauch T., Reindl N. 2019, MNRAS, 483, 5291 Wesson R., Cernicharo J., Barlow M. J. et al. 2010, A&A, 518, 144 Witt A. N., Stecher T. P., Boroson T. A., Bohlin R. C. 1989, ApJ., 336, L21 Wright N. J., Barlow M. J., Ercolano B., Rauch T. 2011, MNRAS, 418, 370 Yuan H. B., Liu X. W. 2013, MNRAS, 436, 718 Zeigler N. R., Zack L. N., Woolf N. J., Ziurys L. M. 2013, ApJ, 778, 16 Zhang Y., Liu X.-W., Luo S. G., Pe´quignot D., Barlow M. J. 2005, A&A, 442, 249 Zijlstra A. A., Van Hoof P. A. M., Perley R. A. 2008, ApJ, 681, 1296

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:57 https://doi.org/10.1007/s12036-021-09720-8

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

UVIT observation of Milky Way satellite galaxy Reticulum II DEVIKA K. DIVAKAR1,2,*, SIVARANI THIRUPATHI2

and VIJAYAKUMAR H. DODDAMANI1 1

Department of Physics, Bangalore University, Bangalore 560 056, India. Indian Institute of Astrophysics, Bangalore 560 034, India. *Corresponding Author. E-mail: [email protected] 2

MS received 7 November 2020; accepted 28 January 2021 Abstract. We present the UV photometry of one of the Ultra-Faint Dwarf (UFD) Milky Way (MW) satellite galaxies, Reticulum II, using images acquired with the Ultra-Violet Imaging Telescope (UVIT) onboard AstroSat. Reticulum II is a dark matter dominated, one of the most metal-poor and old (13.5 Gyr) satellite galaxies. High resolution spectroscopy of red giant branch (RGB) stars in the galaxy showed that the stars are metal-poor and highly enhanced in rapid neutron capture process (r-process) elements, indicating operation of r-process at a very early time in the galaxy. Understanding the stellar population of the galaxy will provide insights to the sites of r-process production and also clues to the formation of UFDs. Here, we present UV and optical color magnitude diagrams (CMD), of Reticulum II using samples selected based on Gaia data release-II proper motions and theoretical isochrone fitting. We identified eight members including the four new members detected for the first time in this study which adds to only 24 confirmed members. These new members are bright enough for follow-up high resolution spectroscopic studies that will be very valuable to probe the early chemical history of the galaxy. We identified three blue horizontal branch (BHB) stars and a possible red horizontal branch (RHB) star in Reticulum II. RHB stars are rare among ultra faint dwarf galaxies. This might indicate the presence of more than one epoch of star formation in Reticulum II, contrary to the earlier studies. Keywords. Ultra-Violet Imaging Telescope—AstroSat—Milky Way satellites—Reticulum II—ultra-faint dwarf—color magnitude diagrams—horizontal branch stars.

1. Introduction Milky Way satellite galaxies are some of the oldest and pristine galaxies in the known Universe (Bovill & Ricotti 2009, 2011). They are faint, least massive, and most dark matter-dominated stellar systems, with very little chemical enrichment (Simon & Geha 2007; Kirby et al. 2013). They are believed to be the remnants of the hierarchical merging of substructures that has originated from the primordial density fluctuations to form our Milky Way (Kravtsov et al. 1998; Kormendy & Freeman 1998). These UltraFaint Dwarf (UFD) satellite galaxies open an exceptional window to the early stages of galaxy formation This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

and chemical evolution (Brown et al. 2014; Frebel & Bromm 2012; Ji et al. 2015). Another aspect of UFDs that makes them critical objects to observe and understand, points toward solving the long-standing ‘‘Missing Satellite Problem’’. According to the Cold Dark Matter (CDM) cosmological model prediction, galaxies like our Milky Way should host several hundreds of satellite galaxies. However, the observations indicate only several tens of such galaxies orbiting around the Milky Way (MW), conflicting the prediction (Kauffman et al. 1993; Klypin et al. 1999; Moore et al. 1999). Large area surveys like Sloan Digital Sky Survey (SDSS), Dark Energy Survey (DES), pan-STARSS have helped to discover ultrafaint dwarf satellite galaxies extending their regime to extremely low luminosities and sizes (Willman et al.

57

Page 2 of 8

2005; Zucker et al. 2006; Laevens et al. 2014; Koposov et al. 2015; Bechtol et al. 2015; Kim & Jerjen 2015). Reticulum II (Ret-II) is one of the most dark matter dominated and metal-poor UFDs discovered as part of DES (Koposov et al. 2015; Bechtol et al. 2015; Walker et al. 2015; Simon et al. 2015). The stars in Ret-II show enhanced r-process abundances: [Eu/ Fe]  1.7 (Ji et al. 2016; Roederer et al. (2016)). Metal-poor r-process rich stars in the MW halo are thought to have formed from gas that is enriched by neutron star mergers or a rare class of supernovae. Though neutron star mergers are viable sites that can make the most neutron rich nuclei such as uranium and thorium, they do not make significant contributions during the early times in the galaxy, which is needed to explain the presence of very metal-poor stars below ½Fe=H\  2:0 of Galactic halo (Coˆte´ et al. 2019). UFDs offer a cleaner fossil record compared to MW halo stars, and hence, Ret-II is an interesting target that can provide clues to possible mechanisms of r-process production in early galaxies. Previous studies (e.g. Simon et al. 2015) used photometric pre-selection of isochrone fitting of optical Color Magnitude Diagram (CMD) for spectroscopic followup. They also identified several possible Horizontal Branch (HB) stars in the photometric preselection and could confirm only three HB stars based on radial velocities. Here, we use broadband UV and optical colors as a possible way to identify a cleaner HB sample to study the HB morphology of the galaxy. Gaia proper motions also significantly improve the target selection, as most of the earlier studies were before the availability of Gaia data.

2. Observations and data Reticulum II was observed on 31st November and 1st December 2016 using the Ultra-Violet Imaging Telescope (UVIT). UVIT is the ultra-violet instrument of multi-wavelength satellite AstroSat launched by the Indian Space Research Organisation (ISRO). UVIT is capable of imaging simultaneously in Near-Ultra Violet (NUV: 200-300 nm) and Far-Ultra Violet (FUV: 130-180 nm) wavelengths with a field-of-view (FoV)  28 arc min (Subramaniam et al. 2016; Tandon et al. 2017, 2020). The data was observed in one of the filters in NUV channel (N219M) of UVIT. The observations presented here are carried out as part of the proposal titled ‘‘UVIT View of Stellar Populations in the Milky Way Ultra Faint Dwarf Satellites: A Pilot

J. Astrophys. Astr. (2021)42:57

Figure 1. The UVIT image of Reticulum II in NUVB15 filter.

Study’’, (proposal ID: A02124) with a total exposure time of 29789 s. The mean wavelength of N219M (NUVB15) filter is 219.6 nm with a bandwidth of 27 nm. The UVIT image of Ret-II in N219M filter is shown in Fig. 1. The initial corrections for the flat field and satellite drift on the NUV images were carried out using CCDLAB software (Postma & Leahy 2017). Aperture photometry was performed using IRAF/PHOT task. The final magnitudes in the AB system have been obtained after aperture and saturation correction. The NUV image has several faint sources and are background limited, hence in order to derive accurate magnitudes we use aperture correction (e.g. Suchkov & Casertano 1997) that are derived from several bright sources in the field. Tandon et al. (2017a) explains the details on saturation correction. These observed magnitudes are corrected for extinction using extinction calculators from NASA/IPAC Extragalactic Database (NED) website1. We corrected for extinction on a star by star basis using reddening maps from Schlegel et al. (1998) to derive EðB  VÞ and calculated effective extinction coefficient for N219M filter by considering Cardelli et al. (1989). We have considered a reddening value range EðB  VÞ ¼ 0:017þ0:002 0:004 mag. It is noted that N219M filter in ˚ bump of the extincUVIT overlaps with the 2175 A tion curve.

1

https://ned.ipac.caltech.edu/help/extinction_law_calc.html.

J. Astrophys. Astr. (2021)42:57

Page 3 of 8

57

Figure 2. Spatial distribution and Color Magnitude Diagram of Reticulum II using DES and UVIT. Top left: UVIT matched sources are shown in red circles. Small gray dots are all the stars in the DES 0:5  0:5 field-of-view (FoV). Top right: The color-magnitude distribution of stars within 1 field. Red curve shows BaSTI isochrone for stellar population with s ¼ 13 Gyr and ½Fe=H ¼ 2:5. The best-fit distance modulus is m  M  17:5. The green shaded area shows the selection of sample around the isochrone with g  r  0:1 and g  0:5. Bottom left: Spatial distribution of UVIT matched stars (blue circle). Red ellipse shows the area considered with a radius of most distant member of Ret-II and half-light radius from Koposov et al. (2015). Bottom right: CMD of UVIT matched sources with isochrone and spatial distribution selection. All the stars in DES field are plotted as small gray points.

3. Sample selection 3.1 Isochrone mask We selected sources within 1 FOV from the Y1A1 DES catalog (Flaugher 2010). The UVIT sources retrieved from NUV image are overlaid on DES sources in Fig. 2 (top left panel). We used weightedaverage magnitudes wavg_mag_psf_fg; r; ig_dered which are extinction corrected for plotting the CMD. We tried to fit different Bag of Stellar Tracks and Isochrones (BaSTI) isochrones on the Hess diagram ðg  r; gÞ by visual inspection. We used the metallicity range 1:7\½Fe=H\  3:2 and ages 11 Gyr\ s\13:5 Gyr. Isochrone with ½Fe=H   2:5 and

s  13 Gyr is identified to be the best-fitting isochorne in agreement with Koposov et al. 2015. We selected all the DES sources that fall in a range of g  r\0:1 and magnitude g\  0:5, with that of the isochrone. The selection of area is shown with green shaded region in the top right panel of Fig. 2. Selected sources based on the isochrone fit are cross matched with UVIT sources. We selected the sources within a spatial area with an elliptical profile extending to a radius just outside the far most Ret-II spectroscopic confirmed members (Simon et al. 2015). The parameters of Ret-II considered for this study are given in Table 1. The position angle  71  1 has been considered for Ret-II from Koposov et al. 2015. The DES samples which satisfied both isochrone color cut

57

Page 4 of 8

J. Astrophys. Astr. (2021)42:57

Table 1. Parameters of Reticulum II. Property

Value

RA (J2000) DEC (J2000) Distance (kpc) m  M (mag) Position Angle (deg)

03:35:41 -54:03:00 32 17:5  0:2 71  1

References: Bechtol et al. (2015) and Koposov et al. (2015).

and spatial profile have been matched with UVIT sources as shown as blue open circles in Fig. 2 (bottom left panel). These cross-matched sources are shown in Optical CMD in Fig. 2 (bottom right panel).

3.2 Proper motion cut-off In this section, we apply candidate selection based on proper motion and spatial location within the elliptical area that is shown in Fig. 2. This exploration will provide an independent sample that does not require isochrone fitting and possibly avoid any biases on the age and metallicity of the stellar population in the isochrone based selection. We use the GAIA DR2 catalog (Gaia Collaboration 2018) to select the samples based on their proper motion. The matched UVIT sources are overlaid on GAIA 0:5  0:5 data has been shown in Fig. 3 (top left panel). Simon (2018) determined a weighted average proper motion of RetII stars by considering the GAIA proper motions for all of the known member stars with G\20. The weighted average proper motion of this sample is

reported to be la cos d ¼ 2:39  0:040 mas yr1 , ld ¼ 1:3  0:048 mas yr1 . GAIA DR2 proper motion accuracy for the sources with G  20 is found to be approximately 1 mas yr1 (Lindegren et al. 2018). We selected all the sources with Ret-II proper motion between the range ðla cos d  1; la cos d þ 1Þ and ðld  1; ld þ 1Þ to include all the candidates within the accuracy limits. These selected sources are plotted in proper motion diagram in Fig. 3 (top right panel). The sources within the Ret-II spatial profile has been selected and shown in Fig. 3 (bottom left panel). Optical CMD of the sources selected based on proper motion cut off and spatial distribution constraint are shown in Fig. 3 (bottom right panel). Clearly the selected candidates are not well constrained by the CMD, indicating need for further investigation. In Fig. 4, we further study the nature of these sources using UV CMD and found that except for three sources, other objects are close to the detection limit of the UVIT data. One of the brighter stars that falls above the HB track could be a low mass post-AGB star, however post-AGB stars are not observed in such old galaxies and it is likely to be a foreground binary system. 4. Results and discussion Accurate ages of UFD satellites are very valuable to understand the pristine objects. Due to low stellar density of these galaxies, the entire CMD is not well populated. In order to derive ages, deep photometry reaching the possible fainter magnitudes of main sequence will be needed and this has been possible for some of the nearby dwarf satellites (\50 kpc).

Table 2. List of Ret-II candidates. RA (deg) 53.9161 54.0530 54.0779 54.0549 53.9044 53.7398 53.9586 53.9292 a

DEC (deg)

NUVB15 (mag)

ga (mag)

ra (mag)

Gb (mag)

Teff (K)

log g (cm s)

[Fe/H]

Class

-54.0828 -53.9339 -53.9625 -53.9844 -54.0670 -54.0920 -54.0275 -53.9624

19.89 20.01 20.04 20.46 21.56 21.61 21.58 21.59

18.29 18.03 18.01 17.99 18.54 18.91 19.68 20.42

18.56 18.21 18.17 17.52 18.01 18.38 19.27 20.12

18.48 18.14 18.12 17.63 18.12 18.50 19.37 20.23

8750 8000 8000 5500 5250 4750

2.5 1.5 1.5 1.5 3 5

-2.5 -2.5 -2 -2 -1.5 -2.5

BHB BHB BHB RHB RGB RGB RGB RGB

Known membership Yes Yes Yes Yes

DES magnitudes, b GAIA magnitude. Teff , log g, [Fe/H] are derived by fitting SED using Kurucz model. UVIT magnitudes are not considered in the case of RGB candidates.

J. Astrophys. Astr. (2021)42:57

Isochrone fitting of the main sequence turn-off suffers from age-metallicity degeneracy. Here, we investigated the samples selected through CMD and proper motion selection, and also used UV photometric data from UVIT. UV photometry has a great advantage to suppress the Red Giant Branch (RGB) stars, hence large contamination of the foreground is suppressed and the entire HB sequence can be probed. This allows accurate estimate of the distance of the satellite galaxy. The positions of three HB stars in the CMD matches well for the assumed distance of 32 kpc of Reticulum II. We identified eight candidate members for Ret-II using optical-UV CMD and GAIA proper motion selection and four HB candidates have consistent UVIT colors for their HB membership. Figure 4 shows the identified eight members on

Page 5 of 8

57

optical-UV CMDs of Ret-II. Four of them are already known spectroscopically confirmed members.The UV magnitudes of RGB stars are close to the sensitivity limit of our UVIT N219M data, they are aligned in a horizontal line in the UV-optical CMD instead of lying close to the RGB of the isochrone. We investigated Spectral Energy Distribution (SED) fitting for the candidates and derived their stellar parameters; temperature (T), surface gravity (log g), and metallicity [Fe/H], using Virtual Observatory SED Analyzer (VOSA) tool (Bayo et al. 2008), and they were found to be consistent with their position in the CMD. VOSA is a Virtual Observatory tool that allows an automated generation of the SEDs from a large number of photometric catalogues ranging from the UV to infrared (IR). The photometric data have been

Figure 3. Spatial distribution and Color Magnitude Diagram of Ret-II using GAIA and UVIT. Top left: UVIT matched sources are shown in red circles. Small gray dots are all the stars in the GAIA 0:5  0:5 FoV. Top right: The proper motion diagram of stars within 1 field. Red circles show the stars with average proper motion of Ret-II members from Simon (2018). Bottom left: Spatial distribution of UVIT matched stars (green circle) with proper motion cut-off. Red ellipse shows the area considered with a radius of most distant member of Reticulum II and half-light radius from Koposov et al. (2015a). Bottom right: Color Magnitude Diagram of UVIT matched sources with proper motion cut-off and spatial distribution selection.

57

Page 6 of 8

compared with various collections of theoretical model SEDs to determine the best fit using this tool. We construct and analyse the SEDs for these eight candidates by combining the flux densities in Galaxy Evolution Explorer (GALEX) (FUV, NUV), UVIT (N219M), GAIA (G), Wide-field Infrared Survey Explorer (WISE; Wright et al. 2010) pass bands. In VOSA tool we used Chi-square minimization option to estimate the best fit stellar parameters for each object from the SEDs using Kurucz models (Castelli et al. 1997). The estimated value of temperatures and surface gravities are presented in Table 2. We note here that, the log g values may not be reliable from the SED fitting, since none of the log g sensitive features are present in the photometry data used here, e.g. Balmer jump. We also used a conservative spatial selection, considering the elongated elliptical structure and presence of tidal structures in several UFDs, we might have missed possible candidates in the outer regions of the galaxy.

4.1 Horizontal branch stars in Reticulum II We identified eight possible HB stars, based on the optical CMD (Fig. 2) and spatial location within the elliptical profile extending to a radius just outside the far most Ret-II spectroscopic confirmed members. While, only five of them were identified based on the

J. Astrophys. Astr. (2021)42:57

GAIA proper motion selection, four of them satisfied both the selection criteria. Three of them are blue horizontal branch (BHB) stars. Two of which are already known HBs (Simon et al. 2015). Presence of BHB stars in these galaxies is consistent with their metallicities. We also identified one possible red horizontal branch (RHB) star, which is very rare among UFDs. Unfortunately, the UV magnitudes of the RHB candidate were close to the sensitivity limit of our UVIT data to firmly define the candidacy. Though, RHBs are rare among UFDs, recent studies of Tucana III (Tuc-III) (Li et al. 2018) showed few dozens of RHB stars and few BHB stars. Vivas et al. (2020) also identified a large number of RR Lyrae stars (which are also HB stars) in Tuc-III, another UFD hosting r-process enhanced stars (Marshall et al. 2019). Tuc-III and Ret-II also show possible tidal disruptions. Tuc-III shows clear distorted structure and Ret-II shows very elongated elliptical morphology. The RR-Lyrae stars in Tuc-III were found mainly on the outskirts of the galaxy. In the case of Ret-II the RR-Lyrae searches based on the GAIA data showed no possible detection. Considering the similarity of Ret-II and Tuc-III in age, metallicity, r-process enhancement and elongated structure, the HB star number counts and morphology are quite different. Unlike, globular clusters the HB morphology in dwarf galaxies may not depend on the helium abundance, as these galaxies do not

Figure 4. Optical-UV CMDs. Optical-UV Color Magnitude Diagrams of UVIT matched sources after iscochrone mask, proper motion cut-off and spatial distribution constraints. Red curve is the best-fit BaSTI isochrone. Green open circles show all the selected sources after GAIA proper motion cut off (same as shown in Fig. 2 (bottom right panel)). Cyan triangles are the finally selected UVIT sources after isochrone match and proper motion cut-off.Small grey dots are the matched sources in DES and UVIT fields.

J. Astrophys. Astr. (2021)42:57

show light element anomaly, such as Na–O and Mg– Al anti correlation (Carretta et al. 2010). Deep UV surveys and spectroscopy will enable understanding of possible multiple epoch of star formations in these galaxies.

5. Conclusion Based on optical and UV color magnitude diagram and GAIA proper motion selection we have identified eight likely members of Ret-II UFD and four of them are new members. Three blue horizontal branch stars could be identified both in the UV and optical color magnitude diagram, however a possible red horizontal branch star was close to the detection limit of our data. All these new members are interesting bright targets for spectroscopic follow-up. This will add to handful of high resolution spectroscopic data that are available for Ret-II. The identification of possible RHB star is interesting as these are rare among UFDs. A detailed study of horizontal branch population in UFD will help understand the possible connection and differences between these population in globular clusters and UFDs.

Acknowledgements We thank the referee for the suggestions and comments that helped to improve the quality of the paper. This publication uses the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). This publication uses UVIT data processed by the payload operations centre at IIA. The UVIT is built in collaboration between IIA, IUCAA, TIFR, ISRO and CSA. This project used public archival data from the Dark Energy Survey (DES). This work presents results from the European Space Agency (ESA) space mission Gaia. The Gaia mission website is https:// www.cosmos.esa.int/gaia. The Gaia archive website is https://archives.esac.esa.int/gaia. This publication makes use of VOSA, developed under the Spanish Virtual Observatory project supported by the Spanish MINECO through grant AyA2017-84089. VOSA has been partially updated by using funding from the European Union’s Horizon 2020 Research and Innovation Programme, under Grant Agreement n 776403 (EXOPLANETS-A).

Page 7 of 8

57

References Bayo A., Rodrigo C., Barrado Y Navascue´s D., Solano E., Gutie´rrez R., Morales-Caldero´n M., Allard F. 2008, A&A, 492, 277, https://doi.org/10.1051/0004-6361: 200810395 Bechtol K., Drlica-Wagner A., Balbinot E., Pieres A., Simon J. D., Yanny B., Santiago B. et al. 2015, ApJ, 807, 50, https://doi.org/10.1088/0004-637X/807/1/50 Bovill M. S., Ricotti M. 2009, ApJ, 693, 1859, https://doi. org/10.1088/0004-637X/693/2/1859 Bovill M. S., Ricotti M. 2011, ApJ, 741, 17, https://doi.org/ 10.1088/0004-637X/741/1/17 Brown T. M., Tumlinson J., Geha M., Simon J. D., Vargas L. C., VandenBerg D. A., Kirby E. N. et al. 2014, ApJ, 796, 91, https://doi.org/10.1088/0004637X/796/2/91 Cardelli J. A., Clayton G. C., Mathis J. S. 1989, ApJ, 345, 245, https://doi.org/10.1086/167900 Carretta E., Bragaglia A., Gratton R. G., Recio-Blanco A., Lucatello S., D’Orazi V., Cassisi S. 2010, A&A, 516, A55, https://doi.org/10.1051/0004-6361/200913451 Castelli F., Gratton R. G., Kurucz R. L. 1997, A&A, 318, 84 Coˆte´ B., Eichler M., Arcones A., Hansen C. J., Simonetti P., Frebel A., Fryer C. L. et al. 2019, ApJ, 875, 106, https://doi.org/10.3847/1538-4357/ab10db Flaugher B. 2010, AAS American Astronomical Society, AAS Meeting #215, id.441.29; Bulletin of the American Astronomical Society, Vol. 42, p.406 Frebel A., Bromm V. 2012, ApJ, 759, 115, https://doi.org/ 10.1088/0004-637X/759/2/115 Gaia Collaboration: Brown A. G. A., Vallenari A., Prusti T., de Bruijne J. H. J., Babusiaux C., Bailer-Jones C. A. L. et al. 2018, A&A, 616, A1, https://doi.org/10. 1051/0004-6361/201833051 Ji A. P., Frebel A., Bromm V. 2015, MNRAS, 454, 659, https://doi.org/10.1093/mnras/stv2052 Ji A.P., Frebel A., Chiti A., Simon J.*D. 2016, Natur, 531, 610, https://doi.org/10.1038/nature17425 Kauffmann G., White S. D. M., Guiderdoni B. 1993, MNRAS, 264, 201, https://doi.org/10.1093/mnras/264.1. 201 Kim D., Jerjen H. 2015, ApJL, 808, L39, https://doi.org/10. 1088/2041-8205/808/2/L39 Kirby E. N., Cohen J.G., Guhathakurta P., Cheng L., Bullock J.S., Gallazzi A., 2013, ApJ, 779, 102, https:// doi.org/10.1088/0004-637X/779/2/102 Koposov S. E., Belokurov V., Torrealba G., Evans N. W. 2015, ApJ, 805, 130, https://doi.org/10.1088/0004-637X/ 805/2/130 Klypin A., Kravtsov A. V., Valenzuela O., Prada F., 1999, ApJ, 522, 82, https://doi.org/10.1086/307643 Kormendy J., Freeman K. C. 1998, AAS, 30, 1281 Kravtsov A. V., Klypin A. A., Bullock J. S., Primack J. R. 1998, ApJ, 502, 48, https://doi.org/10.1086/305884

57

Page 8 of 8

Laevens B. P. M., Martin N. F., Sesar B., Bernard E. J., Rix H.-W., Slater C. T., Bell E. F. et al. 2014, ApJL, 786, L3, https://doi.org/10.1088/2041-8205/786/1/L3 Li, T. S., Simon, J. D., Kuehn, K. et al. 2018, ApJ, 866, 22, https://doi.org/10.3847/1538-4357/aadf91 Lindegren L., Herna´ndez J., Bombrun A., Klioner S., Bastian U., Ramos-Lerate M., de Torres A. et al. 2018, A&A, 616, A2, https://doi.org/10.1051/0004-6361/ 201832727 Marshall J. L., Hansen T., Simon J. D. et al. 2019, ApJ, 882, 177, https://doi.org/10.3847/1538-4357/ab3653 Moore B., Ghigna S., Governato F., Lake G., Quinn T., Stadel J., Tozzi P. 1999, ApJL, 524, L19, https://doi.org/ 10.1086/312287 Postma, J. E., Leahy, D. 2017, PASP, 129, 115002 Roederer I. U., Mateo M., Bailey J. I., Song Y., Bell E. F., Crane J. D., Loebman S. et al. 2016, AJ, 151, 82, https:// doi.org/10.3847/0004-6256/151/3/82 Simon J. D., Geha M. 2007, ApJ, 670, 313, https://doi.org/ 10.1086/521816 Schlegel D. J., Finkbeiner D. P., Davis M. 1998, ApJ, 500, 525, https://doi.org/10.1086/305772 Simon J. D., Drlica-Wagner A., Li T. S., Nord B., Geha M., Bechtol K., Balbinot E. et al. 2015, ApJ, 808, 95, https:// doi.org/10.1088/0004-637X/808/1/95 Simon J. D. 2018, ApJ, 863, 89, https://doi.org/10.3847/ 1538-4357/aacdfb Sneden C., Cowan J. J., Gallino R. 2008, ARA&A, 46, 241, https://doi.org/10.1146/annurev.astro.46.060407.145207

J. Astrophys. Astr. (2021)42:57 Subramaniam A., Tandon S. N., Hutchings J., Ghosh S. K., George K., Girish V., Kamath P. U., et al. 2016, SPIE, 9905, 99051F, https://doi.org/10.1117/12.2235271 Suchkov A., Casertano S. 1997, hstc.work, 378 Tandon S. N., Postma J., Joseph P., Devaraj A., Subramaniam A., Barve I. V., George K. et al. 2020, AJ, 159, 158, https://doi.org/10.3847/1538-3881/ab72a3 Tandon S. N., Subramaniam A., Girish V., Postma J., Sankarasubramanian K., Sriram S., Stalin C. S. et al. 2017, AJ, 154, 128, https://doi.org/10.3847/1538-3881/aa8451 Tandon S. N., Hutchings J. B., Ghosh S. K., Subramaniam A., Koshy G., Girish V., Kamath P. U. et al. 2017, J. Astrophys. Astr., 38, 28, https://doi.org/10.1007/s12036017-9445-x Vivas A. K., Martı´nez-Va´zquez C., Walker A. R. 2020, ApJS, 247, 35, https://doi.org/10.3847/1538-4365/ab67c0 Walker M. G., Mateo M., Olszewski E. W., Bailey J. I., Koposov S. E., Belokurov V., Evans N. W. 2015, ApJ, 808, 108, https://doi.org/10.1088/0004-637X/808/2/108 Willman B., Dalcanton J. J., Martinez-Delgado D., West A. A., Blanton M. R., Hogg D. W., Barentine J. C. et al. 2005, ApJL, 626, L85, https://doi.org/10.1086/431760 Wright E. L., Eisenhardt P. R. M., Mainzer A. K., Ressler M. E., Cutri R. M., Jarrett T., Kirkpatrick J. D. et al. 2010, AJ, 140, 1868, https://doi.org/10.1088/0004-6256/ 140/6/1868 Zucker D. B., Belokurov V., Evans N. W., Kleyna J. T., Irwin M. J., Wilkinson M. I., Fellhauer M. et al. 2006, ApJL, 650, L41, https://doi.org/10.1086/508628

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:46 https://doi.org/10.1007/s12036-021-09695-6

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

A pair of UV nuclei or a compact star-forming region near the active nucleus in Mrk 766? P. P. DEKA1, G. C. DEWANGAN1,* , K. P. SINGH2 and J. POSTMA3 1

Inter-University Centre for Astronomy and Astrophysics (IUCAA), SPPU Campus, Pune 411 007, India. Indian Institute of Science Education and Research Mohali, Knowledge City, Sector 81, Manauli P.O., SAS Nagar, Mohali 140 306, India. 3 Department of Physics and Astronomy, University of Calgary, 2500 University Dr. NW, Calgary AB T2N 1N4, Canada. *Corresponding Author. E-mail: [email protected] 2

MS received 7 November 2020; accepted 16 December 2020 Abstract. We report the discovery of a bright, compact ultraviolet source at a projected separation of 1.1 kpc from the known active galactic nucleus (AGN) in Mrk 766 based on AstroSat/UVIT observations. We perform radial profile analysis and derive the UV flux almost free from the nearby contaminating sources. The new source is about 2.5 and 5.6 times fainter than the AGN in the far and near UV bands. The two sources appear as a pair of nuclei in Mrk 766. We investigate the nature of the new source based on the UV flux ratio, X-ray and optical emission. The new source is highly unlikely to be another accreting supermassive black hole in Mrk 766 as it lacks X-ray emission. We find that the UV/optical flux of the new source measured at four different bands closely follow the shape of the template spectrum of starburst galaxies. This strongly suggests that the new source is a compact star-forming region. Keywords. Galaxies—active galactic nuclei—star formation.

1. Introduction Presence of dual or multiple compact and luminous sources in the central regions of galaxies are rare. Mergers of galaxies can naturally lead to the formation of dual and/or multiple compact sources such as those observed in ultra-luminous infrared galaxies. Simulations of mergers of gas-rich disk galaxies show that very massive, compact and highly luminous star clusters can form from the strongly disturbed gas disks. Consist of young stars, these clusters appear as several bright cores in the central kilo-parsec region of galaxies (Matsui et al. 2012). Another popular interpretation of dual compact sources is the presence of double accreting supermassive black holes in the central regions. It is now widely accepted that almost every galaxy has a This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

supermassive black hole (SMBH) at its center whose mass can be estimated through various techniques like central stellar velocity dispersion observations, reverberation mapping observations, etc. (Schneider 2006). During the merging of galaxies, their respective SMBHs are also expected to come closer due to gravitational attraction and finally coalesce. The whole process can be classified into three phases (Merritt & Milosavljevic 2005): (1) After the two galaxies merge, the two SMBHs move towards the center of the newly formed galaxy and form a binary pair by losing their angular momentum through dynamical friction. (2) The orbit is further hardened by slingshot ejection of stars whose orbits crosses the binary by the process of three body interaction and thereby moving the two SMBHs closer to each other. (3) When the separation between the two SMBHs is small enough such that emission of gravitational

46

Page 2 of 10

waves can dominate the other forms of energy loss, the two black holes merge. There are many key questions to answer, e.g., till when the two SMBHs retain their individual accretion disks and when they start sharing a common accretion disk, at what rate the accretion takes place and the rate of growth of the individual SMBHs etc. Also, after merging, the anisotropically emitted gravitational wave gives a kick velocity to the final merged black hole due to which it gets ejected from the center (Komossa 2012). For sufficient kick velocities, the merged black hole can even get completely ejected from the host galaxy, though its probability is very small (Komossa 2012). Observing multiple luminous, compact sources in the nuclear regions of galaxies and finding their nature is crucial to understand galaxy evolution and mergers of supermassive black holes. The exceptional high spatial resolution of the Chandra X-ray telescope and the Hubble Space Telescope (HST) have led to the first discovery of dual nuclei in galaxies (Komossa et al. 2002; Junkkarinen et al. 2001; Ballo et al. 2004). A number of other techniques have also been used (see Komossa & Zensus 2014 for a review). Here we used high resolution of AstroSat/UVIT and discovered a compact, bright UV source near the well known active nucleus in Mrk 766. In Section 2, we present AstroSat observations and the reduction of UVIT data. In Section 3, we perform spatial analysis of UVIT images and investigate the nature of the new UV source in Section 4, and summarize our findings in Section 5.

2. AstroSat/UVIT observations and the data reduction We observed Mrk 766 with AstroSat as a part of the SXT Guaranteed programme during 4–6 February 2017 with the SXT as the primary instrument for an exposure time of 50 ks. Here we present the UVIT data only. We used the broadband filters FUV/BaF2 (F154W) and NUV/Silica (N242W) and acquired photon counting data. We obtained the Level 1 data from the AstroSat data archive1 and we processed them using the UVIT pipeline CCDLAB (Postma & Leahy 2017). We generated cleaned images for each orbit, aligned them and created merged image for each filter. This resulted in net exposure time of 33.4 ks (NUV/Silica) and 27 ks (FUV/BaF2 ). We derived the astrometric solution transforming the image 1

https://astrobrowse.issdc.gov.in/astro_archive/archive/Home.jsp.

J. Astrophys. Astr. (2021)42:46

coordinates to the world coordinates using the astrometry.net package (Lang et al. 2010). We show the NUV and FUV images of Mrk 766 in Fig. 1. In order to show the relative intensities of the two UV sources, we created 2D histogram views of the NUV image in two different scales, namely linear and square root. We show the histograms in Fig. 2. To identify the position of known active galactic nucleus in Mrk 766 and to look for X-ray counter part of the new UV source, we generated a composite three-color image using FUV/BaF2 , NUV/Silica and Chandra X-ray images. We obtained the processed and cleaned Chandra/ACIS event data (obsID:1597) from the HEASARC archive.2 The composite threecolor image of Mrk 766 is shown in Fig. 1. Clearly, the central bright UV source with strong X-ray emission is the known AGN. We refer the AGN as the primary source and the nearby UV source as the secondary source. We did not find significant X-ray emission at the position of the secondary UV source. We measured the positions of the primary source (AGN) and the secondary source in both the FUV and NUV images using the centroiding algorithm available with the SAOImage/DS9 tool. We list the positions in Table 1.

3. Spatial analysis and measurement of UV flux The separation between the AGN and secondary source is 4.211 arcsec. Though the peak positions of the two sources are well separated, the wings of the two PSFs overlap. Moreover, the two sources are located in the central regions where the diffuse emission from the galaxy is strong. In addition, there are other galaxy features such as the bar and starforming clumps possibly associated with spiral arms in the central regions. The small separation and a number of features complicate measurement of flux from the two sources. Here we use a simplistic approach to separate the emission from the two sources. We extract radial profiles and fit with the PSFs of point sources and profile of the diffuse emission and background level. Given the complexity of the spatial structure, our method will be approximate. Unlike optical images, UV images of galaxies are not smooth due to the presence of star-forming clumps and possible non-uniformity of internal reddening and even the 2D profile fitting using tools such

2

https://heasarc.gsfc.nasa.gov/cgi-bin/W3Browse/w3browse.pl.

J. Astrophys. Astr. (2021)42:46

Page 3 of 10

46

to obtain the count rates of the AGN and the secondary source in the NUV and FUV bands. If we take the ratio of the NUV and FUV count rates for each source, this will represent the slope of the spectrum of the source. In the final step, we compare these ratios with the slopes of spectrum of known sources such as quasars and starburst galaxies. Below, we describe our analysis in these steps.

3.1 Radial profile fitting and count-rate ratio analysis

Figure 1. Composite three color NUV/Silica (red), FUV/ BaF2 (green) and Chandra/HETG X-ray (blue) image of Mrk 766.

as the GALFIT may not yield accurate results. We postpone such a detailed analysis to a future paper. Our analysis consists of the following steps in sequence. First we extract radial profiles centered on one of the two sources in each image. Thus we generate four radial profiles for the two sources in the NUV and FUV images. We then fit the radial profiles

The radial profiles in the FUV and NUV bands centered on each of the two sources were derived using the image display and astronomical data visualization tool SAOImage/DS9. In each case, the source at the center of the radial profile was fitted with a 1D Moffat function (PSF model for UVIT) and the off-centered source was fitted with a 1D Gaussian. The contribution from the galaxy in the form of diffuse emission was fitted with an 1D exponential function and finally the overall constant background was fitted with a constant 1D function. The fitting was performed using Sherpa which is inside Chandra’s data reduction and fitting package CIAO. Before going into the results of

Figure 2. Surface plot of NUV emission from Mrk 766 on linear (left panel) and square root (right panel) scales showing complex spatial structure. Table 1. Positions of the primary and secondary UV sources. FUV Source AGN Secondary

NUV

a(J2000)

d(J2000)

a(J2000)

d(J2000)

12 h 18 m 26.5 s 12 h 18 m 26.8 s

?29 d 48 m 46.1 s ?29 d 48 m 46.4 s

12 h 18 m 26.5 s 12 h 18 m 26.8 s

þ29 d 48 m 46.3 s þ29 d 48 m 46.6 s

46

Page 4 of 10

J. Astrophys. Astr. (2021)42:46

the fitting process, we would like to state here that due to various components present and resolved in the UVIT images of the galaxy (e.g., the bar in the central region, the extended spiral arm etc.), our simple models for fitting the radial profiles didn’t prove to be sufficient. Consequently, to get acceptable values of the fit statistics, we had to add systematic errors to our data, which resulted in increased error bars in the bestfit parameters. Below we give the form of the different profile functions used to fit the components of the radial profile. h  x  x 2 ib 0 ; Moffat: mðxÞ ¼ A 1 þ c h ðx  x0 Þ2 i Gaussian: gðxÞ ¼ A exp  4 ln 2 ; FWHM2 Exponential: eðxÞ ¼ A expðcoeffðx  x0 ÞÞ; Constant: cðxÞ ¼ c0 :

ð1Þ

Table 2. Best-fit parameters from the FUV radial profile analysis for the primary source. Parameters

Type

Best value

m1A m1 x0 m1c m1b c1c0 g1x0 g1A g1FWHM e1x0 e1A e1coeff. g2A g2FWHM g2x0 CPS

Thawed Frozen Frozen Frozen Thawed Thawed Thawed Thawed Thawed Thawed Thawed Thawed Thawed Linked (m1x0 )

187:5  27:3 0.6 2.6 1.9 2.7 ± 0.3 10:1  0:7 6:4  1:5 4:6  1:2 0:6  195:0 130  5338  0:21  0:01 663  6169 0.2 ? 0.3 0.6 0:23  0:03

In what follows, m1 would indicate a Moffat function, g1 or g2 would indicate Gaussian profiles, e1 would indicate an exponential profile and c1 would indicate a constant background profile. 3.1.1 FUV radial profile analysis for the primary source. For fitting the primary source, we fixed the b and c parameters of the Moffat function (m1) by fitting a radial profile extracted from a field star (PSF modelling). The remaining features were fitted with components as described above. Table 2 lists the fitted parameters along with their 89.041% confidence intervals. Also, we had to add 3.5% systematic error to get a good v2 /dof = 21.2/20. Figure 3 shows the model fitted data along with its residual. We integrated the fitted Moffat function from the position of the peak of the Moffat function to a radius of 25 pixels (  10 arcsec) which encompasses greater than 95% of the energy (Tandon et al. 2020) and divide by the exposure time to get the number of counts per second (CPS) from the primary (see Table 2).

3.1.2 FUV radial profile analysis for the secondary source. Again for fitting the secondary source, we fixed the b and c parameters of the Moffat function (m1) from the stellar profile. Table 3 shows the values of the fitted parameters along with their 89.041% confidence intervals. We had to add 7% systematic error to get a reasonable value of reduced v2 ¼ 1:1 for 20 dof. Figure 4 shows the model fitted data along

Figure 3. FUV radial profile of the primary source, the best-fitting model and residuals.

with the residuals. We integrated the fitted function from the position of the peak of the function to a radius of 25 pixels and divide exposure time and obtained the count rate of 0:02 counts s1 for the secondary.

Moffat Moffat by the 0:09 

3.1.3 NUV radial profile analysis for the primary source. Similar procedures as described above for the FUV band were followed for fitting the NUV radial profiles for the primary as well as the secondary. Table 4 lists the best-fit parameters along with their 89.041% confidence intervals. We had to add 2% systematic error to get a reasonable value of v2 /dof = 25.7/24. In this case, we did not freeze the Moffat parameters as the bright AGN was good

J. Astrophys. Astr. (2021)42:46

Page 5 of 10

46

Table 3. Best-fit parameters from the FUV radial profile analysis for the secondary source.

Table 4. Best-fit parameters from the NUV radial profile analysis for the primary source.

Parameters

Type

Best value

Parameters

Type

Best value

m1A m1x0 m1c m1b c1c0 g1x0 g1A g1FWHM e1x0 e1.A e1coeff. CPS

Thawed Frozen Frozen Frozen Thawed Thawed Thawed Thawed Linked (g1x0 ) Thawed Thawed

72:6  16:1 0.6 2.6 1.9 01 10.0 ± 0.2 23.6 ± 3.6 3.7 ± 0.6 10.0 20.9 ± 2.1  0:12 0.01 0:09  0:02

m1A m1x0 m1c m1b c1c0 g1A g1FWHM g1x0 e1x0 e1A e1coeff. CPS

Thawed Thawed Thawed Thawed Thawed Thawed Thawed Thawed Linked (m1x0 ) Thawed Thawed

5353.3 ± 622.8 0.5 ± 0.2 2:6  0:9 2.2 ± 1.1 44:3  5:6 58:4  17:4 3.9 ± 1.1 10:2  0:6 0.5 1384:7  585:4  0:19 0.03 3:9  0:4

Figure 4. FUV radial profile of the secondary source, the best-fitting model and the residuals.

enough to estimate the c and b parameters. Figure 5 shows the model fitted data along with its residual. We derived an NUV count rate of 3:9  0:4 for the primary source using a circular region with a radius of 25 pixels.

3.1.4 NUV radial profile analysis for the secondary source. In this case, we had to fit an additional Gaussian (g2) to account for the AGN. Table 5 below shows the values of the fitted parameters along with their 89.041% confidence intervals. Again, we had to add 2% systematic error to get a reasonable value of reduced v2 ¼ 1:09. In this case, Moffat parameters were fixed based on the stellar radial profile analysis. As before, we derived an NUV count rate of 0:7  0:1 counts s1 for the secondary source. Figure 6 shows the model fitted data along with its residual.

Figure 5. NUV radial profile of the primary source, the best-fitting model and the residuals.

Table 5. Best-fit parameters from the NUV radial profile analysis for the secondary source Parameters

Type

Best value

m1A m1x0 m1c m1b c1c0 g1A g1FWHM g1x0 e1x0 e1A e1coeff. g2FWHM g2A g2x0 CPS

Thawed Frozen Frozen Frozen Thawed Thawed Thawed Thawed Thawed Thawed Thawed Thawed Thawed Thawed

1145:3  215:8 0.6 1.7 1.6 21.7 ± 29.2 355.5 ± 55.7 3:6  0:4 10:0  0:1 11:2  280:3 192 ± 6847  0:12 0.03 7.9 ± 2.7 95:3  52:8 11:4  1:8 0:7  0:1

46

Page 6 of 10

J. Astrophys. Astr. (2021)42:46 Table 7. NUV-to-FUV count ratios. Source

Observed countrate ratio

Internal EðB  VÞ

Predicted ratio

17  4 83

0.36 0.0

18.2 7.8

Primary Secondary

Figure 6. NUV radial profile of the secondary source, the best-fitting model and the residuals.

Table 6 summarizes the results obtained from radial fitting.

3.2 Count-rate ratio analysis The nature of any source can be inferred from its spectrum. In the absence of spectrum, the colors of an object become useful. We calculate the ratio of count rates in the NUV/Silica and FUV/BaF2 bands for the AGN and the secondary source. These ratios are equivalent to colours. Generally two colors are used to classify or infer the type of an object. In addition to the UV colors, we use X-ray/optical flux. In Table 7, we list the UV colours of the two sources. As described earlier, we will compare the ratios listed in Table 7 with the corresponding ratios obtained from standard spectra of quasars and starburst galaxies.

4. Nature of the secondary source 4.1 Comparison with composite quasar spectrum We used the composite quasar spectrum derived by Vanden Berk et al. (2001) using SDSS spectra of over 2200 quasars in the redshift range from 0.044 to 4.789.

Since this spectrum has practically zero extinction, but our observations were reddened by both the Galactic extinction and internal reddening in Mrk 766, we need to redden the composite quasar spectrum before calculating the count rates in NUV and FUV. The ‘non-standard’ extinction caused by the AGN environment of Mrk 766 was applied with the help of empirical formula given in Czerny et al. (2004). We used a color excess of EðB  VÞ ¼ 0:36 derived from the Balmer decrement, i.e., the ratio of observed Ha to Hb flux of Mrk 766 (Gonzalez Delgado & Perez 1996). Then we redshifted this reddened composite spectrum for the redshift of Mrk 766 (z ¼ 0:01293). Finally we reddened the composite spectrum to account for the Galactic extinction using the CCM89 law (Cardelli et al. 1989). We assume this reddened and redshifted quasar composite spectrum as our model spectrum for active nucleus in Mrk 766. We calculated the predicted UVIT count rates for our model spectrum using the effective area of a filter using the following equation: Z fk ð2Þ Aeff ðkÞdk: CPS ¼ ðhc=kÞ Using the effective areas for the FUV/BaF2 and NUV/ Silica filters, we calculated the predicted count-rate ratio for the composite quasar spectrum to be 18.2. We note that the predicted count rate ratio for the composite quasar spectrum is similar to the observed ratio for the AGN but it is very different than the observed ratio for the secondary source. The internal extinction with EðB  VÞ ¼ 0:36 derived from the Balmer decrement is appropriate for the AGN in Mrk 766. It is possible that the

Table 6. Results from radial fitting. Source Primary FUV Secondary FUV Primary NUV Secondary NUV

Systematic error (%)

v2 /dof

3:5 7:0 2:0 2:0

1.0589 1.11381 1.07013 1.0872

CPS 0.23 0.09 3.9 0.7

± ± ± ±

0.03 0.02 0.4 0.1

J. Astrophys. Astr. (2021)42:46

Page 7 of 10

secondary source suffers with a different internal extinction. If we do not redden the composite quasar spectrum with the internal extinction, we predict a count-rate ratio of 7.8. We list the predicted countrate ratios in Table 7. From Table 7, we find that the count-rate ratio of the primary source is within the predicted ratio for the composite quasar spectrum. Thus, our analysis implies that the primary source is indeed an AGN which in turn verifies the correctness of our methodology. The observed count-rate ratio of the secondary source deviates significantly from the expected value for an AGN with similar internal reddening as the primary source. But interestingly, the observed ratio for the secondary source matches well with the AGN ratio if there is no internal reddening. Thus, our analysis clearly rules out the secondary source to be a background AGN or an accreting SMBH embedded in the galaxy Mrk 766.

The observed count rate of 0:7 counts s1 for the ˚ ¼ 1:6  secondary source is converted to fk ð2418 AÞ 1 ˚ . 1016 ergs cm2 s1 A ˚ , we Since we have the observed flux at 2418 A ˚ de-reddened it first from the extinction Ak at 2418 A due to Milky Way using CCM89. Then we simply transferred the wavelength and the corresponding flux to the rest frame of Mrk 766 by multiplying the flux by ð1 þ zÞ (z ¼ 0:01293) and dividing the ˚. wavelength by ð1 þ zÞ which gave us k ¼ 2387 A ˚ In order to estimate the flux density at 2500 A, we used the composite quasar spectrum. We scaled the composite quasar spectrum to have the same flux as ˚ . With this scaled the secondary source at 2387 A ˚ Þ ¼ 1:7  1016 spectrum, we found fk ð2500 A 2 ˚. ergs=cm =s=A The optical to X-ray index is given by (Sobolewska et al. 2009), aox ¼ 

4.2 Estimation of X-ray flux and detectability with Chandra From our analysis in previous sections, it is clear that the count-rate ratio of the secondary source is consistent with the expected ratio for an unabsorbed AGN. This possibility can be tested by estimating the expected count rate in the X-ray band and comparing it with the upper limit from the Chandra data. In order to predict the expected X-ray flux, we use the optical to X-ray flux ratio aox which is the ratio of flux den˚ and 2 keV. We first calculate the sities at 2500 A ˚ 2500 A flux density using the observed count rates in the FUV and NUV bands. We converted the FUV and NUV count rates to flux densities at the mean wavelengths of the filter bandpass using the relation (Tandon et al. 2017): ˚ 1 Þ ¼ CPS  UC: fk ðergs cm2 s1 A

ð3Þ

The unit conversion factor (UC) was derived from the zero point magnitude (ZP) given in Tandon et al. (2020) and using the relation (Tandon et al. 2017): ZP ¼ 2:5 log10 ðUC  k2m Þ  2:407;

ð4Þ

˚. where km is the mean wavelength of the filter in A ˚ With km ¼ 2418 A and ZP ¼ 19:763, the unit conversion factor for N242W is UC ¼ 2:318 ˚ 1 Þ/ðcounts s1 Þ:  1016 ðergs s1 cm2 A

ð5Þ

46

log10 ½ðf2 keV =f2500 A˚ Þ ; 2:605

ð6Þ

where L2 keV and L2500 A˚ are in ergs cm2 s1 Hz1 . ˚ to fm ð2500 AÞ ˚ and using We converted fk ð2500 AÞ aox ¼ 1:37 (Lusso et al. 2010) to find fm ð2 keVÞ ¼ 9:76  1032 ergs cm2 s1 Hz1 or fE ¼ 7:369  106 photons/cm2 /s/keV. Using a power-law model with X-ray photon index C ¼ 1:9 modified with the Galactic absorption NH ¼ 1:8  1020 cm2 along the line-of-sight to Mrk 766, we calculated 0.4–10 keV band X-ray flux, fX ¼ 1:5  1013 ergs cm2 s1 . We converted this flux to Chandra/HETG count rate of 0.0017 counts s1 using the WebPIMMS tool. This is the expected count rate if the secondary source were an unabsorbed AGN. We used the Chandra observation (ObsID:1597) with an exposure time of 89 ks. There is no X-ray source at the location of the secondary source. We calculated the 4r upper limit of 128 counts or 0.0014 counts s1 , which is less than the predicted count rate. Thus, due to the lack of X-ray emission, the secondary is highly unlikely to be another accreting SMBH in Mrk 766. Another possibility is that the secondary source could be a compact region of enhanced star formation. We suspected this possibility based on the ratio image where the secondary source does not stand out. We created a ratio image, shown in Fig. 7, by dividing each pixel value in the NUV image by the

46

Page 8 of 10

corresponding pixel value in the FUV image after scaling the NUV image to have the same exposure time as the FUV image. We find similar ratios at the position of the secondary source as in other parts of the galaxy except at the location of the primary. This suggests that the emission process responsible for the secondary source is likely similar to that of the diffuse emission from the other parts of the galaxy which is likely from a population of young, massive stars resulting from star formation.

J. Astrophys. Astr. (2021)42:46 Table 8. Ratio obtained from UV templates of starburst galaxies. Template Starburst Starburst Starburst Starburst

1 2 3 4

EðB  VÞ

Ratio

EðB  VÞ\0:1 0:11\EðB  VÞ\0:21 0:25\EðB  VÞ\0:35 0:39\EðB  VÞ\0:50

7.36 7.5 10.1 8.3

4.3 Comparison with the spectra of starburst galaxies To investigate further if the secondary source is actually an enhanced star-forming region, we estimated typical NUV-to-FUV count-rate ratios for star burst galaxies. We obtained the template spectra of star burst galaxies from Kinney et al. (1996) derived for different values of the internal extinction, E(B-V). These spectra are already corrected for Galactic extinction. We calculated the count-rate ratios for the template starburst spectra using Equation (2) by using the flux densities at the mean wavelengths of our FUV, NUV filters. We did not apply any additional internal reddening. The predicted ratios are listed in Table 8. We see that the observed count-rate ratio for the secondary source is very similar to that derived for the starburst templates with internal extinction EðB  VÞ\0:21. If the secondary source is indeed a compact star forming region with optical/UV spectrum similar to the starburst templates, we expect significant optical emission.

Figure 8. Selected regions for aperture photometry on the secondary source.

Figure 9. Spectra of starburst galaxies scaled to match the flux density measured with the FUV/F154W filter at the ˚ Filled circles show flux densities mean wavelength 1541 A. ˚ measured with the NUV/N242W filter (kmean ¼ 2418 A), ˚ HST/ACS F330W filter (kcentral ¼ 3362:7 A) and HST/ ˚ are also shown. WFC3 F547M filter (kcentral ¼ 5475 A)

4.4 HST images and calculation of flux Figure 7. Ratio image obtained by dividing the NUV image by the FUV image after the NUV image was scaled to have the same exposure time as that of the FUV image.

We searched for the counterpart of the secondary source in HST images. We detected multiple compact sources near the position of the secondary source in

J. Astrophys. Astr. (2021)42:46

the HST images of Mrk 766 acquired with different instrument and filter combinations. Because of the excellent PSF of the HST, the secondary source that appeared as a point source in UVIT images, now appeared as three distinguishable point sources. Figure 8 shows the secondary source marked with circles in the HST image in the F330W filter. We included the three sources and performed aperture photometry and obtained the count rates for the secondary source in the HST/ACS F330W filter (central ˚ ) and HST/WFC3 F547M filter wavelength 3362.7 A ˚ ). We used a circular (central wavelength = 5475 A region of radius 0.3 arc-seconds to encircle the source as shown in Fig. 8 and another concentric annular region of inner radius 0.35 arc-seconds and outer radius 0.45 arc-seconds to estimate the background count rate. We then calculated the background corrected net count rates and converted to flux densities using the value of the PHOTFLAM keyword in the headers of the HST images. We found ˚ ¼ 6:74  1017 ergs cm2 s1 for F330W fk ð3363 AÞ ˚ ¼ 2:9  1017 ergs cm2 s1 for filter and fk ð5475 AÞ the F547M filter. In order to compare these flux densities measured with the HST and that expected from the secondary source assuming it to be a compact star forming region, we re-scaled the starburst template spectra to ˚ (shown in section the measured flux density at 2418 A 4.2) with the UVIT/NUV. We then compared the UV and optical flux densities with the re-scaled starburst template spectra in Fig. 9. We find that the measured flux at four different wavelengths follow the shape of the starburst template spectra. This clearly suggests that the secondary source is indeed a compact starforming region.

5. Conclusion We identified a bright far and near UV source at a projected distance of  1.1 kpc from the known active nucleus. Such pairs of compact sources can easily be suspected as a pair of accreting SMBHs. We investigated the nature of the secondary source using NUVto-FUV flux ratio, Chandra X-ray observation, and HST images in the near UV and optical bands. The lack of X-ray emission in the Chandra image at the location of the secondary source makes it highly unlikely to be an accreting SMBH. Further, the UV/ optical flux measured at four different bands closely follow the shape of the starburst template spectra.

Page 9 of 10

46

Therefore we conclude that the secondary is most likely a compact star-forming region.

Acknowledgements This publication uses the UVIT data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). This publication uses UVIT data processed by the payload operations centre at IIA. The UVIT is built in collaboration between IIA, IUCAA, TIFR, ISRO and CSA. This research has made use of UVIT pipeline (CCDLAB) developed at University of Calgary for UVIT development and science support. The scientific results reported in this article are based in part on observations made by the Chandra X-ray Observatory, data obtained from the Chandra Data Archive. This research is based on observations made with the NASA/ESA Hubble Space Telescope obtained from the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS 5-26555. This research has made use of the python and julia packages. This research has made use of the SIMBAD database, operated at CDS, Strasbourg, France.

References Ballo L., Braito V., Della Ceca R. et al. 2004, ApJ, 600, 634 Cardelli J. A., Clayton G. C., Mathis J. S. 1989, ApJ, 345, 245 Czerny B., Li J., Loska Z., Szczerba R. 2004, MNRAS, 348, L54 Gonzalez Delgado R. M., Perez E. 1996, MNRAS, 278, 737 Junkkarinen V., Shields G. A., Beaver E. A. et al. 2001, ApJ, 549, L155 Kinney A. L., Calzetti D., Bohlin R. C. et al. 1996, ApJ, 467, 38 Komossa S. 2012, Adv. Astron. 2012, https://doi.org/10. 1155/2012/364973 Komossa S., Burwitz V., Hasinger G. et al. 2002, Astrophys. J., 582, L15 Komossa S., Zensus J. A. 2014, Proc. Int. Astron. Union, 10, 13–25 Lang D., Hogg D. W., Mierle K., Blanton M., Roweis S. 2010, AJ, 139, 1782 Lusso E., Comastri A., Vignali C. et al. 2010, A&A, 512, A34

46

Page 10 of 10

Matsui H., Saitoh T. R., Makino J. et al. 2012, ApJ, 746, 26 Merritt D., Milosavljevic M. 2005, Living Rev. Relat. https://doi.org/10.12942/lrr-2005-8 Postma J. E., Leahy D. 2017, PASP, 129, 115002 Schneider P. 2006, Extragalactic Astronomy and Cosmology, An Introduction (Springer, Berlin, Heidelberg). https://doi.org/10.1007/978-3-642-54083-7

J. Astrophys. Astr. (2021)42:46 Sobolewska M. A., Gierlin´ski M., Siemiginowska A. 2009, MNRAS, 394, 1640 Tandon S. N., Subramaniam A., Girish V. et al. 2017, Astron. J., 154, 128 Tandon S. N., Postma J., Joseph P. et al. 2020, AJ, 159, 158 Vanden Berk D. E., Richards G. T., Bauer A. et al. 2001, AJ, 122, 549

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:36 https://doi.org/10.1007/s12036-020-09687-y

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

UVIT study of UV bright stars in the globular cluster NGC 4147 RANJAN KUMAR1,* , ANANTA C. PRADHAN1, MUDUMBA PARTHASARATHY2,

DEVENDRA K. OJHA3, ABHISEK MOHAPATRA1, JAYANT MURTHY2 and SANTI CASSISI4,5 1

Department of Physics and Astronomy, National Institute of Technology, Rourkela, Odisha 769 008, India. Indian Institute of Astrophysics, Koramangala II-Block, Bangalore 560 034, India. 3 Tata Institute of Fundamental Research (TIFR), Homi Bhabha Road, Mumbai 400 005, India. 4 INAF - Astronomical Observatory of Abruzzo, Via M. Maggini, sn. 64100 Teramo, Italy. 5 INFN - Sezione di Pisa, Universita` di Pisa, Largo Pontecorvo 3, 56127 Pisa, Italy. *Corresponding Author. E-mail: [email protected] 2

MS received 7 November 2020; accepted 13 December 2020 Abstract. We present far ultraviolet (FUV) observations of globular cluster NGC 4147 using three FUV filters, BaF2 (F154W), sapphire (F169M), and silica (F172M) of Ultra-Violet Imaging Telescope (UVIT) onboard the AstroSat satellite. We confirmed the cluster membership of the UVIT observed sources using proper motions from Gaia data release 2 (GAIA DR2). We identified 37 blue horizontal branch stars (BHBs), one blue straggler star (BSS) and 15 variable stars using UV-optical color magnitude diagrams (CMDs). We find that all the FUV bright BHBs are second generation population stars. Using UV-optical CMDs, we identify two sub-populations, BHB1 and BHB2, among the UV-bright BHBs in the cluster with stars count ratio of 24:13 for BHB1 and BHB2. The effective temperatures (Teff ) of BHB1 and BHB2 were derived using color-temperature relation of BaSTI-IAC zero-age horizontal branch (ZAHB). We found that BHB1 stars are more centrally concentrated than BHB2 stars. We also derive physical parameters of the detected FUV bright BSS by fitting younger age BaSTI-IAC isochrones on optical and UV-optical CMDs. Keywords. Galaxy: globular clusters: individual: NGC 4147—stars: horizontal-branch—stars: blue stragglers—stars: Hertzsprung–Russell and colour-magnitude diagrams.

1. Introduction The glimpse of ultraviolet (UV) light in the old stellar population of the Galactic globular clusters (GGCs) is dominated by hot luminous UV-bright stars which are mostly the stars of blue horizontal branch (BHB), blue-stragglers (BS), post-asymptotic giant branch (pAGB), and extreme horizontal branch (EHB) phases having temperature more than 7000 K. Various physical properties of these UV-bright populations have been explored using large sample of GGCs observed by Galaxy Evolution Explorer (GALEX, Schiavon et al. 2012) and Hubble space telescope This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

(HST, Nardiello et al. 2018). Ultraviolet Imaging Telescope (UVIT) on-board the AstroSat satellite (Kumar et al. 2012) has also performed the imaging observations of several GGCs in UV with a better spatial resolution than GALEX enabling in resolving the core of the clusters and also in distinguishing UV bright stars of different evolutionary stages to study their physical parameters in a great detail (Subramaniam et al. 2017; Sahu et al. 2019a; Jain et al. 2019; Kumar et al. 2020, 2021; Rani et al. 2020; Singh et al. 2020). Using the spectroscopic data of a large number of stars in many GGCs along with the UV photometric observations with HST, the evidence for the presence of multiple stellar populations in GGCs is now well established (see reviews Bastian & Lardo 2018; Gratton et al. 2019; Cassisi & Salaris 2020, and

36

Page 2 of 10

references therein). The different chemical evolution within the GGCs seems to be the origin of two distinct sub-population of stars (the first generation (1G) and second generation (2G)) characterized by significant abundance variations in light elements. The stellar populations enriched in He, N, and depleted in O and C are 2G stellar population while stellar populations enriched in C and O and depleted in N and Na with a primordial He-abundance are termed as normal or 1G stellar population (see Milone et al. 2017; Marino et al. 2019, and references therein). We present here the UV study of an old age, metal poor, low density GC NGC 4147, located at a distance of 21 kpc from the Galactic center and 19 kpc from the Sun (l ¼ 252:84 , b ¼ þ77:18 ). We have observed this cluster using three far-UV (FUV) filters of UVIT. Prior to this, the UV imaging and colormagnitude diagrams (CMDs) of the cluster were presented by Schiavon et al. (2012) using GALEX observations. Their study was restricted to the sources in the outer region of the cluster due to the lower resolution (  500 ) of GALEX. However, the cluster is well studied both photometrically and spectroscopically in the optical bands of the electromagnetic spectrum. Several of the photometric observations in the optical bands have explored about specific sources of the cluster such as BHBs, variable stars, red giant branch stars (RGBs), etc., using their optical CMDs (Auriere & Lauzeral 1991; Arellano Ferro et al. 2004; Stetson et al. 2005; Arellano Ferro et al. 2018; Lata et al. 2019). Similarly, many low resolution spectroscopic observations of the cluster have been performed to find its metallicity (½Fe=H ¼ 1:85 dex), and chemical abundances of various alpha-elements (Suntzeff et al. 1988; Martell et al. 2008; Ivans 2009). The latest high-resolution spectroscopy of 18 RGB stars by Villanova et al. (2016) has revealed that most of the RGB stars of the cluster are of 2G population type with 2G to 1G RGB stars’ ratio of 85:15. They found that the red-HB stars (RHBs) are progeny of the 1G population and the BHBs are progeny of the 2G population. They also confirmed an alpha-enhancement of 0.38 dex and a mono-metallicity of ½Fe=H ¼ 1:84 dex among 1G and 2G stars in the cluster. In Section 2, we present the observation details, data reduction process and photometry of the UVIT observation of the cluster. In Section 3, we show the CMDs of all the detected sources. In Section 4, we discuss the sub-population among the UVIT detected 2G BHBs. In Section 5, we estimate various physical properties of the UV bright blue straggler star (BSS). Finally, we summarize our results in Section 6.

J. Astrophys. Astr. (2021)42:36

2. Data reduction We obtained the UVIT observation of the cluster NGC 4147 in three FUV filters, BaF2 , sapphire and silica, having wide, medium and narrow bandwidths, respectively. The UVIT data of the cluster was obtained from AstroSat archival web-page1 in Level 1 format. The instrument and calibration details of UVIT can be found in Tandon et al. (2017, 2020). We reduced the Level 1 data into science image using a customized software package CCDLAB (Postma & Leahy 2017), specially designed for the UVIT data reduction. The UVIT observations have been performed in several orbits and we get science image for each orbit. In order to get a good signal to noise ratio (SNR), we combined all the observed orbits into a single image and then performed the photometry. The UVIT observation details are given in Table 1. In Fig. 1, we provide a color image of the cluster using multi-wavelength observations from infrared (IR) to UV. The sources in blue color are observed in UV using the UVIT BaF2 filter, the green color sources are from the archival catalog of CFHT-4m V band observation (Stetson et al. 2019), and the red color sources are obtained from 2MASS J band observation2 of the cluster. We can see that the hot sources, observed in UV (blue sources) are bright enough and easily distinguishable in the outer region of the cluster. The central region of the cluster also contains many hot sources as visible in cyan color (mixture of blue and green) due to crowding effect at the center. We applied DAOPHOT point spread function (PSF) photometry (Stetson 1987) on the science images to obtain the source positions and their respective apparent magnitudes. The aperture and saturation corrections were applied on the detected sources in all the three FUV filters following the suggestion by Tandon et al. (2017). A typical PSF of 1.500 was obtained in all the science images of the FUV filters. We have excluded the sources within 1000 from the cluster center to avoid contamination. We detected 114, 92, and 65 sources in the BaF2 , sapphire, and silica filters, respectively. The AB-magnitude limits of the detected sources in BaF2 , sapphire and silica filters are 23.0, 22.5, and 21.0, respectively. The magnitudes were corrected for extinction value using EðB  VÞ ¼ 0:0221 mag (Schlafly & Finkbeiner 2011) along the

1

https://astrobrowse.issdc.gov.in/astro_archive/archive/Home.jsp. https://irsa.ipac.caltech.edu/applications/2MASS.

2

J. Astrophys. Astr. (2021)42:36

Page 3 of 10

36

Table 1. Observational details of NGC 4147. Date of observation: 2017, April 17 Telescope pointing: RA ¼ 182:5263 , Channel FUV

DEC ¼ 18:5426

Filters

Mean k ˚) (A

Dk ˚) (A

Exp. time (s)

No. of orbits

BaF2 Sapphire Silica

1541 1608 1717

380 290 125

1536 1648 1209

5 3 2

Figure 1. Color image of NGC 4147; UV: UVIT BaF2 (blue), visible: CFHT-4m V band (green), infrared: 2MASS J band (red).

cluster direction and extinction law of Cardelli et al. (1989) in the UV filters.

3. Color-magnitude diagrams To study UV and UV-optical CMDs of the cluster members, we cross-matched the UVIT observed sources with GAIA data release 2 catalog (Gaia DR2, Gaia Collaboration et al. 2018) with a matching radius of 1.500 . The confirmed cluster members were separated out from the field stars and the background sources (galaxies, quasars, etc.) using Gaia DR2 proper motions. Once the sources were confirmed as cluster members, we cross-matched them to deeper photometric observations, the latest released archival catalog of ground-based observations in UBVRI filters (Stetson et al. 2019). An optical CMD, B-V vs. V, is

shown in panel (a) of Fig. 2 for all the ground-based observed sources within the UVIT field of view (FoV  300 ) in gray solids. The UVIT detections are overplotted in the CMD and are shown in blue and green asterisks and in brown solid circle. In order to obtain the cluster properties, we used the Bag of Stellar Tracks and Isochrones models (BaSTIIAC3, Hidalgo et al. 2018) and generated isochrones and zero-age horizontal branch (ZAHB) with cluster parameters, ½Fe=H ¼ 1:896, age = 13.0 Gyr (Harris 2010) and ½a=Fe ¼ 0:40 (Villanova et al. 2016). Since the alpha-enhanced set of the BaSTI-IAC library has not been still published, we relied on the extension of the BaSTI-IAC library to alpha-enhanced mixture (Pietrinferni et al. 2021) for the present analysis. In panel (a) of Fig. 2, BaSTI-IAC isochrones 3

http://basti-iac.oa-abruzzo.inaf.it/index.html.

36

Page 4 of 10

J. Astrophys. Astr. (2021)42:36

(a)

(b)

(c)

(d)

Figure 2. Optical and UV-optical CMDs of the UVIT observed sources of NGC 4147. Panel (a): The B-V vs. V CMD where ground-based detections from Stetson et al. (2019) are shown in gray solids, the UVIT observed BHBs are shown in blue and green asterisks. The BSS is shown in the brown solid. BaSTI-IAC isochrones and ZAHB with cluster parameter, ½Fe=H ¼ 1:896, ½a=Fe ¼ 0:40, Age = 13.0 Gyr, and distance-modulus ðm  MÞ ¼ 16:40, are over-plotted on the observed sources. Panels (b), (c) and (d): BaF2 –V vs. BaF2 , sapphire-V vs. sapphire, and silica-V vs. silica CMDs, respectively, are shown for the UVIT observed sources with the same colors as mentioned in panel (a), and BaSTI-IAC ZAHB with the same cluster parameters as mentioned in panel (a) are over-plotted on all CMDs.

are over-plotted on the B-V vs. V CMD in red line using a distance-modulus of 16.40, which are fitting very well with the ground-based observations. In panels (b), (c), and (d) of Fig. 2, we have shown the UV-optical CMDs of the cluster using BaF2 , sapphire, and silica filters, respectively, in combination with V magnitude from the optical bands. The ZAHB from BaSTI-IAC library is over plotted on all the UV-optical CMDs. Although all the UV-optical CMDs look similar we have shown them to see the morphological distribution of UV bright stars of the cluster in all the FUV filters. We can see in Fig. 2 that BHBs are major contributors to the UV bright sources of the cluster. If we look at the optical CMD in panel (a) of Fig. 2, we notice that the UVIT observed BHBs have spread along the horizontal line of the HB phase. This spread is better visible in UV-optical CMDs along the diagonal HB line than the optical CMD. In fact, the spread among BHBs in BaF2-V vs. BaF2 CMD (panel (b)) becomes more clear with a gap of 0.8 magnitude in the

BaF2-V color within BHBs. Based upon the gap observed in BaF2-V vs. BaF2 CMD, we divide the UVIT observed BHBs in two groups: BHBs with BaF2-V color brighter than 2.8 magnitude as BHB1 (blue asterisks) and BHBs with BaF2-V color fainter than 3.6 magnitude as BHB2 (green asterisks). These BHBs are lying in left and right portions of the BHB region in B-V vs. V CMD (panel (a)) and they are clearly identified in sapphire-V vs. sapphire and silica-V vs. silica CMDs in panels (c) and (d), respectively. There are 24 sources in BHB1 and 13 sources in BHB2 groups. Villanova et al. (2016) have made a chemical abundance analysis of 18 RGB stars of the cluster NGC 4147 and found that the cluster is composed of 1G stars with ½Na=Fe  þ 0:0 and ½O=Fe  þ 0:3 and 2G stars with ½Na=Fe  þ 0:5 and ½O=Fe   0:2. In Fig. 1 of their paper they have shown that the 2G HB stars are located at B–V \ 0.3. If we compare the UVIT observations with optical

J. Astrophys. Astr. (2021)42:36

data, it appears that all the observed BHBs seem to belong to the 2G population (panel (a) of Fig. 2). In the same figure, one can notice that the BHBs are separated in two sub-groups, which could point to the presence of two sub-populations in the observed 2G BHBs. We discuss this possibility in the following sections. Apart from the BHBs, we have also detected one UV bright BSS in the cluster. The BSS was observed in BaF2 and sapphire filters but not in the silica filter. The brightness of the BBS may be fainter than the detection limit of silica filter (21.0 AB-magnitude). The BSS is shown in Fig. 2 with brown solid. We also found 15 variable stars by cross-matching the UVIT observed sources with the updated variable stars catalog of GCs (Clement 2017). However, we have not shown these variable stars on the CMDs due to the severe crowding conditions affecting these stars, and the fact that we do not have a complete sampling of their light curves.

4. Sub-population in BHBs We derive the effective temperature (Teff ) of the observed BHBs using color-temperature relation of the BaSTI-IAC ZAHB in Fig. 3. The best fitted ZAHB from Fig. 2 is taken to derive the color-temperature relation of the UVIT observed BHBs. As shown in

Page 5 of 10

36

Fig. 3, the Teff of BHB1 is ranging between 10000 K and 11300 K and Teff of BHB2 ranges between 8000 K and 9500 K. The comparison between observations and models suggests the presence of a gap between BHB1 and BHB2 stars with a temperature difference of about 500 K. However, we caution that this width of the observed temperature gap and its actual existence may be affected by the limited number of sample stars (number of BHBs detected) in this cluster. Although a detailed statistical analysis of the reality of such a gap will be carried out in a future work, we note that a puzzling gap or discontinuity in the distribution of BHBs is seen in the color-magnitude diagram (CMD) of many GGCs (Catelan et al. 1998; Piotto et al. 1999; Behr et al. 2000; Brown et al. 2016). It was initially believed to be due to statistically significant under-population of stars but its appearance at similar location for different clusters has substantiated the claim that it is indeed a real feature and the satisfactory explanation for its origin is not yet known. However, speculation is that the gaps may be demarcating the boundaries between separate, discrete populations of HB stars, which differ in their origin or evolution. By relying on the BaF2 –V vs. BaF2 CMDs obtained in the present work, we collect some evidence suggesting the possible presence of a bi-modality among the BHB population. In the following, we provide some suggestions about the possible origin for the observed discontinuity in the color distribution of BHB stars in NGC 4147. The main empirical findings are: (i) the BHB1 and BHB2 are clustered in the optical plane in two quite clear separate regions – although in that plane these two regions appear quite adjacent (panel (a) of Fig. 2); (ii) in the UV-optical plane there exists an evident gap of about 500 K whose width has been evaluated by over-imposing the models.

Figure 3. BaF2 –V vs. BaF2 CMD of the observed BHBs. We use BaSTI-IAC ZAHB color-temperature relation to derive the range of Teff of BHB1 and BHB2. The Teff derived from the ZAHB is given in the upper x-axis of the plot.

These empirical evidence suggests that BHB1 stars could be slightly He enhanced with respect to BHB2 ones and in such a case they would represent two distinct sub-populations in the 2G population; or alternatively they could share with BHB1 more or less the same He abundance, and have experienced different mass loss efficiency during the previous RGB stage (see, e.g., Tailo et al. 2020). In Fig. 4, we have shown the radial distribution of BHB1 and BHB2 stars. We see that the BHB1 stars are spread within 15 pc (with an exception of one BHB1 star at a distance of 43 pc) from the cluster

36

Page 6 of 10

Figure 4. The radial distribution of UVIT detected BHBs. The histogram of BHB1 stars is shown in blue line and histogram of BHB2 stars is shown in green line. The lower and upper x-axes is the distance from the cluster center in parsec and arcminutes, respectively. The y-axes repesents the number of stars present in each distance bin of 5 parsec from the cluster center. The vertical dashed line is the tidal radius of the cluster.

center. However, BHB2 stars show their radial distribution up to 25 pc. We find that the BHB1 stars are having almost double counts than the BHB2 stars up to 10 pc. This indicates that the hotter sources are relatively concentrated at the center than the lower Teff BHBs.

5. FUV bright blue straggler star The blue-stragglers (BS) are generally formed in the old age clusters through two formation channels: mass-transfer in binary system in low density environment (Knigge et al. 2009; Leigh et al. 2013) and merger due to collision between stars in dense environment (Chatterjee et al. 2013). They are relatively more massive hydrogen burning MS stars than the normal MS population of the cluster owning to the higher mass and consequently shifted in the upper MS branch in the H-R diagram. The UV photometry is a great tool to identify the BS population in a cluster. The data from UV telescopes (HST, GALEX and UVIT) have been excellent tool to explore BS population and their formation scenario (Ferraro et al. 1997, 2003; Dieball et al. 2010; Schiavon et al. 2012; Gosnell et al. 2015; Subramaniam et al. 2017; Sahu

J. Astrophys. Astr. (2021)42:36

et al. 2019a; Kumar et al. 2020, 2021; Rani et al. 2020). In particular, UVIT observations of star clusters are able to identify the hot binary companions (WD, HB, EHB, etc.) of BS stars (Subramaniam et al. 2016; Sindhu et al. 2019; Sahu et al. 2019b; Jadhav et al. 2019; Singh et al. 2020). Present observational data-set allows us to identify one FUV bright BSS in the outer region of the cluster, at a distance of 2.150 from the center. The respective colors and magnitudes of the BSS in UV-optical and optical CMDs are shown in Fig. 2. The BS are generally more massive than the stars currently evolving at the MS turn-off and are located along the brighter and hotter extension of the MS locus, mimicking the location of intermediate-age stars. Therefore, we generated relatively younger ages BaSTI-IAC isochrones by adopting the same assumptions concerning metallicity and alpha-element enhancement as the old stellar component. The comparison between model predictions and the empirical data in the B-V vs. V and BaF2-V vs. BaF2 CMDs is shown in Fig. 5. We see that the isochrone with age 3 Gyr is lying closer to the observed BSS. Based on the observed location of the BSS in optical and UV-optical CMDs, we extracted the mass, luminosity, effective temperature of the BSS from the 3 Gyr BaSTI-IAC isochrone which are M ¼ 1:14  0:04M , log L=L ¼ 0:9 0:1, and log Teff =K ¼ 3:93  0:01, respectively. We obtained optical photometry of the FUV bright BSS from various archival catalogs and performed spectral energy distribution (SED) fitting on the observed fluxes to the model generated fluxes for different filters from UV to near-IR bandwidth. We used an online SED fitting tool, VO SED analyszer (VOSA, Bayo et al. 2008), to perform the SED fit. It uses the chi-square minimisation technique to fit the observed fluxes on the model fluxes generated from the theoretical spectra incorporating various stellar atmosphere models. We used the Kurucz stellar atmosphere model, ATLAS9 (Kurucz model, Castelli & Kurucz 2003) grids (spectra) with model parameters in the following range, Teff : 3500 K to 50000 K, ½Fe=H :  2:0 and 1:5 (nearest to the cluster metallicity), and logðgÞ: 0.0 to 5.0, respectively. The observed fluxes of different filters used in the SED fit is given in Table 2. We have used observed fluxes ˚ to from 23 filters in the wavelength range 1800 A ˚ 11800 A to perform the SED fit. In Fig. 6, we have shown the best-fitted Kurucz model spectra of 8500 K on the observed fluxes in gray line. The model fluxes calculated for various filters using the Kurucz model spectra are shown in

J. Astrophys. Astr. (2021)42:36

Page 7 of 10

36

Figure 5. Optical and UV-optical CMDs are shown in left and right panels, respectively. The FUV bright BSS of NGC 4147 is shown in the black solid in both CMDs. The BaSTI-IAC isochrones with the same parameters as mentioned in Fig. 2 but with younger ages, i.e. 2 Gyr (dashed line), 3 Gyr (solid line), 4 Gyr (dash-dotted line), and 5 Gyr (dotted line), are over-plotted in both CMDs. Table 2. List of the telescopes and their filters used in the SED fit. Telescope UVIT/AstroSat GALEX CFHT-3.6m SDSS GAIA PAN-STARRS CTIO/DECam a

Filters

˚ Wavelength range in A

BaF2 , sapphire FUV, NUV U, B, V, R, I u, g, r, i, z G, BP g, r, i, z, y g, r

1350–1800 1350–3000 3000–11800 3000–10800 3300–10600 3900–10800 3925–7233

Reference This paper Schiavon et al. (2012) Stetson et al. (2019) DR9, Ahn et al. (2012) Gaia Collaboration et al. (2018) Chambers et al. (2016) DECam Legacy Survey DR3a

https://www.legacysurvey.org/dr3/description/

Figure 6. The SED fit of BSS using the Kurucz model is shown in the upper panel. The bottom panel shows the residue (deviation) of the observed fluxes from the Kurucz model fluxes.

36

Page 8 of 10

J. Astrophys. Astr. (2021)42:36

Table 3. Various physical parameters of the BSS derived from the SED fit and BaSTI-IAC isochrones are listed in the table. ID

RAJ2000 (degree)

DEJ2000 (degree)

BSS01

182.5633

18.5480

Fit

Teff (K)

Luminosity (L )

Radius (R )

logðgÞ (dex)

Mass (M )

Isochrone SED

8511 ± 196 8500 ± 250

7.94 ± 1.83 10.74 ± 0.35

1.30 ± 0.07 1.51 ± 0.05

– 4.0±0.5

1.14 ± 0.04 1.23 ± 0.02

Table 4. UVIT photometry table of the BHB1, BHB2 and FUV bright BSS containing their respective positions, the extinction-corrected magnitudes and the magnitude errors. ID

RAJ2000

DEJ2000

BaF2

eBaF2

Sapphire

eSapphire

Silica

eSilica

HB01 HB02 HB03 HB04 HB05 HB06 HB07 HB08 HB09 HB10 HB11 HB12 HB13 HB14 HB15 HB16 HB17 HB18 HB19 HB20 HB21 HB22 HB23 HB24 HB25 HB26 HB27 HB28 HB29 HB30 HB31 HB32 HB33 HB34 HB35 HB36 HB37 BSS01

182.5358 182.5117 182.5076 182.5239 182.5231 182.5243 182.5302 182.5229 182.5309 182.5159 182.5288 182.5297 182.5111 182.5240 182.5394 182.5644 182.5427 182.5103 182.5291 182.5228 182.5315 182.5528 182.5381 182.5279 182.5417 182.5150 182.5268 182.5312 182.5171 182.5097 182.5226 182.4897 182.5668 182.5318 182.5307 182.5292 182.5123 182.5633

18.5215 18.5330 18.5353 18.5356 18.5393 18.5396 18.5413 18.5422 18.5458 18.5455 18.5465 18.5473 18.5564 18.5567 18.5611 18.5639 18.6696 18.5344 18.5376 18.5379 18.5503 18.5609 18.5407 18.5506 18.5302 18.5352 18.5388 18.5420 18.5441 18.5476 18.5489 18.5769 18.5771 18.5426 18.5430 18.5481 18.6111 18.5480

19.57 19.79 19.83 19.43 19.46 19.62 19.45 19.53 20.05 19.41 19.70 19.97 19.95 19.85 19.74 19.49 19.52 19.54 19.22 19.51 19.49 19.36 19.28 19.44 20.63 21.78 20.77 21.26 21.84 20.59 20.75 21.73 21.60 20.99 21.96 21.69 21.48 21.80

0.01 0.02 0.02 0.01 0.01 0.01 0.01 0.01 0.01 0.02 0.01 0.01 0.01 0.02 0.01 0.01 0.01 0.02 0.03 0.01 0.01 0.01 0.01 0.02 0.03 0.04 0.02 0.03 0.04 0.02 0.03 0.04 0.03 0.03 0.04 0.04 0.05 0.05

19.37 19.62 19.55 19.23 19.13 19.48 19.27 19.22 19.84 19.13 19.47 19.59 20.07 19.40 19.65 19.42 19.23 19.26 19.09 19.22 19.43 19.11 19.14 19.23 20.54 21.13 20.60 20.84 21.21 20.15 20.13 21.17 21.29 21.05 21.55 21.45 21.68 21.81

0.02 0.02 0.02 0.02 0.01 0.02 0.01 0.01 0.02 0.02 0.02 0.02 0.02 0.01 0.01 0.02 0.02 0.01 0.03 0.01 0.02 0.01 0.01 0.04 0.03 0.03 0.03 0.03 0.04 0.02 0.02 0.04 0.03 0.04 0.05 0.06 0.05 0.05

19.35 19.44 19.41 19.55 19.31 19.29 19.45 19.23 20.08 19.15 19.31 19.70 19.73 19.84 19.41 19.44 19.16 19.33 18.98 19.09 19.36 19.33 19.21 19.38 20.05 20.53 20.26 20.82 20.26 19.89 19.91 20.27 20.59 20.07 – – – –

0.02 0.02 0.02 0.03 0.03 0.03 0.02 0.03 0.03 0.02 0.03 0.03 0.02 0.03 0.03 0.02 0.02 0.03 0.03 0.03 0.02 0.02 0.03 0.02 0.03 0.03 0.04 0.06 0.04 0.05 0.02 0.04 0.04 0.03 – – – –

IDs HB01 to HB24 belong to the BHB1 group and IDs HB25 to HB37 belong to the BHB2 group. The BSS is listed at the end of the table with ID BSS01.

J. Astrophys. Astr. (2021)42:36

solid black circles and the observed fluxes from the UVIT, GALEX, CFHT-3.6 m, SDSS, GAIA, PANSTARRS, and CTIO/DECam telescopes are shown in green, red, blue, brown, violet, magenta, and yellow diamonds, respectively. In the lower panel of Fig. 6, we have shown the fractional deviation of the observed flux from the model flux and indicated as residual [(=Fobs:  Fmod Þ=Fmod Þ]. The best-fitted Kurucz model spectra with physical parameters Teff = 8500 K, ½Fe=H ¼ 2:0 and logðgÞ ¼ 4:0, was obtained with a reduced chi-square value of 9.3. The deviation of the observed fluxes from the theoretical fluxes for most of the filters is less than 15%. But for a few filters (e.g., CTIO/DECam g and Gaia G filters) the deviation is more than or around 15% which might be the reason for the high value of v2 . However, we can see that the overall photometric magnitudes are well within the maximum photometric errors of 20% (0.2 mag). We extracted bolometric luminosity and radius of the FUV bright BSS from the slope of the best fitted spectrum using scaling relation ðR=dÞ2 used in the SED fit, where R is the radius of the source and d is the distance. The distance of the cluster = 19.0 kpc and extinction value Av ¼ 0:068 magnitude were used in the SED fit. We have also estimated the mass of the BSS using the Teff , luminosity and radius derived from the SED analysis. The derived physical parameters from the SED fit and the BaSTI-IAC isochrone fitting are listed in Table 3. We find that the Teff , mass and luminosity of the BSS, derived from the SED fit and BaSTI-IAC isochrone fitting on CMDs are matching well within the error range.

Page 9 of 10

36

by fitting younger BaSTI-IAC isochrones in optical and UV-optical CMDs, and SED fitting of the observed fluxes from UV to near-IR wavelengths with the Kurucz model. The derived physical parameters of the FUV bright BSS are listed in Table 3.

Acknowledgements We would like to thank the referee for her/his valuable suggestions and comments. RK would like to acknowledge CSIR Research Fellowship (JRF) Grant No.09/983(0034)/2019-EMR-1 for the financial support. ACP would like to acknowledge the support by Indian Space Research Organization, Department of Space, Government of India (ISRO RESPOND Project No. ISRO/RES/2/409/17-18). ACP thanks InterUniversity Centre for Astronomy and Astrophysics (IUCAA), Pune, India for providing facilities to carry out his work. DKO acknowledges the support of the Department of Atomic Energy, Government of India, under Project Identification No. RTI 4002. AM thanks DST-INSPIRE (IF150845) for the funding. This publication uses the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Center (ISSDC). The UVIT data used here was processed by the Payload Operations Centre at IIA. The UVIT is built in collaboration between IIA, IUCAA, TIFR, ISRO and CSA.

References 6. Summary and conclusion UV analysis of the GGC NGC 4147 is presented using observations in three FUV filters of UVIT. A catalog of the UVIT detected BHBs, and BSS along with their FUV magnitudes in three UVIT filters is provided in Table 4. It is found that all the FUV bright BHBs belong to the 2G population as defined by Villanova et al. (2016) and they lie in two sub-populations on the UV-optical CMDs with a count ratio of 24:13 (BHB1:BHB2). The derived ranges of Teff of BHB1 and BHB2 are 10000–11300 K and 8000–9500 K, respectively with a gap of 500 K between both the sub-populations. Spectroscopic analysis of the sources is required to investigate further on the distinction between the two sub-population of BHBs. The physical parameters of the FUV bright BSS were derived

Ahn C. P., Alexandroff R., Allende Prieto C. et al. 2012, ApJS, 203, 21 Arellano Ferro A., Are´valo M. J., La´zaro C. et al. 2004, Revista Mexicana de Astronomia y Astrofisica, 40, 209 Arellano Ferro A., Rojas Galindo F. C., Muneer S., Giridhar S. 2018, Revista Mexicana de Astronomia y Astrofisica, 54, 325 Auriere M., Lauzeral C. 1991, A&A, 244, 303 Bastian N., Lardo C. 2018, ARA&A, 56, 83 Bayo A., Rodrigo C., Barrado Y Navascue´s D. et al. 2008, A&A, 492, 277 Behr B. B., Djorgovski S. G., Cohen J. G. et al. 2000, ApJ, 528, 849 Brown T. M., Cassisi S., D’Antona F. et al. 2016, ApJ, 822, 44 Cardelli J. A., Clayton G. C., Mathis J. S. 1989, ApJ, 345, 245 Cassisi S., Salaris M. 2020, A&AR, 28, 5

36

Page 10 of 10

Castelli F., Kurucz R. L. 2003, in Piskunov N., Weiss W. W., Gray D. F., eds, Modelling of Stellar Atmospheres, Vol. 210, A20 Catelan M., Borissova J., Sweigart A. V., Spassova N. 1998, ApJ, 494, 265 Chambers K. C., Magnier E. A., Metcalfe N. et al. 2016, arXiv e-prints, arXiv:1612.05560 Chatterjee S., Rasio F. A., Sills A., Glebbeek E. 2013, ApJ, 777, 106 Clement C. 2017, in Catelan M., Gieren W., eds, European Physical Journal Web of Conferences, Vol. 152, WideField Variability Surveys: A 21st Century Perspective, EDP Sciences, France, 01021 Dieball A., Long K. S., Knigge C., Thomson G. S., Zurek D. R. 2010, ApJ, 710, 332 Ferraro F. R., Sills A., Rood R. T., Paltrinieri B., Buonanno R. 2003, ApJ, 588, 464 Ferraro F. R., Paltrinieri B., Fusi Pecci F. et al. 1997, A&A, 324, 915 Gaia Collaboration, Brown A. G. A., Vallenari A. et al. 2018, A&A, 616, A1 Gosnell N. M., Mathieu R. D., Geller A. M. et al. 2015, ApJ, 814, 163 Gratton R., Bragaglia A., Carretta E. et al. 2019, A&AR, 27, 8 Harris W. E. 2010, arXiv e-prints, arXiv:1012.3224 Hidalgo S. L., Pietrinferni A., Cassisi S. et al. 2018, ApJ, 856, 125 Ivans I. I. 2009, in American Astronomical Society Meeting Abstracts, Vol. 213, American Astronomical Society Meeting Abstracts #213, 341.06 Jadhav V. V., Sindhu N., Subramaniam A. 2019, ApJ, 886, 13 Jain R., Vig S., Ghosh S. K. 2019, MNRAS, 485, 2877 Knigge C., Leigh N., Sills A. 2009, Nature, 457, 288 Kumar A. et al. 2012, in Takahashi T., Murray S. S., den Herder J.-W. A., eds, Proc. SPIE Conf. Ser., Vol. 8443, Space Telescopes and Instrumentation 2012: Ultraviolet to Gamma Ray, SPIE, Bellingham, p. 84431N Kumar R., Pradhan A. C., Parthasarathy M., Ojha D. K., Mohapatra A., Murthy J. 2020, in Bragaglia A., Davies M., Sills A., Vesperini E., eds, Proc. IAU Symp. 351, Star Clusters: From the Milky Way to the Early Universe, Cambridge University Press, Cambridge, UK, p. 464 Kumar R., Pradhan A. C., Mohapatra A. et al. 2021, MNRAS 502, 313

J. Astrophys. Astr. (2021)42:36 Lata S., Pandey A. K., Pandey J. C. et al. 2019, AJ, 158, 51 Leigh N., Knigge C., Sills A. et al. 2013, MNRAS, 428, 897 Marino A. F., Milone A. P., Renzini A. et al. 2019, MNRAS, 487, 3815 Martell S. L., Smith G. H., Briley M. M. 2008, AJ, 136, 2522 Milone A. P., Piotto G., Renzini A. et al. 2017, MNRAS, 464, 3636 Nardiello D., Libralato M., Piotto G. et al. 2018, MNRAS, 481, 3382 Pietrinferni A., Hidalgo S., Cassisi S. et al. 2021, ApJ, in press, arXiv e-print, arXiv:2012.10085 Piotto G., Zoccali M., King I. R. et al. 1999, AJ, 118, 1727 Postma J. E., Leahy D. 2017, PASP, 129, 115002 Rani S., Pandey G., Subramaniam A., Sahu S., Kameswara Rao N. 2020, J. Astrophys. Astr., 41, 35 Sahu S., Subramaniam A., Coˆte´ P., Rao N. K., Stetson P. B. 2019a, MNRAS, 482, 1080 Sahu S., Subramaniam A., Simunovic M. et al. 2019b, ApJ, 876, 34 Schiavon R. P., Dalessandro E., Sohn S. T. et al. 2012, AJ, 143, 121 Schlafly E. F., Finkbeiner D. P. 2011, ApJ, 737, 103 Sindhu N., Subramaniam A., Jadhav V. V. et al. 2019, ApJ, 882, 43 Singh G., Sahu S., Subramaniam A., Yadav R. K. S. 2020, ApJ, 905, 44 Stetson P. B. 1987, PASP, 99, 191 Stetson P. B., Catelan M., Smith H. A. 2005, PASP, 117, 1325 Stetson P. B., Pancino E., Zocchi A., Sanna N., Monelli M. 2019, MNRAS, 485, 3042 Subramaniam A., Sindhu N., Tandon S. N. et al. 2016, ApJL, 833, L27 Subramaniam A., Sahu S., Postma J. E. et al. 2017, AJ, 154, 233 Suntzeff N. B., Kraft R. P., Kinman T. D. 1988, AJ, 95, 91 Tailo M., Milone A. P., Lagioia E. P. et al. 2020, MNRAS, 498, 5745 Tandon S. N., Subramaniam A., Girish V. et al. 2017, AJ, 154, 128 Tandon S. N., Postma J., Joseph P. et al. 2020, AJ, 159, 158 Villanova S., Monaco L., Moni Bidin C., Assmann P. 2016, MNRAS, 460, 2351

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:84 https://doi.org/10.1007/s12036-020-09682-3

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

New results from the UVIT survey of the Andromeda galaxy D. A. LEAHY1,* , J. POSTMA1, M. BUICK1, C. MORGAN1,

L. BIANCHI2 and J. HUTCHINGS3 1

Department of Physics and Astronomy, University of Calgary, Calgary, Canada. Johns Hopkins University, Baltimore, MD, USA. 3 Herzberg Institute of Astrophysics, Victoria, BC, Canada. *Corresponding Author. E-mail: [email protected] 2

MS received 29 October 2020; accepted 4 December 2020 Abstract. The Andromeda galaxy (M31) has been observed with the UltraViolet Imaging Telescope (UVIT) instrument onboard the AstroSat Observatory. The M31 sky area was covered with 19 fields, in multiple UV filters per field, over the period of 2017 to 2019. The entire galaxy was observed in the FUV F148W filter, and more than half observed in the NUV filters. A new calibration and data processing is described which improves the astrometry and photometry of the UVIT data. The high spatial resolution of UVIT (  1 arcsec) and new astrometry calibration (  0.2 arcsec) allow identification of UVIT sources with stars, star clusters, X-ray sources, and other source types within M31 to a much better level than previously possible. We present new results from matching UVIT sources with stars measured as part of the Panchromatic Hubble Andromeda Treasury project in M31. Keyword. UV astronomy—galaxies: M31.

1. Introduction M31 is the closest neighboring large galaxy to our galaxy. It is a large spiral having many similarities to our galaxy and can be used as a template to study the many aspects of our galaxy which are difficult to study because of our location inside it and the resulting high extinction to much of our galaxy. Another advantage of studying objects in M31 is that it is at a well known distance (783 kpc, McConnachie et al. 2005) thus the uncertainty in intrinsic brightness for many objects is better known than for most Galactic sources. M31 has been observed in optical on numerous occasions. The highest resolution observations are carried out with the Hubble Space Telescope, including the Pan-chromatic Hubble Andromeda Treasury (PHAT) survey (Williams et al. 2014). In near and far ultraviolet (NUV and FUV), the GALEX instrument (Martin et al. 2005) has surveyed M31. This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

AstroSat has four instruments, covering NUV and FUV with the UltraViolet Imaging Telescope (UVIT), and soft through hard X-rays with the Soft X-ray Telescope (SXT), Large Area Proportional Counters (LAXPC) and Cadmium–Zinc–Telluride Imager (CZTI) instruments (Singh et al. 2014). We are carrying out a survey of M31 in NUV and FUV with UVIT. UVIT observations have high spatial resolution (’1 arcsec) and have capability of resolving individual stellar clusters and a large number of individual stars in M31. Previous observations of M31 with UVIT were presented in part by Leahy et al. (2017, 2020a) and Leahy and Chen (2020). Those papers presented an M31 UVIT point source catalog, matching M31 UVIT sources with Chandra sources, and analysis of UV bright stars in the bulge, respectively. In this paper we describe new UVIT data processing which gives important improvements in astrometric accuracy and photometric accuracy for UVIT data. We compare the UVIT Field 2 data to

84

Page 2 of 11

J. Astrophys. Astr. (2021)42:84

observations in optical from the PHAT survey to study the properties of the stars northeast of the bulge of M31.

Table 1 here gives the basic properties these four fields, including filters, exposure times and dates of observation.

2. Observations

3. Data analysis

UVIT consists of two 38 cm telescopes, each with field-of-view of  28 arcmin in diameter. The UVIT telescope and calibration are described in Tandon et al. (2017a, b, 2020), Postma et al. (2011), Leahy et al. (2020b) and references therein. One telescope is for far ultraviolet (FUV) (130 to 180 nm) wavelengths and one for near ultraviolet (NUV) (200 to 300 nm) and visible (VIS) (320 to 550 nm) wavelengths. The FUV, NUV and VIS channels each have a number of filters with different bandpasses. The VIS channel is used for spacecraft pointing, so normally science observations are carried out simultaneously in FUV and NUV channels. The pixel scale for UVIT images is 0.4168 arcsec per pixel and point sources in the UVIT images have FWHM ’1 arcsec in the FUV and NUV channels. The survey of M31 with UVIT has been carried out since 2017. With UVIT’s 28 arcmin in diameter field of view, 19 different fields are required to cover the sky area of M31. Exposures in the following filters are being used: F148W (123 to 173 nm), F154W (135 to 173 nm), F169M (146 to 175 nm), F172M (165 to 178 nm), N219M (206 to 233 nm) and N279N (275 to 284 nm). The NUV channel failed in early 2018, so that science observations since then have been carried out in FUV only. Fortunately much of the UVIT survey which overlaps with the PHAT survey was observed prior to the NUV failure, so that most of the PHAT area is covered by both NUV and FUV UVIT observations. The sky positions of the 10 fields covering and adjacent to the area surveyed by HST as part of PHAT are shown in Fig. 1. After the previous work (Leahy et al. 2020a) we developed new detector distortion corrections and new position calibration tools for processing UVIT images. Thus for the current work we are reprocessing the previous data on M31. For this study, we are comparing UVIT FUV and NUV data with NUV (the 275 nm band F275W filter) and optical data from the PHAT survey. The UVIT N279N filter covers nearly the same waveband as the HST F275W filter. Thus for the current analysis we restrict ourselves to the four UVIT fields which overlap with PHAT (see Fig. 1) and also have N279N data: Fields 1, 2, 7 and 13.

The UVIT detector distortion maps are utilized in the CCDLAB UVIT Pipeline (Postma & Leahy 2017). We have modified the pipeline to reduce astrometric errors as well as a reduce the PSF of point sources. Previously the distorion maps were utilized simply at the unit pixel scale of the CMOS given that this is the scale at which they were measured upon, in comparison to the 1/8th pixel scale of the final science centroid-based images. And so although the distortion maps do not have appreciably high-frequency spatial components, there are nonetheless finite differentials in distortion from any given CMOS pixel to the next. Thus, bilinear interpolation of the distortion map is now implemented and is applied to each centroid at the 1/32 pixel scale. Photometric calibration improvements have also been applied with updated filter-wise flat fields. Previously the flats were only at the detector level (Tandon et al. 2017a, b), but ongoing in-orbit calibration has allowed for second-order corrections to these flat fields to be developed for each filter (Tandon et al. 2020). Additionally, the previous WCS solutions were typically solved with approximately only ten sources across the field. With the development of the trigonometric WCS auto-solving algorithm (Postma & Leahy 2020) now implemented in CCDLAB, typically hundreds of sources spread across the field are now used for the plate solution. This provides for a more accurate average WCS solution across the field, with solution residuals commonly falling at 0.2 arcsec standard deviation.

3.1 Point source analysis This study aims to obtain UVIT photometry for stars already identified in the PHAT survey. Thus we searched for an excess above local background, in each UVIT reprocessed image, within 100 of the positions of the known PHAT sources. To avoid confusion of identifying a single UVIT source with either neighboring UVIT sources or with multiple PHAT sources, we only searched at the positions of

J. Astrophys. Astr. (2021)42:84

Page 3 of 11

84

Figure 1. Sky positions of the 10 fields (black circles) that make up the northern half of M31 UVIT survey, in J2000 coordinates. Overlaid on the UVIT fields are the outlines (grey rectangles) of the 23 areas (‘PHAT bricks’) observed by HST in the PHAT project (Williams et al. 2014). Table 1. M31 UVIT observations for fields with N279N data overlapping the PHAT survey. Field

RA (deg)1

Dec (deg)1

1 2 7 13

10.71071 11.03700 11.22142 11.59533

41.25023 41.55735 41.16111 41.53595

Filter2 a, a, a, a,

e, f d, e, f d, e, f d, e, f

Exposure time

Mean BJD3

7872, 7920, 4347 7940, 16022, 7977, 7424 4965, 10693, 10774, 3147 4974, 10609, 10848, 5005

2457671 (?0.8010,?0.8010,?1.2626) 2457704 (?0.1355,?0.4858,?0.1354,?0.4857) 2458071 (?0.1875,?0.4520,?0.4520,?0.1874) 2458092 (?0.2157,?0.4981,?0.4327,?0.2157)

1

RA and Dec are the J2000 coordinates of the nominal pointing center of the observation. 2 Filter labels are a: F148W, b: F154W, c: F169M, d: F172M, e: N219M, f: N279N. 3 Mean BJD is the mean solar-system Barycentric Julian Date of the observation. The common integer part for multiple observations is given as the first number.

PHAT sources which had F275W (275 nm band) magnitudes, and which were separated from eachother by more than 3.7500 1. We used the CCDLAB software (Postma & Leahy 2017) with a threshold Tests showed that separations of 9 pixels (3.7500 ) was enough to

1

obtain good flux measurements for adjacent UVIT sources, and this same separation was more than enough to avoid confusion with nearby PHAT sources.

of 3 sigma excess for a single pixel in the box and 8 sigma above local background for the total counts in the box. The centroids of the excesses were kept in a list of candidate point sources. Then we perform a PSF (point spread function) fit to a 9 pixel  9 pixel box surrounding the position of the source to derive a more accurate position and total counts for each source.

84

Page 4 of 11

The UVIT PSF has extended wings (Fig. 5 and Table 7 of Tandon et al. 2017b). Thus we made measurements of the correction from 9  9 box net counts to larger (17  17 pixels) box counts using fits to a set of bright isolated point sources in each field where we could use the larger box without interfering sources. An additional small correction was added to correct for the counts in a 17  17 pixel box to total counts in a circle of radius 27 pixels using the data in Tandon et al. (2017b). This way counts for sources separated by more than a few pixels were determined with reasonable accuracy. The counts and exposure times were used to convert to AB magnitudes using the updated calibration in Tandon et al. (2020). The position uncertainties for UVIT were determined as follows. The images were registered to optical position calibrators from Gaia as part of the reprocessing the images described in the previous section above. The standard deviation of the resulting Gaia-UVIT offsets ranged from 0.17 arcsec to 0.19 arcsec (see Table 2). Figure 2 shows the distributions of all individual Gaia-UVIT offsets for fields 1, 2, 7 and 13 from the new data processing. Figure 3 shows the distributions of the new-processing Gaia-UVIT offsets and the previous Gaia-UVIT offsets for the point sources combined from fields 1, 2, 7 and 13. The new offsets are smaller than those from the previous (pre-2020) UVIT data processing procedure.

4. Results and discussion Our analysis above produced a catalog of UVIT AB magnitudes in the different observed UVIT filters (see Table 1) for PHAT sources with measured F275W magnitudes, and with separations between PHAT sources [3.7500 . For these individual stellar sources, we have the UVIT AB magnitudes for the different observed UVIT filters (Table 1) and PHAT Vega magnitudes from the PHAT source catalog (Williams et al. 2014). The UVIT bands range from F148W (’150 nm) to N279N ( (’280 nm), and the PHAT bands range from F275W (’275 nm) to F160W (’1600 nm). With this combined photometry, we can study the nature of individual stars in M31. The UVIT Field 2 has the best overlap with PHAT, so we concentrate the current work on Field 2. A three-color image was made using the 280 nm N279N (red), 220 nm N219M (green) and 150 nm F148W (blue) images, which were position registered as part of the data processing. Figure 4 show the UVIT 3

J. Astrophys. Astr. (2021)42:84

color image of Field 2. The hot stars show up as blueish and the coolest stars as yellowish or reddish. There are color artifacts around the edge of the image because we used full field images from all three filters, but the pointing centers for the three images were slightly different. This means that the outer edges of the images can have data from only one or two of the three filters, which results in color errors, but not position distortion. We compare the UVIT images, with  1 arcsecond resolution, to existing GALEX images of M31, which were taken with  5 arcsecond resolution. A concentration of hot stars is seen near RA 00:44:40, Dec. ?41:26 in the Field 2 three-color image in Fig. 4. In Fig. 5 here, we show an expanded view ’1.5 arcmin by 1.5 arcmin across centered at 00:44:40.5 Dec.?41:26:37 (J2000). The left hand panel is the deepest image from the GALEX survey (from the Deep Imaging Survey, DIS), the right panel is the UVIT image in the F148W filter. The advantage of the higher resolution of UVIT is clearly seen, as several blobs of emission in the GALEX image are resolved into separate sources in the UVIT image. Here, the 3 brightest GALEX sources, just left of center in the GALEX image, are resolved into  19 separate sources in the UVIT F148W image. Figure 6 shows an expanded UVIT F148W image, 30 arcsec by 30 arcsec, of the concentration of stars to left of center in Fig. 5. UVIT is sensitive in N279N (same waveband as F275W) only to stars with N279N AB magnitudes brighter than ’22. The conversion factor from AB to Vega magnitudes for the N279N filter is MAB ¼ MVega þ 1:48 using the N279N filter response given in Tandon et al. (2017b). Thus this limit correspond to F275W Vega magnitudes brighter than ’20.5. Overlaid on the UVIT image are the positions of all PHAT stars that have F275W Vega magnitude brighter than 20.5. It is seen that all UVIT F148W (150 nm) detections have a F275W counterpart, but not all F275W stars are detected in F148W. This is expected because only the hottest stars (Teff J10; 000 K) are strong emitters for wavelengths as short as 150 nm. In several locations multiple PHAT F275W stars overlap the UVIT F148W source. This occurs when there is source confusion in the UVIT image. However, there are a significant number of UVIT F148W sources that can be identified with unique PHAT F275W sources. Figure 7 shows the same UVIT F148W image with the concentration of hot stars, but is now overlaid with the stellar clusters listed in Johnson et al. (2015). Four

J. Astrophys. Astr. (2021)42:84

Page 5 of 11

Table 2. Astronometry errors for fields listed in Table 1. Field 1b 2b 7b 13b

Std Dev.a

No. of calibrators

0.192 0.177 0.182 0.176

271 215 273 332

a

Std Dev. is the standard deviation of positions with respect to Gaia positions in units of arcsec. b N279N is the filter image used for calibration with Gaia positions.

50

Number of Sources per Bin

Field 1 Field 2

40

Field 7 Field 13 30

20

10

0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

UVIT_Gaia_Separation (arcsec)

Figure 2. The distributions of Gaia-UVIT point source offsets for fields 1, 2, 7 and 13.

Number of Sources per Bin

160

new UVIT positions old UVIT positions

140 120 100 80 60 40 20 0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

UVIT_Gaia_Separation (arcsec)

Figure 3. The distributions of the new-processing GaiaUVIT offsets (red solid histogram) and the previous GaiaUVIT offsets (blue dashed histogram) for point sources combined from fields 1, 2, 7 and 13.

84

of those clusters are located in the region, with centers of the clusters marked by small red squares and with Rap of each cluster marked by a red circle. Two of the clusters are compact (Rap ¼ 2:66 and 2.63 arcseconds) and two are extended (Rap ¼ 4:67 and 3.81 arcseconds). The compact clusters are not resolved by UVIT, and each compact cluster has several (  10) PHAT F275W stars in it (see Fig. 6). The extended clusters are mostly resolved by UVIT, with more than half of the UVIT sources in each cluster corresponding with single PHAT F275W stars (see Fig. 6). Thus UVIT can resolve the larger clusters, Rap J3 arcseconds, in M31 into individual stars, and thus allow fitting photometry of individual stars. Most hot stars outside clusters are less crowded, so also can be identified with unique UVIT and PHAT identifications. For this work we consider the color-magnitude diagram (CMD) for UVIT sources which can be identified with individual PHAT stars. Model fitting of individual stars UVIT photometry combined with PHAT photometry will be done in future work. To study the CMD, we use our list of all isolated UVIT sources with unique F275W PHAT counterparts, that are not also associated with the known clusters in M31 from Johnson et al. (2015). We found all UVIT sources with N279N detections which match with a PHAT F275W source within 100 in Field 2, then removed any UVIT sources which are within 9 pixels (’3.7500 ) of a second PHAT F275W source. Then we removed those sources with inconsistent N279N and F275W magnitudes, which indicates accidental matches. To have useful photometry for CMDs, we removed those sources from the list with large errors (J0.3) on the UVIT N279N magnitudes by restricting the N279N AB magnitude to be brighter than 22, leaving 851 sources in Field 22. The sky positions of these sources are shown in Fig. 8. We have made a catalogue of the sources from all four fields which were matched to unique PHAT sources (\100 separation, N279N magnitude  22). The catalogue contains the UVIT photometry for all available bands and consists of 1705 UVIT sources. A sample of part of the catalogue is given in Table 3 here, and the full catalogue is available online at https://github.com/denisleahy/ M31-UVIT-sources-at-PHAT-positions.

2

The errors increase rapidly for magnitudes fainter than 22. For example, at magnitude 23, which is roughly the detection limit, typical errors are [0.45 magnitudes.

84

Page 6 of 11

J. Astrophys. Astr. (2021)42:84

Figure 4. UVIT three-color image of the M31 survey Field 2, 28 arcmin in diameter. The N279N image is red, the N219M image is green, and the F148W image is blue. From edge to edge, the image is ’6.4 kpc across at the distance of M31. Coordinates are J2000 RA and Dec.

The UVIT CMD for these sources is shown in Fig. 9, where the temperature indicator is F148WAB magnitude – N279NAB magnitude (hotter stars on the left), and the luminosity indicator is the N279NAB magnitude. The CMD shows two branches on the CMD: a near-vertical branch at F148W-N279N color ’0–1 and and a sloped branch to the right from color F148W–N279N ’ 2, N279N magnitude ’ 22 to color F148W–N279N ’ 6, N279N magnitude ’18. The former likely corresponds to hot main-sequence stars; and the latter may correspond to cooler evolved giant stars. To verify this, a theoretical CMD was calculated for stars at the distance of M31, for main sequence stars

of spectral type O3V through to G0V, giants BOIII through to M0III and supergiants B0I through to M0I. The radii, effective temperatures and log(g) values were taken from the recommended values for Castelli and Kurucz Atlas of Stellar Atmosphere Models at the Space Telescope Science Institute (https://www.stsci. edu/hst/instrumentation/reference-data-for-calibration -and-tools/astronomical-catalogs/castelli-and-kuruczatlas). For example, the O3V star has radius 15R , Teff ¼ 44800 K and logðgÞ ¼ 3:92, the G5III star has radius 10R , Teff ¼ 51500 K and logðgÞ ¼ 2:54, and the K5I star has 459R , Teff ¼ 3850 K and logðgÞ ¼ 0:00. The theoretical CMD is shown in Fig. 10 for 3 different values of extinction. AV ¼ 0:2 corresponds

J. Astrophys. Astr. (2021)42:84

Page 7 of 11

84

Figure 5. Comparison of the GALEX image (left) with the UVIT F148W image (right) for the same region in M31. The area is 1.5 arcmin across and centered at R.A. 00:44:40.5 Dec.?41:26:37 (J2000).

to the standard foreground extinction from the Milky Way in the direction of M31. We have included AV ¼ 0:7 and AV ¼ 1:2, corresponding to additional internal M31 extinction of 0.5 and 1.0. From Fig. 10, it is seen that normal upper main sequence stars (O3V to B5V) in M31 fall in the same region as the observed UVIT sources. B0I, A0I, F0I and B0III stars also fall in the same area as observed UVIT sources. None of the stars in the theoretical CMD fall in the region of the sloped branch on the right side of Fig. 9. However cooler main sequence stars (A0V through F5V) lie at

Figure 6. Expanded view (greyscale) of the UVIT F148W image of Field 2, 30 arcsec across, centered on the bright group of stars east of R.A. 00:44:40.5 Dec.?41:26:37 (J2000) (center of the image in Fig. 3). The red squares mark the positions of all stars from the PHAT survey with F275W Vega magnitude \20.5.

the same color range (F148W–N279N color ’2–8) but much fainter N279N magnitudes. However foreground stars, which are  100–1000 times closer, are brighter by 15 to 10 magnitudes, which places them in the same N279N magnitude range (16 to 21) as the observed UVIT sources. We conclude that most of the sources in the upper-right quadrant of the observed UVIT CMD diagram are foreground stars.

Figure 7. Expanded view (greyscale) of the UVIT F148W image of Field 2, 30 arcsec across, centered on the bright group of stars east of R.A. 00:44:40.5 Dec.?41:26:37 (J2000) (center of the image in Fig. 3). The red circles with squares at their centers mark the stars clusters from Johnson et al. (2015) that are in this region, with the radius of the circle set to the Rap from Johnson et al. (2015).

84

Page 8 of 11

J. Astrophys. Astr. (2021)42:84

Figure 8. Sky positions of the UVIT sources from Field 2 matched with unique PHAT sources.

Table 3. Photometrya;b of unique UVIT sources matched to PHAT (first 10 linesc ). UVIT field 1 1 1 1 1 1 1 1 1 1 F172M (AB mag) 23.35 22.77 22.90 22.67 22.67 22.58 22.88 22.88 23.26 23.29

x

y

RA (deg)

RA err. (deg)

DEC (deg)

DEC err. (deg)

F148W (AB mag)

F148W err. (AB mag)

919.39 2234.93 745.05 2217.53 919.23 2435.96 2130.13 2511.07 653.60 2547.25

1418.74 935.09 2340.21 1583.89 3867.61 1405.44 2629.24 275.40 3727.83 1728.91

10.72342 10.70634 10.70040 10.71760 10.73030 10.71301 10.71587 10.69291 10.73082 10.71788

0.00006 0.00013 0.00009 0.00008 0.00012 0.00021 0.00016 0.00007 0.00022 0.00014

41.27808 41.29259 41.29403 41.29573 41.28674 41.29761 41.26595 41.29296 41.29087 41.30073

0.00014 0.00015 0.00011 0.00012 0.00007 0.00017 0.00008 0.00003 0.00005 0.00014

23.25 23.26 99.99 23.01 23.95 99.99 99.99 22.75 23.84 23.12

0.19 0.19 99.99 0.17 0.25 99.99 99.99 0.16 0.24 0.18

F172M err. (AB mag)

F169M (AB mag)

F169M err. (AB mag)

N219M (AB mag)

N219M err. (AB mag)

N279N (AB mag)

N279N err. (AB mag)

UVIT–PHAT separation (00 )

0.29 0.22 0.23 0.21 0.21 0.20 0.23 0.23 0.28 0.28

22.44 21.37 22.54 22.18 22.54 23.88 99.99 21.50 22.46 22.38

0.14 0.10 0.15 0.13 0.15 0.24 99.99 0.11 0.14 0.14

99.99 23.28 21.88 99.99 22.22 99.99 23.07 22.52 22.80 22.83

99.99 0.35 0.18 99.99 0.21 99.99 0.31 0.25 0.28 0.28

21.25 20.51 21.22 21.43 21.42 21.59 21.89 21.07 20.70 21.87

0.22 0.16 0.21 0.23 0.23 0.25 0.28 0.20 0.17 0.28

0.172 0.557 0.473 0.123 0.124 0.371 0.414 0.546 0.444 0.136

Notes: a The table is 18 columns by 712 rows. For display purposes here, the 18 columns are split into two sets of 9 columns. b Magnitude and error values of 99.99 indicate no flux detected in that filter, blank values means that position not observed in that filter. c The full table is available on-line in csv format as Table3_Full.csv at github.com under repository denisleahy/M31-UVIT-sources-at-PHAT-positions.

J. Astrophys. Astr. (2021)42:84

Page 9 of 11

84

17

N279N

18

19

20

21

22 -2

-1

0

1

2

3

4

5

6

7

8

F148W-N279N

Figure 9. UVIT FUV (150 nm F148W)–NUV (280 nm N279N) color-magnitude diagram for the UVIT sources from Field 2 matched with PHAT sources. Photometry errors are shown by the grey vertical and horizontal lines.

Figure 10. UVIT FUV (150 nm F148W)–NUV (280 nm N279N) color-magnitude diagram for a set of standard stars of luminosity classes I, III and V. Those with extinction AV ¼ 0:2 are marked in blue, with AV ¼ 0:7 in black, and AV ¼ 1:2 in red.

With position matching of the UVIT sources to PHAT sources, we make the PHAT CMD for these same sources. This is shown in Fig. 11, where the temperature indicator is F275WVega magnitude – F336WVega magnitude, and the luminosity indicator is the F336WVega magnitude. The PHAT CMD shows two branches on the CMD: a near-vertical branch at color ’–0.5 and and a sloped branch to the right from

Figure 11. PHAT NUV (275 nm F275W)–optical (336 nm F336W) color-magnitude diagram for the UVIT sources from Field 2 matched with PHAT sources. Photometry errors are shown but are in most cases similar size to the plotted symbols.

Figure 12. PHAT NUV (F275W)–optical (F336W) colormagnitude diagram for a set of standard stars of luminosity classes I, III and V. Those with extinction AV ¼ 0:2 are marked in blue, with AV ¼ 0:7 in black, and AV ¼ 1:2 in red.

color ’–0.5, F336W magnitude ’ 21 to color ’ 2, F336W magnitude ’ 17. To examine the nature of these stars, the theoretical PHAT CMD was calculated for stars at the distance of M31, for the same parameters as for the theoretical UVIT CMD. The theoretical PHAT CMD is shown in Fig. 12 for the 3 different values of extinction AV ¼ 0:2, 0.7 and 1.2. It is seen that normal upper

84

Page 10 of 11

J. Astrophys. Astr. (2021)42:84

For Fig. 13, we plot the model spectrum for the UVIT bands in AB magnitudes and the model spectrum for the PHAT bands in the Vega system, so the model and data show the same offsets between UVIT and PHAT bands. A single star does not fit the data. This is caused by crowding in the longer optical and IR bands (see Leahy et al. (2017) for similar model fits for stars in the bulge of M31), so is not surprising. The crowding is caused by two factors: the number of cool giant stars greatly exceeds that of hot main sequence stars, caused by the combination of initial mass function and stellar lifetimes; the PSF for PHAT is larger for longer wavelengths. A model consisting of a hot star plus a cool star fits the data well. The hot star is consistent with a massive young O-type star (  20M ); the cool star is consistent with a G-type evolved supergiant. Figure 13. Example stellar fit to SED from UVIT (F148W – 150 nm, F172M – 180 nm, N219M – 220 nm, N279N – 280 nm) and PHAT (F275W – 275 nm, F336W – 336 nm, F475W – 475 nm, F814W – 814 nm, F110W – 1100 nm, F160W – 1600 nm). The observed magnitudes and model magnitudes for UVIT are in the AB system. The observed magnitudes and model magnitudes for PHAT are in the Vega system. The data upper and lower limits are the black dashed lines. The plotted total model (blue) is for a hot star (green) plus cool star (red). The fit parameters are: hot star T, radius and AV of 49,000 K, 6.6 Rsun and 1.20; cool star T, radius and AV of 5700 K, 42 Rsun and 1.56.

main sequence stars (O3V to B5V) in M31 fall in the same region of the PHAT CMD as the near vertical branch of UVIT/PHAT sources. As for the UVIT CMD, none of the cool stars are luminous enough to fall in the cool branch of the observed PHAT CMD. However, foreground stars which are  100–1000 times closer (15 to 10 magnitudes brighter) do fall in the same area as the cool branch of the observed PHAT CMD. Thus we find that the cool branch of both UVIT and PHAT CMDs is consistent with foreground stars, rather than stars in M31. The UVIT FUV/NUV (4 filter bands) plus PHAT NUV-optical-IR (6 filter bands) photometry for a star from the main branch in the UVIT CMD is shown in Fig. 13. The drop between PHAT F275W (275 nm) band and the UVIT N279N (280 nm) band is a result of specifying UVIT magnitudes in the AB system and PHAT magnitudes in the Vega system. As noted above, for the 275–280 nm band the difference in the 2 systems is 1.48 mag, with Vega magnitudes brighter.

5. Conclusion The AstroSat Observatory has been carrying out a survey of M31 since 2017, which is nearly complete. The primary goal is to obtain near ultraviolet and far ultraviolet observations with the UVIT on AstroSat. Surveying M31 requirs 19 fields each of  28 arcminute diameter. All 19 fields were observed in the FUV F148W (150 nm) filter, and more than half of the fields observed in the NUV filters. Recently we have developed new calibration and data processing methods which improve the astrometry and photometry of UVIT data. With the new processing, the UVIT data has higher spatial resolution (’1 arcsec) and better astrometry (’0.2 arcsec). This has allowed us to identify point sources detected by UVIT with sources at other wavelengths. In particular, UVIT sources have been identified with stars observed at high resolution with the Hubble Space Telescope as part of the Pan-chromatic Hubble Andromeda Treasury project (PHAT, Williams et al. 2014). We are able to use color magnitude diagrams, using UVIT FUV–NUV colors and using PHAT NUV–visible colors to detect hot main sequence stars and to separate out likely foreground stars. Future work will focus on carrying out UVIT photometry for the case of crowded UVIT sources, including multiple PHAT stars near to eachother and star clusters, and on modelling the multi-band UVIT-PHAT FUV through optical photometry of different sets of stars in M31 to obtain good constraints on their properties.

J. Astrophys. Astr. (2021)42:84

Acknowledgements This project is undertaken with the financial support of the Canadian Space Agency and of the Natural Sciences and Engineering Research Council of Canada. This publication uses data from the AstroSat mission of the Indian Space Research Institute (ISRO), archived at the Indian Space Science Data Center (ISSDC). This publication uses UVIT data processed by the payload operations centre at IIA. The UVIT is built in collaboration between IIA, IUCAA, TIFR, ISRO and CSA.

References Johnson L. C., Seth A. C., Dalcanton J. J. et al. 2015, The Astrophys. J., 802, 127 Leahy D. A., Bianchi L., Postma J. 2017, The Astronom. J., 156, 269

Page 11 of 11

84

Leahy D. A., Postma J., Chen Y., Buick M. 2020a, Astrophys. J. Suppl. Ser., 247, 47 Leahy D. A., Postma J., Hutchings J., Tandon S. N. 2020b, Proceedings of the IAU 2020, pp. 487–491 Leahy D. A., Chen Y. 2020, Astrophys. J. Suppl. Ser., 250, 23 Martin D. C., Fanson J., Schiminovich D. et al. 2005, Astrophys. J., 619, L1 McConnachie A. W., Irwin M. J., Ferguson A. M. N. et al. 2005, MNRAS, 356, 979 Postma J. E., Hutchings J., Leahy D. 2011, PASP, 123, 833 Postma J. E., Leahy D. 2017, PASP, 129, 115002 Postma J. E., Leahy D. 2020, PASP, 132, 05403 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, SPIE, 9144E, 1S Tandon S. N., Hutchings J. B., Ghosh S. K. et al. 2017a, J. Astrophys. Astr., 38, 28 Tandon S. N., Subramaniam A., Girish V. et al. 2017b, The Astronom. J., 154, 128 Tandon S. N., Postma J., Joseph P. et al. 2020, The Astronom. J., 159, 158 Williams B. F., Lang D., Dalcanton J. J. et al. 2014, Astrophys. J. Suppl. Ser., 215, 9

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:33 https://doi.org/10.1007/s12036-021-09699-2

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

Detection of X-ray pulsations at the lowest observed luminosity of Be/X-ray binary pulsar EXO 2030+375 with AstroSat GAURAVA K. JAISAWAL1,* , SACHINDRA NAIK2 , SHIVANGI GUPTA2,

P. C. AGRAWAL3, ARGHAJIT JANA2 , BIRENDRA CHHOTARAY2,4 and PRAHLAD R. EPILI5 1

National Space Institute, Technical University of Denmark, Elektrovej 327-328, DK 2800 Lyngby, Denmark. 2 Astronomy and Astrophysics Division, Physical Research Laboratory, Navrangpura, Ahmedabad 380 009, India. 3 Department of Astronomy and Astrophysics, Tata Institute of Fundamental Research, Colaba, Mumbai 400 005, India. 4 Indian Institute of Technology, Gandhinagar, Palaj 382 355, India. 5 School of Physics and Technology, Wuhan University, Wuhan 430072, China. Corresponding Author. E-mail: [email protected] MS received 28 October 2020; accepted 24 December 2020 Abstract. We present the results obtained from timing and spectral studies of Be/X-ray binary pulsar EXO 2030?375 using observations with the Large Area Xenon Proportional Counters and Soft X-ray Telescope of AstroSat, at various phases of its Type-I outbursts in 2016, 2018, and 2020. The pulsar was faint during these observations as compared to earlier observations with other observatories. At the lowest luminosity of 2.51035 erg s1 in 0.5–30 keV energy range, 41.3 s pulsations were clearly detected in the X-ray light curves. This finding establishes the first firm detection of pulsations in EXO 2030?375 at an extremely low mass accretion rate to date. The shape of the pulse profiles is complex due to the presence of several narrow dips. Though pulsations were detected up to  80 keV when the source was brighter, pulsations were limited up to  25 keV during the third AstroSat observation at lowest source luminosity. A search for quasi-periodic oscillations in 2  104 Hz to 10 Hz yielded a negative result. Spectral analysis of the AstroSat data showed that the spectrum of the pulsar was steep with a power-law index of  2. The values of photon-indices at observed low luminosities follow the known pattern in sub-critical regime of the pulsar. Keywords. Stars: neutron—pulsars: individual: EXO 2030?375—X-rays: stars.

1. Introduction Accreting high mass X-ray binary (HMXB) pulsars are among the brightest X-ray sources in our Galaxy (Nagase 1989). In these binaries, a neutron star and a massive (  10M ) main-sequence star rotate around the common center of mass of the system in a wide and eccentric orbit (Tauris & van den Heuvel 2006). This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

The neutron star accretes matter from the companion star through the capture of stellar wind or Roche-lobe overflow. A majority of the HMXB systems are known to be Be/X-ray binaries (BeXRBs) in which the mass-donor is a non-supergiant B or O spectral type star which shows emission lines in its optical/ infrared spectrum (Reig 2011). Rapid rotation of the companion Be star in the BeXRB system expels its photospheric matter equatorially, forming a circumstellar disk around it. The continuously evolving, equatorial circumstellar disk is known to be the cause

33

Page 2 of 12

of the emission lines and infrared excess in the optical/infrared spectrum of the companion star in the BeXRBs. Significant evolution of the circumstellar disk allows the neutron star to capture copious amount of matter while passing through the periastron. This abrupt accretion of matter by the neutron star enhances its X-ray emission by several orders of magnitude which lasts for several days to tens of days. These events are termed as Type-I X-ray outbursts. Once the neutron star moves away from the periastron, accretion from the circumstellar disk is no more possible and the X-ray source returns to the quiescence. The long term X-ray activity in BeXRBs is characterized by the regular Type-I outbursts with peak luminosity of the order of Lx  1037 erg s1 and irregular rare giant (Type-II) X-ray outbursts with peak luminosity of Lx  1037 erg s1 . The Type-I X-ray outbursts are of short duration, covering 20–30% of orbit and coincide with the periastron passage of the neutron star whereas the Type-II outbursts show no preferred orbital phase dependence but once set in, they tend to cover a large fraction of the orbital period or even several orbital periods (see, e.g., Okazaki & Negueruela 2001; Reig 2011; Jaisawal & Naik 2016; Wilson-Hodge et al. 2018; Jaisawal et al. 2019). EXO 2030?375 is one of the well studied Be/X-ray binary pulsars associated with regular Type-I outbursts during almost every periastron passage. This transient accreting X-ray pulsar was discovered in 1985 with EXOSAT during a giant outburst (Parmar et al. 1989) with  42 s pulsations. The transient behaviour of this pulsar could be traced since its discovery when its initial 1–20 keV outburst luminosity (1:0  1038 d52 ) erg s1 on 1985 May 18 declined by a factor of  2600 within 100 days of the outburst. The associated optical counterpart of EXO 2030?375 is a highly reddened B0 Ve star (Motch & Janot-Pacheco 1987) showing infrared excess and Ha in emission (Coe et al. 1988). Using the relationship between extinction and distance of sources in the Galactic plane, Wilson et al. (2002) estimated the distance of EXO 2030?375 to be 7.1 kpc. The regular Type-I Xray outbursts of EXO 2030?375, occurring almost at every periastron passage of its  46-day orbit (Wilson et al. 2008), have been extensively monitored with the X-ray instruments onboard RXTE, INTEGRAL, XMMNewton, Suzaku and Swift/BAT observatories to understand the characteristic properties of the pulsar (Wilson et al. 2002; Naik et al. 2013; Naik & Jaisawal 2015; Ferrigno et al. 2016; Epili et al. 2017 and references therein).

J. Astrophys. Astr.

(2021) 42:33

In June 2006, EXO 2030?375 was caught for the second time in a giant (Type-II) X-ray outburst with initial flux of 180 mCrab. This surpassed the previous peak flux of about 50 mCrab observed during the entire life of the RXTE/ASM mission (Corbet & Levine 2006). The 2006 Type-II outburst was also followed by Swift/BAT which reported the peak flux steadily increased to 750 mCrab (Krimm et al. 2006). Spin-up trend was observed in the pulsar during the giant X-ray outbursts in 1985 (Parmar et al. 1989) and 2006 (Wilson et al. 2008) whereas spin-down episodes have been observed at low luminous outbursts in 1994–2002 (Wilson et al. 2002, 2005) and during faint outbursts after March 2016 (Kretschmar et al. 2016). The phase-averaged spectra of EXO 2030?375 during normal and giant outbursts prior to 2006 giant outburst were described with various phenomenological and physical models (in some cases) along with an iron emission line at 6.4 keV and interstellar absorption (Epili et al. 2017 and references therein). Apart from the continuum spectrum, several interesting features have also been observed in the pulsar spectrum. Suzaku observations of the pulsar EXO 2030? 375 during less intense Type-I outbursts in 2007 and 2012 did not show any evidence of cyclotron absorption features in its spectrum. However, presence of additional matter locked at certain pulse phases of the pulsar was reported and interpreted as the cause of several prominent absorption dips in the pulse profiles (Naik et al. 2013; Naik & Jaisawal 2015). During the brighter Type-I outburst in 2007, Naik et al. (2013) detected several narrow emission lines (i.e Si XII, Si XIV, S XV) for the first time along with Fe Ka and Fe XVI in the X-ray spectrum. A detailed and comprehensive study of EXO 2030?375 was carried out by using extensive RXTE pointed observations during many Type-I and 2006 Type-II outbursts starting from 1995 till 2011 (Epili et al. 2017). Timing and spectral studies of the pulsar were carried out in 3–30 keV luminosity range from 3:8  1036 to 2:6  1038 erg s1 , covered during the entire duration of RXTE campaign. Timing studies of more than 600 RXTE pointings revealed the evolution of pulse profiles of the pulsar with luminosity a main peak and a minor peak at low luminosity evolved into a two-peaked profile along with minor dips at high luminosity. This study revealed that pulse profiles of the pulsar at a particular luminosity were identical irrespective of the type of X-ray outbursts,

J. Astrophys. Astr.

(2021) 42:33

indicating that the emission geometry depends mainly on the accretion rate. Since the discovery in 1985, the pulsar had been showing regular X-ray outbursts for about 25 years. Since early 2015, however, the Type-I outbursts appeared to be of decreasing intensity and eventually vanished from the light curve towards the end of 2015 or early 2016 (Fu¨rst et al. 2016). The Type-I X-ray outburst activity commenced again in early 2016 and still continuing, though with much fainter peak luminosity (  1036 erg s1 ) than the usual ones. Fu¨rst et al. (2017) reported the detection of pulsation at a minimum luminosity of 6:8  1035 erg s1 in 3–78 keV range, considered to be the lowest luminosity of the pulsar with X-ray pulsations in the light curve. Though the pulsar was observed with Swift/XRT at a fainter phase, the data quality was not good enough for pulsation search. As the pulsar is still showing Type-I X-ray outbursts with fainter peak luminosity, it is interesting to carry out timing and spectral studies with AstroSat to explore whether the pulsar has gone into the propeller regime or still undergoing accretion. Detection of pulsations in the light curve at lower luminosity compared to that during the earlier Swift/ XRT observation (Fu¨rst et al. 2017), would rule out the onset of propeller regime. Further, detection of pulsation at a limiting luminosity may allow us in estimating the magnetic field of the pulsar. The AstroSat observations at lower luminosity, therefore, are important to investigate above properties of the pulsar. In this paper, we investigate the pulsation activities, shape of pulse profiles and spectral properties of the pulsar at a significantly lower luminosity level using five epochs of AstroSat observations. For comparison, data from NuSTAR observation of the pulsar on 25 July 2016, reported in Fu¨rst et al. (2017), were also used in present work. The observations of the pulsar and data reduction procedures are described in Section 2, results obtained from the timing and spectral analysis are presented in Sections 3 and 4, respectively. The implication of our results are discussed in Section 5.

2. Observations and data reduction 2.1 AstroSat The first Indian multi-wavelength astronomical satellite, AstroSat was launched by Indian Space Research Organization on 28 September 2015 (Agrawal 2006;

Page 3 of 12

33

Singh et al. 2014). The observatory is sensitive to photons from optical to hard X-ray ranges by using five sets of instruments such as Ultraviolet Imaging Telescope (UVIT; Tandon 2017), Soft X-ray Telescope (SXT; Singh et al. 2017), Large Area X-ray Proportional Counters (LAXPCs; Agrawal et al. 2017; Antia et al. 2017), Cadmium–Zinc–Telluride Imager (CZTI; Rao et al. 2017), and a Scanning Sky Monitor (SSM; Ramadevi et al. 2018). However, in the present study, five epochs of AstroSat observations of EXO 2030?375 with SXT and LAXPC instruments are used. Details of the observations are summarised in Table 1. As the source was very faint during all five epochs, it was not detected in the CZTI. The UVIT was not operational during these observations. AstroSat caught the source at different phases of the regular Type I X-ray outbursts. Figure 1 shows the MAXI (Monitor of All-Sky X-ray Image, Matsuoka et al. 2009) and Swift/BAT (Burst Alert Telescope, Krimm et al. 2013) monitoring light curves of the pulsar covering the epochs of AstroSat observations. The first, fourth and fifth AstroSat observations were carried out at the declining phase of the Type-I X-ray outbursts at a source luminosity of 15–40 mCrab with BAT. However, during the second observation, monitoring data from MAXI or Swift/BAT were not available. An extremely low level of X-ray intensity,  10 mCrab in 15–50 keV range, was estimated during the third AstroSat observation. A log of these AstroSat pointings of EXO 2030?375 is given in Table 1. The SXT is a soft X-ray focusing telescope onboard AstroSat. It consists of shells of conical mirrors that focus the soft X-ray photons in 0.3–8 keV energy range on a CCD detector. The field-of-view of the SXT is 40 arcmin. The effective area of the telescope is 90 cm2 at 1.5 keV. The energy resolution of the detector is 90 eV at 1.5 keV and 136 eV at 5.9 keV. The source was observed with SXT in the photon counting mode, yielding a time resolution of 2.4 s. We followed standard analysis procedure for the SXT data reduction as suggested by the AstroSat Science Support Cell (ASSC1). The source spectrum was extracted from a 8 arcmin circular region centered at the source coordinate on the SXT chip using XSELECT package. The background spectrum was extracted from the blank sky region on the chip. The LAXPC is a proportional counter detector sensitive to X-ray photons in the 3–80 keV energy range. There are three identical detector units onboard 1

http://astrosat-ssc.iucaa.in/.

33

Page 4 of 12

J. Astrophys. Astr.

(2021) 42:33

Table 1. Log of observations of EXO 2030?375 with AstroSat, NuSTAR and Swift/XRT. Start of observation

Exposure (in ks)

Observation ID

Date

MJD

LAXPC

SXT

G06_089T01_9000000746 G08_081T01_9000002144 G08_081T01_9000002178 G08_081T01_9000002350 T03_244T01_9000003912 90201029002 00030799022

23 October 2016 6 June 2018 19 June 2018 9 September 2018 2 October 2020 25 July 2016 25 July 2016

57684.99 58275.53 58288.88 58370.3 59124.57 57594.36 57594.85

48.5 43.6 46.4 46.5 95

12.1 24.1 23 22.8 8.9

Spin period (s)

Count rateb

41.2895(7) 41.272(9) 41.30(1) 41.2747(8) 41.306(3) 41.287054a –

32 24 11 65 27 – –

AstroSat Obs-1 Obs-2 Obs-3 Obs-4 Obs-5 NuSTAR Swift a

56.7 1

From Fu¨rst et al. (2017). b Average source count rate (in counts s1 ) per LAXPC unit is given in 3–80 keV energy range.

Figure 1. MAXI (2–20 keV, blue data points) and Swift/BAT (15–50 keV, shaded) long-term monitoring light curves of EXO 2030?375 ranging from (a) 21 June 2016 (MJD 57560) to 27 November 2016 (MJD 57719), (b) 12 April 2018 (MJD 58220) to 14 October 2018 (MJD 58405), and (c) 20 July 2020 (MJD 59050) to 13 October 2020 (MJD 59135) are shown in top, middle, and bottom panels, respectively. Arrow marks in the panels represent the epochs of AstroSat and NuSTAR observations of the pulsar.

J. Astrophys. Astr.

(2021) 42:33

AstroSat with an effective area of about 8000 cm2 at 15 keV. The time and energy resolutions of these units are 10 ls and 12% at 22 keV, respectively. Standard data analysis routines (LAXPCsoftware) are used to obtain the source light curves and spectral products from the event mode data. We have used SXT and LAXPC data in our timing study. Depending on the quality of the LAXPC data and instrument gain stability, we have considered events from single or combined LAXPC units. For timing studies, combined data from LAXPC-10, 20 and 30 are used during Obs1, while data from LAXPC-20 only are considered for Obs-2, Obs-3, and Obs-5. The events from LAXPC-10 and 20 are used for timing studies from Obs-4. Background products corresponding to each observation are accumulated from the same data by analysing the Earth occultation period. A systematic uncertainty of 2% is also added in the LAXPC spectra.

2.2 NuSTAR and Swift/XRT In the present study, we also used NuSTAR (Harrison et al. 2013) and Swift/XRT (X-Ray telescope; Burrows et al. 2005) observations on 25 July 2016, at a reported lowest luminosity of EXO 2030?375 till date (Fu¨rst et al. 2017), to compare the results obtained from the AstroSat observations. For NuSTAR observation, we used the NuSTARDAS 1.6.0 software in HEASoft version 6.24. Unfiltered events from the FPMA and FPMB were reprocessed by using the nupipeline routine in the presence of CALDB of version 20191219. Source products were then extracted by selecting circular region of 120 arcsec radius with souce coordinates as center by using the nuproducts task. Background products were also accumulated in a similar manner by selecting a source-free region. Data from the Swift/XRT observation in photon counting mode, with an effective exposure of 1 ks, are also used. We obtained XRT products by using the online standard tool provided by the UK Swift Science Data Centre2 (Evans et al. 2009).

3. Timing analysis We extracted source and background light curves from the SXT and LAXPC event data at 2.4 s and 0.1 s binning time, respectively. After subtracting the 2

http://www.swift.ac.uk/user_objects/.

Page 5 of 12

33

background, X-ray pulsations were searched in the barycentric corrected light curves of EXO 2030?375 from all five observations. We applied the chi-square maximization technique using efsearch task of FTOOLS package (Leahy 1987). The spin period of the pulsar is estimated to be 41.2895(7) s, 41.272(9) s, 41.30(1) s, 41.2747(8) s, and 41.306(3) s from first, second, third, fourth, and fifth AstroSat/LAXPC observations, respectively. The spin period and its error are also estimated by using the Lomb–Scargle and Clean techniques in the publicly available PERIOD package (Currie et al. 2014). This package has been used for period estimation in several other binary X-ray pulsars e.g. 4U 2206?54 (Torrejo´n et al. 2018), 2S 1417-624 (Gupta et al. 2019), Swift J0243.6?6124 (Beri et al. 2021). The results obtained from these methods are found to agree with the above quoted values. The evolution of pulse period with luminosity using AstroSat observations is presented in Fig. 2. As the source was observed at a low luminosity level, a few measurements have large errors on the spin period. A marginal spinning-up with increasing luminosity can be seen in the figure, though it is not adequate to draw any significant claim on this. The light curves in 0.3–7 keV and 3–80 keV ranges from the SXT and LAXPC data from each epoch of observations are folded with the corresponding estimated pulse period to obtain the pulse profiles of the pulsar. The pulse profiles obtained from the SXT and LAXPC data for all five AstroSat observations are shown in Figures 3 and 4, respectively. Phases of the pulse profiles are adjusted manually to align the minima at phase zero. The profiles obtained from the SXT data (Fig. 3) appears single peaked. This is possibly due to the the fact that the soft X-ray photons are largely affected by absorption due to the material along the line of sight and low source count rate in SXT. The profiles from the LAXPC data, however, are found to be complex due to the presence of multiple structures at various pulse phases of the pulsar during the first, third, fourth and fifth observations (see Fig. 4). Sharp dip-like features were detected in 0.1–0.2 and 0.60–0.85 phase ranges during these observations. Pulse profile of the pulsar from the second epoch of observation, however, appears relatively simpler. To investigate the observed features in the LAXPC profiles with energy, barycenter corrected light curves in 3–10, 10–25, 25–50 and 50–80 keV ranges are extracted from the LAXPC data from all epochs of observations and folded with the respective spin period and shown in Fig. 5. The energy resolved pulse

33

Page 6 of 12

J. Astrophys. Astr.

(2021) 42:33

pulsating component. In our study, we define the pulse fraction as the ratio between the difference and sum of maximum and minimum intensities observed in the pulse profile of the pulsar. For all the observations, we found that the pulse fraction decreases with energy. A maximum value of pulse fraction of  60% is detected in the profiles below 20 keV during the first observation. A relatively lower value is observed in rest of the data sets.

4. Spectral studies

0 0.5 1 1.5 2 Pulse Phase

0 0.5 1 1.5 2 Pulse Phase

0 0.5 1 1.5 Pulse Phase

Obs−5

1 0.5

1

1.5

Obs−4

0.5

1

1.5

Obs−3

0.5

1

1.5

Obs−2

0.5

1

1.5

Obs−1

0.5

Normalized Intensity

profiles are found to be strongly energy dependent. The presence of dip-like features are seen up to higher energies in the profiles from all observations. The observed dips are evident up to  50 keV, especially during the first observation, whereas during second, fourth and fifth observations, the features are present up to  25 keV. However, during the third observation, the dips are present up to  10 keV (Fig. 5). We checked the significance of pulsations in the hard Xray band by taking a ratio between the peak count rate and the standard deviation of the low or minimum intensity interval observed in the pulse profile. It is found that the significance of detection of pulsation in 50–80 keV range is more than 15r during all observations except the third observation. We calculated the pulse fraction of the pulsar using the pulse profiles in various energy bands and presented in Fig. 6. It is done to determine the nature of

The spectral properties of EXO 2030?375 are studied using data from all five AstroSat observations. Using the source and background spectra extracted from the SXT and LAXPC data (as described above) and response files provided by the instrument teams, we carried out spectral fitting of the data in 0.5–7 keV from SXT and 3.5–25 keV from LAXPC by using XSPEC package of version 12.10.0 (Arnaud 1996). The LAXPC data were limited to 25 keV in our spectral fitting because of background uncertainties at high energies. Various standard models such as power-law, cutoff power-law, high energy cutoff power-law are attempted to fit the 0.5–25 keV spectrum, along with a component for photo-electric absorption (TBabs, Wilms et al. 2000). We found that a cutoff based model is necessary to describe the spectra obtained from the first and fourth observations when the pulsar was relatively brighter (Fig. 1). However, a simple absorbed power-law model can describe the spectra from second, third and fifth observations, satisfactorily. These models provided goodness of fit per degree of freedom close to v2m ¼ v2 =m 1 in all cases. In place of a power-law component, we also tried to fit the spectra from second,

1.5

Figure 2. Spin period evolution of the pulsar with luminosity during AstroSat observations. L36 represents the 3–30 keV unabsorbed luminosity in unit of 1036 erg s1 .

0 0.5 1 1.5 2 Pulse Phase

0 0.5 1 1.5 2 Pulse Phase

Figure 3. Pulse profiles of EXO 2030?375 in 0.3–7 keV range with SXT instrument are shown for all five AstroSat observations (left to right). These profiles are obtained by folding the light curves at the respective pulse period determined from LAXPC data. Two pulses are shown for clarity.

J. Astrophys. Astr.

(2021) 42:33

1.5

Page 7 of 12

Obs-1

1.9L36

Obs-2

1L36

Obs-3

0.25L36

Obs-4

3L36

Obs-5

1L36

1 0.5 1.25

Normalized Intensity

1 0.75 1.1 1 0.9 1.1 1 0.9 1.2 1 0.8 0

0.5

Pulse Phase

1.5

2

Figure 4. Pulse profiles of EXO 2030?375 in 3–80 keV range are shown for all five AstroSat observations (top to bottom). L36 denotes the 0.5–30 keV unabsorbed luminosity of the pulsar in 1036 erg s1 at a distance of 7.1 kpc. Two pulses are shown for clarity.

third and fifth observations with a thermal blackbody component. This yielded a poor fit with a v2 =m of more than 5. We do not detect any signature of cyclotron absorption scattering feature(s) (Jaisawal & Naik 2017; Staubert et al. 2019) in 0.5–25 keV spectral range. The spectra obtained from SXT and LAXPC data also do not show any iron emission line(s) in 6–7 keV range. Spectral parameters estimated from these fittings are given in Table 2. In our fitting, the relative instrument normalization for SXT was found to be in the range of 0.65–0.80 with respect to LAXPC. We fitted the NuSTAR data from FPMA and FPMB detectors along with the Swift/XRT data in 1–79 keV energy range with a high energy cutoff power-law model along with the interstellar absorption. This model fitted the spectrum well. The spectral parameters such as column density, power-law photon index, cutoff and folding energies obtained from our fitting

33

are found to be consistent with the values reported in Table 1 of Fu¨rst et al. (2017). The energy spectra corresponding to each observation along with best-fit model and corresponding residuals are shown in Fig. 7. The cflux convolution model is used for flux estimation in our study. We would like to mention here that the quoted flux and luminosity in Table 2 are estimated in 0.5–30 and 3–30 keV ranges though 0.5–25 keV data were used in spectral fitting with AstroSat. Estimation of flux and luminosity in 3–30 keV range was done for comparison with the earlier values reported in literature. It is done by extrapolating the bestfit model up to 30 keV. We attempted to find any correlation between the power-law photon index and luminosity of the pulsar during the AstroSat observations. For this, we plotted the photon index with the observed 3–30 keV luminosity of the pulsar from AstroSat and combined NuSTAR and Swift/XRT observations in Fig. 8. Along with this, corresponding data from the RXTE observations of EXO 2030?375, as reported in the top left panel of Fig. 6 of Epili et al. (2017), are also shown for comparison. The figure shows that the photon index and luminosity are anti-correlated during all five epochs of AstroSat and combined NuSTAR and XRT observations. The results obtained from the present work extended the observed spectral behaviour of EXO 2030?375 in sub-critical regime at much lower luminosity.

5. Discussion Be/X-ray binary pulsars are expected to show X-ray enhancements, termed as Type-I X-ray outbursts, at the periastron passage of the neutron star (Okazaki & Negueruela 2001; Reig 2011). However, in most of the cases, such enhancements are not always observed which has been interpreted as due to the lack of significant evolution of the equatorial circumstellar disk around the Be star companion. An alternative interpretation of the lack of X-ray activities (Type-I or Type-II outbursts) is related to the Be-disk dynamics due to Kozai–Lidov effect (Laplace et al. 2017). EXO 2030?375 is unique in the sense that the pulsar shows Type-I X-ray outburst almost at every periastron passage of binary. The pulsar had been observed and studied extensively with the RXTE for 606 epochs, spanning over 15 years, during Type-I and Type-II (giant) X-ray outbursts (Epili et al. 2017) though there are many pointing observations with other observatories used to study the characteristics of the source. Long term monitoring data from RXTE/ASM,

33

Page 8 of 12

J. Astrophys. Astr.

50−80 keV

Normalized Intensity 25−50 keV 10−25 keV

3−10 keV

1.5

1.2

1.5

Obs−1

1 0.5 1.5

1

0.5

0.8

1.2

1.1

1

1

0.5

0.8

1.2

1.1

1

1

1

0.9

0.9

0.8

1 0.9 1.1

1.2

1.1

1

1

1

0.9

0.9

0.8 0.5 1 1.5 Pulse Phase

50−80 keV

Normalized Intensity 25−50 keV 10−25 keV

3−10 keV

0

1.5

1.1

0

2

Obs−3

Obs−2

1

(2021) 42:33

0.5 1 1.5 Pulse Phase 1.5

Obs−4

1

1

0.5 1.1

0.5

2

0

0.5 1 1.5 Pulse Phase

2

Obs−5

1.2 1

1

0.8 0.9 1.05

1.1

1

1

0.95

0.9

1.1

1.1 1

1

0.9

0.9 0

0.5 1 1.5 Pulse Phase

2

0

0.5 1 1.5 Pulse Phase

2

Figure 5. Energy resolved pulse profiles of EXO 2030?375 obtained by folding the energy resolved light curves from LAXPC instrument(s) onboard AstroSat at the respective estimated spin period(s). Two pulses are shown in each panel for clarity. The error-bars in the figure represent 1r uncertainties.

Pulse fraction

0.6

Swift/BAT and MAXI/GSC show regular Type I X-ray outbursts at the periastron passage of the pulsar. However, it has been noticed that the intensity at the peak of the Type-I X-ray outbursts has been in decline since last several years (Naik & Jaisawal 2015; Fu¨rst et al. 2016; Laplace et al. 2017) along with an extended period of low X-ray activity without any Type-I outbursts from 57200 MJD (27 June 2015) to 57600 MJD (31 July 2016) (Kretschmar et al. 2016). Following the extended low state, the transient activities started with the appearances of outbursts, though of lower peak intensities to date.

Obs−1 Obs−2 Obs−3 Obs−4 Obs−5

0.4

0.2

20

40 Energy (keV)

60

80

Figure 6. Pulse fraction variation of EXO 2030?375 with energy, obtained from the pulse profiles in multiple energy bands from five AstroSat observations.

5.1 Detection of X-ray pulsations at the lowest observed luminosity The NuSTAR and Swift/XRT observations on 25 July 2016 were reported to be carried out at the

J. Astrophys. Astr.

(2021) 42:33

Page 9 of 12

33

Table 2. Best-fitting spectral parameters (with 90% errors) of EXO 2030?375 during AstroSat and NuSTAR?XRT observations. Parameters NH (1022 cm2 ) Photon index (C) Norm (102 ) Efold (keV) Ecut (keV) Fluxa (3–30 keV) Fluxa (0.5–30 keV) Luminosityb (1036 erg s1 ) v2m (m)

Obs-1

Obs-2

Obs-3

Obs-4

Obs-5

NuSTAR?XRT

4:5 0:3 1:61 0:07 3:7 0:4 7:2 0:7 36þ8 6 2:8 0:1 3:1 0:1 1.9 1.06 (363)

4:7 0:3 1:92 0:05 3:4 0:3 – – 1:5 0:1 1:7 0:1 1.0 1.28 (258)

3:5 0:5 2:1 0:2 1:3 0:5 – – 0:42 0:03 0:50 0:05 0.25 0.91 (105)

4:6 0:3 1:54 0:1 6:5 1 6 2 32þ11 8 5 0:1 5:5 0:1 3.3 1.18 (454)

4 0:4 1:85 0:05 2:9 0:3 – – 1:51 0:1 1:71 0:1 1.0 1.48 (100)

5:8 0:6 1:64 0:05 2:5 0:2 6:5 0:4 27 2 1:67 0:1 2 0:1 1.21 0.96 (808)

Unabsorbed flux in 1010 erg s1 cm2 ; b 0.5–30 keV unabsorbed luminosity at a distance of 7.1 kpc. Note: By fitting NuSTAR and XRT data, we estimated the unabsorbed flux of EXO 2030?375 in 3–10 and 0.5–79 keV ranges to be 8:8  1011 and 2:3  1010 erg s1 cm2 , respectively. This is for comparison with the quoted values in Fu¨rst et al. (2017).

keV (Photons cm−2 s−1 keV−1)

Obs−1

0.01

10−3

10−4

10−3

10−4

5

0

0

−5

Obs−2

0.01

5 χ

χ

keV (Photons cm−2 s−1 keV−1)

a

−5 0.5

1

2

10

5 Energy (keV)

20

0.5

1

2 5 Energy (keV)

10

20

Obs−3

keV (Photons cm−2 s−1 keV−1)

keV (Photons cm−2 s−1 keV−1)

0.01

10−3

10−4

5 χ

χ

0

1

2 5 Energy (keV)

10

10−4

0

20

2

1

5

10

20

Energy (keV)

Obs−5

keV (Photons cm−2 s−1 keV−1)

keV (Photons cm−2 s−1 keV−1)

10−3

−5 0.5

10−3

−4

10

5

0.01

NuSTAR+XRT

10−3

5

0

χ

χ

Obs−4

5

−5

0.01

0.01

0 −5

−5 1

2

5 Energy (keV)

10

2

5

10 Energy (keV)

20

50

Figure 7. Best-fitting energy spectra obtained from first, second, third, fourth and fifth AstroSat observations of EXO 2030?375. Broadband energy spectra from NuSTAR and Swift/XRT data in 2016 July are also shown.

33

Page 10 of 12

Photon Index

2

1.5

0.01

0.1 1 10 Luminosity L37 (3−30 keV)

Figure 8. Variation of power-law photon index with 3–30 keV luminosity are shown for AstroSat and NuSTAR observations of EXO 2030?375 with solid bullets and star symbols (red points), respectively, along with corresponding data from the RXTE observations (black points) as shown in Fig. 6 of Epili et al. (2017). The power-law photon index obtained from the present study follows the anti-correlated pattern with the luminosity in the subcritical regime of the pulsar. L37 denotes the 3–30 keV unabsorbed luminosity of the pulsar in 1037 erg s1 at a distance of 7.1 kpc.

J. Astrophys. Astr.

(2021) 42:33

In accretion powered X-ray pulsars, material is channeled from the disk to magnetic poles. Decrease in the mass accretion rate decreases the ram pressure, eventually leading to the increase in the size of the magnetosphere (Illarionov & Sunyaev 1975; Nagase 1989). In case the magnetosphere exceeds beyond corotation radius, the centrifugal barrier prohibits accreting material to fall onto the neutron star. This leads to the cessation of pulsations of the pulsar and is referred as propeller effect (Illarionov & Sunyaev 1975). Though, EXO 2030?375 was detected at a lowest luminosity level of  1034 erg s1 using Swift/ XRT data (Fu¨rst et al. 2017), non-detection of X-ray pulsations and presence of softer thermal component with a temperature of 1.22 keV suggest the neutron star surface to be the source of observed emission, which occurs when the neutron star enters into the propeller phase (see, e.g., Wijnands & Degenaar 2016; Tsygankov et al. 2016; Fu¨rst et al. 2017). This allowed us to consider the luminosity of 2:5  1035 erg s1 (third AstroSat observation) as the lowest during which pulsations are seen. Assuming above luminosity as the upper limit for the onset of propeller effect, we can calculate the pulsar magnetic field as follows (Campana et al. 2002; Fu¨rst et al. 2017): 2=3

Llim ¼ 7:3  k7=2 P7=3 R56 B212 M1:4 lowest luminosity of EXO 2030?375 during which pulsations were detected in the light curves (Fu¨rst et al. 2017). Though the pulsar was observed at even lower luminosity of  1034 erg s1 with the Swift/XRT in 3–10 keV range, poor data quality refrained from pulsation search (Fu¨rst et al. 2017). On reanalysis of the NuSTAR and Swift/XRT data, the 0.5–30 keV range luminosity of the pulsar on 25 July 2016 was estimated to be 1.211036 erg s1 (Table 2). On comparing the luminosities during five AstroSat observations with the NuSTAR and Swift/XRT observations, it is interesting to point out that the pulsar was caught at even lower luminosities during second, third and fifth AstroSat observations. Among these, the lowest luminosity of 2:5  1035 erg s1 in 0.5–30 keV range was estimated during the third epoch of AstroSat observation. The LAXPC data from this observation showed a clear pulsation at 41.3 s in the light curve. Since the discovery in 1985, the luminosity of 2:5  1035 erg s1 in 0.5–30 keV range, observed with AstroSat/LAXPC on 19 June 2018 is the lowest luminosity of EXO 2030?375 at which X-ray pulsations are detected in the light curves.

 1037 erg s1 ; ð1Þ

where P is spin period in s, B12 is magnetic field in unit of 1012 G, R6 is the neutron star radius in unit of 106 cm, M1:4 is the mass of the neutron star in unit of 1.4M . The factor k is related to the accretion geometry with a value of k ¼ 0:5 and 1 in case of disk and spherical wind accretions, respectively. Using above equation and assuming disk accretion scenario for EXO 2030?375, we obtain a range of equatorial magnetic field between ð315Þ  1012 G for a minimum luminosity of 1  1034 and 2:5  1035 erg s1 , respectively. Based on the detection of cyclotron line, the polar magnetic field of the neutron star is tentatively estimated to be 1  1012 G (Wilson et al. 2008) and 5  1012 G (Klochkov et al. 2008). However, later studies did not confirm the cyclotron feature in a broad energy range (Naik et al. 2013; Naik & Jaisawal 2015; Fu¨rst et al. 2017). In absence of firm detection of cyclotron line, we calculate the magnetic field by putting standard parameters of a neutron star in Equation (1) and found that EXO 2030?375 hosts a highly magnetized neutron star with a field strength between ð315Þ  1012 G.

J. Astrophys. Astr.

(2021) 42:33

5.2 Pulse profiles and spectroscopy Extensive studies of RXTE observations of EXO 2030?375 revealed that the pulse profiles of the pulsar strongly depend on the luminosity or mass accretion rate (Epili et al. 2017). Irrespective of the type of X-ray outbursts, whether regular Type-I or giant Type-II, or phases of the outbursts, the morphology of the pulse profiles remain same at a certain luminosity. In the present study, the pulse profiles of the pulsar during all five epochs of AstroSat observations are characterised with the presence of narrow dip-like features, though the features are prominent in the lowest luminosity phase on 19 June 2018. These features are commonly seen in Be/X-ray binary pulsars (see, e.g., Devasia et al. 2011; Jaisawal et al. 2016, 2018; Gupta et al. 2018). At a luminosity an order of 1036 erg s1 , Ferrigno et al. (2016) and Fu¨rst et al. (2017) have detected a sharp absorption feature in the pulse profile of EXO 2030?275. This feature is interpreted as due to the effect of obscuration through accretion column along the line-of-sight. This is supported by the phaseresolved spectroscopy which revealed a high column density and effectively harder spectrum due to reprocessing of the emission. In our study, the low luminosity of the pulsar and limited understanding of the background and spectral calibration of the instruments at high energies (Antia et al. 2017) prevented us to investigate the cause of the prominent dips in the pulse profiles through pulse-phase resolved spectroscopy. From energy resolved pulse profiles (Fig. 5), clear pulsations up to  50 keV are seen during all five epochs of observations. Significance of pulsation can also be seen from the values of pulse fraction with energy (Fig. 6). The fraction of number of photons contributing towards pulsation can be found to decrease with energy as well as luminosity. Broad-band energy spectrum of accretion powered X-ray pulsars originates due to thermal and bulk Comptonization of soft X-rays photons from the thermal mound on the neutron star surface (Becker & Wolff 2007). In spite of complex processes taking place in the accretion column, the observed spectrum can be described by high energy cutoff power-law, exponential cutoff power-law models along with components for emission lines and absorption due to the interstellar medium. We have studied five AstroSat and NuStar?XRT observations between 2016 and 2020 after renewed activities from EXO 2030?375. Spectral analysis of these observations revealed the dependence of power-law photon index with luminosity. Extensive

Page 11 of 12

33

studies of available RXTE observations of the pulsar established the relation between the power-law photon index with source luminosity (Epili et al. 2017). From above study, the photon index are found to be distributed in three distinct regions depending on the 3–30 keV luminosity, suggesting the spectral transition from sub-critical to super-critical regimes through the critical luminosity of ð24Þ  1037 erg s1 for EXO 2030? 375 at constant photon index. The source spectrum became harder with the luminosity in the sub-critical regime. A softening in the spectral emission was thereafter detected in the super-critical regime of the pulsar. As quoted, AstroSat observations were carried out at lower luminosities compared to the RXTE observations. In this study, we found that the power-law photon index is anti-correlated with the luminosity of the pulsar (Fig. 8) in the same manner as reported by Epili et al. (2017) at lower luminosities. This confirms that the the spectral shape of the pulsar depends on the mass accretion rate.

6. Conclusion In this paper, we carried out timing and spectral studies of EXO 2030?375 using five AstroSat observations at various phases of its Type-I X-ray outbursts. The source luminosity was detected to as low as 2:5  1035 erg s1 in 0.5–30 keV range at which clear pulsations are detected. This is the first time when pulsations at such a low luminosity level is detected in the pulsar. Considering this as an limiting luminosity for propeller regime, we calculated the magnetic field of the neutron star. We have also studied pulse profiles of the pulsar. The pulse morphology is found to be complex due to presence of multiple absorption like features. The energy spectrum of EXO 2030?375 can be described by a high energy cutoff power-law model during brighter (first and fourth) AstroSat observations. The power-law photon index shows an anti-correlation with the source luminosity which is expected when the source is below the critical luminosity.

Acknowledgements We thank the anonymous reviewer for suggestions on the paper. This publication uses the data from AstroSat mission of the ISRO, archived at the Indian Space Science Data Centre. We thank members of SXT and LAXPC instrument teams for their

33

Page 12 of 12

contribution to the development of the instruments and analysis software. The SXT and LAXPC Payload Operations Centers (POCs) at TIFR are acknowledged for verifying and releasing the data via the ISSDC data archive and providing the necessary software tools for data analyses. We also acknowledge the contributions of the AstroSat project team at ISAC and IUCAA. This research has made use of data obtained through HEASARC Online Service, provided by the NASA/ GSFC, in support of NASA High Energy Astrophysics Programs. This work used the NuSTAR Data Analysis Software (NuSTARDAS) jointly developed by the ASI Science Data Center (ASDC, Italy) and the California Institute of Technology (USA).

References Agrawal P. C. 2006, Adv. Space Res., 38, 2989 Agrawal P. C. et al. 2017, J. Astrophys. Astr., 38, 30 Antia H. M. et al. 2017, ApJS, 231, 10 Arnaud K. A. 1996, in Jacoby G. H., Barnes J., eds, ASP Conf. Ser. Volume 101, Astronomical Data Analysis Software and Systems V, Astron. Soc. Pac., San Fransisco, p. 17 Becker P. A., Wolff M. T. 2007, ApJ, 654, 435 Beri A. et al. 2021, MNRAS, 500, 565 Burrows D. N. et al. 2005, Space Sci. Rev., 120, 165 Campana S., Stella L., Israel G. L. et al. 2002, ApJ, 580, 389 Coe M. J., Payne B. J., Longmore A., Hanson C. G. 1988, MNRAS, 232, 865 Corbet R. H. D., Levine A. M. 2006, ATel, 843 Currie M. J., Berry D. S., Jenness T., Gibb A. G., Bell G. S., Draper P. W. 2014, in ASP Conf. Ser., Volume 485, Astron. Soc. Pac., San Francisco, p. 391 Devasia J., James M., Paul B., Indulekha K. 2011b, MNRAS, 414, 1023 Epili P., Naik S., Jaisawal G. K., Gupta S. 2017, MNRAS, 472, 3455 Evans P. A. et al. 2009, MNRAS, 397, 1177 Ferrigno C., Pjanka P., Bozzo E., Klochkov D., Ducci L., Zdziarski A. A. 2016, A&A, 593, A105 Fu¨rst F., Wilson-Hodge C. A., Kretschmar P., Kajava J., Kuehnel M. 2016, Atel, 8835 Fu¨rst F. et al. 2017, A&A, 606, 89 Gupta S., Naik S., Jaisawal G. K., Epili P. R. 2018, MNRAS, 479, 5612 Gupta S., Naik S., Jaisawal G. K. 2019, MNRAS, 490, 2458 Harrison F. A. et al. 2013, ApJ, 770, 103 Illarionov A. F., Sunyaev R. A. 1975, A&A, 39, 185

J. Astrophys. Astr.

(2021) 42:33

Jaisawal G. K., Naik S., Epili P. 2016, MNRAS, 457, 2749 Jaisawal G. K., Naik S. 2016, MNRAS 461, 97 Jaisawal G. K., Naik S. 2017, in Serino M., Shidatsu M., Iwakiri W., Mihara T., eds, 7 years of MAXI: monitoring X-ray Transients, held on 5–7 December 2016 at RIKEN, RIKEN, Saitama, Japan, p. 153 Jaisawal G. K., Naik S., Chenevez J. 2018, MNRAS, 474, 4432 Jaisawal G. K. et al. 2019, ApJ, 885, 18 Klochkov D., Santangelo A., Staubert R., Ferrigno C. 2008, A&A, 491, 833 Kretschmar P. et al. 2016, Astron. Telegram, 9485 Krimm H., Barthelmy S., Gehrels N., Markwardt C., Palmer D., Sanwal D., Tueller J. 2006, Astron. Telegram, 861 Krimm H. A. et al. 2013, ApJSS, 209, 14 Laplace E., Mihara T., Moritani Y. et al. 2017, A&A, 597, A124 Leahy D. A. 1987, A&A, 180, 275 Matsuoka, M. et al. 2009, PASJ, 61, 999 Motch C., Janot-Pacheco E. 1987, A&A, 182, L55 Nagase F. 1989, PASJ, 41, 1 Naik S., Maitra C., Jaisawal G. K., Paul B. 2013, ApJ, 764, 158 Naik S., Jaisawal G. K. 2015, Res. Astron. Astrophys., 15, 537 Okazaki A. T., Negueruela I. 2001, A&A, 377, 161 Parmar A. N., White N. E., Stella L., Izzo C., Ferri P. 1989, ApJ, 338,359 Ramadevi M. C. et al. 2018, J. Astrophys. Astr., 39, 11 Rao A. R., Bhattacharya D., Bhalerao V. B., Vadawale S. V., Sreekumar S. 2017, Curr. Sci., 113, 595 Reig P. 2011, Ap&SS, 332, 1 Singh K. P. et al. 2014, SPIE, 9144, 15 Singh K. P. et al. 2017, J. Astrophys. Astr., 38, 29 Staubert R. et al. 2019, A&A, 622, A61 Tandon S. N. et al. 2017, AJ, 154, 128 Tauris T. M., van den Heuvel E. P. J. 2006, in Lewin W., Klis M. V. D., eds, Formation and Evolution of Compact Stellar X-Ray Sources, Cambridge University Press, Cambridge, UK, p. 623 Tsygankov S. S., Lutovinov A. A., Doroshenko V. et al. 2016, A&A, 593, A16 Wijnands R., Degenaar N. 2016, MNRAS, 463, L46 Wilms J., Allen A., McCray R. 2000, ApJ, 542, 914 Wilson C. A., Finger M. H., Coe M. J., Laycock S., Fabregat J. 2002, ApJ, 570, 287 Wilson C. A., Fabregatet J., Coburn W. 2005, ApJ, 620, L99 Wilson C. A., Finger M. H., Camero-Arranz A. 2008, ApJ, 678, 1263 Wilson-Hodge C. A. et al. 2018, ApJ, 863, 9

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:72 https://doi.org/10.1007/s12036-021-09701-x

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

AstroSat observations of eclipsing high mass X-ray binary pulsar OAO 1657-415 GAURAVA K. JAISAWAL1,* , SACHINDRA NAIK2 , PRAHLAD R. EPILI3,

BIRENDRA CHHOTARAY2,4, ARGHAJIT JANA2

and P. C. AGRAWAL5

1

National Space Institute, Technical University of Denmark, Elektrovej 327-328, 2800 Lyngby, Denmark. Astronomy and Astrophysics Division, Physical Research Laboratory, Navrangapura, Ahmedabad 380 009, India. 3 School of Physics and Technology, Wuhan University, Wuhan 430072, China. 4 Indian Institute of Technology, Gandhinagar 382 355, India. 5 Department of Astronomy and Astrophysics, Tata Institute of Fundamental Research, Colaba, Mumbai 400 005, India. *Corresponding Author. E-mail. [email protected] 2

MS received 28 October 2020; accepted 1 January 2021 Abstract. We present the results obtained from analysis of two AstroSat observations of the high mass Xray binary pulsar OAO 1657-415. The observations covered 0.681–0.818 and 0.808–0.968 phases of the  10.4 day orbital period of the system, in March and July 2019, respectively. Despite being outside the eclipsing regime, the power density spectrum from the first observation lacks any signature of pulsation or quasi-periodic oscillations. However, during July observation, X-ray pulsations at a period of 37.0375 s were clearly detected in the light curves. The pulse profiles from the second observation consist of a broad single peak with a dip-like structure in the middle across the observed energy range. We explored evolution of the pulse profile in narrow time and energy segments. We detected pulsations in the light curves obtained from 0.808–0.92 orbital phase range, which is absent in the remaining part of the observation. The spectrum of OAO 1657-415 can be described by an absorbed power-law model along with an iron fluorescent emission line and a blackbody component for out-of-eclipse phase of the observation. Our findings are discussed in the frame of stellar wind accretion and accretion wake at late orbital phases of the binary. Keywords. Stars: neutron—pulsars: individual: OAO 1657-415—X-rays: stars.

1. Introduction OAO 1657-415 is an accreting high mass X-ray binary pulsar, discovered with Copernicus satellite in 1978 (Polidan et al. 1978). The X-ray pulsations from the neutron star were detected at 38.2 s using HEAO A-2 observations (White & Pravdo 1979; Parmar et al. 1980). Later, the system was identified as an eclipsing binary by using 1991 and 1992 observations with the Burst and Transient Source Experiment (BATSE) onboard Compton Gamma Ray This article is part of the Special Issue on: ‘‘AstroSat: Five Years in Orbit’’.

Observatory (CGRO; Chakrabarty et al. 1993). This study also revealed the orbital period of the system to be Porb ¼ 10:4 days along with the eclipse duration of about 1.7 days. Despite dedicated efforts in the optical band, no counter part was detected up to a limitings magnitude of V\23. A highly reddened B-type supergiant star was subsequently discovered as the optical companion of the pulsar within Chandra X-ray error box (Chakrabarty et al. 2002). The spectral class of the donor star was refined to be a Ofpe/WNL type star which is thought as a transitional object between the main sequence and Wolf-Rayet stars (Mason et al. 2009, 2012). Observed X-ray eclipses and massive nature of the companion established the binary system

72

Page 2 of 11

as an eclipsing high mass X-ray binary. The source distance is measured to be 4.4–12 kpc (Chakrabarty et al. 2002; Mason et al. 2009), consistent with the measurement of 7.1±1.3 kpc provided by Audley et al. (2006). The recent Gaia data suggests a relatively lower distance of 2.2þ0:5 0:7 kpc based on the parallax angle of the optical companion (Malacaria et al. 2020). In most of the high mass X-ray binary (HMXB) systems, a neutron star resides as the compact object. The mass accretion onto the neutron star in these systems takes place either via stellar wind accretion, Be-disk accretion, or Roche–Lobe overflow from the optical companion (see e.g. Reig 2011; Walter et al. 2015). Based on the mass accretion mechanisms and the nature of donor star, the HMXBs can be classified into Be-X-ray binaries (BeXBs) and supergiant X-ray binaries (SGXBs). The compact object in BeXBs corotates around a non-supergiant companion of class III–V, and accretes directly from a circumstellar disk of the Be star. On the other hand, the SGXBs consist of a super-giant optical companion of luminosity class I–II. The mass accretion in SGXB systems occurs through the stellar wind accretion (wind-fed accretion) or via an accretion disk formed after the capture of stellar wind or Roche-Lobe overflow (disk-fed accretion). Some of the SGXBs are also known to be eclipsing systems due to edge-on periodic obscuration of the compact object by the super-giant companion. Typical X-ray luminosity of the neutron star in SGXBs is in the range of 1034 –1037 erg cm2 s1 (Martı´nez-Nu´n˜ez 2017). The source luminosity can also vary by a factor of 5–100 within a ks time-scale due to flaring activities (see, e.g. Fu¨rst 2010; Naik et al. 2011; Jaisawal et al. 2020). Various sub-classes of HMXBs represent a unique position in spin period vs. orbital period diagram (Corbet diagram; Corbet 1986). The BeXB systems show a strong correlation between the orbital and spin periods, whereas the wind-fed SGXBs are distributed in a horizontal line. There are, however, three disk-fed SGXBs viz. Cen X-3, SMC X-1 and LMC X-4, that exhibit an anti-correlation between the orbital and spin periods in the Corbet diagram. In exception to the known pattern, OAO 1657-415 occupies an intermediate position in the Corbet diagram, between the wind-fed and disk-fed SGXBs (Chakrabarty et al. 1993; Jenke et al. 2012). Since discovery, the pulse period of OAO 1657415 changes stochastically at a rate of m_  8:5 1013 Hz s1 (Baykal 1997, 2000; Bildsten et al. 1997).

J. Astrophys. Astr.

(2021) 42:72

Steady spin-up and spin-down patterns, as seen in Cen X-3, are also observed in the system (Bildsten et al. 1997). The changes in the spin episodes, however, can not be explained by the theory of torque reversal without considering the formation of a transient accretion disk (Baykal 1997). Based on a positive correlation between X-ray luminosity and the spin frequency during a spin-down phase in 1997, the presence of a prograde disk was suggested (Baykal 2000). On investigation of almost two decades of long-term spin evolution with BASTE and GammaRay Burst Monitor (Fermi/GBM) observations, two accretion modes are inferred in this system (Jenke et al. 2012). The first mode is due to the disk-wind accretion (formation of an accretion disk after stellar wind accretion) where a stable accretion disk produces a correlation between flux and spin-up of the neutron star. In the latter case, a direct stellar wind accretion produces almost no correlation between the flux and spin-down parameter of OAO 1657-415 at a lesser accretion rate (Jenke et al. 2012). Recently, Kim and Ikhsanov (2017) proposed a magnetic levitating disk hypothesis to explain the spin evolution in OAO 1657-415 . Being located at a low galactic latitude, OAO 1657415 is highly absorbed with a column density of 1023 cm2 (Polidan et al. 1978; Kamata et al. 1990). The energy spectrum of the pulsar can be described by a power law continuum with an exponential high energy cutoff along with a soft excess and prominent emission lines at 6.4, 6.7 and 7.1 keV (Kamata et al. 1990; Audley et al. 2006; Barnstedt et al. 2008; Pradhan et al. 2014, 2019; Jaisawal & Naik 2014). ASCA observations were performed on 1994 March 22 and 1997 September 17 between orbital phases from –0.001 to 0.074 and from –0.21 to 0.11 (mid-eclipse time as phase zero), respectively (Audley et al. 2006). Presence of a dust scattered X-ray halo in OAO 1657415 was suggested from these data. As expected, the observed source flux was very low near the mideclipse region with a dominant 6.4 keV line compared to the 6.7 keV line in the data from the first ASCA observation. The second ASCA observation covered entire eclipse along with out-of-eclipse phases before and after eclipse. The energy continuum in this phase was found to be absorbed along with detection of 6.4 and 7.1 keV iron emission lines (Audley et al. 2006). Suzaku observation of the source in September 2011, covering 0.12–0.34 orbital phase range, revealed flaring activities in the soft and hard X-ray light curves (Jaisawal & Naik 2014; Pradhan et al. 2014). Detailed

J. Astrophys. Astr.

(2021) 42:72

time-resolved spectroscopy of the Suzaku data suggested the accretion of clumpy material as the cause of flare-like episodes during the observation. Strong 6.4 and 7.1 keV lines were also detected. These lines mostly originated from the neutral and ionized iron atoms within the accretion radius of 19 lt-sec (Jaisawal & Naik 2014). Using BeppoSAX observation, a presence of a cyclotron absorption line was suggested at 36 keV (Orlandini et al. 1999). However, later studies with INTEGRAL and Suzaku did not confirm the feature (Barnstedt et al. 2008; Pradhan et al. 2014; Jaisawal & Naik 2014). In the present paper, we study the properties of the source by using two AstroSat observations, carried out at mid and late orbital phases of the orbit (with mid-eclipse time as phase zero) in 2019 . Data analysis is presented in Section 2, followed by timing and spectral results in Sections 3 and 4. The discussion and conclusion are summarized in Section 5.

2. Observations and data analysis AstroSat is the first Indian multi-wavelength astronomical satellite launched by Indian Space Research Organization on 28 September 2015 (Agrawal 2006; Singh 2014). It provides a broad-band coverage from optical to X-ray bands for exploring the nature of the cosmic sources. There are five sets of instruments such as Soft X-ray Telescope (SXT) (Singh 2017), Large Area X-ray Proportional Counters (LAXPCs) (Agrawal 2017; Antia 2017), Cadmium Zinc Telluride Imager (CZTI) (Rao et al. 2017), a Scanning Sky Monitor (SSM) (Ramadevi 2018), and Ultraviolet Imaging Telescope (UVIT) (Tandon 2017), onboard the satellite. In this paper, we study two observations of OAO 1657-415 with SXT and LAXPC instruments. These observations were performed on 31 March and 4 July 2019, covering orbital phase ranges of 0.681–0.818 and 0.808–0.968, respectively. The orbital phase range is calculated based on the ephemeris of the binary system provided by Jenke et al. (2012). In our study, the CZTI data from both the epoch of observations are not used as the source was faint for the detector. The UVIT was not operational during these epochs. The log of the observations are given in Table 1. The SXT is a soft X-ray focusing telescope consisting of a CCD detector and sensitive in 0.3–8 keV energy range. The effective area and energy resolution of SXT is 128 cm2 and 5–6% at 1.5 keV and 22 cm2 and 2.5% at

Page 3 of 11

72

Table 1. Log of observations of OAO 1657-415 with AstroSat. ObsID

Start date (MJD)

Expo. (ks)

/orb

XX_9000002824 XX_9000003012

58573.50 58668.85

59.8 62.2

0.681–0.818 0.808–0.968

Here XX stands for A05_205T01. /orb represents the orbital phase of the binary system.

6 keV, respectively. The observations of OAO 1657415 were carried out with SXT operating in photon counting mode, yielding a time resolution of 2.4 s. We used standard pipeline for SXT data reduction and merging tool sxtevtmergertool provided by the AstroSat Science Support Cell (ASSC1). As the pulsar was faint for SXT during the first observation, the data were used for spectral analysis only. The source spectrum was extracted from a 5 arcmin circular region centered at the source coordinate on the SXT chip using XSELECT package. On the other hand, the light curves and spectra were extracted from the second observation by considering a source circular region of 5 arcmin. The background spectrum was obtained from a source free region on the SXT chip. The three LAXPCs are sensitive to X-ray photons in the 3–80 keV range and provide a total effective area of 8000 cm2 at 15 keV. The time and energy resolution of the LAXPC units are 10 ls and 12% at 22 keV, respectively. During both the observations, data from LAXPC20 were considered in our analysis. The data from LAXPC10 and LAXPC30 units were not used due to the presence of high background and gain issues with the instrument during observations (Antia 2017). Using the standard data analysis routines (LAXPCsoftware), the event mode data are analyzed to obtain the source light curves and spectral products. The LAXPC background products are obtained from the observation using standard routines, recommended by the team. A systematic uncertainty of 2% is also added in the LAXPC spectrum.

3. Timing studies Swift/BAT (Burst Alert Telescope, Krimm 2013) and MAXI (Monitor of All-sky X-ray Image, Matsuoka 2009) long term monitoring light curves of OAO 1657-415 in 15–50 keV and 2–20 keV ranges, 1

http://astrosat-ssc.iucaa.in/.

72

Page 4 of 11

respectively, are shown in Fig. 1 to examine the overall activity of the pulsar during both the epochs of AstroSat observations. During the first observation on 31 March 2019, OAO 1657-415 was observed in a low intensity phase. However, during the second observation on 4 July 2019, the pulsar was observed to be brighter in the beginning and gradually entered the low flux level in the later part. The orbital phases covered during both the epochs of observations are 0.681–0.818 and 0.808–0.968 (Table 1). Background subtracted light curves obtained from LAXPC20 data of both the observations of the pulsar are shown in Fig. 2. The top panel (a) and middle panel (b) represent the source light curves in 3–10 keV and 10–80 keV energy ranges. The hardness ratio (HR), the ratio between the light curves in 10–80 keV and 3–10 keV ranges are also shown in the bottom panels of the figure. On comparison of light curves in 3–10 keV and 10– 80 keV ranges (top two panels of both sides of Fig. 2), it can be noticed that the pulsar was in a low intensity phase during the first AstroSat observation, compared to that during the initial part of the second observation. During the first epoch of observation, the pulsar did not show any significant time variability in soft and hard X-ray light curves, despite being observed away from the eclipse regime. The hardness ratio was also found to be constant (  1). However, the second data set (right side of Fig. 2) shows a gradual evolution in the 3–10 keV and 10–80 keV light curves observed between 0.808–0.968 orbital phase of the binary. In the early part of this observation (out-ofeclipse phase), the pulsar was relatively brighter than that during the first AstroSat observation. The hardness ratio also found to change during the second observation. A search for X-ray pulsations was performed in the 3–80 keV barycentric corrected light curves, binned at 0.1 s, from both the observations. For this, the power-density spectra (PDS) were generated from the light curves using the Fast Fourier Transformation technique with the powspec task of FTOOLS. Absence of sharp narrow peaks in the PDS from the first observation (Fig. 3) suggests the nondetection of X-ray pulsations in the light curve of the pulsar. Pulsations were again searched in light curves in different energy bands such as 3–10 keV, 3–25 keV, and 10–80 keV ranges. However, we failed to detect any pulsating signal from the neutron star during the first observation. The second observation of OAO 1657-415 was carried out by covering outof-eclipse and eclipse phases of the binary orbit. For

J. Astrophys. Astr.

(2021) 42:72

the pulsation search, the entire observation was divided into four different segments (Seg-I, II, III and IV) on the basis of source intensity. These segments are marked in different colors in the top right panel of Fig. 2. It should be noted that the fourth segment (Seg-IV) represents the duration of the eclipse of the neutron star during the second AstroSat observation. The PDS obtained from the 3–80 keV segmented light curves show signatures of strong pulsations along with its harmonics for Seg-I, II and III (Fig. 4). The PDS obtained from Seg-IV, however, did not show any such signature at a frequency corresponding to the spin period of the pulsar (right bottom panel of Fig. 4). This suggests that pulsations are present in the light curves during the out-of-eclipse phase of the pulsar and absent during the eclipsing phase. We then applied the chi-square maximization technique (Leahy 1987) using efsearch task to determine the pulsation period. From the PDS analysis and the chi-square maximization technique, the spin period of the pulsar is estimated to be 37.0375(8) s from the out-of-eclipse phases (Seg-I, II and III) of the second AstroSat observation. We checked the long term spin frequency history of OAO 1657-415 using Fermi/GBM2 data. The GBM instrument did not detect any pulsation in the source from MJD 58571.37 to MJD 58574.35 during which the first AstroSat observation was carried out. However, pulsations were detected during the second AstroSat observation as well as with Fermi/GBM. It is also worth to note that the count rate observed in the first observation is relatively higher than the same in third segment of second observation. The presence of pulsations at lower source count rate in the third segment confirms that the source was not into the propeller regime during the first observation. The background subtracted light curves in 3–80 keV energy range from the LAXPC20 data for Seg-I, II and III were folded at the estimated spin period of the pulsar. The pulse profiles from these segments are shown in the top panels of Fig. 5 (left to right). A single peaked profile with a dip at the peak is observed in first three segments. Non-detection of pulsations in the light curve from Seg-IV, as the neutron star entered into the eclipsing phase of the binary (Fig. 2), pulse profiles for this segment were not generated. Evolution of the pulse profile with energy was investigated by generating profiles in different energy ranges. The light curves were extracted in several 2

https://gammaray.nsstc.nasa.gov/gbm/science/pulsars.html.

J. Astrophys. Astr.

(2021) 42:72

Figure 1. Long-term monitoring light curves of OAO 1657-415 with MAXI (blue) and BAT (shaded) in 2–20 keV and 15–50 keV ranges, respectively. Arrows in both the panels represent the dates of the AstroSat observations. The pulsar appears to be weak during the first AstroSat observation (top panel). However, the second observation caught the source while changing from a relatively bright to faint phase of the binary orbit (bottom panel).

energy ranges such as 3–10 keV, 10–25 keV, 25–50 keV, and 50–80 keV, for Seg-I, II and III. These light

Page 5 of 11

72

curves were folded with the estimated spin period and presented in Fig. 5 in second, third, fourth, and fifth panels (from top to bottom) for respective segments. In our study, pulsations are effectively detected up to 80 keV in first and second segments. The dip in the peak of the profile was also found to present up to higher energies. The third segment which is in 0.89– 0.92 orbital phase range (out-of-eclipse phase), appears like the quiescent phase of the pulsar (Fig. 2, top right panel). Significant decrease in pulsar intensity during this segment, affected the soft X-ray pulse profile more. A single peaked structure with a dip appears only in the hard X-rays above 10 keV. The pulsation is detected up to  50 keV in this case. To quantify the nature of these pulsating components, pulse fraction from the pulse profiles of OAO 1657-415 is calculated and shown in Figure 6. In our study, the pulse fraction is defined as the ratio between the difference and sum of maximum and minimum intensities observed in the pulse profile. The pulse fraction from Seg-IV is not estimated as the pulsation was not detected in this segment. From the figure, a decreasing trend in the value of pulse fraction with energy, from a maximum of  40%, was seen for Seg-I and II. It suggests that the fraction of lower energy X-ray photons contributing towards pulsation is higher compared to the hard X-ray photons. In contrast, a marginally increasing trend between the pulse fraction and energy was seen for the Seg-III. Though a decreasing trend in the pulse fraction with

Figure 2. Light curves from first and second AstroSat/LAXPC20 observations of OAO 1657-415 on 31 March 2019 (left panels) and 4 July 2019 (right panels) are shown. The top and middle panels on both sides represent light curves in 3–10 keV and 10–80 keV ranges, respectively. The hardness-ratio, the ratio between light curves in 10–80 keV (middle panel) and 3–10 keV (top panel) energy bands, are shown in the bottom panels. The data from second observation are divided into four segments for further analysis and represented with different colors (top panel of the right side of the figure). The y-axis in top and middle panels represents the source rate in counts per second unit.

72

Page 6 of 11

J. Astrophys. Astr.

(2021) 42:72

values of pulse fraction with energy is anticipated when the source gets obscured by dense matter in the form of clumps of stellar wind, accretion wake, or the binary companion (e.g. in Seg-III). Pulse profiles of the pulsar were also generated from the SXT data. However, as the source is heavily absorbed in soft X-ray ranges, the profiles were not suitable to draw any meaningful information and not shown here.

4. Spectral studies

Figure 3. Power density spectrum of OAO 1657-415 obtained from the light curves in 3–80 keV range from the LAXPC20 data of first AstroSat observation. Absence of peaks corresponding to the spin period of the pulsar can be seen.

energy can be clearly seen for all segments, the pulse fraction values in 3–10 keV range for Seg-II and SegIII are found to be less than corresponding values in 10–25 keV ranges, whereas for Seg-I, it is comparable. This can be due to the fact that the soft X-ray photons are subjected to absorption by the particles in the interstellar medium. The gradual decrease in the

To understand the cause of non-detection of pulsations in the first AstroSat observation and the intensity variation during the second observation, we carried out spectral analysis by using SXT and LAXPC20 data from both the observations. The spectral fitting package XSPEC of version 12.10.0 (Arnaud 1996) was used. In the beginning, we extracted observed source?background and background spectra using laxpc_make_spectra and laxpc_find_back tasks of LAXPCsoftware package, respectively, from LAXPC20 data of Obs-1. The observed, background, and source spectra from LAXPC20 of Obs-1 are shown in Fig. 7. It can be seen that the source spectrum is limited up to  22 keV. As we aim to carry out our spectral analysis using data from both

Figure 4. Power density spectra (PDS) of OAO 1657-415 obtained from the segmented light curves of second AstroSat observation in 3–80 keV range. Presence of peaks at frequency corresponding to the spin period of the pulsar and its harmonics can be seen in the first, second and third segments of the observation. These peaks are, however, absent in the PDS corresponding to the fourth segment (right bottom panel).

J. Astrophys. Astr.

(2021) 42:72

Page 7 of 11

72

Figure 5. Energy resolved pulse profiles of OAO 1657-415 obtained by folding the light curves from LAXPC20 instrument during Seg-I, Seg-II and Seg-III of the second observation (as marked in the top right panel of Fig. 2) at the estimated spin period. Top panels show the pulse profiles of the pulsar in entire LAXPC energy range for different segments, whereas the other panels show the pulse profiles in narrow energy ranges (quoted in each panel). Two pulses are shown in each panel for clarity. The error-bars represent 1r uncertainties.

Figure 6. Pulse fraction variation of the pulsar with energy obtained from pulse profiles in multiple energy bands.

the observations, we restricted ourselves to fit the data up to 20 keV in our fitting. In our analysis, we considered 0.5–7 keV spectrum from SXT and 3.5–20 keV spectrum from LAXPC20. The 0.5–20 keV energy spectrum obtained from the first observation was fitted with several standard models such as power law, high energy cutoff power law, power law with black body component etc., along with the Galactic absorption column density TBabs (Wilms et al. 2000). Only an absorbed power law with a black body component (bbodyrad in XSPEC) was able to fit the spectrum better, though excess residuals were observed in 6–7 keV range. This excess was

resolved into two iron emission lines at 6.4 keV and 6.7 keV while fitting SXT data alone. Similar multiple iron emission lines have also been seen in other accretion powered X-ray pulsars such as GX 1?4 (Naik et al. 2005; Yoshida 2017), Cen X-3 (Naik et al. 2011), Swift J0243.6?6124 (Jaisawal et al. 2019). In the present study, these lines could not be distinguished in the LAXPC data due to relatively poor spectral resolution of the instrument. Therefore, the parameters of the Gaussian components for emission lines were fixed at values obtained from fitting SXT data alone, in the joint fitting of SXT and LAXPC data. Because of poor energy resolution of LAXPC, these parameters were not allowed to vary. This resulted in improving the goodness of fit per degree of freedom to v2m ¼ v2 =m  1. The best-fit spectral parameters obtained from our fitting are quoted in Table 2. The iron line parameters quoted in Table 2, such as line energy, width, equivalent width and line flux are obtained by fitting the SXT data alone. The LAXPC20 light curve of the pulsar during the second AstroSat observation shows different intensity phases of the source (Fig. 2). During the normal phase (Seg-I), the source intensity was maximum which gradually decreased (Seg-II) to low intensity phases (Seg-III and IV). As the source flux was extremely low during Seg-III and IV and the duration of Seg-II is very short, spectral analysis was carried out by considering Seg-I and added Seg-II, III and IV,

72

Page 8 of 11

J. Astrophys. Astr.

Figure 7. Comparison between source and background energy spectra obtained from the first AstroSat observation of OAO 1657-415. Table 2. Best-fitting spectral parameters (with 90% errors) of OAO 1657-415. Obs-2 Parameters NH1 a NH2 a CF C Norm (104 ) kT (keV) BBnorm

Obs-1

Seg-I

II?III?IV

15 ± 3 – – 1.1 ± 0.2 2.5 ± 1 1.2 ± 0.1 0.2 ± 0.1

6.6 ± 0.5 – – 1.2 ± 0.1 341 ± 1 1.7 ± 0.1 1.4 ± 0.4

4.9 ± 0.4 166 ± 29 0.56 ± 0.06 1.6 ± 0.1 172 ± 1 – –

6.37 ± 0.1 0.2 ± 0.2 2.3 ± 1 1.1 ± 0.7 6.68þ0:3 0:1 0.4þ0:3 0:4 1.9 ± 1 0.8 ± 0.4 1.6 ± 0.3 0.77 (61)

– – – – – – – – 55 ± 2 1.03 (310)

6.4 ± 0.1 0.1 1 ± 0.6 0.12 ± 0.07 – – – – 13 ± 3 1.01 (142)

Iron lines E1 (keV) W1 (keV) Flux1b Eqw1 (keV) E2 (keV) W2 (keV) Flux2b Eqw2 (keV) Fluxc v2m (m)

In 1022 cm2 unit; b Unabsorbed line flux in 1012 erg cm2 s1 ; c 0.5–30 keV unabsorbed flux in 1011 erg cm2 s1 unit.

a

separately. While fitting the SXT and LAXPC spectra for Seg-I, the absorbed power-law model with a black body component provided acceptable fit. However,

(2021) 42:72

there was no clear signature of iron emission line in the residuals. While fitting the spectra corresponding to added Seg-II, III, and IV, a simple absorbed powerlaw model or an absorbed power-law with black body model did not fit the data well. Considering the low intensity nature of the neutron star at late orbital phases, a partial covering component was tried along with the absorbed power-law model. Addition of partial covering component to the absorbed power-law model improved the fitting. We also detected a 6.4 keV iron fluorescence line in the spectra for this added segment. The best-fitted spectral parameters obtained from the spectral fitting of data from both the AstroSat observations are given in Table 2, whereas the energy spectra along with corresponding residuals are shown in Figure 8. We used cflux convolution model for flux estimation in our study.

5. Discussion and conclusions We studied two AstroSat observations of OAO 1657415 carried out in March and July 2019 covering 0.681–0.818 and 0.808–0.968 phase ranges of the 10.4 day orbital period of the binary system, respectively. During the first observation, no pulsation was detected in the SXT and LAXPC data. In this observation, despite being significantly away from the eclipse, the source was found weak in the soft and hard X-ray light curves. The 0.5–30 keV unabsorbed flux was estimated to be 1:6  1011 erg cm2 s1 , corresponding to a source luminosity of  0:93  1034 and 9:4  1034 erg s1 at a distance of 2.2 and 7 kpc, respectively. The spectral analysis of the data from this observation revealed a high value of column density NH of about 1:5  1023 cm2 . Strong iron lines at 6.4 keV and 6.7 keV were also detected in the SXT spectrum obtained from the first observation. The equivalent width of the lines are estimated to be as high as 1 keV. Detection of such strong iron lines with high equivalent width suggests the presence of abundant material around the pulsar for reprocessing. A high value of equivalent width (  1 keV) is only possible when a dense absorbing medium faces directly the X-ray source (Inoue 1985). There are at least two possibilities of absorbing medium one can presume in front of the neutron star. First one, is an accretion wake. Using simulation study of accretion onto neutron stars in wind-fed sources, Blondin et al. (1990) suggested the formation of accretion wake at the late orbital phases. The density inside the non-

J. Astrophys. Astr.

(2021) 42:72

Figure 8. Best-fitting energy spectra of OAO 1657415 during the first (top panel) and second (middle and bottom panels) AstroSat observations of the source. The data from SXT and LAXPC instruments in 0.5–7 and 3.5–20 keV are used in the spectral fitting, respectively.

steady accretion wake could be 100 times higher than the undisturbed stellar wind (Blondin et al. 1990). The dense wake can efficiently absorb the X-ray photons and hence reducing the source flux to lower values as

Page 9 of 11

72

observed in this case. Similar structure has been reported in case of 4U 1700–37 during the out-ofeclipse part of the binary between 0.63–0.73 phases (Boroson et al. 2003; Jaisawal & Naik 2015). An increase in the column density is usually observed in these orbital phases. In contrast to the present study, Jaisawal & Naik (2015) did not find any significant increase in the iron line parameters in case of 4U 1700–37. The study of OAO 1657-415 in 0.12–0.34 orbital phase range with Suzaku revealed the presence of eclipse-like segment in the light curve. Time-resolved spectroscopy of the Suzaku data corresponding to 0.19–0.23 phase showed a high column density as well as strong 6.4 keV and 7.1 keV iron lines with equivalent width of about 1 keV and 0.3 keV, respectively (Jaisawal & Naik 2014; Pradhan et al. 2014). The presence of a dense blob of material across the line of sight of neutron star was speculated within an accretion radius (Jaisawal & Naik 2014). In the present study, the existence of a dense blob of material or the accretion wake can not be denied. If the material is dense, it can absorb the X-ray photons, henceforth no pulsation is observed as seen during the first observation. Nonetheless, the detection of strong iron emission lines favors the hypothesis of absorption through a clumpy stellar wind in our study. Alternatively, the cessation of the pulsations can be possible in X-ray pulsars when the source enters into propeller regime (Illarionov & Sunyaev 1975) at extremely low mass accretion rate (see, e.g., in case of GX 1?4 and GRO J1744-28 reported by Cui 1997). If this is the case, the observed X-ray luminosity can be assumed as a limiting luminosity of the pulsar. However, it is unlikely that the pulsar was in the propeller regime during first AstroSat observation. This is due to the detection of iron emission lines during this observation. Considering the presence of high column density and iron emission lines at late orbital phases of the binary orbit (first AstroSat observation), the disappearance of pulsations can be interpreted as due to the presence of accretion wake. From the second AstroSat data, pulsations were clearly detected in the out-of-eclipse phase. The pulse profiles of OAO 1657-415 were found to be singly peaked with a dip like structure at the peak of the profile. Usually this source is subjected to strong and dense stellar wind of the supergiant companion as it is in a close binary system. Inhomogenous distribution of stellar wind as well as clumpy wind accretion in SGXBs can affect the pulsed emission. The profile can be affected mostly in the soft X-rays due to absorption. In the present and previous studies of

72

Page 10 of 11

OAO 1657-415, the presence of a dip in the profile is seen up to higher energies (Pradhan et al. 2014). Such type of profiles with absorption-like feature(s) at certain pulse phases are common in transient BeXRB pulsars (Maitra et al. 2012; Naik et al. 2013; Jaisawal et al. 2016). The complex shape of the pulse profiles in these Be/X-ray pulsars is interpreted as an effect of coupling between accreted material and magnetic field lines. We also carried out spectral investigation of the source during the second AstroSat observation by dividing the data into two parts. The first part of the data (Seg-I, Fig. 2) was relatively brighter by a factor of 34 than the first AstroSat observation. We did not detect any iron emission line in the data from this segment. The SXT and LAXPC spectra from the second part of the second observation (added Seg-II, III and IV, Fig. 2), covering 0.87–0.968 orbital phase range, were highly absorbed. Only a weak iron emission line at 6.4 keV was detected in this segment. As the spectral fitting is limited to 0.5–20 keV range, detection of the cyclotron absorption features (Jaisawal & Naik 2017; Staubert 2019) is not expected in the data. In summary, we have studied two AstroSat observations in March and July 2019 at late orbital phases of the binary system. We did not detect any strong pulsation at the spin period of the pulsar in the first data set, observed in 0.681–0.818 phase range. The presence of relatively high column density and strong iron emission lines at 6.4 keV and 6.7 keV with an equivalent width of 1 keV, suggest the clumpy wind material close to the neutron star. A possibility of accretion wake can not be denied at a late orbital phases of wind-fed SGXBs. The pulsations are detected in the light curves obtained from the second observation. From the energy and time resolved pulse profiles, it is found that the pulsations are detected in the data before the source gets eclipsed by the optical companion.

Acknowledgements We thank the anonymous referee for suggestions on our paper. This publication uses the data from AstroSat mission of the ISRO, archived at the Indian Space Science Data Centre. We thank members of SXT, LAXPC, and CZTI instrument teams for their contribution to the development of the instruments and analysis software. LAXPC data were processed by

J. Astrophys. Astr.

(2021) 42:72

the Payload Operation Centre at TIFR, Mumbai. This work was performed utilizing the calibration databases and auxiliary analysis tools developed, maintained and distributed by the AstroSat-SXT team with members from various institutions in India and abroad, and the SXT Payload Operation Center (POC) at the TIFR, Mumbai (https://www.tifr.res.in/ *astrosat_sxt/index.html). SXT data were processed and verified by the SXT POC. SN, BC and AJ acknowledges the support from Physical Research Laboratory which is funded by the Department of Space, India.

References Agrawal P. C. 2006, Adv. Space Res., 38, 2989 Agrawal P. C. et al. 2017, J. Astrophys. Astr., 38, 30 Antia H. M. et al. 2017, ApJS, 231, 10 Arnaud K. A. 1996, in Jacoby G. H., Barnes J., eds, ASP Conf. Ser., vol. 101, Astronomical Data Analysis Software and Systems V, Astron. Soc. Pac., San Fransisco, p. 17 Audley M. D., Nagase F., Mitsuda K., Angelini L., Kelley R. L. 2006, MNRAS, 367, 1147 Barnstedt J., Staubert R., Santangelo A. et al. 2008, A&A, 486, 293 Baykal A. 1997, A&A, 319, 515 Baykal A. 2000, MNRAS, 313, 637 Bildsten L., Chakrabarty D., Chiu J. et al. 1997, ApJS, 113, 367 Blondin J. M., Kallman T. R., Fryxell B. A., Taam R. E. 1990, ApJ, 356, 591 Boroson B., Vrtilek S. D., Kallman T., Corcoran M. 2003, ApJ, 592, 516 Chakrabarty D., Grunsfeld J. M., Prince T. A. et al. 1993, ApJ, 403, L33 Chakrabarty D., Wang Z., Juett A. M., Lee J. C., Roche P. 2002, ApJ, 573, 789 Corbet R. H. D. 1986, MNRAS, 220, 1047 Cui W. 1997, ApJ, 482 163 Fu¨rst F. et al. 2010, A&A, 519, 37 Illarionov A. F., Sunyaev R. A. 1975, A&A, 39, 185 Inoue H. 1985, Space Sci. Rev., 40, 317 Jaisawal G. K., Naik S. 2014, BASI, 42, 147 Jaisawal G. K., Naik S. 2015, MNRAS, 448, 620 Jaisawal G. K., Naik S., Epili P. 2016, MNRAS, 457, 2749 Jaisawal G. K., Naik S. 2017, in Serino M., Shidatsu M., Iwakiri W., Mihara T., eds, 7 years of MAXI: Monitoring X-ray Transients, held on 5–7 December 2016 at RIKEN, RIKEN, Saitama, Japan, p. 153 Jaisawal G. K. et al. 2019, ApJ, 885, 18 Jaisawal G. K. et al. 2020, MNRAS, 498, 4830

J. Astrophys. Astr.

(2021) 42:72

Jenke P. A., Finger M. H., Wilson-Hodge C. A., CameroArranz A. 2012, ApJ, 759, 124 Kamata Y., Koyama K., Tawara Y. et al. 1990, PASJ, 42, 785 Kim V. Yu., Ikhsanov N. R. 2017, J. Phys.: Conf. Ser., 929 012005 Krimm H. A. et al. 2013, ApJSS, 209, 14 Leahy D. A. 1987, A&A, 180, 275 Maitra C., Paul B., Naik S. 2012, MNRAS, 420, 2307 Malacaria C., Jenke, P., Roberts O. J., Wilson-Hodge C. A., Cleveland W. H., Mailyan B. 2020, ApJ, 896, 90 Mason A. B., Clark J. S., Norton A. J., Negueruela I., Roche P. 2009, A&A, 505, 281 Mason A. B., Clark J. S., Norton A. J. et al. 2012, MNRAS, 422, 199 Matsuoka, M. et al. 2009, PASJ, 61, 999 Martı´nez-Nu´n˜ez S. et al. 2017, Space Sci. Rev., 212, 59 Naik S., Paul B., Callanan P. J. 2005, ApJ, 618, 866 Naik S., Paul B., Ali Z. 2011, ApJ, 737, 79 Naik S., Maitra C., Jaisawal G. K., Paul B. 2013, ApJ, 764, 158 Orlandini M., dal Fiume D., del Sordo S. et al. 1999, A&A, 349, L9

Page 11 of 11

72

Parmar A. N., Branduardi-Raymont G., Pollard G. S. G. et al. 1980, MNRAS, 193, 49 Pradhan P., Maitra C., Paul B., Islam N., Paul B. C. 2014, MNRAS, 442, 2691 Pradhan P., Raman G., Paul B. 2019, MNRAS, 483, 5687 Polidan R. S., Pollard G. S. G., Sanford P. W., Locke M. C. 1978, Nature, 275, 296 Ramadevi M. C. et al. 2018, J. Astrophys. Astr., 39, 11 Rao A. R., Bhattacharya D., Bhalerao V. B., Vadawale S. V., Sreekumar S. 2017, Curr. Sci., 113, 595 Reig P. 2011, Ap&SS, 332, 1 Singh K. P. et al. 2014, SPIE, 9144, 15 Singh K. P. et al. 2017, J. Astrophys. Astr., 38, 29 Staubert R. et al. 2019, A&A, 622, A61 Tandon S. N. et al. 2017, AJ, 154, 128 Walter R., Lutovinov A. A., Bozzo E., Tsygankov S. S. 2015, A&AR, 23, 2 White N. E., Pravdo S. H. 1979, ApJ, 233, L121 Wilms J., Allen A., McCray R. 2000, ApJ, 542, 914 Yoshida Y. et al. 2017, ApJ, 838, 30

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:96 https://doi.org/10.1007/s12036-021-09760-0

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

Understanding the inner structure of accretion disk in GX 17+2: AstroSat’s outlook K. SRIRAM1,* , P. CHIRANJEEVI1, S. MALU1 and V. K. AGRAWAL2 1

Department of Astronomy, Osmania University, Hyderabad 500 007, India. Space Astronomy Group, ISITE Campus, U R Rao Satellite Center, Bangalore 560 037, India. *Corresponding Author. E-mail: [email protected] 2

MS received 3 November 2020; accepted 6 March 2021 Abstract. We performed the timing and spectral studies of a Z source GX 17?2 observed from Astrosat LAXPC instrument. Cross-Correlation Function (CCF) was performed using soft (3–5 keV) and hard (16–40 keV) X-ray bands across the hardness intensity diagram and found correlated/anti-correlated hard and soft lags which seems to be a common feature in these sources. We performed spectral analysis for few of these observations and found no consistent variation in the spectral parameters during the lags, however 10–40% change was noticed in diskbb and power-law components in few of observations. For the first time, we report the detection of HBOs around  25 Hz and  33 Hz along with their harmonics using AstroSat LAXPC data. On comparison with spectral results of HB and other branches, we found that inner disk front is close to the last stable orbit and as such no systematic variations are observed. We suggest that the detected lags are readjustment time scales of corona close to the NS and constrained its height to be around few tens to hundreds of km. The detected lags and no significant variation of inner disk front across the HID strongly indicate that structural variation in corona is the most possible cause of Z track in HID. Keywords. Accretion—accretion disk—binaries: close—stars: individual (GX 17?2)—X-rays: binaries.

1. Introduction Neutron Star (NS) Low Mass X-ray Binaries (LMXBs) that are highly luminous (close to or more than Eddington luminosity) exhibiting a Z-shaped track on the Hardness Intensity Diagram (HID)/Colour–Colour Diagram (CCD) are termed as Z sources (Hasinger & van der Klis 1989). These are further classified into Sco X-1 like and Cyg X-2 like sources based on the particular Z track traced by them (Kuulkers et al. 1994, 1997). The various branches of this track are characterized by varying temporal and spectral features along them. The three main branches are the horizontal, normal and flaring branch (HB, NB and FB). The two existing contrary pictures to explain the motion of the source along the HID are the varying

This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

_ rate hypothesis (Hasinger et al. mass accretion (m) 1990; Vrtilek et al. 1990) and the constant mass accretion rate hypothesis where other physical or radial instabilities cause the source to trace out the Z-shaped track (Homan et al. 2002; Lin et al. 2012). One of the fundamental characteristics of each branch is the signature of Quasi Periodic Oscillations (QPOs) exhibited in each of them (Hasinger & van der Klis 1989). QPOs in the horizontal branch (HB) termed as horizontal branch oscillations (HBOs) are found in the 15–60 Hz frequency range, while those in the normal branch (NB) termed as normal branch oscillations (NBOs) are found in the 5–8 Hz frequency range (van der Klis 2006). Flaring branch Oscillations, so far detected only in two sources Sco X-1 and GX 17?2 (Priedhorsky et al. 1986; Penninx et al. 1988; Homan et al. 2002), are found in the frequency range of 10–25 Hz. Each of these oscillations are considered to have different physical origins. Alpar and Shaham (1985) proposed a model where HBOs could be associated with

96

Page 2 of 17

the Keplerian orbital frequency of the inner edge of the disk and the spin frequency of the neutron star. Stella & Vietri (1999) and Stella et al. (1999) proposed the Relativistic Precession Model (RPM) that associates HBOs to the nodal precession of tilted orbits near the neutron star. NBOs and FBOs could possibly be oscillations in the optical depth of accretion flow in the inner disk region as proposed by Lamb et al. (1989) and Fortner et al. (1989) or they could be oscillations associated with the sound waves in a thick disk (Alpar & Shaham 1985). Another model for NBOs were proposed by Titarchuk et al. (2001), where these oscillations are considered to be acoustic oscillations of a spherical viscous shell around the NS. Spectral modeling has been ambiguous in these sources, especially when it comes to the origin of soft and hard energy photons. As per the eastern spectral model, soft photons emanating from the disk are modeled using a multi-temperature black body emission (MCD, Mitsuda et al. 1984) and these photons are the seed for inverse Comptonization in the compact corona producing hard X-rays. The western model on the other hand uses a single temperature blackbody emission from the NS surface (or the immediate surrounding region) and a high energy power-law model for describing the hard emission (White et al. 1986). Another model proposed by Popham and Sunyaev (2001) considers a hot, low density boundary layer around the NS surface to be responsible for the hot Comptonized spectrum and the optically thick accretion disk to be responsible for the black body spectrum. Though both the models fairly describe the X-ray spectrum of Z sources, the picture is not clear in terms of understanding the location and size of corona and whether the accretion disk is truncated or not. A vital timing tool to resolve some of these ambiguities in understanding the accretion disk corona geometry is a Cross Correlation Function (CCF) study between soft and hard X-ray energy bands (e.g. Vaughan et al. 1999; Sriram et al. 2007, 2012, 2019). Previous CCF studies of Z sources have led to the detection of lags of the order of hundred seconds in the HB and NB branch of the Z track while FBs are mostly seen to have strong positively correlated CCFs (Lei et al. 2008; Sriram et al. 2012, 2019). It was suggested that these lags are attributed to the readjustment of the coronal structure which could in effect help in constraining the size of the coronal structure (Sriram et al. 2019). Here we present an extensive spectro-temporal study of the source GX 17?2 using the AstroSat LAXPC archival data of the source. GX 17?2 is a burster Z source with no confirmed optical counterparts, located at a distance of 13

J. Astrophys. Astr.

(2021) 42:96

kpc (Galloway et al. 2008), having a spin frequency of 293.2 Hz (Wijnands et al. 1997). GX 17?2 is a low inclination system i\40 (Cackett et al. 2010; Ludlam et al. 2017; Malu et al. 2020). Cackett et al. (2010) used a relativistic diskline model on the Suzaku data of the source, which led to an estimate of 7–8 GM/c2 for the inner disk radius. Using NuStar spectra of GX 17?2 (3–30 keV), Ludlam et al. (2017) determined that the disk extends to 1.0–1.02 ISCO. Sriram et al. (2019) based on the RXTE spectrum of the source estimated an inner disk radius of 20–35 km. Using the AstroSat spectrum of the source, Agrawal et al. (2020) found a decreasing powerlaw component along the Z track from HB to NB and this component was found to be increasing from the NB to FB. They estimated a 28–42 km inner disk radius from their spectral analysis. Based on the AstroSat SXT?LAXPC spectra, Malu et al. (2020) found an inner disk radii  12–16 km (5.7–8.0 Rg), along the NB, which is close to ISCO. Hence the inner disk can be considered to be almost at the last stable orbit without much change in the position of the disk front. Energy-dependent CCF studies of GX 17?2 using RXTE and NuStar data (3–5 keV and 16–30 keV) performed by Sriram et al. (2019) led to the detection of a few hundred second delays when the source was in the NB and HB. Using these lags the coronal height was constrained to be around 20–35 km. A similar study using the SXT (0.8–2 keV) and LAXPC (3–5 keV, 16–20 keV, 20–40 keV and 20–50 keV) data performed by Malu et al. (2020) again revealed lags of the order few hundred seconds in the NB of the Z track and the height of the corona was constrained to few tens of km. Here the spectral analysis revealed only a varying power-law index across the NB thus leading to the conclusion that only the hot comptonized region was varying. Earlier CCF study of GX 17?2 using AstroSat (Malu et al. 2020) was focused only on the NB and an extensive CCF study along with a correlated spectral and timing study of the source across the complete HID can help in resolving the inconsistencies related to the accretion disk corona geometry. Hence we report here, the CCF, PDS and spectral study of the source GX 17?2 using AstroSat’s LAXPC data.

2. Data reduction and analysis AstroSat Large Area X-ray Proportional Counter (LAXPC) archival data of the source was used in this study. Datasets used for the study include the Obs. ID

0.63–0.72 0.62–0.68 0.60–0.65 0.58–0.64

29/50

422  85 369  30 306  70 571  37 555  43

281  31

6  36 78  55 9  26 604  63

0:37  0:05 0:55  0:02 0:57  0:03 0:46  0:03 0:38  0:05

0:56  0:02

0:37  0:04 0:4  0:04 0:59  0:03 0:27  0:03

540–570, 530–560, 520–550, 500–540,

Figure 1. HID for GX 17?2 using AstroSat LAXPC observations. LAXPC 20 data was used. Hard colour is defined as 10.5–19.7/7.3–10.5 keV and Intensity is that in the 7.3–19.7 keV range. G05-112T01-9000000452 is represented using colours red, blue, green and yellow, T02087T01-9000002352 is represented using colours orange. G08-037T01-9000002256 is represented using colour violet.

Detected HBO. a

2016-05-14, 11:17:27

2018-07-26, 18:56:51

2018-09-10, 23:55:00

2018-07-26, 08:54:03

2018-09-10, 04:24:32

G05-112T01-9000000452 Section Aa Section B Section C Section D Section E G08-037T01-9000002256 Section F T02-087T01-9000002352 Section G Section H Section I Section J

2016-05-11, 12:27:21

3100

2300 2700 3200 3600

Lags ± error (s)

2400 3000 2900 3100 1700

96

29/50 32/50 36/51 25/31

Page 3 of 17

470–520, 0.58–0.62

550–600, 0.81– 0.73 540–580, 0.77–0.66 520–550, 0.75–0.67 460–500, 0.68–0.59 400–450, 0.62–0.54

HID location (hard color, intensity)

31/40 72/54 27/37 24/47 25/32

v2 /dof

(2021) 42:96

CC ± CCerr Exposure time (s) Stop time (UT) Start time (UT) ObsID

Table 1. CCF information of LAXPC soft (3–5 keV) vs. hard (16–40 keV) lightcurves. Here CC means cross-correlation coefficient. Lags refer to the CCF lags obtained and HID location gives the position of the lightcurve segment on the HID.

J. Astrophys. Astr.

G05-112T01-9000000452 observed from 2016, May 11 to 2016, May 14 (  100 ks), Obs. ID T02-087T019000002352 observed on 2018, September 10 (  49 ks) and Obs. ID G08-037T01-9000002256 observed on July 26, 2018 (  18 ks). LAXPC data in the EA mode was used for this study and this has a time resolution of 10 ls. LAXPC onboard AstroSat has three proportional counter units that are co-aligned – LAXPC 10, 20 and 30. The combined effective area is 6000 cm2 at 15 keV and it is operational in the 3–80 keV energy range with a moderate energy resolution (Yadav et al. 2016; Antia et al. 2017). The Level 1 data was analysed using LAXPC software (Format A May 19, 2018) which is provided by the AstroSat Science Support Center (ASSC). LAXPC_MAKE_EVENT, LAXPC_MAKE_STDGTI, LAXPC_MAKE_LIGHTCURVE and LAXPC_ MAKE_SPECTRA modules were used to generate the event file, GTI file, corresponding lightcurves and

96

Page 4 of 17

J. Astrophys. Astr.

(2021) 42:96

Figure 2. The background subtracted LAXPC soft (3–5 keV) and hard (16–40 keV) light curves on the top panels and the corresponding CCF lag observed are on the bottom panels. Energy bands used are mentioned in the light curves (top panel). Bottom panels show the cross correlation function (CCF) of each section of the light curve and shaded regions show the standard deviation of the CCFs. Bottom panel inset figure gives the Gaussian fit of the lag portion.

J. Astrophys. Astr.

(2021) 42:96

Page 5 of 17

Figure 2. Continued.

96

96

Page 6 of 17

spectra respectively. LAXPC_MAKE_BACKLIGHT CURVE and LAXPC_MAKE_BACKSPECTRUM were used to generate the corresponding background light curves and spectrum. The corresponding response files were generated along with it. Spectrum in the 3–50 keV band was used for the study and 1% systematic error was considered (Agrawal et al. 2020). LAXPC10 data has been used for spectral analysis (otherwise mentioned) as it is better calibrated and have less background issues when compared to other LAXPC units. LAXPC 20 data has been used for timing analysis for observations during 2018, due to low gain operation of LAXPC 10. For remaining observations LAXPC 10 and 20 has been used.

3. Timing analysis Timing analysis was performed on the background subtracted binned LAXPC10 and LAXPC20 light curves of GX 17?2. Using the hard colour 10.5–19.7 keV/7.3– 10.5 keV and intensity in the range 7.3–19.7 keV, the HID was obtained for sections that exhibited CCF lags (see Table 1). The light curves were corrected for dead-time effects (van der Klis 1988); for LAPXC the dead time is 42.5 ls (Yadav et al. 2016). For all the datasets, LAXPC 20 data was used to obtain the HID (Fig. 1). Crosscor tool of XRONOS package was used to perform the cross correlation function (CCF) analysis (Sriram et al. 2007; Lei et al. 2008). We used the direct slow method (fast = no) to compute the correlation coefficients (CC) as a function of lags between the two light curves. The error bars of CCF are obtained by propagating the theoretical error bars of the cross correlations from the individual intervals and the cross correlations are normalized by dividing them with the square root of the product of newbins of the two light curves in each interval.1 CCF analysis was performed between the 3–5 keV (soft) and 16–40 keV (hard) 32 s binned light curves (Sriram et al. 2007, 2012, 2019; Lei et al. 2008; Malu et al. 2020). The top panel of Fig. 2 shows the soft and hard energy band light curves of each section and the bottom panel shows the CCF along with the Gaussian fit (inset) of each section where lag was observed. In order to estimate the CCF lags, we adopted a procedure wherein several different segments were considered 1

https://heasarc.gsfc.nasa.gov/docs/xandau/xronos/help/crosscor. html.

J. Astrophys. Astr.

(2021) 42:96

around the hypothetical centroid and Gaussian functions were fitted. The minimum v2 fit was considered to estimate the CCF lags and their errors were estimated with a 90% confidence level using the criterion of Dv2 = 2.71 (see Table 1). Figure 3 shows few of the representative lag vs. Dv2 ð¼ v2  v2min Þ figures obtained from the above mentioned procedure. In the Obs. ID G05-112T01-9000000452, lags were found in five different light curve segments. CCF lags varying from  306–571 s were found in HB and NB with a correlation coefficients (CC) varying from  0.4–0.6 for the lags (see Table 1). This is consistent with that obtained from previous studies (Sriram et al. 2019). Out of the five segments, one was a positively correlated soft lag and remaining were anti-correlated hard lags. For Obs. ID G08-037T01-9000002256, lag was found in one section with a lag of 281  31 s (CC = 0.58 ± 0.03, positively correlated soft lag) in the HB. For Obs. ID T02-087T01-9000002352, again four segments were found to exhibit lags in the HB varying from 78–604 s, with four of them showing anti-correlated lags. One segment was associated with hard lags (see Table 1 and Fig. 2). Two of the sections (G and I) show anticorrelated CCFs but with almost no significant lags. Power Density Spectrum (PDS) was obtained using a 1/2048 s binned light curves in the 3–20 keV energy band, using LAXPC10 and LAXPC20 in order to improve the statistics. According to the normalization method by Miyamoto et al. (1991), power density spectra (PDS) were normalized in units of (rms/ mean)2 /Hz. PDS were fitted with a one or two Lorentzian and power-law model (see Table 2). PDS has revealed a  25 Hz and  33 Hz HBO in the G05112T01-9000000452 (orbit 03352, 03353, 03355), when the source was in the HB (see Table 2). For one of the light curve segments (orbit 03352), PDS revealed a 25.04 ± 0.44 Hz HBO with a 50.37 ± 0.43 Hz harmonic. For orbit 03353, a 32.34 ± 0.23 Hz HBO with a harmonic of 61.74 ± 4.75 Hz was observed. This section also exhibits an anti-correlated hard lag of 426 ± 38 s. Another segment (orbit 03355) revealed a 35.24 ± 0.33 Hz HBO with no harmonic (see Fig. 4 and Table 2). 4. Spectral analysis In order to check if any associated spectral variations are present, spectral analysis was performed for the first and last 1000 s of light curve segments (marked (a), (b) in Table 3, Fig. 2) which exhibited significant

J. Astrophys. Astr.

(2021) 42:96

Page 7 of 17

96

Figure 3. The CCF lag vs. Dv2 ð¼ v2  v2min ) plot obtained from the procedure mentioned in Section 3.

lags and separately for those segments which exhibited HBOs (Table 4, Fig. 4). Spectra were fitted in the 3–50 keV energy range using XSPEC v12.10.1f

(Arnaud 1996) and uncertainties were estimated with a 90% confidence level (Dv2 = 2.71). Figures 5, 6, 7 and 8 show the top panel with the unfolded spectra

96

Page 8 of 17

J. Astrophys. Astr.

Table 2. HBO parameter values. Observation Id:

G05-112T01-9000000452

Orbit

HBO m (Hz)

Dm (Hz)

RMS %

03352

25.04±0.44 50.37±0.43 32.34±0.23 61.74±4.75 35.24±0.33

5.97±1.91 5.67±2.75 1.77±0.28 13.74±6.13 4.41±1.18

1.21±0.11 1.60± 0.14 0.96±0.08 1.53±0.14 1.06 ±0.10

03353 03355

(thick line) along with the model components (dashed line) and the bottom panel with the residuals to the fit. Absorption column density was modeled using the Tbabs model (Wilms et al. 2000) and was fixed at  2.2  1022 cm2 (Cackett et al. 2010). Using the continuum model DiskBB ? bbody ? power-law (Cackett et al. 2008, 2010; Lin et al. 2007), the spectra were fitted which led to residuals around 6.7 keV which was then modeled using a Gaussian function centered at 6.7 keV (fixed) resulting in lower reduced v2 /dof values (see Table 3, Fig. 4). Gaussian and power-law component models are needed to unfold the spectra as indicated by the values of F-test probability (Table 3). v2 /dof values before adding the power-law component was noted to be 93.69/39, 112.21/39 for Section A(a), (b) which is significantly higher compared to that obtained after incorporating the power-law component, i.e. 29.72/37, 47.54/37. Spectral parameters are found to be non-varying (within error bars) for the first and last 1000 s sections of each segment that exhibited lags (see Table 3). During these observations, the inner disk temperature (kTin ) was different, ranging from 1.66–2.02 keV. It should be noted that kTin and kTBB are affected by the spectral hardening factor (fcol ; for more details, see Davis et al. 2005). The effective temperature is given by Teff ¼ kTin =fcol and Teff was found to be in the range of 1.03– 1.26 keV for fcol ¼ 1:6 (Davis et al. 2005). We did not find any significant variation in the spectral parameters between the final and last sections of each observation. There is a slight change in total flux in B and D sections (Table 3). The spectral analysis of sections exhibiting HBOs (Table 5, Fig. 7) suggest that the disk is close to the last stable orbit during the HB and we did not find any significant variation in the spectral parameters when the source varied from  25 Hz to  33 Hz HBOs. We have also unfolded the spectra of various sections using a thermal comptonization model viz.

(2021) 42:96

NthComp ? Gaussian ? power-law) (Agrawal et al. 2020). In the NthComp model, soft seed photons are assumed to be a black body emission (Zdziarski et al. 1996). Based on the F-test probability in all the sections, we find that the power-law model is necessary for the data (Tables 4 and 6). Unfolded spectra for various sections are shown in Figures 6 and 8. For the fits, we fixed the power-law index, C at 3.0 in order to constrain the other parameters. The reason to fix C came from the study of Agrawal et al. (2020), as in most of the branches, spectra exhibit C around 3. We noted that even if we vary C from 2.9 to 3.1, there is no noticeable variation in other spectral parameters. We found that for e.g. Section A, CNthComp varied slightly from 1.80 to 1.94 with in 90% confidence level. Similar small changes were observed in kTe for Sections A and D. Total fluxes were found to be slightly varying in subsections i.e. (a) and (b) of Sections B and D (Table 4).

5. Results and discussion 5.1 Constraining the inner disk radius Based on the PDS analysis, we found HBOs with a centroid frequency  25 Hz and 33 Hz. Based on the relativistic precession model (RPM), higher frequency kHz QPOs, lower frequency kHz QPOs and HBOs correspond to the Keplerian, periastron precession and nodal precession frequencies, respectively (Stella & Vietri 1999; Stella et al. 1999). Based on the equation given by Di Matteo and Psaltis et al. (1999), we can estimate an upper limit to the inner disk radius. It should be noted that the below relation is based on the empirical correlation between HBOs and the upper kHz QPO mHBO ¼ 63  (mupp /1 kHz)1:9 reported by Psaltis et al. (1999). This relation should be verified for GX 17?2 by considering all the available detections of QPOs. " #2=3 Rin M 0:35  27m : ð1Þ 2M Rg With m ¼ 25 Hz, we found R in to be  23 km and with m ¼ 33 Hz, we found R in to be  21 km.

5.2 Constraining the inner disk radius: TL model We constrained the inner disk radius using the transition layer (TL) model (Titarchuk & Osherovich

J. Astrophys. Astr.

(2021) 42:96

Page 9 of 17

96

"    Rin X 2=3 X sin d 2 ¼ 220 p mHBO p Rg #" # 2=3 1=3 M 1 ; 10M

ð2Þ

where d is the angle between rotational angular velocity (X) and normal to the Keplerian oscillation (d ¼ 6:30 for GX 17?2; Wu 2001). We found Rin to be  21Rg (  44 km; Rg ¼ GM=c2 ) for 25 Hz HBO. For m  25 Hz, we found Rin to be  22Rg (  46 km) and for m  33 Hz, Rin  18Rg  38 km).

5.3 Inner disk radii from spectral modeling

Figure 4. HBOs seen in Observation Id G05-112T019000000452 for the light curve in 3–20 keV energy range using AstroSat LAXPC observations fitted using a Lorentzian and powerlaw model to the PDS.

1999; Osherovich & Titarchuk 1999a, b). Using the correlation equation between HBO and kHz QPO in Z sources given by Wu (2001) and the relation between kHz QPO and inner disk radius given by Matteo and Psaltis (1999), Sriram et al. (2009) arrived at an equation connecting HBO frequency and the inner disk radius as below:

Based on the DiskBB model normalization (Mitsuda et al. 1984), NdBB is proportional to the inner disk pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi radius. Using the equation, Rin ðkmÞ ¼ ðN= cos iÞ  D=10 kpc, and taking i ¼ 28 (Malu et al. 2020), we estimated the inner disk radii Rin to be 9.79 km (A(a)), 7.91 km (A(b)), 8.78 km (B(a)), 10.59 km (B(b)), 11.68 km (C(a)), 11.02 km (C(b)), 13.12 km (D(a)), 11.75 km (D(b)). Since these radii needs to be corrected for spectral hardening using the relation, Reff ¼ j2 nRin (Kubota et al. 2001) where j  1.7–2.0 (Shimura & Takahara 1995) and n ¼ 0:41 (Kubota et al. 1998), we estimate the Rin to be 11.55–16.06 km (A(a)), 9.34–12.97 km (A(b)), 10.37–14.41 km (B(a)), 12.49–17.36 km (B(b)), 13.79–19.16 km (C(a)), 13.01–18.08 km (C(b)), 15.49–21.52 km (D(a)) and 13.87–19.27 km (D(b)), which is close to that estimated by Cackett et al. (2009), Ludlam et al. (2017), Agrawal et al. (2020) and Malu et al. (2020) and for GX 17?2. The derived Rin values suggest that disk front is found to be near or almost at the last stable orbit. Along the HID track we find no change in the inner radius, hence suggesting that it could be the coronal structure which is responsible for the source traversing along the HID and not the mass accretion rate. This is similar to the results by Lin et al. (2012) and Homan et al. (2002).

5.4 Coronal height estimation from CCF lags An extensive timing and spectral study using the archival data of AstroSat LAXPC for GX 17?2 is reported here. Based on the CCF study between LAXPC soft (3–5 keV) and hard (16–40 keV) light

369 ± 30 s 78–392 km

B

8.68107

2.16105

1.78±0.13 58.56±19.6 10.59 km 12.49–17.36 km 2.80±0.08 0.075±0.006 3.11±0.14 7.01±2.62 6.7 0.75±0.47 0.016±0.008 0.65±0.06 0.60±0.06 0.26±0.05 1.53±0.01 3.08 ± 0.02 1.65 ±0.01 31.96/37 67.96/39

(b)

2.13107

1.57105

1.66±0.15 71.34±31.12 11.68 km 13.79–19.16 km 2.68±0.07 0.080±0.007 2.89±0.12 6.07±2.36 6.7 0.84±0.45 0.020±0.011 0.55±0.06 0.64±0.05 0.35±0.06 1.57±0.01 3.16±0.02 1.69±0.01 38.02/37 87.22/39

(a)

306 ± 70 s 58–292 km

C

3.451015

4.346105

1.74±0.15 63.51±24.31 11.02 km 13.01–18.08 km 2.73±0.08 0.075±0.007 2.98±0.13 6.20±2.40 6.7 0.84±0.52 0.018±0.010 0.63±0.07 0.59±0.06 0.30±0.07 1.55±0.01 3.12 ±0.02 1.67±0.01 34.98/37 211.60/39

(b)

1:74  107

1:28  106

1.66 ± 0.12 90.04 ± 33.12 13.12 km 15.49–21.52 km 2.60 ± 0.07 0.072±0.007 2.88±0.21 3.96±2.47 6.7 0.78±0.63 0.017±0.011 0.70±0.07 0.57±0.01 0.23±0.08 1.52±0.01 3.06±0.02 1.64±0.01 30.31/37 70.30/39

(a)

–571±37 s 104–519 km

D (b)

3:56  106

2:17  103

1.80 ± 0.09 72.15 ± 15.23 11.75 km 13.87–19.27 km 2.79 ± 0.08 0.063±0.005 3.24±0.20 3.06±1.95 6.7 0.78±0.54 0.014±0.009 0.83±0.11 0.50±0.04 0.26±0.04 1.46±0.01 2.94±0.02 1.57±0.01 32.98/37 64.97/39

Temperature of the DiskBB model. b Normalization of the DiskBB model. c Inner disk radii from DiskBB normalization. d Effective radius obtained by using DiskBB normalization and the spectral corrections 1.18–1.64. e Temperature of the BB model. f Normalization of the BB model. g Power-law index. h Normalization of the PL model. i Line energy of the Gaussian model for iron line. j Line width of the Gaussian model for iron line. k Normalization of the Gaussian model for iron line. l Coronal height estimated from lag.

1.14108

1.25107

5.951010

1.87±0.18 40.32±16.12 8.78 km 10.37–14.41 km 2.84±0.10 0.069±0.007 3.01±0.10 7.65±2.17 6.7 0.56±0.53 0.014±0.008 0.56±0.08 0.55±0.06 0.36±0.03 1.48±0.01 2.98±0.02 1.60±0.01 28.31/37 76.06/39

(a)

1.02105

2.02±0.15 32.71±9.42 7.91 km 9.34–12.97 km 3.04±0.10 0.074±0.006 3.42±0.22 12.02±3.81 6.7 0.86±0.48 0.015±0.008 0.66±0.09 0.59±0.07 0.25±0.06 1.53±0.02 3.08 ±0.04 1.65±0.02 47.54/37 112.21/39

(b)

1.37103

422 ± 85 s 98–489 km

A

1.39106

1.74±0.16 50.10±23.2 9.79 km 11.55–16.06 km 2.84±0.07 0.083±0.006 2.93±0.11 6.80±2.17 6.7 0.91±0.40 0.021±0.011 0.50±0.06 0.66±0.06 0.36±0.06 1.55±0.01 3.12 ± 0.02 1.67 ± 0.01 29.72/37 93.69/39

(a)

Page 10 of 17

a

kTin (keV)a NdBB b Rin (i ¼ 28 Þc Reff (i ¼ 28 Þd kTBB (keV)e NBB f Cpl g Npl h EFe (keV)i rFe (keV)j NFe k DiskBB flux bbody flux Powerlaw flux Total flux L350 m_ v2 /dof v2 /dof (without power-law) Lag observed Hc l F-test (Gaussian) prob. F-test (powerlaw) prob.

Parameter

Table 3. Best-fit spectral parameters for sections which exhibited lags using DiskBB ? Gaussian ? bbody ? Power-law model. The subscript BB represents the bbody model and dBB represents DiskBB model. The flux in units of 108 ergs cm2 s1 is calculated in the energy band 3–50 keV. Errors are quoted at a 90% confidence level. Luminosity is in units of 1038 erg s1 assuming the distance 13 kpc for GX 17?2. Mass accretion rate is in units of 1018 g s1 .

96 J. Astrophys. Astr. (2021) 42:96

3.0 6.52±0.95 6.7 0.95±0.29 0.022±0.006 1.94±0.04 3.45±0.07 0.42±0.07 0.72±0.18 1.17±0.04 0.28±0.05 1.52±0.01 3.06 ±0.02 1.64±0.01 44.47/38 48.04/39 1.23106 8.87102

(b) 3.0 6.45±1.02 6.7 0.80±0.32 0.018±0.007 1.88±0.04 3.19±0.08 0.66±0.08 0.68±0.16 1.16±0.04 0.30±0.05 1.48±0.01 2.98±0.02 1.60±0.01 26.18/38 107.35/39 1.79107 3.331013

(a) 3.0 3.42±0.99 6.7 0.76±0.29 0.019±0.006 1.98±0.03 3.25±0.07 0.60±0.06 1.09±0.22 1.35±0.04 0.16±0.05 1.53±0.01 3.08 ± 0.02 1.65 ±0.01 30.04/38 57.43/39 1.16107 8.15107

(b) 3.0 7.73±0.90 6.7 0.67±0.31 0.018±0.007 1.87±0.04 3.07±0.07 0.69±0.07 0.64±0.13 1.18±0.04 0.36±0.04 1.56±0.01 3.14±0.02 1.68±0.01 42.52/38 188.0/39 1.17105 7.871014

(a)

C

3.0 5.59±0.91 6.7 0.81±0.35 0.20±0.007 1.95±0.04 3.13±0.07 0.66±0.06 0.82±0.15 1.27±0.04 0.26±0.04 1.55±0.01 3.12 ±0.02 1.67±0.01 34.40/38 114.27/39 1.95106 1.881011

(b)

3.0 5.20±0.77 6.7 0.74±0.37 0.019±0.008 2.04±0.04 2.98±0.07 0.70±0.05 0.82±0.12 1.26±0.03 0.24±0.04 1.52±0.01 3.06±0.02 1.64±0.01 35.67/38 131.06/39 3.67106 2.721012

(a)

D

3.0 5.36±0.86 6.7 0.97±0.21 0.027±0.006 2.15±0.02 3.21±0.04 0.59±0.04 0.92±0.22 1.21±0.04 0.28±0.05 1.46±0.01 2.94±0.02 1.57±0.01 31.50/38 101.30/39 8.78109 3.521011

(b)

Powerlaw index. bNormalization of the PL model. cLine energy of the Gaussian model for iron line. dLine width of the Gaussian model for iron line. eNormalization of the Gaussian model for iron line. fNthComp power-law index. gElectron temperature (NthComp). hSeed photon temperature (NthComp). iNormalization of the NthComp model.

3.0 7.27±1.02 6.7 0.77±0.30 0.019±0.007 1.80±0.04 3.22±0.07 0.64±0.09 0.64±0.12 1.19±0.05 0.34±0.05 1.55±0.01 3.12 ± 0.02 1.67 ± 0.01 35.64/38 139.46/39 9.91107 8.161013

(a)

B

(2021) 42:96

a

Cpl Npl 6b EFe (keV)c rFe (keV)d NFe e CNthComp f kTe (keV)g kTbb (keV)h NNthComp i NthComp flux Powerlaw flux Total flux L350 m_ v2 /dof v2 /dof (without power-law) F-test (Gaussian) prob. F-test (powerlaw) prob.

a

Parameter

A

Table 4. Best-fit spectral parameters for sections exhibiting lags using NthComp ? Gaussian ? Power-law model. The flux in units of 108 ergs cm2 s1 is calculated in the energy band 3–50 keV. Errors are quoted at a 90% confidence level. Luminosity is in units of 1038 erg s1 : assuming the distance 13 kpc for GX 17?2. Mass accretion rate is in units of 1018 g s1 :

J. Astrophys. Astr. Page 11 of 17 96

96

Page 12 of 17

J. Astrophys. Astr.

(2021) 42:96

Figure 5. The LAXPC 10 spectral fit for the the DiskBB ? Gaussian ? bbody ? Powerlaw model. The top panel gives the unfolded spectra (thick line) with the component models (dashed lines) and the bottom panel gives the residuals obtained from the fit. Here in the top panel, the light green colour line gives the Gaussian model component, cyan gives the Powerlaw component, red gives the diskBB component and dark blue gives the bbody component.

J. Astrophys. Astr.

(2021) 42:96

Page 13 of 17

96

Figure 6. The LAXPC 10 spectral fit for the NthComp ? Gaussian ? Power-law model. The top panel gives the unfolded spectra (thick line) with the component models (dashed lines) and the bottom panel gives the residuals obtained from the fit. Here in the top panel, the light green colour line gives the Gaussian model component, red gives the Power-law component and dark blue gives the NthComp component.

96

Page 14 of 17

J. Astrophys. Astr.

(2021) 42:96

Figure 7. The LAXPC 10 spectral fit for the segments exhibiting HBOs using the DiskBB ? Gaussian ? bbody ? Powerlaw model. The top panel gives the unfolded spectra (thick line) with the component models (dashed lines) and the bottom panel gives the residuals obtained from the fit. Here in the top panel, the light green colour line gives the Gaussian model component, cyan gives the power-law component, red gives the diskBB component and dark blue gives the bbody component.

Figure 8. The LAXPC 10 spectral fit for the segments exhibiting HBOs using the NthComp ? Gaussian ? Powerlaw model. The top panel gives the unfolded spectra (thick line) with the component models (dashed lines) and the bottom panel gives the residuals obtained from the fit. Here in the top panel, the light green colour line gives the Gaussian model component, red gives the Powerlaw component and dark blue gives the NthComp component.

curves, we detect correlated and anti-correlated lags of the order of few hundred seconds (see Table 1). This is in accordance with previously observed results for

this source (Sriram et al. 2019; Malu et al. 2020). The hard component could be arising from a coronal structure or a hot boundary layer component. Previous

J. Astrophys. Astr.

(2021) 42:96

Page 15 of 17

96

Table 5. Best-fit spectral parameters for sections which exhibited HBOs using DiskBB ? Gaussian ? bbody ? Powerlaw model. The subscript BB represents the bbody model and dBB represents DiskBB model. The flux in units of 108 ergs cm2 s1 is calculated in the energy band 3–50 keV. Errors are quoted at a 90% confidence level. Luminosity is in units of 1038 erg s1 assuming the distance 13 kpc for GX 17?2. Mass accretion rate is in units of 1018 g s1 : Parameter kTin (keV)a NdBB b Reff ði ¼ 28 Þc kTBB (keV)d NBB e Cpl f Npl g EFe (keV)h rFe (keV)i NFe j DiskBB flux bbody flux Powerlaw flux Total flux L350 m_ v2 /dof v2 /dof (without Gaussian) v2 /dof (without power-law) F-test (Gaussian) prob. F-test (Powerlaw) prob.

For HBO 1

For HBO 2

For HBO 3

2.10±0.21 20.53±7.89 7.39–10.28 km 3.05±0.08 0.077±0.007 2.96±0.29 5.99±4.01 6.7 0.86±0.29 0.02±0.007 0.50±0.07 0.62±0.06 0.39±0.04 1.54±0.01 3.10 ± 0.02 1.66 ± 0.01 47.44/37 79.12/39 155.14/39 7.76105 3.021010

1.92±0.14 33.52±11.32 9.45–13.13 km 2.93±0.07 0.077±0.005 2.97±0.31 4.33±2.12 6.7 0.82±0.40 0.019±0.008 0.56±0.06 0.61±0.05 0.33±0.04 1.53±0.01 3.08 ± 0.02 1.65 ± 0.01 42.60/37 71.40/39 118.78/39 7.09105 5.77109

1.87±0.13 39.54±12.75 10.26–14.27 km 2.91±0.06 0.077±0.005 3.09±0.08 8.53±2.13 6.7 0.88±0.39 0.019±0.009 0.56±0.06 0.62±0.04 0.33±0.04 1.53±0.01 3.08± 0.02 1.65 ± 0.01 41.77/37 73.59/39 324.97/39 2.81105 3.281017

a

Temperature of the DiskBB model. b Normalization of the DiskBB model. c Effective radius obtained by using DiskBB Normalization and the spectral correction 1.18–1.64. d Temperature of the BB model. e Normalization of the BB model. f Powerlaw index. g Normalization of the PL model. h Line energy of the Gaussian model for iron line. i Line width of the Gaussian model for iron line. j Normalization of the Gaussian model for iron line.

studies have shown that the causative factor for these large lags could only be the readjustment of the coronal structure, especially since here the inner disk radius is found to be almost at the last stable orbit at all times along the HID (for a detail spectral modeling see Agrawal et al. 2020), ruling it out as the agent for change causing the large CCF lags. All the lags were found in the HB and NB. Sriram et al. (2019), derived the equation (shown below) for determining the height of the coronal structure based on the condition that the CCF lags are the readjustment timescale of the corona. # " tlag m_ ð3Þ  Rdisk  b cm: Hcorona ¼ 2pRdisk Hdisk q 3=20 9=8

Here Hdisk ¼ 108 a1=10 m_ 16 R10 f 3=20 cm, q ¼ 7  108 a7=10 m_ 11=20 R15=8 f 11=20 g cm3 , f ¼ ð1  ðRs =RÞ1=2 Þ1=4 and b ¼ vcorona =vdisk (Shakura & Sunyaev 1973; Sriram et al. 2019).

For the sections A, B, C and D, the coronal height was constrained to be 98 km, 78 km, 58 km and 104 km (for b ¼ 0:1) and 489 km, 392 km, 292 km and 519 km for b ¼ 0:5, for b ¼ 0:1  0:5 (Manmoto et al. 1997; Pen et al. 2003; McKinney et al. 2012), an average Rdisk estimated from DiskBB normalization for each segment and m_ estimated from the _ luminosity considering the equation L ¼ GM m=R. Here a was taken to be 0.1. The lack of spectral variation among the sections and the presence of lags, along with the detection of inner disk radius almost at the last stable orbit, indicate towards the possibility of a non varying disk front and varying coronal structure, which leads to the conclusion that the mass accretion rate is not the factor causing the source to move along the HID but the varying comptonized corona. This was independently reported by Homan et al. (2002) and Lin et al. (2012), where upon studying GX 17?2 they conclude

96

Page 16 of 17

J. Astrophys. Astr.

(2021) 42:96

Table 6. Same model as used to obtain Table 4 and Best-fit spectral parameters for sections which exhibited HBOs using NthComp ? Gaussian ? Power-law model. The flux in units of 108 ergs cm2 s1 is calculated in the energy band 3–50 keV. Errors are quoted at a 90% confidence level. Luminosity is in units of 1038 erg s1 : assuming the distance 13 kpc for GX 17?2. Mass accretion rate is in units of 1018 g s1 . Parameter a

Cpl Npl b EFe (keV)c rFe (keV)d NFe e CNthComp f kTe (keV)g kTbb (keV)h NNthComp i NthComp flux Powerlaw flux Total flux L350 m_ v2 /dof F-test (Gaussian) prob. F-test (Powerlaw) prob.

For HBO 1

For HBO 2

For HBO 3

3.0 6.96±0.89 6.7 0.88±0.27 0.021±0.006 1.72±0.03 3.31±0.06 0.55±0.11 0.70±0.23 1.19±0.05 0.32±0.05 1.54±0.01 3.10 ± 0.02 1.66 ± 0.01 47/38 3.14106 1.461012

3.0 4.88±0.77 6.7 0.82±0.26 0.020±0.006 1.85±0.02 3.30±0.05 0.53±0.09 1.03±0.3 1.28±0.04 0.22±0.05 1.53±0.01 3.08 ± 0.02 1.65 ± 0.01 43/38 1.19106 7.701011

3.0 4.93±0.74 6.7 0.84±0.25 0.020±0.004 1.86±0.02 3.28±0.05 0.54±0.08 1.03±0.3 1.28±0.04 0.23±0.05 1.53±0.01 3.08± 0.02 1.65 ± 0.01 44/38 6.74107 2.391011

the same to address the evolution of the source across the HID. Hard lags suggest that the corona is decreasing in size, supported by most of the observations that hard count rates are often found to be decreasing in the light curves (Figures 2 and 3) and vice-verse for soft lags i.e. corona is increasing in height. However as it can be seen from CCFs that it is not always the scenario and more studies are required in this direction. The decreasing corona size should result in narrow equivalent width of iron line (Kara et al. 2019) but similar studies require better spectral resolution. 6. Conclusion (1) Based on energy-dependent CCF studies, we detect correlated and anti-correlated soft and hard lags of the order of few tens to few hundred seconds in the HB and NB. We interpret these lags as the readjustment timescales of the compact corona close to the neutron star. These lags constrain the coronal height to few tens to few hundred kms. (2) PDS study has led to the detection of HBOs of  25 Hz and  33 Hz, along with their harmonics, which are characteristic features of these type of

sources. Based on the RPM and TLM, we find the inner disk radius to be 10–21 Rg. (3) Spectral studies of the segments associated with lags show no significant variation in any of the spectral parameters, though we noticed small variations in fluxes for few sections. The inner disk radius was found to be close to the last stable orbit along the HB–NB/FB, suggesting that the disk is not truncated. (4) The lack of variation in the disk front suggests that the contributing factor for the source traversing along the HID could be a variation in the corona or boundary layer. Hence we can conclude that the mass accretion rate is not the primary factor for the movement along the HID. This result is in congruence with that of Homan et al. (2002) and Lin et al. (2012). The readjustment velocity factor of the corona b could play a significant role in this regard.

Acknowledgements We thank the referee for the comments that has improved the quality of the paper. K.S. and C.P. acknowledge the financial support of ISRO under AstroSat archival Data utilization program. This

J. Astrophys. Astr.

(2021) 42:96

publication uses data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). K.S. also acknowledges the financial support from SERB CRG program. M.S. acknowledges the financial support from DSTINSPIRE fellowship. V.K.A. thanks GH SAG; DD PDMSA, and Director URSC for encouragement and continuous support to carry out this research. Authors sincerely acknowledge the contribution of the LAXPC team toward the development of the LAXPC instrument on-board the AstroSat. This work uses data from the LAXPC instruments developed at TIFR, Mumbai, and the LAXPC POC at TIFR is thanked for verifying and releasing the data via the ISSDC data archive. Authors thank the AstroSat Science Support Cell hosted by IUCAA and TIFR for providing the LAXPC software that was used for LAXPC data analysis.

References Agrawal V. K., Nandi A., Ramadevi M. C., 2020, Ap&SS, 365, 41 Alpar M. A., Shaham J. 1985, Nature, 316, 239 Antia H. M., Yadav J. S., Agrawal P. C. et al. 2017, ApJS, 231, 10 Cackett E. M. et al. 2008, ApJ, 674, 415 Cackett E. M. et al. 2009, ApJ, 690, 1847 Cackett E. M., Miller J. M., Ballantyne D. R. et al. 2010, ApJ, 720, 205 Davis S. W., Blaes O. M., Hubeny I., Turner N. J. 2005, ApJ, 621, 372 Fortner B., Lamb F. K., Miller G. S. 1989, Nature, 342, 775 Galloway D. K., Muno M. P., Hartman J. M., Psaltis D., Chakrabarty D. 2008, ApJS, 179, 360 Hasinger G., van der Klis M. 1989, A&A, 225, 79 Hasinger G., van der Klis M., Ebisawa K., Dotani T., Mitsuda K. 1990, A&A, 235, 131 Homan J., van der Klis M., Jonker P. G. et al. 2002, ApJ, 568, 878 Kara E. et al. 2019, Nature, 565, 198 Kuulkers E., van der Klis M., Oosterbroek T., Asai K., Dotani T., van Paradijs J., Lewin W. H. G. 1994, A&A, 289, 759 Kuulkers E., van der Klis M., Oosterbroek T., van Paradijs J., Lewin W. H. G. 1997, MNRAS, 287, 495 Kubota A., Tanaka Y., Makishima K., Ueda Y., Dotani T., Inoue H., Yamaoka K. 1998, PASJ, 50, 667 Lamb F. K. 1989, in Hunt J., Battrick B., eds, Proceedings of the 23rd ESLAB Symposium on Two Topics in X-Ray

Page 17 of 17

96

Astronomy, Vol. 1: X Ray Binaries, Vol. 2: AGN and the X Ray Background, ESA, Noordwijk, p. 215 Lei Y. J., Qu J. L., Song L. M. et al. 2008, ApJ, 677, 461 Lin D., Remillard R. A., Homan J. 2007, ApJ, 667, 1073 Lin D., Remillard R. A., Homan J., Barret D. 2012, ApJ, 756, 34 Ludlam R. M. et al. 2017, ApJ, 836, 140 Malu S., Sriram K., Agrawal V. K. 2020, MNRAS, vol. 499, Issue 2 (2009.11002v2) Manmoto T., Mineshige S., Kusunose M. 1997, ApJ, 489, 791 McKinney J. C., Tchekhovskoy A., Blandford R. D. 2012, MNRAS, 423, 3083 Mitsuda K., Inoue H., Koyama K. et al. 1984, PASJ, 36, 741 Miyamoto S., Kimura K., Kitamoto S., Dotani T., Ebisawa K. 1991, ApJ, 383, 784 Osherovich, V., Titarchuk, L. 1999a, ApJL, 522, 113 Osherovich, V., Titarchuk, L. 1999b, ApJL, 523, 73 Pen U.-L., Matzner C. D., Wong S. 2003, ApJ, 596, L207 Penninx W., Lewin W. H. G., Zijlstra A. A. et al. 1988, Nature, 336, 146 Priedhorsky W., Hasinger G., Lewin W. H. G., Middleditch J., Parmar A., Stella L., White N. 1986, ApJ, 306, L9 Psaltis et al. 1999, ApJ, 520, 763 Shakura N. I., Sunyaev R. A. 1973, A&A, 24, 337 Shimura T., Takahara F. 1995, ApJ, 445, 780 Sriram K., Agrawal V. K., Pendharkar J. K., Rao A. R. 2007, ApJ, 661, 1055 Sriram K., Agrawal V. K., Rao A. R. 2009, RAA, 9, 901 Sriram K., Rao A. R., Choi C. S. 2012, A&A, 541, A6 Sriram K., Malu S., Choi C. S. 2019, ApJS, 244, 5S Stella L., Vietri M. 1999, Phys. Rev. Lett., 82, 17 Stella L., Vietri M., Morsink S.M. 1999, ApJL, 524, 63 Titarchuk L., Bradshaw C. F., Geldzahler B. J., Fomalont E. B. 2001, ApJ, 555, 45 Titarchuk L. G., Osherovich V. A. 1999, ApJL, 518, 95 van der Klis M. 2006, in Lewin W., van der Klis M., eds, Compact Stellar X-ray Sources, Cambridge Univ. Press, Cambridge, p. 39 van der Klis M. 1988, in Ogelman H., van den Heuvel E. P. J., eds, NATO Advanced Science Institutes (ASI) Series C, Vol. 262, NATO Advanced Science Institutes (ASI) Series C, p. 27 Vaughan B. A., van der Klis M., Lewin W. H. G., van Paradijs J., Mitsuda K., Dotnai T. 1999, A&A, 343, 197 Vrtilek S. D., Raymond J. C., Garcia M. R. et al. 1990, A&A, 235, 162 White N. E., Peacock A., Hasinger G., Mason K. O., Manzo G., Taylor B. G., Branduardi-Raymont G. 1986, MNRAS, 218, 129 Wijnands R. et al. 1997, ApJ, 490, L157 Wu, X. B. 2001, ApJ, 552, 227 Yadav J. S., Agrawal P. C., Antia H. M. 2016, Proc. SPIE, 9905, 99051D Zdziarski A. A., Johnson W. N., Magdziarz P. 1996, MNRAS, 283, 193

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:39 https://doi.org/10.1007/s12036-021-09696-5

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

Multi-wavelength view of the galactic black-hole binary GRS 1716–249 SANDEEP K. ROUT1,2,* , SANTOSH V. VADAWALE1, E. AARTHY1,2,

SHASHIKIRAN GANESH1, VISHAL JOSHI1, JAYASHREE ROY3, RANJEEV MISRA3 and J. S. YADAV4 1

Physical Research Laboratory, Navarangpura, Ahmedabad 380 009, India. Indian Institute of Technology, Palaj 382 355, India. 3 Inter-University Center for Astronomy and Astrophysics, Ganeshkhind, Pune 411 007, India. 4 Department of Physics, Indian Institute of Technology, Kanpur 208 016, India. *Corresponding Author. E-mail: [email protected] 2

MS received 3 November 2020; accepted 22 December 2020 Abstract. The origins of X-ray and radio emissions during an X-ray binary outburst are comparatively better understood than those of ultraviolet, optical and infrared radiation. This is because multiple competing mechanisms – emission from intrinsic and irradiated disk, secondary star emission, synchrotron emission from jet and/or non-thermal electron cloud, etc – peak in these mid-energy ranges. Ascertaining the true emission mechanism and segregating the contribution of different mechanisms, if present, is important for correct understanding of the energetics of the system and hence its geometry and other properties. We have studied the multi-wavelength spectral energy distribution of the galactic X-ray binary GRS 1716-249 ranging from near infrared (5  104 keV) to hard X-rays (120 keV) using observations from AstroSat, Swift, and Mount Abu Infrared Observatory. Broadband spectral fitting suggests that the irradiated accretion disk dominates emission in ultraviolet and optical regimes. The near infrared emission exhibits some excess than the prediction of the irradiated disk model, which is most likely due to Synchrotron emission from jets as suggested by radio emission. Irradiation of the inner disk by the hard X-ray emission from the Corona also plays a significant role in accounting for the soft X-ray emission. Keywords. Black holes—X-ray binary—NIR/optical/UV—synchrotron emission—jets.

1. Introduction During outburst, a low mass X-ray binary brightens by several orders of magnitude in the entire electromagnetic spectrum. While the origins of radio and X-ray emission are extensively studied, the mid-energy emission in ultra-violet (UV), optical, and infrared (IR) poses a certain level of ambiguity (van Paradijs & McClintock 1995; Charles & Coe 2006). Soft X-rays are produced from the inner-most regions of the hot accretion disk due to thermal radiation, while hard Xrays, up to a few 100 keV, could be produced by inverse Comptonisation of the disk photons by an This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

optically thin hot electron cloud known as ‘‘Corona’’ (Done et al. 2007). Radio and sub-mm emissions are believed to originate through synchrotron processes in a bipolar jet (Falcke & Biermann 1996, 1999). Synchrotron emission has also been found to contribute in hard X-rays in some sources (Markoff et al. 2001; Vadawale et al. 2001). The origins of UV, optical, and IR emission are difficult to discern as there are multiple contenders for the same. The thin accretion disk, which is almost always approximated as a multitemperature black body, emits X-rays close to the compact object (within tens of gravitational radii) owing to its inner temperature reaching as high as  107 K and in longer wavelengths as one moves farther out (Shakura & Sunyaev 1973). This emission is further escalated by irradiation of the outer parts of

39

Page 2 of 11

the disk by X-rays from the inner disk and/or back scattered photons from the Corona (Hameury 2020). The strength of this irradiation depends on the geometry of the Corona and scale height of the outer disk (Cunningham 1976; van Paradijs & McClintock 1994). The companion star, which in case of low-mass X-ray binaries is a late type M, K, or G class star, peaks in optical or IR (OIR) wavelengths. This emission from the companion is also enhanced by irradiation of X-rays from the accretion disk by a few to several percentage. Synchrotron emission from relativistic jets can also dominate the OIR flux (Corbel & Fender 2002; Russell et al. 2006). In fact, the crucial break frequency dividing the optically thick and thin portion is believed to lie in the OIR bands which can help in quantifying the total energy content of a jet (Russell et al. 2013). However, this break frequency is observed only in a very few sources with good significance (Coriat et al. 2009). Deciphering the correct emission mechanism behind UV, optical, and IR emission remains challenging. Many techniques have been developed over the years to ascertain their true origin. Emissions from jet are known to show rapid variability in short time scales whereas those from disk or irradiated disk remain stable (Gandhi et al. 2008). High cadence observations in OIR bands can shed some light on the mechanism involved in its emission (Curran & Chaty 2013; Kosenkov et al. 2020). The study of correlations between contemporaneous OIR and X-ray emission can also help in picking the dominant mechanism (Russell et al. 2006; Bernardini et al. 2016). Fitting a simultaneous broadband spectral energy distribution (SED) spanning radio to X-rays is also a well-established method. Furthermore, there are several models which attempt to explain the low frequency radiation by synchrotron emission from a hot plasma above the disk (Veledina et al. 2013). In these models, the hard X-ray emission is produced by thermal Comptonisation of soft synchrotron photons. Optical excess is also thought to be produced from magnetic reconnections in flares on the disk (Merloni et al. 2000) or from gravitational energy release near the circularisation radius (Campana & Stella 2000). The problem, however, remains that of degeneracy wherein multiple models are able to satisfactorily explain the emissions whereas the available data are unable to discriminate between various competing models. Here we attempt to discern the origins of the near infrared (NIR), optical, and UV emissions observed during outburst of a black-hole binary GRS 1716–249 by evaluating the broadband SED.

J. Astrophys. Astr.

(2021) 42:39

GRS 1716–249 (aka GRO J1719–24 and Nova Oph 1993; hereafter, to be referred as GRS1716) went into outburst on 18 December 2016 after more than 20 years of quiescence (Negoro et al. 2016). During its discovery outburst in 1993, GRS1716 was measured to be located at a distance of 2:4  0:4 kpc along with harboring a K type (or late) companion in a 14.7 hr orbit (della Valle et al. 1994). The lower limit on black hole mass was estimated to be 4:9M (Masetti et al. 1996). During 2016, GRS1716 underwent a ‘‘failed’’ outburst wherein it did not transition into high/soft or soft-intermediate state (Bassi et al. 2019). It also lied in the outlier branch of radio/X-ray cor1 relation plot with LR / L1:4 X . Bharali et al. (2019) found minimal to no disk truncation along with detecting type C quasi periodic oscillations whose frequency increased with time. Jiang et al. (2020) also concluded similarly about the inner disk radius along with providing constraints on the disk density. With joint Swift and NuSTAR spectroscopy, Tao et al. (2019) constrained the spin and inclination of the source. The spin was constrained to a high value with a  0:92 and inclination was estimated to lie within 40 –50 . GRS1716 was also observed by AstroSat as a Target of Opportunity on three epochs: (1) 15 February 2017 (57799), (2) 6 April 2017 (57849), and (3) 13 July 2017 (57947). It was also observed from Mount Abu Infrared Observatory during March to May 2017 for 24 nights in optical and NIR bands. We present the joint spectral analyses of all three instruments of AstroSat along with a multi-wavelength SED study to find the origin of NIR/optical/UV emission. The UV data were observed from Swift/UVOT (Ultraviolet and optical Telescope). We have also utilised radio data from the Australian Large Baseline Array in the SED. The observation logs for this work are noted in Table 1. All the observations are further marked alongside the full MAXI light curve in Fig. 1.

2. Observations and data reduction 2.1 Soft X-ray telescope (SXT/AstroSat) SXT (Singh et al. 2016, 2017) data were analysed with the standard analysis software and other auxiliary tools developed by the Payloads Operations Center (POC2). The tool sxtpipeline was run to generate 1

LR and LX refer to radio and X-ray luminosity respectively.

2

http://www.tifr.res.in/astrosat_sxt.

J. Astrophys. Astr.

(2021) 42:39

Page 3 of 11

39

Table 1. List of all observations used in this work. The wavelengths in UV, optical and NIR bands corresponds to the filters in the third column. Bands

Observatory

Instrument

Energy/Wavelength

DateH

X-ray UV Optical Optical NIR Radioy

Astrosat Swift Swift MIRO MIRO LBA

SXT, LAXPC, CZTI UVOT (W2, M2, W1) UVOT (U, V, B) CCD (B, V, R, I) NICS (J, H, Ks) –

1–120 keV ˚ 1928, 2246, 2600 A ˚ 3465, 4392, 5468 A ˚ 4353, 5477, 6349, 8797 A 1.25, 1.64, 2.15 lm 8.4 GHz

15 February, 06 April, 13 July 28 January–13 August 31 January–20 October 22 March–28 May 17 April–25 May 22 April

H

All observations were made in the year 2017. y The data for the radio observation was reported by Bassi et al. (2019).

Figure 1. MAXI on-demand light curve of GRS1716 in 2–20 keV band (black circles; Matsuoka et al. 2009). The blue vertical bars mark the three epochs of AstroSat observations. UVOT observations in all six bands are depicted with violet bars. Optical (B, V, R, I) and NIR (J, H, Ks) observations from Mt. Abu are represented with red and green vertical bars.

orbit-wise Level 2 event files. This extraction takes care of most of the elementary data cleaning and good-time interval (GTI) selection. The event files for every orbit were then merged using the script sxtevtmergerjl. Final products were generated using the FTool xselect after incorporating custom GTIs to remove flaring regions and drop outs from the light curve. SXT spectra are known to be affected by pile up for rates above  40 counts/s (or 200 mCrab) in Photon Counting (PC) mode. The count rates for GRS1716 during the first two epochs were  42 and  59 counts/s respectively. In order to verify the presence of pile up, annular regions with outer radius fixed at 120 and inner radius varying from 10 upwards were used to generate spectra. Each of these spectra was fitted with an absorbed multi-color disk and power-law model and then the variation of photon

index studied. The spectra were found to be piled up and thus annular regions with inner radii of 10 and 2:50 were selected for source extraction for epoch 1 and 2 observations respectively. The epoch 3 spectrum, with a count rate of  28 counts/s, was not piled up and hence, a circular region of 150 radius was opted. To incorporate the changes in the effective area due to the annular region and vignetting caused by the off-axis positioning of the PSF (Point Spread Function), the default ARF (Auxiliary Response File) was scaled using the script sxtARFModule. The response matrix and the background spectra were provided by the POC. The spectra were grouped to have a minimum 30 counts per energy bin and a systematic error of 2% was added.

2.2 Large Area X-ray Proportional Counter (LAXPC/AstroSat) The analysis of LAXPC (Yadav et al. 2016a, b; Agrawal et al. 2017) data were carried out using the Format A – LAXPCsoftware_Aug4 package.3 LAXPC30 module was affected by gas leakage resulting in continuous gain instability (Antia et al. 2017). Hence, only LAXPC10 and LAXPC20 were chosen for spectral analysis in this work. All the layers of both the LAXPCs were opted to maximise the signal as there were more than 15% of counts in the bottom four layers. Level 2 event files were generated using the tool laxpc_make_event which was followed by the usage of laxpc_make_spectra and laxpc_make_backspectra for source and background spectral extractions. The channels in the spectra were grouped by a factor of 5 and a systematic error of 2% was added. 3

http://astrosat-ssc.iucaa.in/.

39

Page 4 of 11

J. Astrophys. Astr.

(2021) 42:39

2.3 Cadmium Zinc Telluride Imager (CZTI/AstroSat)

2.5 Mount Abu Infrared Observatory (PRL)

The reduction of CZTI (Vadawale et al. 2016; Bhalerao et al. 2017) level 1 files to level 2 and final products were carried out using the tool cztpipeline (version 2.1). Spectrum from only quadrant 0 was used for analysis as the other quadrants are affected by higher systematics. The spectra were grouped to have a minimum of 30 counts per energy bin and no systematic error was added.

The Mount Abu Infrared Observatory (MIRO), managed by Physical Research Laboratory, Ahmedabad, houses a 1.2-m f/13 telescope with Cassegrain focus. Out of the many back-end instruments in operation, two were used in this work: (1) Near-Infrared Camera/ Spectrograph (NICS; Anandarao et al. 2008), and (2) Optical CCD. The optical filters (B, V, R, I) are in Johnson–Cousins photometric system and the NIR (J, H, Ks) filters are in MKO system. Four campaigns of observations were undertaken spanning 17 March to 28 May 2017 starting from a mid-plateau region to the first quarter of outburst decay of GRS1716 (Joshi et al. 2017). Standard aperture photometry was carried out for optical and NIR observations using the IRAF package. NIR observations were acquired in 5 dithered positions separated by  3000 . These frames were median combined to produce a sky-frame which was subtracted from individual raw frames. Individual frames of the raw images for each day was subjected to bias subtraction for optical observations. Variable pixel response was corrected by the standard procedure of flat-field correction for both optical and NIR observations. The instrumental magnitudes of 3 to 4 field stars were then compared with corresponding apparent magnitudes from the standard catalogues viz. SDSS and 2MASS for optical and NIR, respectively, to find a zero point. The R and I band magnitudes of the comparison stars were in Sloan filters. The differences in Sloan and Johnson-Cousins filters were taken into account using transformation equations calculated by Jordi et al. (2006). Similarly, the 2MASS magnitudes of the comparison stars were converted to MKO system.4 This zero-point factor was, in turn, considered to calculate the apparent magnitude and standard deviation of the source. The Vega flux for magnitude to flux conversion for all 7 filters were taken from Bessell et al. (1998). The optical and NIR lightcurves with MIRO are shown in Fig. 3.

2.4 Ultraviolet and Optical Telescope (UVOT/Swift) GRS1716 was observed several times during the outburst with UVOT onboard Swift satellite (Breeveld et al. 2011). Photometry for all the filters (W2, M2, W1, U, B, V) was done using the tool uvotsource. A circular region of 500 radius was considered for source while a source-free aperture of 2000 radius was chosen for background extraction. The U, B, and V magnitudes of UVOT were converted to Johnson system using the conversion factors described by Poole et al. (2008). The flux conversions were also done using the zero-point values for Vega flux given by Poole et al. The lightcurves with all six UVOT filters are shown in Fig. 2.

3. Analyses and results 3.1 Joint AstroSat spectral fitting Figure 2. Light curves in UV and optical bands using Swift/UVOT. The top three panels display the light curves in UV filters (W2, M2, W1) while the bottom three show the optical light curves (U, V, B). The blue dots in V and B bands represent the same filters as observed from MIRO.

Spectral analysis was carried out by jointly fitting SXT, LAXPC10, LAXPC20, and CZTI using xspec 4

https://sites.astro.caltech.edu/jmc/2mass/v3/transformations/.

J. Astrophys. Astr.

(2021) 42:39

Figure 3. Optical and NIR light curve of GRS1716 as observed from MIRO. From top to bottom: the black points represent the optical filters of B, V, R, I respectively (also labelled on the plots). Similarly, the red points represent the NIR bands of J, H, and Ks respectively.

(Arnaud 1996). The energy ranges of SXT and LAXPC spectra were restricted to 1–7 keV and 5–60 keV respectively so as to avoid higher systematic errors outside these ranges. CZTI spectrum was fitted in the full range of 30–120 keV. Thus, the combination of the three instruments resulted in a contiguous and wide energy coverage from 1 to 120 keV (henceforth, the 4 AstroSat spectra would be referred to as one X-ray spectrum). For SXT, an additional gain correction was added using the command gain fit in xspec. The best fit offset was found to be  40 eV, for a unit slope, which improved the fits significantly.5 To adjust for the cross-calibration discrepancy among the different instruments, a constant was multiplied to the model. It was fixed to unity for LAXPC10 and was left free to vary for the rest of the instruments. The average value of the best-fit constant factor for SXT and CZTI varied around 21% of LAXPC10’s factor whereas LAXPC20 varied between 5%. These are within the expected uncertainties in the effective areas of the three instruments. 5

https://www.tifr.res.in/*astrosat_sxt/instrument.html

Page 5 of 11

39

The AstroSat spectrum was fitted with an absorbed multi-temperature accretion disk (diskbb; Mitsuda et al. 1984; Makishima et al. 1986) and a thermal Comptonisation model (nthComp; Zdziarski et al. _ 1996; Zycki et al. 1999) – TBabs*(diskbb? nthComp) – for all three epochs. The abundance for ISM absorption in TBabs was set to Wilms et al. (2000). The fit for all epochs were statistically good with v2m staying around 1.04. The hydrogen column density (NH ) was constrained to 0:7  1022 cm2 , varying by 0.1 across the three epochs. The inner disk temperature and photon index were constrained to  0.3 keV and  1.7 respectively. The spectrum for all three epochs is depicted in Fig. 4 and the residuals of the above fit are shown in the top panel of Fig. 5. Despite good statistics, the residuals seem to be of somewhat wavy nature indicating the presence of reprocessed Coronal emission from the disk. Although reflection is a common candidate for reprocessing, we did not find telltale features like broad Fe line or Compton hump in the residuals. Nevertheless, we checked for the presence of reflection in the spectra by adding a relativistic reflection component – relxillCp – to the existing model (Dauser et al. 2014; Garcı´a et al. 2014). The photon index and cutoff energy in relxillCp were tied to the corresponding parameters in nthComp. The emissivity index was fixed to the Newtonian value and the Fe abundance was set to that of the Sun. The spin parameter was fixed to 0.998 and the innerdisk radius was left free to vary. The possible values of inclination vary between 30 –50 as reported by various authors (Bharali et al. 2019; Tao et al. 2019). Since inclination could not be constrained, we fixed it to a rough average of 40 . Rest of the parameters were left to float. There was only a marginal improvement in fit compared to the previous model with Dv2 of 7.7, 10.4, and 24.1 per degree of freedom (dof) for epochs 1, 2, and 3 respectively. The inner radius could not be strongly constrained but hinted towards a possible disk truncation. Fixing spin to intermediate and Schwarschild values like a ¼ 0:7, 0.3 and 0 also did not have any effect. The cutoff energy was only weakly constrained in the first epoch at 74:2þ52:8 14:3 . For the other two cases, it pegged at the maximum limit and could not be constrained. This is in contrast to the values obtained by Bassi et al. (2019) who strongly constrained kTe to  48 and  52 keV for the first two epochs, and weakly at  74 keV for the third. This is probably due to competition of the Compton hump to fit the curvature around 40–50 keV which resulted in the cutoff energy being unconstrained.

39

Page 6 of 11

Figure 4. AstroSat spectra of all three epochs and instruments. From top to bottom: Epoch 1: 15 February 2017 (black), Epoch 2: 06 April 2017 (red), and Epoch 3: 13 July 2017 (blue). The spectrum LAXPC20 is represented by a lighter shade on all three days.

In the hard state of black-hole binaries, thermalisation of Comptonised photons in the inner disk can also substantially contribute to the disk emission (Gierlin´ski et al. 2008). To test this, the spectrum was fitted with the diskir model (Gierlin´ski et al. 2008). diskir is a hybrid of both blackbody and Comptonisation components. It parameterises irradiation by two additional components: (1) fraction of Compton tail that is thermalised (fin ), and (2) radius of Compton illuminated disk (rirr ); along with calculating the ratio of luminosities in Compton tail and unilluminated disk (Lc =Ld ). It also has two more parameters for effects of

J. Astrophys. Astr.

(2021) 42:39

irradiation at the outer disk. These parameters were frozen to nominal values as they would not affect Xray emission and would be considered in the next exercise where broadband SED fitting is undertaken. First, irradiation was turned off by freezing fin to 0. The best-fit parameters and fit statistics were almost identical to the previous fits with TBabs* (diskbb?nthComp). The values of Lc =Ld were constrained around 5 indicating strong reprocessing which would also affect the thermal disk emission (Gierlin´ski et al. 2008). To include this effect, fin was fixed to 0.1 and rirr was left free to vary while fitting. For epochs 2 and 3, the fits improved significantly while for epoch 1 it deteriorated a bit. fin , for epoch 1, was found to be smaller by roughly a factor of 3. rirr was constrained to  1.01Rin (where Rin is the innerdisk radius). The improvement in fit after including irradiation was verified by F-Test, wherein the probability that this advancement would be random was found to be less than 1015 for all three epochs. The best-fit parameters are listed in Table 2 and the residuals are represented in bottom panel of Fig. 5. There is a clear improvement in the fit after including irradiation. Fits with disk irradiation were also statistically better compared to reflection model with Dv2 decreasing as 32.2, 11.6, and 26.1 per dof for the three epochs. Thus, the data favors the case for disk irradiation and does not require any reflection component. In reality, however, the spectrum could have contribution from both reflection and irradiation. However, deciphering the exact fraction of contribution from each of the two would be extremely difficult given the modest resolutions of SXT and LAXPC.

Figure 5. The top panel displays the residuals of the AstroSat spectra with diskir model assuming no irradiation in the inner disk for all three epochs. The bottom panel displays residuals after including irradiation. The color coding is the same as in Fig. 4. Dots, up-triangles, down-triangles, and squares are used to represent SXT, LAXPC10, LAXPC20, and CZTI respectively.

J. Astrophys. Astr.

(2021) 42:39

Page 7 of 11

39

Table 2. Best-fit parameters of the joint X-ray fit with SXT, LAXPC, and CZTI using TBabs*diskir model. Date (MJD)

NH (10 cm2 )

57799 57849 57947

0:64  0:05 0:66  0:02 0:52  0:02

H

22

kTin (keV)

C

kTe (keV)

Lc =Ld

fin

8:2  0:6 0:03H 0:47  0:02 1:60  0:01 46þ7 5 0:44  0:01 1:68  0:01 22  1 2:54  0:04 0:1H 0:48  0:01 1:65  0:01 23þ1 0:1H 2:73þ0:05 2 0:14

rirr (Rin )

Norm (103 )

1:012  0:002 2:26þ0:79 0:49 1:011  0:001 5:80þ0:04 0:03 1:010  0:001 1:87þ0:03 0:01

v2m (v =dof) 2

0.93 0.92 0.91

Fixed during fit.

3.2 Broadband spectral energy distribution Although there were many observations of GRS1716 in the low energy bands, there were not any strictly simultaneous with AstroSat observations. It was only during the second epoch (06 April) of AstroSat when a number of observations with MIRO and UVOT were in temporal proximity with it. The closest were the NIR observations which were made on 7 April while the optical measurements were made a few days later on 11 April. Similarly, the UV observations in W1, U, B, and V bands were scattered within a few days of 6 April. Moreover, around the date of the X-ray observation, the optical and UV flux did not vary significantly allowing multi-waveband spectroscopy (Figures 2 and 3). The other two UV bands of UVOT (M2 and W2) were much farther away in time from the X-ray observation and also suffered from heavy extinction leading to huge uncertainty in flux measurement. Therefore, they were not included in the SED. As hinted in Section 3.1, the model diskir also calculates the effect of irradiation in the outer disk where the emission is predominantly in UV, optical, and NIR bands (Gierlin´ski et al. 2009). This is parameterised as fout and Rout which represents the fraction of total flux thermalising the outer disk and the outer-disk radius respectively. The NIR, optical, and UV magnitudes were converted to flux in physical units and then incorporated into PHA files, one each for MIRO and UVOT. Diagonal response matrices were created such that the convolved model would be in the same unit as the spectra. All spectra were loaded into xspec and fitted simultaneously with the model TBabs*redden*diskir. Since MIRO and UVOT spectra were not corrected for interstellar absorption, the xspec routine redden was employed to calculate extinction (Cardelli et al. 1989). It has one free parameter, the color excess (EBV ), and it was left to float. NH and EBV were fixed to 0 for UV/optical/NIR

and X-ray spectrum, respectively, to avoid intermixing of the effects. EBV was constrained to 0:75  0:04 and NH was constrained to 0:70  0:04  1022 cm2 making the ratio EBV =NH  1:1  1022 cm2 mag. This ratio is fully consistent with the recent findings of Lenz et al. (2017). The cumulative line of sight galactic reddening is  0.93 (Schlafly & Finkbeiner 2011) and our result is also consistent with this limit. The best-fit value of fout was 0:04  0:01 while Rout was constrained to  105:5 Rin . Although the irradiated disk model explained the optical and UV bands well, it failed to account for the excess in the NIR emission (see Fig. 6). To verify whether the NIR excess could be due to intrinsic absorption we multiplied another reddening component to the model. We fixed one EBV to the best-fit value of 0.75 and let the other to float. The extra reddening component could not explain the NIR excess and was weakly constrained to a small value of  0.05.

4. Discussion GRS1716 exhibited ‘‘failed’’ outburst and never transitioned to the canonical soft state (Bassi et al. 2019). This was also suggested by the broadband rms variability of the source in the 3-30 keV band. For epoch 1, the variability was 24% whereas for epochs 2 and 3 the variability remained around 20%. During the three AstroSat observations, spanned across  5 months, GRS1716 remained in a powerlaw dominant state with the luminosity ratio (Lc =Ld ) remaining [2. The spectrum of the source was significantly affected by irradiation of the back scattered Compton flux in the inner disk (Gierlin´ski et al. 2008). For hard states of black-hole binaries, the fraction of the thermalising flux (fin ) is about 0.1 (Poutanen et al. 1997). This fraction is a function of the geometry of the electron cloud and angle-averaged albedo of the thin disk. While epochs 2 and 3 confirmed the expected value of

39

Page 8 of 11

Figure 6. SED of GRS1716 after fitting with TBabs* diskir to the quasi-simultaneous observations on epoch 2. The X-ray spectrum, in steel blue color, is unfolded and rebinned by a factor of 4 for clarity. The SXT, LAXPC10, LAXPC20, and CZTI are represented by dots, up-triangles, down-triangles, and squares respectively. In lower energies, the NIR spectrum is represented with red circles, optical points with green diamonds, and UV points with violet stars. The black dashed line represents the best-fit model.

0.1, fin for the first epoch was constrained to 0.03. This suggests a possible change in geometry, and hence covering fraction, of the overlying electron cloud. The complete outburst of GRS1716 is marked by 3 small softening episodes, also characterised by increase in flux. The epoch 2 AstroSat observation (MJD 57849) was done 5 days before the second peak (MJD 57854). The effect of this softening was reflected in the spectral fits wherein, there was an increase in the disk flux and decrease in the luminosity ratio (Table 2). We monitored GRS1716 in the optical and NIR bands from MIRO during mid-March to May 2017. The optical light curve (Fig. 2) remained constant throughout the observations while the NIR light curve was marked by a drop in flux (1–2 magnitudes) around MJD 57850. The UVOT light curve (Fig. 3), which spanned a longer duration, displayed constant flux in all bands up to MJD 57900 and a gradual decrease in the UV bands thereafter. Using quasi-simultaneous Xray, UV, optical and NIR spectra we carried out a broadband spectral study to decipher the origin of the low energy emission using an irradiated disk model. The irradiated disk perfectly explained the optical and UV flux while slightly underestimating the NIR

J. Astrophys. Astr.

(2021) 42:39

spectrum (Fig. 6). Without irradiation, the model underpredicted the flux by almost 3 orders of magnitude in all the bands. From the best-fit norm of diskir (4181þ1451 887 ), the inner-disk radius was calculated to be  22 km, which was consistent with radii obtained by Bassi et al. (2019). The best-fit value of fout was found to be 0:03  0:01 and that of Rout was constrained to 6:27  106 km. This Rout is slightly larger than the Roche lobe radius of 1:56  106 km estimated from the reported orbital parameters and assuming a 10M black hole and 45 disk inclination.6 Considering the uncertainties involved in the above calculation, we can infer that the outer disk has to be as big as the Roche lobe in order to describe the UV/optical spectrum. Trying to increase Rout to account for the NIR spectrum would make the disk unrealistically large and also overestimate the optical spectrum. A natural alternative for the this excess is emission from a secondary star. We searched for the images of the source in All-Sky surveys to quantify the flux during quiescence which would be predominantly from the companion. However, the 2MASS H-band image of the field does not have any object at the position of GRS1716 (Fig. 7). The H-band image from MIRO, on the other hand, shows a bright object at the source position. Hence, the NIR brightening which the source had undergone during the outburst is not due to the secondary star. It is also possible for the emissions from the binary to be absorbed and reemitted in the IR regime from a dust envelope/cloud covering the binary. Taranova and Shenavrin (2001) reported such a scenario where the X-ray binary XTE J1118þ480 showed excess in mid-IR regime which could be explained by a 900 K circumstellar dust envelope. To test this hypothesis, we added the bbodyrad model to the SED only in the UV/optical/ NIR region. We fixed the temperature to 900 K and fitted for the norm. The best-fit radius of the cloud was found to be 3:3  108 km. The cloud, if present, could as well be hotter and smaller in size or cooler and larger. Without observations in longer wavelengths, it is difficult to constraint any of these properties robustly. Although the primary source of heating of a bright dust cloud is disk emission during an outburst, it can also be moderately excited by emission from the secondary star. The object was, 6

The Roche lobe radius—distance of the inner Lagrangian point to the primary object (black hole)—was estimated using the following tool: http://www.orbitsimulator.com/formulas/Lagran gePointFinder.html.

J. Astrophys. Astr.

(2021) 42:39

Page 9 of 11

39

Figure 7. Left panel shows the 2MASS image of the source during quiescence phase. The purple crossbar marks the source position. The right panel shows the H band image of the same field-of-view as observed with NICS instrument at MIRO during the outburst. Inside the purple circle, a bright object is clearly seen.

however, not detected in the mid-IR bands of the WISE (Wide-field Infrared Survey Explorer) catalogue observed during the 2010–11 epoch. Moreover, the source was also not detected in NIR, or even optical bands, after we resumed observations postmonsoon during September–October 2017. During these months the source had decayed substantially in X-rays but was still bright in UV (see UVOT light curves in Fig. 3) This suggests that the NIR and optical brightening was exclusively tied to the X-ray activity. The possibility of a NIR emitting dust cloud that engulfs the entire system is, therefore, highly speculative. The other most viable candidate for the NIR excess is Synchrotron emission from a compact jet. As reported by Bassi et al. (2019), GRS1716 was detected in radio wavebands throughout the outburst by ATCA, LBA, and VLA from 9 February 2017 till 13 August 2017 (see their Table 3). The closest radio observation to our SED was made on 22 April using LBA. The flux in the 8.4 GHz band, with a bandwidth of 64 MHz, was reported to be 1:13  0:11 mJy. The observation before 22 April was made about two months earlier, on 9 February, during which the flux was 1:28  0:15 mJy. This means the flux would not have changed much during the intervening period. The correlation between radio and X-ray luminosities provide an useful tool to study the emission properties of black-hole binaries. These are known to follow two distinct power-law relations in the log-log space. The radio-loud systems follow a relation LR / LX1:4

whereas the radio-quiet systems (the so-called ‘‘outliers’’) follow LR / L0:6 (Corbel et al. 2013). Using x the radio luminosity from LBA in the 8.4 GHz band (LR 6:5  1028 erg s1 ) and the X-ray luminosity in the 1–10 keV band from AstroSat (LX 4:5  1036 erg s1 ) we obtained a radio/X-ray luminosity relation of LR / LX1:45 . Here, we have assumed a distance of 2.4 kpc and a proportionality constant of 1.85 (Corbel et al. 2013). Thus, GRS1716 adds to the pool of sources in the ‘‘outlier’’ branch of the radio/X-ray plane, consistent with the findings of Bassi et al. (2019). To have a cursory idea of the full radio to X-ray SED, we also add the 22 April radio observation in the SED (Fig. 8). Unfortunately, with the data available with us, it is not possible to identify the exact position of the break frequency. We tried fitting the radio to NIR spectrum with a broken powerlaw but could not constrain the parameters, especially the break frequency. An approximate spectral index of the radio spectrum is þ0:5, obtained by fixing the spectral break at the Ks band of the NIR spectrum. Such highly inverted optically thick part of the radio spectrum has been seen earlier for a few sources such as MAXI J1836–194 (Russell et al. 2014), XTE J1118þ480 (Fender et al. 2001), etc. Although standard jet models, as that of Blandford and Ko¨nigl (1979), assuming a conical geometry predict a shallower slope, steeper spectrum can be expected for a rapidly flaring jet geometry (Dinc¸er et al. 2018).

39

Page 10 of 11

J. Astrophys. Astr.

(2021) 42:39

the HEASARC data archives and were analyzed using the information provided by the UK Swift Science Data Center (University of Leicester). This research has also made use of MAXI data provided by RIKEN, JAXA, and the MAXI team. SKR thanks Sushree S. Nayak for valuable feedback on the manuscript.

References

Figure 8. Complete SED of GRS1716 along with the 22 April radio observation with LBA. The X-ray spectrum is rebinned by a factor of 4 for clarity. The X-ray, UV, optical, and NIR spectra are the same as in Fig. 6. The radio observation is marked with a black open triangle. The grey dashed line joining the radio and Ks band is just for representation and not a fitted model.

5. Conclusion We have presented the results of a multi-wavelength spectral analysis of the galactic X-ray binary GRS 1716–249 using data in X-rays from AstroSat, NIR/ optical from MIRO and UV from Swift/UVOT. Broadband X-ray spectral analysis of all three epochs of AstroSat spectra show that the source was in a power-law dominant state. Irradiation of X-rays in the inner regions of the accretion disk significantly contribute to the soft X-ray flux of the source on all three epochs. Using multi-wavelength SED analysis, we found the optical and UV flux to originate from the irradiated outer accretion disk while parts of the NIR emission is most likely emitted from a jet.

Acknowledgements This work was supported by Physical Research Laboratory, a unit of Department of Space, Government of India. It uses data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). We thank the POCs of SXT (TIFR), LAXPC (TIFR), and CZTI (IUCAA) for verifying and releasing the data via ISSDC data archive and providing the necessary data analysis software through the AstroSat Science Support Cell. UVOT data were obtained from

Agrawal P. C. et al. 2017, J. Astrophys. Astr., 38, 30 Anandarao B. et al. 2008, SPIE, 7014, 70142Y Antia H. M. et al. 2017, ApJS, 231, 10 Arnaud K. A. 1996, ASPC, 101, 17 Bassi T. et al. 2019, MNRAS, 482, 1587 Bernardini F. et al. 2016, ApJ, 826, 149 Bessell M. S., Castelli F., Plez B. 1998, A&A, 333, 231 Bhalerao V. et al. 2017, J. Astrophys. Astr., 38, 31 Bharali P. et al. 2019, MNRAS, 487, 3150 Blandford R. D., Ko¨nigl A. 1979, ApJ, 232, 34 Breeveld A. A. et al. 2011, AIPC, 1358, 373 Campana S., Stella L. 2000, ApJ, 541, 849 Cardelli J. A., Clayton G. C., Mathis J. S. 1989, ApJ, 345, 245 Charles P. A., Coe M. J. 2006, csxs.book, 39, 215 Corbel S., Fender R. P. 2002, ApJL, 573, L35 Corbel S. et al. 2013, MNRAS, 428, 2500 Coriat M. et al. 2009, MNRAS, 400, 123 Cunningham C. 1976, ApJ, 208, 534 Curran P. A., Chaty S. 2013, A&A, 557, A45 Dauser T. et al. 2014, MNRAS, 444, L100 della Valle M., Mirabel I. F., Rodriguez L. F. 1994, A&A, 290, 803 Dinc¸er T. et al. 2018, ApJ, 852, 4 Done C., Gierlin´ski M., Kubota A. 2007, A&ARv, 15, 1 Falcke H., Biermann P. L. 1996, A&A, 308, 321 Falcke H., Biermann P. L. 1999, A&A, 342, 49 Fender R. P. et al. 2001, MNRAS, 322, L23 Gandhi P. et al. 2008, MNRAS, 390, L29 Garcı´a J. et al. 2014, ApJ, 782, 76 Gierlin´ski M., Done C., Page K. 2008, MNRAS, 388, 753 Gierlin´ski M., Done C., Page K. 2009, MNRAS, 392, 1106 Hameury J. M. 2020, AdSpR, 66, 1004 Jiang J. et al. 2020, MNRAS, 492, 1947 Jordi K., Grebel E. K., Ammon K. 2006, A&A, 460, 339 Joshi V., Vadwale S., Ganesh S. 2017, ATel, 10196, 1 Kosenkov I. A. et al. 2020, A&A, 638, A127 Lenz D., Hensley B. S., Dore´ O. 2017, ApJ, 846, 38 Makishima K. et al. 1986, ApJ, 308, 635 Markoff S., Falcke H., Fender R. 2001, A&A, 372, L25 Masetti N. et al. 1996, A&A, 314, 123 Matsuoka M. et al. 2009, PASJ, 61, 999 Merloni A., Di Matteo T., Fabian A. C. 2000, MNRAS, 318, L15 Mitsuda K. et al. 1984, PASJ, 36, 741

J. Astrophys. Astr.

(2021) 42:39

Negoro H. et al. 2016, ATel, 9876, 1 Poole T. S. et al. 2008, MNRAS, 383, 627 Poutanen J., Krolik J. H., Ryde F. 1997, MNRAS, 292, L21 Russell D. M. et al. 2006, MNRAS, 371, 1334 Russell D. M. et al. 2013, MNRAS, 429, 815 Russell T. D. et al. 2014, MNRAS, 439, 1390 Schlafly E. F., Finkbeiner D. P. 2011, ApJ, 737, 103 Singh K. P. et al. 2016, SPIE, 9905, 99051E Singh K. P. et al. 2017, J. Astrophys. Astr., 38, 29 Shakura N. I., Sunyaev R. A. 1973, A&A, 500, 33 Tao L. et al. 2019, ApJ, 887, 184 Taranova O. G., Shenavrin V. I. 2001, AstL, 27, 25

Page 11 of 11

39

Vadawale S. V., Rao A. R., Chakrabarti S. K. 2001, A&A, 372, 793 Vadawale S. V. et al. 2016, SPIE, 9905, 99051G van Paradijs J., McClintock J. E. 1994, A&A, 290, 133 van Paradijs J., McClintock J. E. 1995, xrbi.nasa, 58 Veledina A., Poutanen J., Vurm I. 2013, MNRAS, 430, 3196 Wilms J., Allen A., McCray R. 2000, ApJ, 542, 914 Yadav J. S. et al. 2016, SPIE, 9905, 99051D Yadav J. S. et al. 2016, ApJ, 833, 27 Zdziarski A. A., Johnson W. N., Magdziarz P. 1996, MNRAS, 283, 193 _ Zycki P. T., Done C., Smith D. A. 1999, MNRAS, 309, 561

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:77 https://doi.org/10.1007/s12036-020-09677-0

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

Observations of bright stars with AstroSat soft X-ray telescope K. P. SINGH1,* , G. STEWART2, S. CHANDRA3, G. C. DEWANGAN4,

S. BHATTACHARYYA5, N. S. KAMBLE5, S. VISHWAKARMA5 and J. G. KOYANDE5 1

Indian Institute of Science Education and Research Mohali, Sector 81, P.O. Manauli, SAS Nagar 140 306, India. 2 Department of Physics and Astronomy, The University of Leicester, University Road, Leicester LE1 7RH, UK. 3 Center for Space Research, North-West University, Potchefstroom 2520, South Africa. 4 Inter-University Centre for Astronomy and Astrophysics, Ganeshkhind, Pune 411 007, India. 5 Department of Astronomy and Astrophysics, Tata Institute of Fundamental Research, Homi Bhabha Road, Mumbai 400 005, India. *Corresponding author. E-mail: [email protected] MS received 5 November 2020; accepted 4 December 2020 Abstract. We present observations of four bright stars observed with the AstroSat Soft X-ray Telescope (SXT). Visible light from bright stars like these can leak through the very thin filter in front of the CCD in the focal plane CCD camera of the SXT and thus making the extraction of X-ray events difficult. Here, we show how to extract the X-ray events without contamination by the visible light. The procedure applied to four bright stars here demonstrates how reliable X-ray information can be derived in such cases. The sample of bright stars studied here consists of two A spectral types (HIP 19265, HIP 88580), one G/K giant (Capella), and a nearby M-type dwarf (HIP 23309). No X-ray emission is observed from the A-type stars, as expected. X-ray spectra of Capella and HIP 23309 are derived and modeled here, and compared with the previous Xray observations of these stars to show the reliability of the method used. We find that optical light can start to leak in the very soft energy bands below 0.5 keV for stars with V ¼ 8 mag. In the process, we present the first X-ray spectrum of HIP 23309. Keywords. Stars: individual: HIP 19265—HIP 88580—Capella—HIP 23309—stars: coronae—X-rays: stars.

1. Introduction The focal plane camera in the Soft X-ray Telescope (SXT) (Singh et al. 2016, 2017) aboard the AstroSat (Singh et al. 2014) carries a very thin optical light blocking filter in front of the CCD to block visible light but to allow the transmission of soft X-rays. The filter consists of a single fixed polyimide film which is 1840 Angstroms thick and coated with 488 Angstroms of aluminum on one side. The filter is similar to the This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

thin filter aboard the European Photon Imaging Camera (EPIC) (Struder et al. 2001) used in the XMM-Newton and the X-ray Telescope (XRT) aboard the Swift Observatory (Burrows et al. 2005). The CCD used in the SXT is identical to the one used in the cameras of XMM-Newton and Swift. The filter has to be thin to allow the transmission of soft X-rays while blocking the visible light from the cosmic X-ray sources. The X-ray transmission of the filter is shown in Fig. 1. The typical optical transmission of the filter is less than 5  103 (similar to the XMM-Newton thin filter and the filter onboard Swift X-ray telescope). The filter design provides  7 magnitude of optical

77

Page 2 of 9

Figure 1. X-ray transmission efficiency through the thin optical blocking filter in the SXT.

extinction over the visible band. For the Swift XRT with a PSF of  15 arcsec a 6th magnitude star gives an optical loading of a few e-per pixel, at which point the quality of the X-ray data begins to be affected. For the SXT with a  7–8 times larger PSF and  2 times larger angular size of the pixel the safe optical limit is expected to be closer to a  4th magnitude star, but is needed to be verified by post-launch observations and to check when the visible light can start leaking through the filter. This is specially important, as the SXT is occasionally pointed towards very bright stars with V  8 which can have a significant contribution to the events registered in the CCD due to visible light photons, thus contaminating the X-ray data in the very soft bands. For this purpose, we describe here SXT observations of a few bright stars two of which are non-X-ray emitting stars and the other two are bright X-ray emitting stars. We describe how to handle data from such observations and to obtain reliable X-ray information.

J. Astrophys. Astr. (2021)42:77

are X-ray dark because they neither have an active corona or strong colliding and shocked winds to produce X-ray emission – the two processes known to produce X-rays in stars. A very small number of A type stars that have been detected in X-rays are all suspected to harbor a late type companion and thus not single or they have a very peculiar chemistry or/and magnetic field (Ap or Am stars). One of the stars in our sample is Capella, a nearby G type giant that is known to be highly coronally active with copious X-ray emission and has been studied extensively in the past. Finally, we have an active M-type dwarf which was detected in the ROSAT All-Sky Survey and has not been looked at with X-ray observatories since then. There is very little information available on HIP 19265 and HIP 88580, other than what has been given in Table 1, except that they may have infrared excesses (McDonald et al. 2012). Capella, apart from being a very bright visible star system, believed to be a spectroscopic binary consisting of a K0 III star plus a rapidly rotating (period  8 days) G1 III star in a 104 day orbit (Hummel et al. 1994; Strassmeier & Fekel 1990). Both the stars are bright X-ray emitting coronal stars (Ayres et al. 1983; Linsky et al. 1998; Brickhouse et al. 2000; Gu et al. 2006; Raassen & Kaastra 2007). HIP 23309 is a high proper motion star and a member of the b Pictoris moving group (Mamajek & Bell 2014). Anomalous high proper motion in nearby stars in Hipparcos and Gaia catalogs are likely to be a signature of possible substellar companions, and therefore important targets for further studies (Kervella et al. 2019). It has rotational velocity, v sin i ¼ 5:8  1:5 km s1 , and is very young with age estimated to be 10±3 Myr (Weise et al. 2010). It was detected in soft X-rays in the ROSAT All-Sky Survey II (RASS II) with a flux of 2.30  1012 ergs cm2 s1 in the 0.5–2.0 keV energy band (Schwope et al. 2000). Photometric variability in optical has been reported from this star by Kiraga (2012).

2. The sample of stars

3. Observations

We have selected four bright single stars: HIP 19265, HIP 88580, Capella and HIP 23309 for our study here. Some of their important properties are listed in Table 1. Two of these have A0 spectral type which are generally known to be X-ray dark, and have never been detected in X-rays. Stars with A-type spectrum

All observations were carried out in the photon counting (PC) mode of the SXT. Each source was observed continuously in an orbit of the satellite keeping the Sun avoidance angle  45 and RAM angle (the angle between the payload axis to the velocity vector direction of the spacecraft) [ 12 to

J. Astrophys. Astr. (2021)42:77

Page 3 of 9

77

Table 1. Properties of stars observed with SXT. Names Other names

HIP 19265 HD 24716

HIP 88580 HD 165505

Capella aAur

HIP 23309 CD-57 1054

A0 8.07 7.94 7.94* 325

A0 8.05 7.96 7.96* 226.6

G3 III 0.88 0.08 - 0.52 13.1

M0Ve 11.36 9.98 9.89 26.90

Parameters Spectral type B (mag) V (mag) R=G (mag) Distance (pc)

Distances are based on parallax measurements given in Gaia DR2 (Bailer-Jones et al. 2018), and at Gaia website: http:// gaia.ari.uni-heidelberg.de/tap.html.

Table 2. Log of observations. Star name

Observation ID

Start time (UT) Y:M:D:H:M:S

Stop time (UT) Y:M:D:H:M:S

Effective exposure (s)

Count rate 0.3-3.0 keV

HIP 19265 HIP 88580 Capella HIP 23309

9000000076 9000000266 9000000298 9000001720

2015:11:04:15:17:59 2016:01:12:17:52:39 2016:01:27:16:15:54 2017:11:24:02:08:24

2015:11:05:03:14:24 2016:01:13:03:42:12 2016:01:28:23:41:12 2017:11:24:20:27:07

9296 1947 30720 17970

\0.01 \0.02 0.66±0.005 0.069±0.003

ensure the safety of the mirrors and the detector. A log of the observations is given in Table 2. Level 1 Data from individual orbits received at the SXT POC (Payload Operation Centre) from the ISSDC (Indian Space Science Data Center) were further processed with the sxtpipeline available in the SXT software (AS1SXTLevel2, version 1.4b). The source events were calibrated, extracting Level-2 cleaned event files for the individual orbits were extracted. The cleaned event files of the individual orbits were merged into a single cleaned event file to avoid the time-overlaps in the events from consecutive orbits using Julia based merger tool. The XSELECT (V2.4d) package built-in HEAsoft was used to extract the images, spectra and to examine light curves from the processed Level-2 cleaned event files. The useful exposure times for each source thus obtained are listed in Table 2.

4. Data analysis and results 4.1 Extracting X-ray events X-ray images were extracted from the observations of the stars shown in Tables 1 and 2. These images, extracted in the energy band of 0.3–3.0 keV, are

shown in Figures 2, 3, 4 and 5 for HIP 19265, HIP 88580, Capella, and HIP 23309. There were no events detected from the position of HIP 19265, showing lack of any signal from visible light or soft X-rays. Several events were registered in the SXT data from the position of HIP 88580 showing a strong detection as can be seen in Fig. 3 (left panel). This star, like most A-stars, is not expected to show any X-ray emission. An examination of the pulse-height information shown in the left panel of Fig. 6 showed that almost all these events were confined to pulse-heights corresponding to energies below 0.7 keV and resembled split events (events with charge split onto neighbouring pixels, known as events with grades [1) (see Burrows et al. 2005). These grades are used to distinguish between X-ray photons and charged particles (and Compton-scattered high energy photons) in CCD based cameras. These grades can range from 0 type (single pixel events where the X-ray photon is absorbed in a single pixel of the CCD) to 36 types depending on the pattern of the charge splitting registered in the CCD. Grades from 0–12 only are identified as due to X-rays, while grade zero events are generally pure X-ray events. The default setting in the processing to Level 2 data is to use grades 0–12 to maximise the number of events registered and thus

77

Page 4 of 9

J. Astrophys. Astr. (2021)42:77

Figure 2. SXT image of HIP 19265 in 0.3–3.0 keV energy band for event grades 0 to 12. The source region is shown as a circle centered on the position of the star and used for extraction of photons used in the analysis.

Figure 3. SXT images of HIP 88580 in 0.3–3.0 keV energy band grade 0 to 12 on the left, and grade 0 only on the right. The green circle shows the extraction region for getting the spectra of HIP 88580.

improve the signal-to-noise ratio as most X-ray sources are weak. This default selection of events led to the signal seen in the left panel of Fig. 3 and the red data points in the left panel of Fig. 6. Selection of events with different grades can be made while using XSELECT. We found that by selecting events with grade 0 (single pixel events) the source practically disappeared, as can be seen in the image shown in the right panel of Fig. 3 and black data point in the left panel of Fig. 6. The few single pixel events

correspond to energies below 0.3 keV, the lowest recommended threshold for the SXT. It, therefore, appears that leaked optical photons resemble higher grades (  1) and are most likely due to arrival of several visible light photons within the readout time of 2.3775 s of the CCD. The same technique applied to an extremely bright star like Capella, however, is not sufficient to get X-ray events. The pile up is extremely large in the low pulse-height channels that it overwhelms the CCD

J. Astrophys. Astr. (2021)42:77

Page 5 of 9

77

Figure 4. SXT images of Capella in 0.3–3.0 keV energy band grade 0 to 12 on the left, and grade 0 only on the right. The magenta circles define the annular extraction region used for getting the spectra of capella.

Figure 5. SXT images of HIP 23309 in 0.3–3.0 keV energy band grade 0 to 12 on the left, and grade 0 only on the right. The cyan circle defines extraction region used for getting the spectra of HIP 23309, while the cyan rectangular box defines the extraction region for the background.

electronics leading to overflow and registering zero counts in the centre portion leading to a dark patch shown in Fig. 4, irrespective of the grades chosen. In this case, we excluded central dark patch extending to a radius of 8 arcmin. The X-ray events were then extracted from an annular region with radii of 8 arcmin and 16 arcmin. A comparison of the spectrum from such events extracted for all grades 0–12 and grade 0 is shown in the middle panel of Fig. 6, which shows that there is a pile-up of events with all grades

0–12 even after the exclusion of central portion, which almost disappears for single pixel events. In such cases, a combination of avoiding the central region and extracting only single pixel events can work quite well to extract X-ray events. This is further corroborated by the modeling of X-ray spectra thus obtained, as described below. The fourth star in our sample, HIP 23309, is only moderately bright, and a comparison of images extracted for the grades 0–12 and grade 0 is shown in

77

Page 6 of 9

J. Astrophys. Astr. (2021)42:77

Figure 6. Comparison of spectra extracted using events with grade 0 and events with all grades from 0 to 12: HIP 88580 (left), Capella (right), HIP 23309 (centre).

Fig. 5, while the corresponding spectra for such events recorded are shown in the right panel of Fig. 6. There seems to be a pileup for all grades but is confined below our low threshold of 0.3 keV, while the spectra above that energy are almost identical. Modelling of the X-ray spectra from single pixel events recorded from Capella and HIP 23309 is described below.

4.2 Modeling of X-ray spectra X-ray spectra were extracted for the entire observation as described above (grade 0 only) for the four stars in our sample, after checking that there are no variations in the count rates. The useful exposure times and the average count rates for the sources are given in Table 2. The X-ray counts from the sources in the spectra were grouped using the grppha tool to ensure a minimum of 25 counts per bin, prior to further analysis here and below. The response matrix,

sxt_pc_mat_g0.rmf, calculated for only single pixel events was used, and is available at the SXT POC website https://www.tifr.res.in/astrosat_sxt/index.html. For Capella, we used a specially made ancillary response file (ARF): sxt_pc_excl00_v04_ann8to16arcm_ 20190608.arf made by using the tool sxtmkarf appropriate for the source location and the annular extraction size on the CCD plane. For HIP 23309, we used the standard ARF file: sxt_arf_excl00_v04_20190608.arf, available at the SXT POC website. The background for the Capella was estimated from a deep exposure of 123900 s on a source free region with observation ID of 9000000298, and using the grade 0 events only. The background file name is bg_id190_12am_g0.pha, and this will be made available to public from the SXT POC website. The background for the HIP 23309 was extracted from a rectangular box region of the same observation as the source, shown in Fig. 5. Single pixel events were used here, as well. The source extraction region for the HIP 23309 was circular with a radius of 11 arcmin.

J. Astrophys. Astr. (2021)42:77

Page 7 of 9

77

Figure 7. X-ray spectra of Capella (left) and HIP 23309 (right) with best-fit optically-thin plasma models (vapec) with variable abundances.

Table 3. Spectral parameters for Capella obtained from SXT data (0.3–5.0 keV). Parameters a

Spectral model

tbabs*apec tbabs*apec tbabs*(apec ? apec) tbabs*(apec ? apec) tbabs*vapec tbabs*(vapec ? vapec)

kT1apec (keV)

Zb

A1 c

kT2apec

A2 c

v2m /dof

Fluxd

0.78 0.78þ0:008 0:008 0.73þ0:014 0:014 0.69þ0:04 0:03 0.75þ0:013 0:013 0.73þ0:014 0:015

1.0 0.32þ0:02 0:02 1.0 0.44þ0:06 0:04

4.61 11.3 3.79 6.5 8.59 6.66

– – 1.625þ0:12 0:14 1.14þ0:25 0:10 – 1.95þ0:70 0:40

– – 1.89 3.19 – 1.37

3.79/190 1.596/189 1.827/188 1.229/187 1.358/184 1.115/182

1.10 1.20 1.20 1.21 1.21 1.21

e f

NH in tbabs is kept fixed at 2  1018 cm2 for all spectral models; b abundance, Z, is relative to solar values for all the elements; c A1 and A2 are normalisations in units of 102 photons cm2 s1 ; d fluxes are in units of 1010 ergs cm2 s1 and are þ0:26 þ0:10 þ0:09 þ0:34 quoted for total energy of 0.3–5.0 keV; e O ¼ 0:83þ0:23 0:21 , Ne ¼ 0:980:25 , Mg ¼ 0:540:09 , Si ¼ 0:440:08 , S ¼ 0:800:31 , f Fe ¼ 0:39þ0:040 O ¼ 0:94þ0:25 0:034 ; 0:30 , Ne ¼ 1:25  0:35, Mg ¼ 0:55  0:11, Si ¼ 0:43  0:11, S ¼ 0:55  0:35, Fe ¼ þ0:07 0:500:06 ; all errors quoted are with 90% confidence.

a

The single pixel spectra for the two stars are shown in Fig. 7. X-ray spectra of Capella and HIP 23309, were fitted with optically-thin plasma emission models apec using xspec program (version 12.9.1; Arnaud 1996) distributed with the heasoft package (version 6.20). The atomic data base used was AtomDB version 3.0.7 (http://www.atomdb.org). An absorber model Tbabs was used as a multiplicative model with the model parameter NH , i.e., the equivalent Galactic neutral hydrogen column density, which was fixed at a low value of 2  1018 cm2 . The elemental abundance table aspl given by Asplund et al. (2009) was used in our analysis. We used v2 minimisation technique to find

the best fit parameters of the plasma emission models. We tried single temperature apec as well as two temperature apec models, with solar as well as non-solar elemental abundances. The normalisation and temperature (kT) for the plasma component(s) were kept free. The abundances of all the elements were tied together and could be varied together with respect to the solar values as one parameter. The results of our modelling are presented in Table 3 for Capella and Table 4 for HIP 23309. Single temperature plasma models with solar abundances did not fit the spectrum of Capella as the reduced v2 , henceforth v2m , was unacceptably high. Similarly, two temperature models with solar

77

Page 8 of 9

J. Astrophys. Astr. (2021)42:77

Table 4. Spectral parameters for HIP 23309 obtained from SXT data (0.3–7.0 keV). Parameters a

Spectral model

tbabs*apec tbabs*apec tbabs*(apec ? apec) tbabs*(apec ? apec) tbabs*vapec

b

kT1apec (keV)

Z

0.84 0.90þ0:06 0:08 0.79þ0:05 0:06 0.83þ0:09 0:06 0.92þ0:06 0:1

1.0 0.13þ0:06 0:04 1.0 0.15þ0:11 0:05 e

A1

c

1.18 5.30 0.85 4.5 3.3

kT2apec

A2 c

v2m /dof

Fluxd

– – 3.1þ3:0 1:0  1.6 –

– – 1.39 0.86 –

2.247/56 1.040/55 1.213/54 0.98/53 0.948/50

2.8 3.8 4.0 4.3 4.0

NH in tbabs is kept fixed at 21018 cm2 for all spectral models; b abundance, Z, is relative to solar values for all the elements; c A1 and A2 are normalisations in units of 103 photons cm2 s1 ; d fluxes are in units of 1012 ergs cm2 s1 and are þ0:5 quoted for total energy of 0.3–7.1 keV; e O ¼ 2:32þ2:7 1:6 , Ne\1:0, Mg\0:5, Si ¼ 0:40:3 , S\3:0, Fe ¼ 0:2  0:1. All errors quoted are with 90% confidence.

a

abundances also gave a poor fit with v2m of 1.827 for 188 degrees of freedom. The fits were considerable improved when the elemental abudances were varied, either together for all the elements or individually based on the vapec models (see Table 3). The best fit was obtained with two temperature vapec models þ0:70 with temperature of 0.73þ0:014 0:015 keV and 1.950:40 keV and with abundances of O ¼ 0:94þ0:25 0:30 , Ne ¼ 1:25 0:35, Mg ¼ 0:55  0:11, Si ¼ 0:43  0:11, S ¼ 0:55 0:35, Fe ¼ 0:50þ0:07 0:06 , relative to the solar values. The emission measures (EM) of the two components obtained from this best-fit are 1:4  1053 and 2:52  1052 for the low and high temperature components, respectively. The best fit models are shown as histograms in the left panel of Fig. 7. The contributions of the two temperature components are also shown individually. The low temperature component dominates the emission in the best-fit model. The X-ray spectrum of HIP 23309 could not be fitted with single temperature solar abundance plasma models. Varying the abundance of all the elements to a very low sub-solar values gave an acceptable fit for a single temperature plasma. Two temperature plasma models with solar abundances were also able to fit the data as shown by acceptable values of the v2m shown in Table 4, which improved further with sub-solar abundances. The best fit with the lowest v2m was, however, obtained by varying the abundances of the individual elements and using a single temperature models. We estimate the elemental abundances in the optically-thin coronal plasma as: O ¼ 2:32þ2:7 1:6 , þ0:5 Ne\1:0, Mg\0:5, Si ¼ 0:40:3 , S\3:0, Fe ¼ 0:2  0:1 times solar. This best-fit model is shown as a histogram in the right panel of Fig. 7.

5. Discussion Capella has been studied quite extensively with a similarly low resolution CCD in the ASCA observatory (Brickhouse et al. 2000) and also with very high spectral resolution instruments like Low Energy Transmission Grating (LETG) and High Energy Transmission Gratings (HETG) aboard Chandra Xray Observatory (Gu et al. 2006; Raassen & Kaastra 2007). It is known to have a very complex X-ray spectrum that has been used to refine atomic data used in the plasma codes and also shows long term variations of  30–50% (Brickhouse et al. 2000; Gu et al. 2006; Raassen & Kaastra 2007; Gu et al. 2020). In almost all these studies, one sees a continuous distribution of emission measures with a range of temperatures from very low (kT = 0.3 keV) to very high (kT = 4 keV) (Gu et al. 2006; Raassen & Kaastra 2007), with two peaks: one at kT  0.55–0.70 keV and another broader peak at kT  1.7–2.2 keV. The best fit two temperature vapec model obtained here has temperatures very close to these values. The EM value of 6.74pD 2  1012 cm3 , where D is the distance of Capella, for the low temperature component is comparable to the peak value obtained by Gu et al. (2006). The EM value of 1.44pD 2  1012 cm3 , for the high temperature component is lower than the peak value obtained by Gu et al. (2006). It should, however, be noted that we have used two discrete temperature components and not the differential emission measure (DEM) analysis adopted by Gu et al. (2006) and these values are very much dependent on the atomic database used and the abundances thus derived. Our low resolution spectrum is not sufficient to carry out DEM analysis here.

J. Astrophys. Astr. (2021)42:77

We provide the first detailed spectroscopy of HIP 23309, an M0Ve star. The X-ray flux measured by us in the energy band of 0.5–2.0 keV of the ROSAT is 3.01012 ergs cm2 s1 , which is about 30% higher than the value given in the RASS II. We measure the X-ray emission measure of the star as 2.91052 cm3 and its X-ray luminosity as 3.51030 ergs s1 in the energy band of 0.3–7.1 keV, for the adopted distance of 26.9 pc. These values firmly place this star as a group of extremely active M dwarfs (Singh et al. 1999), which could be the result of its young age and possibly a very rapid rotation.

6. Conclusions We have shown how using single pixel events from the data recorded in the SXT observations of moderately bright stars of V  8 mag can be used to extract X-ray spectral information above the low threshold of 0.3 keV, despite the leakage of visibly light photons through the thin filter of the SXT. For stars that are extrmely bright, like Capella, one needs to disregard the photons from the central core of the point spread function of the SXT as well while using the single pixel event data. X-ray spectra of Capella and HIP 23309 have been thus extracted reliably as compared with the past measurements. In the process, we have provided the first detailed X-ray spectrum of a nearby young active M dwarf.

Acknowledgements The authors thank the Indian Space Research Organisation for scheduling the observations and the Indian Space Science Data Centre (ISSDC) for making the data available. This work has been performed utilizing the calibration data-bases and auxillary analysis tools developed, maintained and distributed by AstroSat-SXT team with members from various institutions in India and abroad and the SXT Payload Operation Center (POC) at the TIFR, Mumbai for the pipeline reduction. The work has also made use of software, and/or web tools obtained from NASA’s High Energy Astrophysics Science Archive Research Center (HEASARC), a service of the Goddard Space Flight Center and the Smithsonian Astrophysical Observatory.

Page 9 of 9

77

References Arnaud K. A. 1996, in Jacoby G., Barnes J., eds, Astronomical Data Analysis Software and Systems V, p. 17, ASP Conf. Series, vol. 101 Asplund M., Grevesse N., Sauval A. J., Scott P. 2009, ARAA, 47, 481 Ayres T. R., Schiffer F. H., Linsky J. L. 1983, ApJ, 272, 223 Bailer-Jones C. A. L., Rybizki J., Fouesneau M., Mantelet G., Andrae R. 2018, AJ, 156, 58 Brickhouse N. S., Dupree A. K., Edgar R. J., Liedahlm D. A., Drake S. A., White N. E., Singh K. P. 2000, ApJ, 530, 387 Burrows D. N., Hill J. E., Nousek J. A., Kennea J. A., Wells A., Osborne J. P., Abbey A. F. et al. 2005, SSRv, 120, 165B Gu L., Shah C., Mao J., Raassen T. et al. 2020, A&A, 641, A93 Gu M. F., Gupta R., Peterson J. R., Sako M., Kahn S. M. 2006, ApJ, 649, 979 Hummel C. A., Armstrong J. T., Quirrenbach A., Buscher D. F., Mozurkewich D., Elias N. M. II. 1994, AJ, 107, 1859 Kervella P., Arenou F., Mignard F., The´venin F. 2019, A&A, 623, A72 Kiraga M. 2012, Acta Astronomica, 62, 67 Linsky J. L., Wood B. E., Brown A., Osten R. A. 1998, ApJ, 492, 767 McDonald I., Zijlstra A. A., Boyer M. L. 2012, MNRAS, 427, 343 Mamajek E. E., Bell C. P. M. 2014, MNRAS, 445, 2169 Peterson E., Littlefield C., Garnavich P. 2019, AJ, 158, 131 Raassen A. J. J., Kaastra J. S. 2007, A&A, 461, 679 Roberts D. H., Lehar J., Dreher J. W. 1987, AJ, 93, 968 Schwope A. D., Hasinger G., Lehmann I., Schwarz R., Brunner H., Neizvestny S., Ugryumov A., Balega Yu, Trueu¨mper J., Voges W. 2000, Astron. Nachr., 321, 1 Singh K. P., Drake S. A., Gottehlf E. V., White N. E. 1999, ApJ, 512, 874 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, Proc. SPIE, Space Telescopes and Instrumentation 2014: Ultraviolet to Gamma Ray. 9144, 91441S, https://doi. org/10.1117/12.2062667 Singh K. P., Stewart, G. C., Chandra S. et al. 2016, Proc. SPIE, in Space Telescopes and Instrumentation 2016: Ultraviolet to Gamma Ray. 9905, p. 99051E, https://doi. org/10.1117/12.2235309 Singh K. P., Stewart G. C., Westergaard N. J. et al. 2017, J. Astrophys. Astr., 38, 29 Strassmeier K. G., Fekel F. C. 1990, A&A, 230, 389 Struder L., Briel U., Dennerl K. et al. 2001, A&A, 365, L18 Weise P., Launhardt R., Setiawan J., Henning T. 2010, A&A, 517, A88

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:55 https://doi.org/10.1007/s12036-021-09734-2

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

An alternative scheme to estimate AstroSat/LAXPC background for faint sources RANJEEV MISRA1, JAYASHREE ROY1,*

and J. S. YADAV2

1

Inter-University Centre for Astronomy and Astrophysics, Pune 411 007, India. Department of Physics, Indian Institute of Technology Kanpur, Kanpur 208 016, India. *Corresponding Author. E-mail: [email protected] 2

MS received 5 November 2020; accepted 11 February 2021 Abstract. An alternative scheme is described to estimate the layer 1 LAXPC 20 background for faint sources where the source contribution to the 50–80 keV count rate is less than 0.25 counts/s (15 milli-crabs or 6  1011 ergs/s/cm2 ). We consider 12 blank sky observations and based on their 50–80 keV count rate in 100 second time-bins, generate four template spectra which are then used to estimate the background spectrum and lightcurve for a given faint source observation. The variance of the estimated background subtracted spectra for the 12 blank sky observations is taken as the energy dependent systematic uncertainty which will dominate over the statistical one for exposures longer than 5 ks. The estimated 100 second time bin background lightcurve in the 4–20 keV band with a 3% systematic error matches with the blank sky ones. The 4–20 keV spectrum can be constrained for a source with flux ’1 milli-crab. Fractional rms variability of 10% can be determined for a  5 milli-crab source lightcurve binned at 100 seconds. To illustrate the scheme, the lightcurves and spectra of three different blank sky observations, three AGN sources (Mrk 0926, Mrk 110, NGC 4593) and LMC X-1 are shown. Keywords. AstroSat/LAXPC—instrument background—calibration.

1. Introduction The sensitivity of the LAXPC instrument (Antia et al. 2017) onboard AstroSat (Yadav et al. 2016; Agrawal et al. 2017) to extract spectral and long-term temporal information of faint sources depends critically on how well the background of the instrument is characterized. The background variation is primarily due to the changing response of the instrument to a varying local charged particle distribution. The cosmic X-ray background from the  0.25 square degree field of view contributes less than 10% of the observed background and hence its cosmic variance does not contribute significantly to the background variation. The standard method (Antia et al. 2017) to estimate the background involves blank sky spectra obtained as a function of latitude and longitude of the satellite. For This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

a given science observation, a blank sky observation is chosen which is typically one that is closest in time to the science observation. The background is estimated as that expected for the latitude and longitude covered during the science observation based on the blank sky observation after taking into account gain variation between the blank sky and source observations. This method provides background spectra which differ from the true background by roughly 3% and has been extensively used for science analysis. This systematic uncertainty exceeds the statistical one for exposures longer than 5 ks. It is prudent to have different and independent methods to estimate the background to provide confidence on the scientific results obtained. Here we describe such a scheme which assumes that for faint sources the detected flux in the high energy band (5080 keV) can be attributed to the background alone and hence can act as a proxy to measure the background level as a function of time. As shown in this work, the

55

Page 2 of 7

technique is applicable to sources that contribute less than 15 milli-Crab of flux in the 50–80 keV band. The scheme named as ‘‘faint source background estimation’’ has been incorporated in the LAXPC software laxpcsoft available at the AstroSat Science Support Cell.1 It has been used for scientific analysis of several faint sources (e.g. LMC X-1 (Mudambi et al. 2020), RGB J0710?591 Goswami et al. 2020; Yadav et al. 2021).

2. Estimating LAXPC background We consider twelve blank sky observations that are listed in Table 1. For these observations, the lightcurve in 100 seconds for LAXPC 20 Layer 1, were computed in energy bands 4–20 keV and 50–80 keV bands. The count rates of these two energy bands are plotted against each other in Fig. 1. Most of the data lie within 50–80 keV count rate of 14 to 19 counts/s. We find that it is prudent to consider data only in this range, and divide the range into four parts corresponding to 14–15, 15–16, 16–17 and 17–19 counts/s. The average spectra corresponding to these selections are shown in Fig. 2. The spectra have been normalized such that 50–80 keV flux levels are nearly equal, in order to highlight the different spectral shapes at low energies.2 We note that the spectral shapes are different at low energies and treat these four spectra as templates for estimating the background spectrum and lightcurve for a source. The procedure for estimating the background for a source is as follows: (1) Collect the 50–80 keV lightcurve in 100 second time bins. (2) Select GTIs based on the count rate of the 50–80 keV energy range being between 14 and 19 counts/s. (3) For each time bin of the 50–80 keV, use the observed count rate to estimate the complete background energy spectrum using the four templates. The four templates are assigned to the mid point of their count rate in 50–80 keV, i.e. 14.5, 15.5, 15.5 and 18 counts/s. The templates are then interpolated to obtain the corresponding spectrum appropriate for the observed count rate. 1

http://astrosat-ssc.iucaa.in/?q=laxpcData. Due to gain variation, the channel to energy conversion for the blank sky observations are different. Here, we consider one observation spectrum as the reference and we interpolate the other spectra such that all of them have the same energy bin channels as that of the reference.

2

J. Astrophys. Astr.

(2021) 42:55

(4) Integrate the estimated background energy spectrum for each time bin over the desired energy band. (5) Combine the estimated background energy spectra to estimate the time averaged background spectrum. To test the efficacy of the method, the background spectra was estimated for each of the blank sky observations using the method described above. The estimated background was then compared with the observed spectrum and at each energy bin. The twelve blank sky observations were used to get the standard deviation of the estimated background and the observed spectra as a function of energy. The standard deviation in units of counts/s/keV are shown as a function of energy in Fig. 3. This standard deviation can be used as the energy dependent systematic error on an estimated background spectrum obtained from this method. The standard deviation is compared with a typical background spectrum shown in Fig. 3 and hence the systematic error on the background is of the order of a few percent. Thus, the systematics attained from this method are of the same order as that from the standard technique. The standard deviation of the background subtracted count rates in 4–20 keV band for the 12 blank sky observations is around 0.3 counts/s, indicating a 3-sigma detection of a source to be  1 counts/s in this energy band. Also shown is the 1 milli-crab source spectrum, which reveals that although the source count rate is a factor of few below the background, it should be detected using this method. The systematics at 25 keV is comparable to the source flux from a 1 mCrab source. For comparative reference, the typical Poisson noise levels for an exposure of 5 and 50 ks are shown. Note that the systematic error dominates over the Poisson one for exposures longer than 5 ks. The software includes this error in the background spectrum file. The estimated background lightcurve in any energy band is based on the count rate in the 50–80 keV band. The deviation of the estimated lightcurve from the true background is due to the systematic limitations of the technique and the typical Poisson error of the 50–80 keV band count rate in 100 second time bin which is  2.5%. As shown in the next section, an addition of a uniform 3% error on the estimated lightcurve at each 100 second time bin, leads to consistent estimates of the expected variance for blank sky observations. The spectral templates used in this technique correspond to 50–80 keV count rates separated by 1 count/s and which is assumed to be only from the background. Hence this scheme is limited to cases when the source count rate is significantly less than the template

J. Astrophys. Astr.

(2021) 42:55

Page 3 of 7

55

Table 1. Details of the twelve blank sky observations used to generate the template spectra. Target Sky-9_75_50 Sky-5 Sky-6 Sky-10 Sky-6 Sky-3 Abell3535 Sky-9_75_50 Sky_4u1626 Sky-8 Sky-9_75_50 Blank Sky 5 255-50

Observation ID

R.A. (deg)

Decl. (deg)

Date

Exposure (ks)

G05_156T09_9000000604 T01_132T01_9000000636 C01_015T01_9000000668 G06_115T01_9000000734 C02_011T01_9000000850 C02_003T01_9000000924 A02_108T01_9000001024 G07_044T09_9000001334 G07_049T02_9000001354 C02_021T01_9000001482 G08_046T09_9000001600 A04_198T01_9000001708

237.37 57.37 7.65 321.22 7.65 129.48 194.45 237.37 250.00 237.39 237.37 57.37

47.10 –47.10 12.55 –48.68 12.55 –27.89 –28.49 47.10 –70.00 70.35 47.10 –47.10

2016 Aug 16 2016 Aug 30 2016 Sep 15 2016 Oct 16 2016 Dec 03 2016 Dec 24 2017 Feb 11 2017 Jun 24 2017 Jul 04 2017 Aug 21 2017 Oct 11 2017 Nov 21

32.0 27.4 43.0 35.9 39.4 32.1 49.1 32.9 22.4 39.5 28.9 35.6

Figure 1. Count rates for the 4–20 keV and 50–80 keV plotted against each other for LAXPC 20 Layer 1. The timescale for integration is 100 s.

separation rate. Thus it is applicable for sources with count rate \0.25 counts/s which translates to 15 millicrabs or 6  1011 ergs/s/cm2 in the 50–80 keV band. The estimated background lightcurves and spectra for LAXPC 10 can also be estimated using the same technique. However, since LAXPC 10 has an higher background with larger uncertainty than LAXPC 20, it is recommended that LAXPC 20 should be primarily used for such analysis and LAXPC 10 results to be taken as a corroboration.

3. Verification and examples To illustrate the method, lightcurves and spectra were generated for three blank sky observations from 2017, 2018 and 2019 which were not part of the

Figure 2. Average blank sky spectrum corresponding to when the 50–80 keV count rates are in the range 14–15,15–16,16–17 and 17–20 counts/s. The spectra have been normalized such that 50–80 keV flux levels are nearly equal, in order to highlight the different spectral shapes at low energies. These four spectra are used as templates to estimate the background spectrum and lightcurve of a source observation.

observations used to obtain the templates, three Active Galactic Nuclei (Mrk 0926, Mrk 110 and NGC 4593), and for the extra-galactic X-ray binary LMC X-1. The lightcurves were generated for a time bin of 100 s and for 4–20 keV energy range. The left panels of Figures 4 and 5 show the total lightcurve (i.e. source with background marked as Src?Bkg), the estimated background lightcurve and the subtraction of the two for the blank sky observations from 2017 and 2019. The 2019 blank sky observation shows increased count rate for two times just before the satellite entered the SAA. If the two

55

Page 4 of 7

Figure 3. The standard deviation of the estimated background as compared to blank sky observations. Also shown for comparison are a typical blank sky spectrum, the spectrum for a 1 milli-crab source and the typical Poisson level for the blank sky spectrum.

increases are removed from the data then the resultant lightcurves are similar to ones obtained for the 2017 data (middle panel of Fig. 5). While the reason for these higher counts rates is not clear, such variations just before entry into SAA should be treated with caution in science analysis. These are most likely due to local variation/fluctuations in geomagnetic field. This is to emphasize that proper time-selection by inspection is required to obtain reliable results. The right panels of Figures 4 and 5, show the residuals of the observed spectra over the background for the 2017 and 2019 blank sky observations. Although the residuals are shown for a wide energy band, note that the technique is only valid in 4–30 keV range. This shows that the systematics included in the background

J. Astrophys. Astr.

(2021) 42:55

spectra are adequate and no significant residuals are seen in the 4–30 keV band. Residuals of the order of 0.02 counts/s/keV should be attributed to systematics. Since a 3% uncertainty has been included in the estimated background lightcurve, the error on the background subtracted lightcurve is a combination of the Poisson noise in the observed lightcurve and this background uncertainty. Table 2 lists the average count rate, the variance (r2 ), the standard deviation pffiffi ( ðr2 Þ), the expected variance (r2EV ) and the expecpffiffi ted standard deviation ( ðr2EV Þ) for the background subtracted lightcurves binned in 100 and 1000 seconds. For these three blank sky observations the average count rate is within the 0.3 counts/s deviation found for the twelve blank sky used as the template pffiffi for the technique. The standard deviation ( ðr2 Þ) and pffiffi the expected one ( ðr2EV Þ) are similar. This implies that the background estimation technique can be applied to study lightcurves in 100 to 1000 second binning over a time-span of  40 ks. Note that the errors quoted in the Table 2 for the measured and expected variance are statistical ones and hence maybe underestimated. The recommendation is that an excess variance of more than 1 counts/s, should be considered as evidence of variability. The background subtracted spectra for the AGN sources were fitted using a power-law and for LMC X-1 a disk emission model and power-law was used. Figure 6 shows the background subtracted spectrum along with the expected background spectrum (top panel) and residuals (bottom panel) for Mrk 0926. We emphasize that the spectra here are shown for the full energy range of 4–80 keV for illustration, and science analysis should be limited to 30 keV, since the higher

Figure 4. Left panels shows the lightcurves of the blank sky observations from 2017 (top panel: the total source with background count rate, middle panel: the background and the bottom panel: the background subtracted count rate from the source). Blank spaces between the lightcurves pertains to the SAA passages. Right figure shows the residuals of the energy spectrum.

J. Astrophys. Astr.

(2021) 42:55

Page 5 of 7

55

Figure 5. Left figure shows the lightcurves of the blank sky observations from 2019 (top panel: the total source with background count rate, middle panel: background and the bottom panel: the background subtracted count rate from the source). Right figure shows the same lightcurves when the two increases due to entry of the satellite in SAA passages are removed from the data. Bottom figure shows the residuals of the energy spectrum.

energy 50–80 keV spectrum has been used to calibrate the background. Similar results were obtained for the other sources and the residuals are of the order of 0.02 counts/s/keV as expected from the blank sky observations. Table 2 lists the properties of the background subtracted lightcurves for the sources and the flux in the 4-20 keV band. Mrk 0926 and Mrk 110 are clearly detected with a background subtracted average count rate of 8 and 6 counts/s (Table 2), but show no variability with the observed standard deviation of the same order as the expected one. On the other hand NGC 4593 has a lower count rate of 4 counts/s but shows slight evidence of variability with pffiffi a rms of ðr2  r2EV Þ  0:7 counts/s. The AstroSat observation of extra galactic source LMC X-1 has been reported by Mudambi et al. (2020). It shows a count rate of 19 counts/s with clear evidence of variability with rms of  1.2 counts/s. These results show that a LAXPC observation of a source with a 4–20 keV band flux ’2  1011 ergs/s/cm2 (i.e. ’2 counts/s or ’1 milli-crab), will be able to constrain the source spectrum. For 100 second binned lightcurves a variability of  1.0 counts/s can be detected which translates to a fractional rms of 10% for a 5 milli-crab source.

4. Summary and discussion We have presented an alternate scheme to estimate the background spectrum and lightcurve for AstroSat LAXPC 20 based on using the detected count rate in the 50–80 keV as a measure of the background. A software that incorporates the scheme is available at AstroSat Science Support Cell3. The software also can compute the estimated background lightcurves and spectra for LAXPC 10 using the same technique. However, since LAXPC 10 has an higher background with larger uncertainty than LAXPC 20, it is recommended that LAXPC 20 should be primarily used for such analysis and LAXPC 10 results to be taken as a corroboration. The scheme can be perhaps be improved by exploring possibilities to better estimate the background rate. This includes considering only single events to see if the correlation between the low and high energy counts in the blank sky spectra is tighter. The correlation between the low and high energy count rate does vary for different blank sky observations and one can examine if any satellite system parameter can be used to predict the variation. A related idea would be to study the correlation as a function of latitude and 3

http://astrosat-ssc.iucaa.in/?q=laxpcData.

70.35

70.35

–8.69

237.39

237.39

346.18

141.30

84.91

189.91

Mrk 110

LMC X-1

NGC 4593

–5.34

–69.74 2016 Jul 16

2016 Nov 25

2017 Apr 15

2016 Nov 28

2019 Feb 01

2018 Apr 11

2017 Jan 10

Date

39.9

54.2

26.2

9.9

16.7

40.5

3.98

Exposure (ks) 100 1000 100 1000 100 1000 100 1000 100 1000 100 1000 100 1000

–0.05±0.03 0.03 ±0.04 –0.05±0.03 –0.02±0.05 –0.16±0.05 –0.12±0.06 7.73 ±0.07 7.64 ±0.10 5.90 ±0.04 5.94 ±0.06 18.31±0.03 18.42±0.04 4.08±0.03 4.28±0.04

Average counts/s 0.40±0.03 0.17±0.03 0.24±0.02 0.06±0.01 0.26±0.03 0.10±0.03 0.35±0.05 0.06±0.02 0.38±0.03 0.16±0.04 5.18±0.32 4.05±0.67 0.96±0.07 1.04±0.20

Variance (r2 ) (counts/s)2 0.63±0.02 0.42±0.04 0.49±0.02 0.25±0.02 0.51±0.03 0.32±0.05 0.59±0.04 0.24±0.04 0.62±0.03 0.40±0.05 2.28±0.07 2.01±0.17 0.98±0.04 1.02±0.10

pffiffiffi2 r (counts/s)

0.36±0.03 0.10±0.02 0.40±0.03 0.10±0.02 0.40±0.04 0.09±0.03 0.47±0.07 0.16±0.06 0.42±0.04 0.12±0.03 0.55±0.04 0.14±0.02 0.40±0.03 0.10±0.02

Expected variance (r2EV ) (counts/s)2

0.60±0.02 0.31±0.03 0.64±0.03 0.32±0.03 0.62±0.04 0.29±0.04 0.69±0.05 0.40±0.07 0.65±0.03 0.35±0.04 0.74±0.02 0.37±0.03 0.63±0.02 0.31±0.03

pffiffiffiffiffiffiffiffi2 rEV (counts/s)

3.4

10.1

4.9

– – – – – – 6.6

Flux 4–20 keV 1011 ergs cm2 s1

Page 6 of 7

52.29

22.80

183.48

Decl. (deg)

Blank Sky-8 (2017) Blank Sky-8 (2018) Blank Sky-8 (2019) Mrk 0926

Source

R.A. (deg)

Time bins (s)

Table 2. The average counts, variance and expected variance of the background subtracted lightcurves and the flux in the 4–20 keV band for sources used to verify the scheme.

55 J. Astrophys. Astr. (2021) 42:55

J. Astrophys. Astr.

(2021) 42:55

Page 7 of 7

55

Figure 6. Left panel shows the lightcurves of the source Mrk 0926 (top panel: total source with background count rate, middle panel: the background and bottom panel: the background subtracted count rate). Right panel shows the energy spectrum along with the expected background spectrum (top) and residuals (bottom) after fitting with a power-law model.

longitude to see if there is any predictable trend. These improvements may lead to a background estimation closer to the Poisson limit for a 30 ks exposure. The scheme is applicable to faint sources where the source contribution to the 50–80 keV count rate is less than 0.25 counts/s (15 milli-crabs or 6  1011 ergs/s/ cm2 ) and is limited to the energy range 4–30 keV. The systematic uncertainty in the background spectra will dominate over the statistical error for exposures larger than 5 ks and is of the same order as that from the standard technique. Thus, it will be prudent to confirm spectral results using both techniques. The technique allows for background lightcurve estimation on timescales larger than 100 seconds. The spectrum of a  1 milli-crab source (i.e. 4–20 keV band flux of ’2  1011 ergs/s/cm2 ) can be constrained by LAXPC observations. Since the systematics dominate the Poisson statistics for exposures greater than 5 ks, the sensitivity of the instrument to measure the spectrum of a source does not improve for exposures longer than  30 ks. Variability of a lightcurve binned at 100 seconds can be detected for a level greater than 1 c/s in the 4–20 keV band, which translates to 10% fractional r.m.s. of a 5 milli-crab source. Since the background lightcurve is estimated using the observed variability seen in the high energy band, it is expected to be more reliable than the standard technique. Thus, using this technique, LAXPC can be used to study both the spectral and temporal properties of sources with flux greater than 5 milli-crab.

Acknowledgements This publication uses the data from the AstroSat mission of the Indian Space Research Organization (ISRO), archived at the Indian Space Science Data Centre (ISSDC). We thank members of LAXPC instrument team for their contribution to the development of the LAXPC instrument. This research has made use of software provided by the High Energy Astrophysics Science Archive Research Center (HEASARC), which is a service of the Astrophysics Science Division at NASA/GSFC. The authors thank the referee, Keith M Jahoda, for suggestions and comments which substantially improved the manuscript.

References Antia H. M., Yadav J. S., Agrawal P. C. et al. 2017, ApJS, 231, 10 Agrawal P.C., Yadav J. S., Antia H. M. et al. 2017, J. Astrophys. Astr., 38, 30 Goswami P., Sinha A., Chandra S. et al. 2020, MNRAS, 492(1), 796 Mudambi S. P., Rao A., Gudennavar S. B. 2020, MNRAS, 498(3), 4404 Yadav J. S., Agrawal P. C., Antia H. M. et al. 2016, Proc. of SPIE, 9905, id. 99051D 15 pp. Yadav J. S. et al. 2021, J. Astrophys. Astr., 42. https://doi. org/10.1007/s12036-021-09717-3

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:44 https://doi.org/10.1007/s12036-020-09676-1

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

Results from AstroSat LAXPC observations of Hercules X-1 (Her X-1) D. A. LEAHY*

and Y. CHEN

Department of Physics and Astronomy, University of Calgary, Calgary, Canada. *Corresponding Author. E-mail: [email protected] MS received 20 October 2020; accepted 19 November 2020 Abstract. The Large Area Proportional Counter (LAXPC) instruments onboard the AstroSat Observatory has observed the X-ray binary system Her X-1 during AstroSAT observing sessions AO2, AO3 and TO2. These include observations while Her X-1 is in different stages of its 35 day cycle: Low State, Turn-On to Main High State, peak of Main High State and early decline of Main High State. These observations also include a number of dips and one egress of neutron star eclipse. Here we present light curves and softness ratio analysis for these observations and discuss new features of the spectral changes with 35-day phase and orbital phase. We find a new phenomenon for dips during Main High State: about half of the dips show constant softness ratio as count rate decreases, which has not been seen before, and could be caused by highly ionized matter or very dense cold matter. The other half of the dips show the normal decrease of softness ratio as count rate decreases. These are caused by cold matter absorption and were previously known. Keywords. HZ Hercules binary stars—neutron stars—X-ray photometry.

1. Introduction Her X-1/HZ Her is a persistent X-ray pulsar which continues to give new information about X-ray binary astrophysics (Leahy & Wang 2020) during its 35-day cycle using Swift/BAT and RXTE/ASM long-term monitoring data. (Leahy 2015) found a large-scale electron scattering corona using eclipses of the neutron star. Accurate measurement of the companion radius using eclipses was carried out by Leahy and Abdallah (2014). The pulse period is being tracked with Swift/BAT (Klochkov et al. 2009). Emission lines from the Her X-1 accretion disk (Ji et al. 2009) have been analyzed to learn about the state of the disk atmosphere. This binary, with neutron star (Her X-1) and companion (HZ Her), has masses ’1.5M and ’2.2M , respectively, measured by Leahy & Abdallah (2014) and Reynolds et al. (1997). Her X-1/HZ Her is a

This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

strong emitter at optical, ultraviolet, EUV, and X-rays, which enables studies of the binary in detail. For example, the 1.7-day optical light curve of the companion shows there is a Roche-lobe filling accretion disk which precesses (Gerend & Boynton 1976). The inner disk and from the irradiated surface of the companion star emit in the EUV (Leahy & Marshall 1999; Leahy et al. 2000; Leahy 2003). Mass accretion onto the neutron star generates hard X-rays ([1 keV). The X-ray pulsations are determined by the hot spot geometry and by gravitational lightbending (Leahy 2004a, b). The soft X-ray pulsations were shown to be reprocessing by the inner disk (McCray et al. 1982). The precessing accretion disk causes the 35-day flux cycle in flux and also causes the pulse shape changes with the 35-day phase (Scott et al. 2000). The accretion disk and the cycle were measured and modeled by Shakura et al. (1998), Scott & Leahy (1999), Leahy (2002, 2004c) and Leahy & Igna (2010, 2011). A main result is that the neutron star is directly seen in Main High State, partly obscured by the disk during Short High State, and greatly obscured during Low State.

44

Page 2 of 8

The accretion stream causes the well-known absorption dips (Igna & Leahy 2011, 2012). The light curve and spectral absorption for eclipse ingresses and egresses are determined by the companion star atmosphere (Day et al. 1988; Leahy & Yoshida 1995). During Low State and Short High State, reflected X-rays from the companion star are detected (Abdallah & Leahy 2015). For Low State and eclipses, a low level of X-rays is measured (Choi et al. 1994; Leahy 1995a), which is likely scattered by extended plasma in the binary. Overall, Her X-1 is well measured and show patterns which are explained by the geometry of the accretion disk, the accretion stream from HZ Her, and a large scale corona. India’s first space-borne Astronomy Observatory is AstroSat (Singh et al. 2014). The mission was conceived starting in 1996, built over two decades, then launched in September 2015. AstroSat has four science instruments that are co-aligned and provide simultaneous observations over a wide energy range. The energy bands are: hard X-rays with the Large Area Proportional Counter (LAXPC) and Cadmium–Zinc– Telluride Imager (CZTI) instruments; soft X-rays with the Soft X-ray Telescope (SXT); and optical, near- and far-ultraviolet (NUV and FUV) with the UltraViolet Imaging Telescopes (UVIT). LAXPC is a large area proportional counter, sensitive to the 3–100 keV band, and is described in Yadav et al. (2016) and (Antia et al. 2017). CZTI is a coded mask imager in the 25–150 keV band, and is described in Bhalerao et al. (2017). SXT covers the energy ranges 0.3–8 keV and is described in Singh et al. (2017). UVIT and its calibration are described in Tandon et al. (2017) and Postma et al. (2011). The 35-day cycle has been characterized previously and our general goal is to improve our understanding of the system geometry which causes the 35-day cycle. The notation for the states is summarized in Scott and Leahy (1999): Turn-On, Main High, Low State, Short High and Low State, in time order. The Turn-On is defined as 35-day phase 0, and is marked by the time when the fast rise to Main High reaches 20% of the peak flux of Main High. The Main High and Short High are the two time intervals when the X-ray flux is bright, and the two Low States are when the X-ray flux is faint. The durations of the states in 35-day phase were best measured by Leahy and Igna (2011) using the entire set of RXTE/PCA observations of Her X-1. They are revised from Scott and Leahy (1999) as Main High: 0–0.30, Low State 1: 0.30–0.57, Short High: 0.57–0.75, and Low State 2: 0.75–1.0. The length of the 35-day cycle is somewhat variable

J. Astrophys. Astr. (2021)42:44

and the boundaries between states are also variable. The most comprehensive characterization of the variability of the shape and timing of the 35-day cycle is given by Leahy and Wang (2020). To understand Her X-1 better, and the states during the 35-day cycle, we analyze AstroSat LAXPC observations of Her X-1 in the current work. In Section 2, we describe the observations and the resulting Her X-1 light curves. In Section 3, we carry out a softness ratio analysis and discuss the results. We conclude in Section 4 with a short summary.

2. Observations and data analysis The LAXPC is one of the main instruments on AstroSat and consists of 3 separate proportional counter detectors, each with geometric area of 3600 cm2 (Antia et al. 2017) and time resolution of 10 microseconds. The description of the LAXPC and its calibration is given by Antia et al. (2017). The observations of Her X-1 with LAXPC were carried out in standard event mode as part of three separate proposals for AstroSat observing sessions AO2, AO3 and TO2. The observation dates were March 15–16, 2017 for AO2, June 29–30, 2017 for AO3 and September 20–21, 2017 for TO2. For these observations only units 1 and 2 of the 3 LAXPC detectors produced useful data. Hereafter, we label these LAXPC detectors 1 and 2 as LX10 and LX20, respectively. To determine the part of the 35-day cycle that Her X-1 was in at the time of the AstroSat LAXPC observations, we used publically available long-term monitoring light curves. The data on Her X-1 from the Swift/BAT Hard X-ray Transient Monitor data archive (Krimm et al. 2013) was downloaded and analyzed. The resulting light curves for the  10-day period around the times of the LAXPC observations are shown in Figures 1, 2 and 3. For TO2, there is a data gap in the Swift/BAT light curve where there is no data between MJD58371 and 58378, so we show a  20-day period. Comparison with light curves for longer timescales (35 days or more) shows that the peak of the plotted Swift/BAT light curves is peak of Main High State at 35-day phase 0.10 (see Fig. 2 of Leahy and Igna (2011) for the combined view of the Main High and Short High from all RXTE/PCA observations). Thus we see that the three AstroSat observations occured roughly around Main High State. The Swift/BAT data (Fig. 1) shows that the AO2 observation occurred during Low State prior to Main

J. Astrophys. Astr. (2021)42:44

Figure 1. Swift/BAT light curve of Her X-1 around the time of the TO2 AstroSat observation. The x-axis is in MJD, converted from Swift mission time. The AstroSat Her X-1 observation for AO2 covered the time interval MJD57827.68 to MJD57828.97, thus covering Low State and early Turn-on to Main High State.

Page 3 of 8

44

High and the early part of the Turn-On-to-Main-High period. For AO3, the Swift/BAT data (Fig. 2) shows that the AstroSat data occurs during peak of Main High or just before peak but well after Turn-On. For TO2, the Swift/BAT data (Fig. 3) shows the AstroSat data occurs during Main High, likely after peak, and includes the early part of the Main High decline that starts at ’MJD58382.5 and continues down to Low State at ’MJD58387. The Turn-On (hereafter TO) times of Her X-1 for each Main High State can be measured to better than 1 day using the SWIFT/BAT light curves (when there is data), which results in an error in the 35-day phase of .0.03. We used the TO of Her X-1 measured from the Swift/BAT light curves (or nearest TO for the AstroSat TO2 observation), to set 35-day phase 0.0 for the AstroSat LAXPC observations and calculated 35-day phase using a mean length of the 35-day cycle of 34.9 days (Leahy & Wang 2020). We determined orbital phase using the ephemeris of Staubert et al. (2009). Table 1 summarizes the observations, including MDJ, orbital phase and 35-day phase. The data was downloaded from the AstroSat Data Archive at the AstroSat Science Support Cell (hereafter ASSC) website. The data analysis was carried using the LAXPC software package from the ASSC. Data segments corresponding to the time intervals during earth occultation of source and satellite passage through the South Atlantic Anomaly (SAA) region were removed before the creation of light curves and spectra. We used the LAXPC data from all layers combined.

2.1 Light-curve analysis

Figure 2. Swift/BAT light curve of Her X-1 around the time of the AO3 AstroSat observation. The X-axis is in MJD, converted from Swift mission time. The AstroSat observation for AO3 covered the time interval MJD57933.53 to MJD57934.27, thus covering the peak of Main High State.

Background subtraced light curves were created for each of AO2, AO3 and TO2 in the energy bands 3–5 keV, 5–9 keV, 9–20 keV and total band of 3–80 keV. The 3–80 keV light curves extracted from AO2 LAXPC unit 1 and LAXPC unit 2 are shown in Fig. 4. Here we plot vs. orbital phase because of the importance of timing of eclipses which block direct X-rays from the neutron star. For AO2, the observation starts after end of eclipse at orbital phase 0.065 and ends prior to start of the following eclipse at orbital phase 0.935. The temporary drop in count rate seen at orbital phase ’0.83 is identified as an absorption dip. Points

44

Page 4 of 8

J. Astrophys. Astr. (2021)42:44

beginning (orbital phase 0.05 to 0.08). The remainder of the observation is Main High State. Points during eclipse and egress are marked in green and points during Main High State are marked in blue. In order to test whether the first 2 hours after eclipse are different than the rest of Main High, we analyzed the X-ray colors later for that time period separately, and mark those points in light blue in Fig. 6.

3. Results and discussion

Figure 3. Swift/BAT light curve of Her X-1 around the time of the TO2 AstroSat observation. The X-axis is in MJD, converted from Swift mission time. The AstroSat Her X-1 observation for TO2 covered the time interval MJD58381.83 to MJD58382.82, thus covering the peak and start of decline of Main High State.

corresponding to Low State are marked in pink in this plot and points during the Turn-On phase are marked in red. The 3–80 keV light curves extracted from AO3 for LAXPC units 1 and 2 are shown in Fig. 5, plotted vs. orbital phase. For AO3, the observation includes no eclipses but exhibits strong and numerous dips between orbital phase 0.65 and the end of the observation at orbital phase 0.80. Points during Main High State are marked in blue and points during the dips are marked in orange. For TO2, the 3–80 keV light curves were for LAXPC units 1 and 2, and are very similar. We show the LAXPC unit 2 light curve in Fig. 6, plotted vs. orbital phase. The TO2 observation includes an eclipse and eclipse egress at the

Her X-1 has a spectrum (reviewed in Leahy & Chen 2019), which during Main High State, is characterized by a broken power law with flat low energy index (a ’ 0:9) below a break energy of  10 keV and steeper index (a ’ 1:5) above that energy. The power law is modified by an exponential cut-off with cut-off energy  22 keV and e-folding energy  12 keV. Other components include: a cyclotron absorption line at  38 keV, iron emission lines at 6.4 and 6.6 keV, an iron L emission complex at 0.9 keV with width  0.2 keV, and a blackbody with kT ¼ 0:09 keV. On top of this spectrum is a partial covering absorber, with variable covering fraction and with absorption column density of  5  1023 cm2 . Leahy and Chen (2019) found from AstroSat SXT observations that the presence of an additional partially ionized absorber was strongly indicated by the 2–8 keV SXT spectrum during Main High State. Spectral changes in Her X-1 over the 35-day cycle or over the orbital period are primarily driven by the system geometry. As reviewed in Igna and Leahy (2011) (see also, Leahy 1995b, 2019), the changes in spectrum can be summarized by changes in softness ratio.The changes in system geometry include eclipse of neutron star by Her X-1 (Leahy & Abdallah 2014), absorption dips from the accretion stream (Igna & Leahy 2011), and absorption re-emission and

Table 1. LAXPC Observations of Her X-1. Observation AO2 AO3 TO2 a

MJD (start)

Exposure

/aorb;1

/aorb;2

/a35d;1

/a35d;2

57827.68 57933.53 58381.83

51523 s 26786 s 36380 s

0.09 0.36 0.04

0.87 0.80 0.63

0.974 0.052 0.168

0.012 0.073 0.197

1 and 2 indicate observation start and end for orbital phase, /orb , and 35-day phase, /35d .

J. Astrophys. Astr. (2021)42:44

Page 5 of 8

44

3.1 Softness ratio analysis

Figure 4. AstroSAT LAXPC 3–80 keV light curve of Her X-1 for the AO2 AstroSat observation from LAXPC unit 1 (top: labelled LX10) and from LAXPC unit 2 (bottom: labelled LX20). The points are color-coded by 35-day state: pink for Low State and red for Turn-On. The top panel in Fig. 1 of Leahy and Chen (2019) showed the SXT light curve.

Figure 5. AstroSAT LAXPC 3–80 keV light curve of Her X-1 for the AO3 AstroSat observation from LAXPC unit 2. The points are color-coded by 35-day state: blue for Main High State (MH) and orange and grey for MH with dips (see text for description). The light curve for LAXPC1 looks similar, as in Fig. 4. The bottom panel in Fig. 1 of Leahy and Chen (2019) showed the SXT light curve.

scattering by the accretion disk (Leahy 2002). Here we study the time variability of Her X-1 in the LAXPC observations using softness ratio, with detailed spectral analysis to be carried out in future work.

Previous analyses of the 35-day cycle and the properties of the different states (Turn-On, Main High, Low State, Short High, Eclipse, Eclipse Ingress and Egress, and Dips) were carried out by Leahy and Igna (2011) and Leahy (1995b). The former used RXTE/ PCA observations and the latter used GINGA observations of Her X-1. X-ray color (softness ratio) vs. intensity plots were made for the different states, showing characteristic features of the different states. For example, Fig. 4 in Leahy (1995b) showed eclipse and eclipse egress in Main High State, and dipped during Main High State. Figure 6 in Leahy (1995b) showed a summary plot with all of the points for different states labelled by state, illustrating the utility of a color-intensity diagram as a diagnostic for the Her X-1 states. Here we have created color-intensity diagrams for the three observations (AO2, AO3 and TO2) using 5 to 9 keV count rate for intensity and (5–9 keV)/(9–20 keV) softness ratio for color. The results are colorcoded by state as noted above in the light curve plots of Figures 4, 5 and 6. Figure 7 shows these colorintensity plots. The Low State (pink), the deepest part of dips (orange points at count rate at \5 counts/s) and eclipse (green points at count rate .10 counts/s) all have the same softness ratio (1.0 within errors). With the current background subtraction for LAXPC, the Low State is significantly brighter than deep dips or eclipse. Next we compare the Turn-On phase (red), dip ingress and egress (orange and grey points at count rate J10 counts/s) and eclipse egress (green points at count rate J10 counts/s). The softness ratio increases from 0.25 to 0.8 for Turn-On (red points in Fig. 7), with final value of 0.8 the same as for the Main High State (blue points in the lower two panels of Fig. 7). Dip ingress and egress (the time period between onset of dip and when the count rate reaches its lowest value during dip) shows two different branches in softness ratio (middle panel of Fig. 7, orange and grey points). The lower branch changes in softness ratio from 0.8 (normal Main High softness ratio) down to 0.25 when the count rate reaches its minimum (  5 counts/s). The upper branch is, in contrast, nearly constant in softness ratio at 0.8 (normal Main High softness ratio) while the count rate decreases from its maximum (  170 counts/s) to its minimum (  5 counts/s). Here, we chose the line of softness ratio (SR) given by SR ¼ Rð5  9Þ=800 þ 0:5 with

44

Page 6 of 8

J. Astrophys. Astr. (2021)42:44

3.2 Implications

Figure 6. AstroSat LAXPC 3–80 keV light curve of Her X-1 for the TO2 AstroSat observation for LAXPC unit 2. The points are color-coded by 35-day state: blue for Main High, green for eclipse egress and light blue for the 2 hours immediately after Main High eclipse egress. The light curve for LAXPC1 looks similar, as in Fig. 4.

Rð5  9Þ, the 5–9 keV count rate, to divide the upper branch set from the lower branch set of dip points. This dividing line is shown in the middle panel of Fig. 7 by the red line. We note that at high 5–9 keV count rate (above  100 counts/s), the line is not able to separate the two branches because they overlap too closely. At low count rates, the softness ratio errors are too large to separate the two branches. Referring back to the light curve for MH and MH with dips (Fig. 5), it is seen that there are 3 clear dips during the observation. The two dips with constant softness ratio (grey points) are the sharp dips at orbital phase ’0.66 and ’0.68. The dip with softness ratio decreasing with decreasing count rate is the wider and slower dip which occurs from orbital phase ’0.78 to ’0.80. Eclipse egress (green points in lower panel) starts at softness ratio  0.25 and rises smoothly with count rate to the normal 0.8 value for Main High State. Figure 6 shows that the final part of egress from 5–9 keV count rate of 70 to 125 counts/s is missing in the AstroSat data. The two blue points at count rate  80 and 105 are not part of egress, rather small short dips, as seen from Fig. 6. The two hours of data just after eclipse egress (see Fig. 6) are separately plotted in Fig. 7 as the light blue points. It is seen that they cannot be distinguished from the rest of Main High State. This indicates that the effects of eclipse are finished within a short time (.50 min) after egress.

The light curves obtained during AO2, AO3 and TO2 exhibit the various known states of Her X-1 for the parts of the 35-day cycle that were observed. The Low State, Turn-On, Main High State, Main High dips and one eclipse and its egress were well measured. A color-intensity diagram analysis, similar to that carried out for GINGA data by Leahy (1995b) was carried out. The AstroSat LAXPC color-intensity diagram has features that agree with those from the color-intensity diagram for GINGA. However the new observations and analysis with AstroSat have smaller errors than those from the GINGA analysis. One new feature that we have found comes from the extensive dipping period during peak of Main High State during the AstroSat AO3 observation. For the first time, we detect two types of dips which can be separated by the count-rate dependence of their softness ratios. As seen in the middle panel of Fig. 7, we separated the dip data into two sets: dips which show a decrease softness ratio as the count rate decreases (orange points); and dips which show a constant softness ratio as the count rate decreases (grey points). The first type of dips behave the same way as eclipse egress or as Turn-On. Previous spectral analyses of eclipse egress (Day et al. 1988; Leahy & Yoshida 1995) have shown that this is caused by cold matter absorption changing both intensity and softness ratio. Similarly analysis of Turn-On has shown that the changes are caused by cold matter absorption (Kuster et al. 2005). The newly discovered dips have nearly constant softness ratio as the count rate decreases. These dips could be caused by highly ionized matter blocking the X-rays to the neutron star, so that the spectrum does not change as count rate decreases. Alternately, they could be caused by partial coverage by very dense matter (column density Z1024 g cm2 ), which does not change the spectrum as count rate decreases, but requires a sharp physical edge to the matter.

4. Conclusion For the current work, we have analyzed observations of Her X-1 from the Large Area Proportional Counter (LAXPC) instruments onboard the AstroSat Observatory. The observing sessions include observations while Her X-1 is in different stages of its 35 day cycle: Low State, Turn-on to Main High State, peak of Main

J. Astrophys. Astr. (2021)42:44

Page 7 of 8

44

Figure 7. AstroSat LAXPC softness ratio vs. count rate for the AO2 (top), AO3 (middle) and TO2 (bottom) AstroSat observations, shown for LAXPC unit 2. The points are color-coded by 35-day state: pink for Low State and red for TurnOn (top panel), blue for Main High, orange and grey for dips (middle panel, see text for description), green for eclipse egress and light blue for the 2 hours immediately after Main High eclipse egress (bottom panel). The red line in the middle panel is used to divide the dips data into two parts: those with high softness ratio (grey) and those with low softness ratio (orange).

High State and early decline of Main High State. The observations also include a number of dips and one egress of neutron star eclipse. We have presented light curves and a softness ratio analysis. Many results confirm previous findings for Her X-1 on the behaviour of the different states. A fraction  1/2 of the dips show decreasing softness ratio as count rate decreases. These are the previously known type of dips and are caused by cold matter absorption. However, a new phenomenon is found for dips during Main High State: the other fraction  1/2

of dips are new: they show constant softness ratio as count rate decreases. These dips could be caused by highly ionized matter or by very dense cold matter with a sharp edge. Future work will focus on verifying and understanding the new type of dips by carrying out a spectral analysis with both LAXPC and SXT instruments. New observations are planned for Short High State to characterize it better, including a search for dips to test if Short High dips behave in the same way as Main High dips.

44

Page 8 of 8

Acknowledgements This project is undertaken with the financial support of the Canadian Space Agency and of the Natural Sciences and Engineering Research Council of Canada. This publication uses data from the AstroSat mission of the Indian Space Research Institute (ISRO), archived at the Indian Space Science Data Center (ISSDC). LAXPC data were processed by the Payload Operation Centre at TIFR, Mumbai.

References Abdallah M. H., Leahy D. A. 2015, MNRAS, 453, 4222 Antia H. M., Yadav J. S., Agrawal P. C. et al. 2017, The Astrophys. J. Suppl., 231, 10 Bhalerao V., Bhattacharya D., Vibhute A. et al. 2017, J. Astrophys. Astr., 38, 31 Choi C., Dotani T., Nagase F., Makino F., Deeter J., Min K. 1994, The Astrophys. J., 427, 400 Day C. S. R., Tennant A. F., Fabian A. C. 1988, MNRAS, 231, 69 Gerend D., Boynton P. 1976, The Astrophys. J., 209, 652 Igna C. D., Leahy D. A. 2012, MNRAS, 425, 8 Igna C. D., Leahy D. A. 2011, MNRAS, 418, 2283 Ji L., Schulz N., Nowak M., Marshall H. L., Kallman T. 2009, The Astrophys. J., 700, 977 Klochkov D., Staubert R., Postnov K., Shakura N., Santangelo A. 2009, A&A, 506, 1261 Krimm H. A., Holland S. T., Corbet R. H. D. et al. 2013, The Astrophys. J. Suppl., 209, 14 Kuster M., Wilms J., Staubert R. et al. 2005, A&A, 443, 753 Leahy D., Wang Y. 2020, arXiv:2009.07246 Leahy D. 2019, Proceedings of the IAU Symposium 346, pp. 235–238

J. Astrophys. Astr. (2021)42:44 Leahy D. A., Chen Y. 2019, The Astrophys. J., 871, 152 Leahy D. A. 2015, The Astrophys. J., 800, 32 Leahy D. A., Abdallah M. H. 2014, The Astrophys. J., 793, 79 Leahy D. A., Igna C. 2011, The Astrophys. J., 736, 74 Leahy D. A., Igna C. 2010, The Astrophys. J., 713, 318 Leahy D. A. 2004a, The Astrophys. J., 613, 517 Leahy D. A. 2004b, MNRAS, 348, 932 Leahy D. A. 2004c, Astron. Nachr. 325, 205 Leahy D. A. 2003, MNRAS, 342, 446 Leahy D. A. 2002, MNRAS, 334, 847 Leahy D. A., Marshall H., Scott D. M. 2000, The Astrophys. J., 542, 446 Leahy D. A., Marshall H. 1999, The Astrophys. J., 521, 328 Leahy D. A., Yoshida A. 1995, MNRAS, 276, 607 Leahy D. A. 1995a, The Astrophys. J., 450, 339 Leahy D. A. 1995b, A&A Suppl., 113, 21 McCray R., Shull M., Boynton P., Deeter J., Holt S., White N. 1982, The Astrophys. J., 262, 301 Postma J., Hutchings J. B., Leahy D. 2011, PASP, 123, 833 Reynolds A., Quaintrell H., Still M., Roche P., Chakrabarty D., Levine S. 1997, MNRAS, 288, 43 Scott D. M., Leahy D. A., Wilson R. B. 2000, The Astrophys. J., 539, 392 Scott D. M., Leahy D. 1999. The Astrophys. J., 510, 974 Shakura N., Postnov K., Prokhorov M. 1998, A&A, 331, L37 Singh K. P., Stewart G. C., Westergaard N. J. et al. 2017, J. Astrophys. Astr., 38, 29 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, SPIE, 9144E, 1S Staubert R., Klochkov D., Wilms J. 2009, A&A, 500, 883 Tandon S. N., Subramaniam A., Girish V. et al. 2017, The Astronom. J., 154, 128 Yadav J. S., Agrawal P. C., Antia H. M. et al. 2016, Proceedings of the SPIE, 9905, 99051D

 Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:47 https://doi.org/10.1007/s12036-020-09683-2

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

UOCS. V. UV study of the old open cluster NGC 188 using AstroSat SHARMILA RANI1,2,*, ANNAPURNI SUBRAMANIAM1, SINDHU PANDEY3,

SNEHALATA SAHU1, CHAYAN MONDAL1 and GAJENDRA PANDEY1 1

Indian Institute of Astrophysics, Koramangala II Block, Bengaluru 560 034, India. Pondicherry University, R.V. Nagar, Kalapet 605 014, India. 3 Aryabhatta Research Institute of Observational Sciences (ARIES), Manora Peak, Nainital 263 001, India. *Corresponding author. E-mail: [email protected] 2

MS received 7 November 2020; accepted 30 November 2020 Abstract. We present the UV photometry of the old open cluster NGC188 obtained using images acquired with Ultraviolet Imaging Telescope (UVIT) on board the AstroSat satellite, in two far-UV (FUV) and one near-UV (NUV) filters. UVIT data is utilised in combination with optical photometric data to construct the optical and UV colour-magnitude diagrams (CMDs). In the FUV images, we detect only hot and bright blue straggler stars (BSSs), one hot subdwarf, and one white dwarf (WD) candidate. In the NUV images, we detect members up to a faintness limit of  22 mag including 21 BSSs, 2 yellow straggler stars (YSSs), and one WD candidate. This study presents the first NUV-optical CMDs, and are overlaid with updated BaSTIIAC isochrones and WD cooling sequence, which are found to fit well to the observed CMDs. We use spectral energy distribution (SED) fitting to estimate the effective temperatures, radii, and luminosities of the UV-bright stars. We find the cluster to have an HB population with three stars (Teff ¼ 4750–21000 K). We also detect two yellow straggler stars, with one of them with UV excess connected to its binarity and X-ray emission. Keywords. Galaxy: open clusters: individual: NGC 188—stars: horizontal-branch—stars: blue stragglers—stars: Hertzsprung–Russell and colour-magnitude diagrams.

1. Introduction Old open clusters (OC) provide the ideal environments for studying the formation and evolution of single and binary stellar populations. OCs are extensively used to understand the evolution of the Galactic disk, and chemical and dynamical evolution of the Galaxy. Most of the OCs are studied in the optical band, but only a handful of them are studied in the Ultraviolet (UV) band; this is due to the lack of spacebased UV telescopes, which has improved during recent times. Looking into the properties of OCs in the UV region is crucial in understanding the cluster’s hotter stellar populations, which emit a significant fraction of flux in the UV domain; hence, an essential This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

tool in discovering hot stars. The UV light of stellar populations is contributed by exotic populations such as Blue Straggler Stars (BSSs), Horizontal Branch (HB) stars, hot White Dwarfs (WDs), subdwarfs, cataclysmic variables, etc. All the stars mentioned above are detectable in optical and infrared bands, but as these stars are hot, the bolometric corrections make them optically faint and indistinguishable from cooler stars. In the old OCs M67, NGC 188, and NGC 6791, hot stars have been detected and studied by Landsman et al. (1998) using Ultraviolet Imaging Telescope (UIT) data. De Martino et al. (2008) have obtained FUV and NUV images of the old OC NGC 2420 from Galaxy Evolution Explorer (GALEX) and cross-matched with Sloan Digital Sky Survey (SDSS) u, g, r, i, z photometric data in search of WDs in the cluster. The wide field photometry for nearby clusters M67, NGC 188, NGC 2539, and M79 using Swift Ultraviolet

47

Page 2 of 11

Optical Telescope (UVOT) by Siegel Michael et al. (2014) shows that the UV colour-magnitude diagrams (CMDs) can easily identify the unusual UV bright stars. Browne et al. (2009) used GALEX observations to check UV variability in OCs Hyades and the Pleiades and detected 16 UV variable sources. Gosnell et al. (2015) utilised Hubble Space Telescope (HST) data in far-ultraviolet (FUV) passbands to detect WD companions to BSSs in old OC NGC 188. Thus, UV imaging of OCs provides a unique window for identifying and characterising hot stellar populations to further shed light on their formation and evolution. The importance of studying OCs in the UV, as outlined above, has lead to the formation of the Ultra Violet Imaging Telescope (UVIT) OC study (UOCS), as described in Jadhav et al. (2019). The program was initiated to understand the properties of single and binary stars in OCs. BSSs are defined as the stars that are observed to be brighter and bluer than the corresponding main-sequence (MS) turnoff in an optical CMD in star clusters (Sandage 1953). The origin of these stars is unexplained by the standard theory of single-star evolution, but they are found to be confirmed members of the clusters. The primary scenarios proposed to explain the formation of BSSs in clusters are mass transfer in close binary systems (McCrea 1964) and stellar mergers resulting from direct stellar collisions (Hills & Day 1976). Yellow Straggler Stars (YSSs) are brighter than stars in the sub-giant branch (SGB), and bluer than the red giant branch (RGB). Observationally, they are found to lie between blue straggler and RGB regions on the CMD (see Sindhu et al. 2018 and references therein). These stars are also described as evolved BSSs. The hot subdwarfs of B and O-type (sdB, sdO) represent the late stages of the evolution of low mass stars. sdB stars are considered as the metalrich counterparts of the extreme HB (EHB) stars in globular clusters (GCs). Till now, a few hot subdwarfs and HB stars are detected in two metal-rich and old OCs, i.e., NGC 6791 and NGC 188 (Kaluzny & Udalski 1992; Liebert et al. 1994; Green et al. 1997). Only one sdB star is found in NGC 188, whereas NGC 6791 hosts five sdB stars. Schindler et al. (2015) investigated 15 OCs to understand formation of EHB stars by taking NGC 188 and NGC 6791 as template clusters. They identified only red giant clump stars but no EHB stars were detected. In this paper, we study one of the oldest and well studied OC, NGC 188, known in the Galaxy. The cluster is studied widely due to its richness, metallicity, age and its location in the galactic plane

J. Astrophys. Astr. (2021)42:47

d2000 ¼ þ85 140 4900 ; l ¼ (a2000 ¼ 0h 47m 12s :5,   122 :85; b ¼ þ22 :38). The age of this cluster is determined to be 7 Gyr (Sarajedini et al. 1999), and the reddening of the cluster is 0:036  0:01 mag (Jiaxin et al. 2015). This cluster is located at a distance of 1866 pc (Gao 2018) and the metallicity is found to be solar (Sarajedini et al. 1999). In this study, we present the results of the UV imaging of the NGC 188 in two FUV and one NUV filters using an ultraviolet imaging telescope (UVIT) on AstroSat. We characterise the UV bright stars identified in this cluster by analysing SEDs to throw light on their formation and evolution. The paper is organised as follows: Section 2 describes the observations and data analysis. In Section 3, Optical, and UV CMDs, along with the results obtained from SED analysis, are described. All the results are discussed in Section 4 The results are summarised in Section 5.

2. Observations and data analysis The data used in this study were acquired with the UVIT instrument on-board the AstroSat satellite. UVIT is one of the payloads in AstroSat, the first Indian multi-wavelength space observatory. The UVIT instrument has two 38-cm telescopes: one observes only in FUV region (k ¼ 130–180 nm), and the other in NUV (k ¼ 200–300 nm) and VIS (k ¼ 320–550 nm) region. UVIT can simultaneously generate images in the FUV, NUV and VIS channels in a circular field of view of  280 diameter. Every channel consists of multiple filters covering a different wavelength range. The data obtained in VIS passband is used for drift correcting the images. The detectors used for the UV and VIS passbands operate in photon counting mode and integration mode, respectively. Further details about UVIT instrument can be found in Tandon et al. (2017a, 2020), along with calibration results. The magnitude system adopted for UVIT filters is similar to the used for GALEX filters, and hence the estimated magnitudes will be in the AB magnitude system. NGC 188 was observed as UVIT’s ‘‘first light’’ object on 30 November 2015. The cluster was observed every month to track the variation in UVIT senstivity over the first six months of its operation. The observations of NGC 188 were done from December 2015 to March 2016. All data used in the analysis were taken in three filters: one NUV (N279N) and two FUV (F148W and F172M). Images were

J. Astrophys. Astr. (2021)42:47

Page 3 of 11

47

corrected for distortion, flat field illumination, and spacecraft drift utilising the customised software package CCDLAB (Postma & Leahy 2017). The images created for each orbit were then aligned and combined to produce a science ready image. The UVIT observational details of the NGC 188 UV images are tabulated in Table 1. The UVIT image of the cluster created using one FUV F148W and one NUV N279N filter is shown in Fig. 1 with blue and yellow corresponding to FUV and NUV detections, respectively.

2.1 Photometry Aperture photometry was carried out on the FUV F172M image to get the estimate of counts. We performed psf photometry on one FUV F148W and NUV N279N images using the IRAF/NOAO package DAOPHOT (Stetson 1987). A curve of growth technique was used to determine the aperture correction value and then applied to all magnitudes estimated using aperture and psf photometry. To obtain final magnitudes in each filter, saturation correction was applied according to the method described in Tandon et al. (2017b). The instrumental magnitudes are calibrated to the AB magnitude system using zero-point (ZP) magnitudes reported in Tandon et al. (2017b). The PSF fit errors in estimated magnitudes are shown in Fig. 2 as a function of magnitude for FUV and NUV filters. The stars as faint as 22 mag are detected both in FUV and NUV images with typical error 0.2 mag and 0.3 mag, respectively. The observed UVIT magnitudes are corrected for extinction. In order to compute the extinction value in visible band (AV ), We have adopted reddening EðB  VÞ ¼ 0:036 mag (Jiaxin et al. 2015) and the ratio of total-to-selective extinction as Rv ¼ 3:1 from Whitford (1958) for the Milky Way. The Fitzpatrick reddening relation (Fitzpatrick 1999) is used to determine extinction co-efficients Ak for all bandpasses.

Table 1. Details of observations of NGC188. Filter

kmean ˚) (A

Dk ˚) (A

ZP (AB mag)

Exposure time (s)

F148W F172M N279N

1481 1717 2792

500 125 90

18.016 16.342 16.50

3216 4708 9365

Figure 1. UVIT image of NGC 188 obtained by combining images in NUV(N279N) and FUW(F148W) channels. Yellow and blue colour corresponds to NUV and FUV detections, respectively.

3. Results 3.1 UV and optical colour-magnitude diagrams To identify and classify the stars detected with UVIT into various evolutionary phases. We have used ground-based optical photometric data (Sarajedini et al. 1999) of NGC 188 to cross-identify with the UVIT detected stellar sources. First, we have selected stars with proper motion (PM) membership probability more than 50% as most likely members of the cluster from the catalog given by Platais et al. (2003), which gives us 562 member stars. Then we have cross-matched optical photometric data taken from Sarajedini et al. (1999) with PM catalog. Geller et al. (2008) presented the results of radial-velocity (RV) survey of NGC 188 using WIYN data. They measured the radial velocities for 1046 stars in the direction of NGC 188, and further calculated the RV membership probability for all the stars. Out of 1046 stars, 473 stars are found to be likely cluster members. Further, we cross-matched optical data with RV catalog to check for the RV membership probability. After choosing all the PM and RV members in the cluster, we further cross-matched with UVIT detected stars in all the three filters. We have also included the stars for which PM membership probability is given, but RV membership is not known. We have cross-identified 356 stars of NGC 188, as a cluster members, in NUV passband and 10 cluster members in both FUV passbands. The accuracy of this cross-match is within

47

Page 4 of 11

J. Astrophys. Astr. (2021)42:47

Figure 2. PSF fit errors as a function of magnitude for the UVIT observations of NGC 188. Top panel shows the errors for FUV F148W filter, whereas the bottom panel shows for the NUV N279N filter.

1:500 . The 24 BSS candidates have been cataloged by Ahumada and Lapasset (2007). Geller et al. (2008), identified 20 BSSs to be confirmed members based on the radial velocity membership. As there are many stars detected in the NUV, the detected members are segregated based on their detection in the NUV. In the NUV filter, we detect the MS, turn-off, SGB, RGB, BS and YS stars, whereas in FUV images, only hot and bright stars are detected. We have detected 21 BSSs previously known in literature, out of which sixteen are RV members, and five are PM members. One previously known hot subdwarf is also detected in both FUV and NUV images, but RV membership for this star is not known as it is fainter than 16 mag in optical CMD. Two YSSs are selected on the basis of their position in the optical CMD. We have also detected one very bright and hot star in NUV and one FUV image, probably a WD candidate. Both PM and RV memberships are unknown for this star. For only cross-matched member stars, we have created the optical and UV-optical CMDs as shown in Figures 3 and 4. The parameters adopted in this study are listed in Table 2. The optical and UV-optical CMDs are over-plotted with updated 1

http://basti-iac.oa-abruzzo.inaf.it/.

BaSTI-IAC isochrones (Hidalgo et al. 2018). The updated BaSTI-IAC1 isochrones are generated for an age 7 Gyr, a distance modulus of 11.44 mag (Sarajedini et al. 1999) and, a solar metallicity with helium abundance Y ¼ 0:247, ½a=H ¼ 0, including overshooting, diffusion, and mass loss efficiency parameter g ¼ 0:3. The BaSTI-IAC model also provides HB model, which includes zero age HB (ZAHB), postZAHB tracks and end of the He phase known as terminal age HB (TAHB) with or without diffusion for a selective mass range. We generated the ZAHB and TAHB tracks for a solar metallicity including diffusion. The overlaid BaSTI-IAC isochrones are fitting well to the observations in both optical and NUV-optical CMDs as shown in Fig. 3. This is probably the first UV CMD for this well studied cluster in the NUV, the overlaid isochrones match the observed sequence more or less satisfactorily. The CMDs shown in Fig. 3, suggests that this cluster probably has a HB population, with stars in the HB. Two stars marked with red colour are lying along HB tracks, implying that these stars may belong to the HB evolutionary phase. In fact, they are lying close to TAHB track, indicating that they are about to evolve off the HB phase. The previously known subdwarf is found close

J. Astrophys. Astr. (2021)42:47

Page 5 of 11

47

Figure 3. Left panel: Optical CMD of NGC 188 of all member stars co-detected with UVIT N279N filter and optical photometric data. Right panel: NUV-optical CMD of NGC 188 of member stars cross-identified using UVIT N279N data with optical photometric data. The meaning of all the symbols are marked in the figures. Previously known binary and single BSSs are shown as open blue circles and open blue diamonds, respectively. The BSSs with WD detections are shown with a magenta plus symbol. The BSS outlined with magenta square symbol is bright in both NUV and FUV CMDs. The over-plotted black colour dots represent updated BaSTI-IAC model isochrones generated for an age 7 Gyr and a solar metallicity. The solid and dashed lines shown along the HB track correspond to zero-age HB (ZAHB) and terminal-age HB (TAHB), respectively. The dashed-dotted black line indicates the WD cooling sequence for a WD with mass 0:5M .

Figure 4. FUV-optical CMDs of NGC 188 of member stars cross-identified using UVIT FUV and ground-based optical photometric data. Other details are same as in Fig. 3.

to the blue end of the ZAHB track in the optical CMD, and fainter than the blue end in UV CMDs. We have also shown a WD cooling curve for a 0:5M WD with hydrogen envelope with dashed-dotted black line in Figures 3 and 4 provided by Tremblay et al. (2011) to define the location of WDs in all CMDs (P. Bergeron, private communication). Note that star marked with green colour in all CMDs is lying along the WD cooling track, indicating that the star is a possible WD candidate, although its membership is uncertain. We have also detected two stars which are bright in NUV CMDs with respect to other giants shown with yellow colour in the optical as well as in NUV-optical CMD (see Fig. 3). These two stars are bluer than the RGB and brighter than SGB track in the optical CMD, but

they are bright in the NUV CMD. These two stars are classified as YSSs.

3.2 Spectral energy distributions of UV bright stars In this section, our main goal is to check the evolutionary status of stars, which appear bright in NUV and FUV CMDs, by determining their stellar parameters using spectral energy distribution (SED) fit technique. We constructed the SEDs to estimate the stellar parameters such as effective temperature (Teff ), luminosity (L), and radius (R) of the stars which appear bright and hot in the UV CMDs. To create the SEDs, We have used observed photometric data points

47

Page 6 of 11

J. Astrophys. Astr. (2021)42:47

Table 2. Parameters of NGC 188 used in this study. Parameters Metallicity (Z) Age Distance modulus, ðm  MÞV Reddening, EðB  VÞ

Value

Reference

0.02 7  0:5 Gyr 11:44  0:08 mag 0:036  0:01 mag

Jiaxin et al. (2015) Sarajedini et al. (1999) Sarajedini et al. (1999) Jiaxin et al. (2015)

spanning a wavelength range from FUV-to-Infrared and then fitted with selected theoretical models. The virtual observatory tool, VOSA (VO Sed Analyser, Bayo et al. 2008) is used for SED analysis. The filter transmission curves are employed to calculate the synthetic photometry of the selected theoretical models. VOSA utilises the fixed distance to the cluster to scale the synthetic fluxes with the observed fluxes. After constructing the synthetic SED, a v2 minimisation test is performed to compare the observed with the synthetic photometry to find the best fit parameters of the SED. The reduced v2red value for the best-fit is evaluated using the expression given below: v2red ¼

N 1 X ðFo;i  Md Fm;i Þ2 ; N  Nf i¼1 r2o;i

where N is the number of photometric data points, Nf is the number of free parameters in the model, Fo;i is the observed flux, Md Fm;i is the model flux of the star, Md ¼ ðDR Þ2 is the scaling factor corresponding to the star (where R is the radius of the star and D is the distance to the star) and ro;i is the error in the observed flux. The number of observed photometric data points N for stars varies from 6 to 25 depending upon their detection in different available filters. The number of free parameters (Nf ) used to fit SED depend on the selected theoretical model. In general, the free parameters are [Fe/H], log(g) and Teff . The radius of the stars were calculated using the scaling factor, Md . Kurucz stellar atmospheric models are used to construct SEDs for two HB, one subdwarf, and two YS stars which are bright in NUV band when compared to RGB stars (Castelli et al. 1997; Castelli & Kurucz 2003). The model provides the temperatures ranging from 5000–50000 K, log g from 0–5 and metallicity from -2.5 to 0.5 dex. Since this cluster has solar metallicity, we fixed the metallicity ½Fe=H ¼ 0 close to the cluster metallicity. We have given a Teff and log g range from 5000–50000 K and 2–5 dex, respectively for the adopted Kurucz models to fit the SEDs of above mentioned stars. We have combined

three UVIT photometric data points with two GALEX, five ground photometry, three GAIA, five PANSTARRS, three 2MASS, and four WISE photometric data points to generate SEDs for UV bright stars. The number of observed photometric data points used for fitting will be equal to or less than the above mentioned data points as some stars are not detected in all the filters. The data points, which were not fitting well to the theoretical model, are also excluded from the fit. We have also constructed SED for a possible WD detected with UVIT in both FUV and NUV images using a Koester WD model (Koester 2010; Trembley & Bergeron 2009). The free parameters of the Koester model are Teff and log g. The value for the Teff for the Koester model ranges from 5000 to 80000 K and log g from 6.5–9.5 dex. We have used two UVIT photometric data points along with two GALEX and two ground photometric data points to fit the SED of the WD. VOSA makes use of Fitzpatrick reddening relation (Fitzpatrick 1999; Indebetouw et al. 2005) to correct for extinction in observed data points. Best SED fits are obtained for six stars, out of which two are HB, two are YSSs, one is subdwarf, and one is WD. The SEDs of all the stars are shown in Figures 5 and 6. We can notice in Figures 5 and 6 that all the data points are fitted well to the models. The estimated values of stellar parameters from the SED fit along with errors are tabulated in Table 3. VOSA estimates uncertainties in the effective temperatures as half the grid step, around the best-fit value. The high temperature of subdwarf (21000 K) suggests that it belong to the class of subdwarf B (sdB) stars (Heber 1986). The effective temperature of the WD turns out to be 20000 K with a radius 0.02R , which confirms the star to a possible WD candidate. In order to check the evolutionary status of HB stars identified with UVIT, we have plotted the theoretical evolutionary tracks using the updated BaSTI-IAC models presented by Hidalgo et al. (2018). We have selected the model with metallicity close to the cluster metallicity. The evolutionary tracks starting from MS

J. Astrophys. Astr. (2021)42:47

Page 7 of 11

47

Figure 5. Spectral energy distribution (SED) of a hot subdwarf and WD candidate detected with UVIT after correcting for extinction. The best-fit parameters are shown in the figure.

Figure 6. Spectral energy distribution (SED) of two HB stars (upper panels) and two YSSs (lower panels) detected with UVIT after correcting for extinction. The best-fit parameters are displayed in the figure.

turn-off to the moment when a star has entered to the post-HB phase are shown in the H-R diagram in Fig. 7. To show the WD cooling tracks, we used WD models for a 0.5M from Tremblay et al. (2011). The ZAHB and post-ZAHB tracks are shown for a mass range from 0:475M to 0:8M . The TAHB is shown with dash-dotted line in Fig. 7. For WDs, cooling sequence for a 0:5M DA type WD is shown with dash-dotted black line. The parameters estimated from the SED fit for six stars are plotted in H-R diagram.

The stars are marked with same colour as in Fig. 3. We can see in Fig. 7 that HB stars are lying along the HB tracks suggesting that these two stars belong to the HB evolutionary phase. The location of two YSSs on the H-R diagram is near the theoretical RGB sequence. It indicates that these two stars might belong to a giant evolutionary stage. The location of WD on the H-R diagram is near the theoretical WD cooling sequence, implying that the star is likely to be a WD candidate.

47

Page 8 of 11

J. Astrophys. Astr. (2021)42:47

Table 3. SED fit parameters of bright stars detected with UVIT in this cluster. Column 1 lists the star ID used in this work. Column 2 represents the identification number (WOCS ID) according to Platais et al. (2003). Star ID Subdwarf WD BHB RHB YSS1 YSS2

PKM ID WOCS WOCS WOCS WOCS WOCS WOCS

4918 4073 3856 5027 4705 4346

RA (deg)

DEC (deg)

Model

11.96753 10.32108 9.77341 11.98741 11.34416 10.64712

85.31897 85.23638 85.15604 85.24889 85.21065 85.22385

Kurucz Koester Kurucz Kurucz Kurucz Kurucz

Teff (K)

L L

R R

21000  500 6:71  0:22 0:19  103 20000  312 0:039  0:003 0:02  104 7500  125 49:13  7:49 4:13  0:01 4750  125 73:83  3:97 12:56  0:03 5000  125 10:28  1:33 4:26  0:01 5000  125 13:63  1:95 4:94  0:01

v2red

Nfit Ntot

14.23 0.95 32.92 30.85 38.99 5.6

19/23 6/6 22/25 20/20 21/22 20/21

Columns 3 to 5 show the RA, DEC and model used for SED fit, respectively. Rest of the columns give the estimated values of various parameters along with errors. The last column contains the ratio of the number of photometric data points used for SED fitting and the number of total data points available for fitting.

Figure 7. H-R diagram of UV-bright stars in NGC 188 compared to theoretical evolutionary tracks. The evolutionary tracks starting from MS turn-off to the moment when a star has entered to the post-HB phase (Hidalgo et al. 2018) are shown. Along the HB phase, post-ZAHB tracks span a mass range from 0.475–0.8M . In the plot, magenta, blue and green colours correspond to the sequences populating the extreme, blue and red parts of the HB. The black solid and dashed lines indicate the ZAHB and TAHB, respectively. The dash-dotted black line corresponds to cooling track for a 0.5M WD with hydrogen envelope. The SED fit parameters of HB stars, subdwarf, WD and YSSs identified with UVIT are shown with red, cyan, green and yellow square filled symbols, respectively.

4. Discussion In the NUV passband, we have detected 2 HB, 1 sdB, 21 BSSs, 2 YSSs, one WD candidate along with MS, MS turn-off, SGB and RGB stars, whereas in FUV, only hot and bright BSSs, WD and HB stars are

detected. For comparison with theoretical predictions, we overlaid the CMDs with updated BaSTI-IAC and WD model isochrones generated for respective UVIT and Ground based filters. The UV magnitude distribution of all the detected member stars with UVIT is reproduced well with theoretical isochrones. Out of 21

J. Astrophys. Astr. (2021)42:47

BSSs detected with UVIT, 15 BSSs are characterised by Gosnell et al. (2014, 2015) using SED fitting technique. They found hot and young WD companions to 7 BSSs, out of which 5 are detected in our UVIT images. 7 BSSs binaries are likely formed from mass transfer. Other 6 BSSs are not studied in detail in literature till now. We detected one BSS, which is both FUV and NUV bright. This BSS is a PM member of the cluster, but RV membership is not known. Detailed SED analysis is required to check whether the BSS has any hot companion. We plan to characterise the hot companions of the BSSs of this cluster in a separate study and locate them in the H-R diagram (Fig. 7). Out of three UV-bright stars identified with UIT, one star II-91 (numbered by Sandage 1962) is detected in both FUV and NUV images, and here it is designated as a subdwarf. This star was first identified by Sandage (1962), and used for calibration. Later, Dinescu et al. (1996) confirmed its membership based on the proper motion. Green et al. (1997) performed the spectroscopy of UV-bright stars in open clusters NGC 6791 and NGC 188. They reported that II-91 star is an sdB star and also a spectroscopic binary. In fact, it is a close binary of 2.15 days orbital period (Green et al. 2004). Landsman et al. (1998) estimated the effective temperature of II-91 about 30000 K from its UIT magnitude and m152  V colour. Our UV photometry confirms that this star is a UV bright. The Teff measured from SED fitting also confirms it to be an sdB or EHB star. The Teff of one HB star derived from SED analysis is found to be 7500 K, which is in the range of the Teff of a BHB star. For the other HB star, derived temperature is 4750 K, hence likely to be a RHB star in the cluster. The observed UV magnitudes of HB stars match well with theoretical TAHB isochrones, which indicates that these stars are about to evolve off the HB phase, assuming that these are in fact HB stars. Geller et al. (2008) classified BHB star as a rapid rotator (RR) and binary likely non-member (BLN) as they were unable to measure its radial velocity because of its high rotation. But this star is PM member of the cluster with membership probability more than 80% (Platais et al. 2003). According to Dinescu et al. (1996), RHB (D719) is one of the brightest giants in NGC 188. Belloni et al. (1998) identified X-ray sources in OCs M67 and NGC 188 using ROSAT observations. RHB (X29) star is detected as an X-ray source in the cluster. The spectroscopic study of this star (Harris and McClure 1985) found it to be a fast rotator with rotation velocity

Page 9 of 11

47

 24 km s1 and also this star exhibits emission in the Ca H and K and Ha lines. The absence of radialvelocity variations suggest that this star is a single, rapidly rotating giant, an FK Comae-type star. Our SED fit of RHB star also suggests that it is a single star. From V mag and B  V colour, the Teff of this star was estimated to be  4800 K. Our temperature measurement for RHB star is in close agreement with previous estimations. We suggest that the chromospheric emission of this star might make it bright in NUV region. Two YSSs characterised in this work are also RV members, and YSS2 star is previously identified as a giant and single member. Geller et al. (2008) suggested that YSS1 is a double-lined spectroscopic binary (SB2). Mazur and Kaluzny (1990) identified YSS1 as a variable star (V11). They also suggested that YSS1 might be an RS CVn-type binary. Gondoin (2005) studied the X-ray sources in NGC 188 using XMM-Newton observations, and he identified YSS1 (S18) star as an X-ray source. He also estimated the bolometric luminosity and effective temperature of this star about  8L and 5110 K, respectively. Landsman et al. (1997) obtained spectra of the yellow giant S1040 in the OC M67, and found that the star is a single-lined spectroscopic binary. They estimated effective temperature of cool component as 5150 K with a radius 5:1R . Our estimation of luminosity (  10L ), radius (  5R ) and Teff (5000 K) of the YSS1 from SED fitting is close to this value. As YSS1 is a double-lined spectroscopic binary, it might have two components with similar temperature. From SED fitting, we can not separate two components with a similar temperature present in a binary system. The SED fit parameters determined for two YSSs indicate that these two stars may belong to the giant phase. Thus UV emission in the NUV region in the case of YSS1 is likely due to chromospheric activity, probably connected to its binarity and X-ray emission. von Hippel and Sarajedini (1998) studied the WDs in NGC 188 using the WIYN 3.5-meter telescope at Kitt Peak National Observatory. They identified 9 WD candidates, of which 3-6 are expected to be cluster members. Andreuzzi et al. (2002) identified 28 candidate white dwarfs in the cluster using the data in HST WFPC2 F555W and F814W filters, but their membership was not certain. We have also identified one possible WD candidate in one NUV and one FUV band. We checked with the previous available catalog, but this star is not reported in the earlier studies.

47

Page 10 of 11

5. Summary The important results from this study are summarised below: • In this study, we employed UVIT observations on-board AstroSat to identify UV-bright stars in the well-known old open cluster NGC 188. We further created the optical and UV-optical CMDs of member stars co-detected using UVIT and Ground-based data in this cluster. • Stars belonging to the different evolutionary stages such as MS, SGB and RGB are detected in NUV image, but only hot stars are detected in FUV images. • To compare the observations with theoretical predictions, optical and UV-optical CMDs are overlaid with updated BaSTI-IAC and WD models generated for respective UVIT and ground-based filters. The theoretical isochrones reproduce the features of the CMDs very well. • This study presents the first NUV CMD for this well studied cluster. The CMDs suggest the presence of HB in this cluster, with an RHB, a BHB and an EHB/sdB star populating a temperature range of 4750–21000 K. • We suggest two YSSs in this cluster, based on their location in the CMDs. YSS1 is found to have excess flux in the UV, may be connected to its binarity and X-ray emission. • We detect a candidate WD from the UV images and is found to have parameters similar to that of a 0:5M WD.

Acknowledgements We are grateful to the anonymous referee for the comments and suggestions which improved the quality of the manuscript. We warmly thank Pierre Bergeron for providing us the WD cooling models for UVIT filters. S. Rani thanks Vikrant Jadhav for fruitful discussions. This publication uses the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). This publication uses UVIT data processed by the payload operations centre at IIA. UVIT project is a result of collaboration between IIA, Bengaluru, IUCAA, Pune, TIFR, Mumbai, several centres of ISRO, and CSA. This research made of VOSA, developed under the Spanish Virtual Observatory project supported by the Spanish

J. Astrophys. Astr. (2021)42:47

MINECO through grant AyA2017-84089. This research also made use of Topcat (Taylor 2005, 2011), Matplotlib (Hunter 2007), NumPy (Van der Walt et al. 2011), Astropy (Astropy Collaboration et al. 2018) and Pandas (McKinney 2010).

References Ahumada J. A., Lapasset E. 2007, A&A, 463, 789 Andreuzzi G. et al. 2002, in Thibault Lejeune and Joa˜o Fernandes, eds, Observed HR Diagrams and Stellar Evolution, ASP Conference Proceedings, vol. 274, ISBN: 1-58381-116-8, San Francisco: Astronomical Society of the Pacific, 2002, p. 349 Astropy Collaboration et al. 2018, AJ, 156, 123 Belloni T. et al. 1998, A&A, 339, 431 Browne S. E., Welsh B. Y., Wheatley J. 2009, PASP, 121, 450 Castelli F., Kurucz R. L. 2003, in Piskunov N., Weiss W. W., Gray D. F., eds, IAU Symposium, vol. 210, Modelling of Stellar Atmospheres. p. A20 (arXiv:astroph/0405087) Castelli F., Gratton R. G., Kurucz R. L., 1997, A&A, 318, 841 De Martino C., Bianchi L., Pagano I., Herald J., Thilker D., 2008, Mem. Soc. Astron. Italiana, 79, 704 Dinescu D. I., Girard T. M., van Altena W. F., Yang T.-G., Lee Y.-W. 1996, AJ, 111, 1205 Fitzpatrick E. L., 1999, PASP, 111, 63 Gao, X.-H. 2018, Ap&SS, 363, 232 Geller A. M., Mathieu R. D., Harris H. C., McClure R. D., 2008, AJ, 135, 2264 Gondoin P., 2005, A&A, 438, 291 Gosnell N. M. et al. 2014, ApJ, 783L, 8 Gosnell N. M., Mathieu R. D., Geller A. M., Sills A., Leigh N., Knigge C., 2015, ApJ, 814, 163 Green E. M., Liebert J. W., Peterson R. C., Saffer R. A. 1997, in Philip A. G. D. et al., ed., The Third Conf. on Faint Blue Stars (Schenectady, NY: L. Davis), 271 Green E. M., For B., Hyde E. A. et al. 2004, Ap&SS, 291, 267 Harris H., McClure R. 1985, PASP, 97, 261 Heber U. 1986, A&A, 155, 33 Hidalgo S. L. et al. 2018, ApJ, 856, 125 Hills J. G., Day C. A. 1976, Astrophys. Lett., 17, 87 Hunter J. D. 2007, Computing in Science and Engineering, 9, 90 Indebetouw R. et al. 2005, ApJ, 619, 931 Jeffery E., Platais I., Williams K. A. 2013, American Astronomical Society, AAS Meeting #221, id.250.40 Jiaxin W., Jun M., Zhenyu W., Song W., Xu Z. 2015, AJ 150, 61 Kaluzny J., Udalski A. 1992, AcA, 42, 29

J. Astrophys. Astr. (2021)42:47 Koester D. 2010, Memorie della Societa Astronomica Italiana, 81, 921 Landsman W., Paricio J. A., Ergeron P. B., Stefano R. Di, Stecher T. P. 1997, ApJ, 481L, 93 Landsman W., Bohlin R. C., Neff S. G., O’Connell R. W., Roberts M. S., Smith A. M., Stecher T. P. 1998, AJ, 116, 789 Liebert J., Saffer R. A., Green E. M. 1994, AJ, 107, 1408 Mazur, B., Kaluzny J. 1990, AcA, 40, 361 McCrea W. H., 1964, MNRAS, 128, 147 McKinney W. 2010, in van der Walt S., Millman J., eds, Proceedings of the 9th Python in Science Conference, pp. 51–56 Platais I., Kozhurina-Platais V., Mathieu R. D., Girard T. M., van Altena W. F. 2003, AJ, 126, 2922 Postma J. E., Leahy D. 2017, PASP, 129, 115002 Sandage A. R. 1953, AJ, 58, 61 Sarajedini A., von Hippel T., Kozhurina-Platais V., Demarque P. 1999, AJ, 118, 2894 Sandage A. 1962, ApJ, 135, 333

Page 11 of 11

47

Schindler J.-T., Green E. M., Arnett W. D. 2015, ApJ, 806, 178 Siegel Michael H. et al. 2014, The Astronom. J., 148(6), article id. 131, 13 pp. Sindhu N. et al. 2018, MNRAS, 481, 226 Stetson P. B. 1987, PASP, 99, 191 Tandon S. N. et al. 2017a, J. Astrophys. Astr., 38, 28 Tandon S. N. et al. 2017b, AJ, 154, 128 Tandon S. N. et al. 2020, AJ, 159, 158 Taylor M. B. 2005, TOPCAT & STIL: Starlink Table/ VOTable Processing Software Taylor M. 2011, TOPCAT: Tool for OPerations on Catalogues And Tables (ascl:1101.010) Trembley P.-E., Bergeron P. 2009, ApJ, 696, 1755 Tremblay P.-E., Bergeron P., Gianninas A. 2011, ApJ, 730, 128 van der Walt S., Colbert S. C., Varoquaux G. 2011, Computing in Science and Engineering, 13, 22 Jadhav Vikrant V. et al. 2019, ApJ, 886, 13 von Hippel T., Sarajedini A. 1998, AJ, 116, 1789 Whitford A. E., 1958, AJ, 63, 201

J. Astrophys. Astr. (2021) 42:50 https://doi.org/10.1007/s12036-021-09761-z

Ó Indian Academy of Sciences Sadhana(0123456789().,-volV)FT3 ](0123456789().,-volV)

SCIENCE RESULTS

A tale of two nearby dwarf irregular galaxies WLM and IC 2574: As revealed by UVIT CHAYAN MONDAL1,2,* , ANNAPURNI SUBRAMANIAM1 and KOSHY GEORGE3 1

Indian Institute of Astrophysics, 2nd Block, Koramangala, Bangalore 560 034, India. Pondicherry University, R.V. Nagar, Kalapet, Puducherry 605 014, India. 3 Faculty of Physics, Ludwig-Maximilians-Universita¨t, Scheinerstr. 1, 81679 Munich, Germany. *Corresponding Author. E-mail: [email protected] 2

MS received 7 November 2020; accepted 15 April 2021 Abstract. We present an ultra-violet study of two nearby dwarf irregular galaxies WLM and IC 2574, using the Far-UV and Near-UV data from the Ultra-Violet Imaging Telescope (UVIT). We used the F148W band Far-UV images and identified 180 and 782 young star-forming clumps in WLM and IC 2574, respectively. The identified clumps have sizes between 7–30 pc in WLM and 26–150 pc in IC 2574. We noticed more prominent hierarchical splitting in the structure of star-forming regions at different flux levels in IC 2574 than WLM. We found that the majority of the clumps have elongated shapes in the sky plane with ellipticity () greater than 0.6 in both the galaxies. The major axis of the identified clumps is found to show no specific trend of orientation in IC 2574, whereas in WLM the majority are aligned along south-west to north-east direction. We estimated (F148W–N242W) colour for the clumps identified in WLM and noticed that the younger ones (with (F148W–N242W) \  0:5) are smaller in size (\10 pc) and are located mostly in the southern half of the galaxy between galactocentric radii 0.4–0.8 kpc. Keywords. Galaxies: dwarf irregular—galaxies: individual—galaxies: star formation—ultra-violet.

1. Introduction Dwarf galaxies show wide variation in their star formation history (Tolstoy et al. 2009; McQuinn et al. 2015; Cignoni et al. 2018). They are typically gas-rich systems with a range of gas-to-mass ratios observed among different sub-classes. Interaction with a massive companion or even a nearby dwarf can produce episodes of starburst in dwarf galaxies and thus can sustain star formation for a longer time (Patton et al. 2013; Stierwalt et al. 2015). The duration of such bursts also varies among different dwarf systems. Dwarf galaxies are also unique in terms of their physical characteristics. These systems have a low gravitational potential well, metal-poor environment, and a rigid-body like rotation with no density waves. Such unique nature of dwarf galaxies provides a scope This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

to study star formation in extreme conditions. As dwarf galaxies mostly have low metallicity, studying these systems will be important to know how star formation proceeded in the earlier epochs (Weisz et al. 2011). The absence of spiral density waves and shear force in dwarfs further offers an opportunity to study the nature of star-forming clumps in a completely different environment than the massive spiral galaxies. With growing evidence from both observation and simulation, we understand that star formation is a hierarchical process from smaller cores to larger complexes (Elmegreen et al. 2000; Grasha et al. 2017). Especially, the young massive stars are found in loosely bound groups with no specific length scale (Garcia et al. 2010). Studying young star-forming regions in dwarf galaxies hence is crucial to know the properties of young stellar groups and how they continue to form in such environments. Apart from the external perturbation, star formation in dwarfs is also

50

Page 2 of 11

J. Astrophys. Astr.

controlled by the stochastic self-propagating mode and the internal stellar feedback (Gerola et al. 1980; Hunter 1997). The feedback from massive stars that formed earlier can inhibit or induce secondary star formation (Cignoni et al. 2019). The distribution of young star-forming regions in isolated dwarfs hence partly depends on the impact of stellar feedback across the galaxy. As young and massive OB type stars emit mostly in the far-UV wavelength, observations in Far-UV (FUV) band can directly trace young star-forming regions in a galaxy (Kennicutt & Evans 2012). The (FUV–NUV) colour also serves as an important quantity to constrain the age of stellar clumps up to a few hundred Myr (Goddard et al. 2010; Mondal et al. 2019b). Several studies have used UV data from the Hubble Space Telescope (HST) to understand the properties of star-forming regions from the analysis of resolved stellar population (Bianchi et al. 2012a, 2014; Calzetti et al. 2015). Bianchi et al. (2012b) used HST observations from FUV to I band to study active star-forming regions in six nearby dwarfs and highlighted the importance of UV observations. The UV space telescope Galaxy Evolution Explorer (GALEX) has made a phenomenal contribution to the study of star-forming regions in nearby galaxies with FUV and Near-UV (NUV) broad band imaging (Gil de Paz et al. 2007; Thilker et al. 2007; Kang et al. 2009; Melena et al. 2009; Goddard et al. 2010). Melena et al. (2009) used the GALEX FUV and NUV data combined with multi-band optical UBV and infrared JHK observations of young star-forming knots in 11 dwarf galaxies to study their properties. In this paper, we aim to understand the properties of young star-forming clumps in two nearby dwarf irregular galaxy WLM and IC 2574. The galaxy WLM, a member of the Local Group, is located at a distance of 995 kpc (Urbaneja et al. 2008). It is a

(2021) 42:50

relatively smaller, less massive, and metal-poor gasrich system (parameters listed in Table 1). The galaxy has an isolated location in the sky and does not show any signature of interaction (Leaman et al. 2012). The other galaxy IC 2574, located at a distance of 3.79 Mpc, is relatively large, massive, and metal-rich than WLM (parameters listed in Table 2). IC 2574 is a member of the M81 group, and it also does not have signature of interaction (Yun 1999). Both the galaxies have been studied in UV with observations from different telescopes. Bianchi et al. (2012b) studied WLM with HST multi-band observations including FUV and NUV and noticed active star formation during the last 10 Myr. Melena et al. (2009) have performed a photometric study of starforming knots in WLM using GALEX data. Mondal et al. (2018) studied the (FUV–NUV) colour demographics of young star-forming regions with multiband imaging observations from the Ultra-Violet Imaging Telescope (UVIT). The galaxy IC 2574 has been studied by Mondal et al. (2019a) to understand the connection between expanding H I holes and the triggered star formation using UVIT FUV observations. In this work, we studied the physical properties of the FUV-bright young star-forming clumps in the galaxies WLM and IC 2574 using UVIT broadband imaging observations. The UVIT PSF (  1.400 ) could resolve star-forming clumps up to  7 pc in WLM and  26 pc in IC 2574. The earlier studies on these galaxies with UVIT have not performed quantitative measurement of the individual star-forming clump up to the scales resolvable by the telescope. Here, we used the UVIT FUV intensity map to identify starforming clumps and compared their size, shape, orientation, FUV magnitude, and UV colour. As the two galaxies have significantly different size and mass, we further compared the properties of star-forming

Table 1. Properties of WLM. Property RA DEC Distance Metallicity log M ðM Þ Major axis Inclination PA of major axis

Value

References

00 01 57.8 -15 27 51.0 0.995 Mpc 0.003 6.88 5.70 69° 181°

Gallouet et al. (1975) Gallouet et al. (1975) Urbaneja et al. (2008) Urbaneja et al. (2008) Lee et al. (2006) de Vaucouleurs et al. (1991) de Vaucouleurs et al. (1991) de Vaucouleurs et al. (1991)

J. Astrophys. Astr.

(2021) 42:50

Page 3 of 11

50

Table 2. Properties of IC 2574. Property RA DEC Distance Metallicity (Z) log M ðM Þ Major axis Inclination PA of major axis

Value

References

10 28 23.5 ?68 24 43.7 3.79 Mpc 0.006 8.39 6.70 63° 55°

Skrutskie et al. (2006) Skrutskie et al. (2006) Dalcanton et al. (2009) Cannon et al. (2005) Lee et al. (2006) de Vaucouleurs et al. (1991) Pasquali et al. (2008) Pasquali et al. (2008)

clumps identified in both the system. In the next section, we present the data and observations used in this study. The analysis of the work is discussed in Section 3, results and discussions in Section 4, followed by summary in Section 5.

2. Data and observation We have used FUV and NUV imaging observations from the UVIT in this work. UVIT on AstroSat is composed of two telescopes (Kumar et al. 2012). One telescope observes in the FUV waveband (1300–1800 ˚ whereas the other one operates in both NUV A), ˚ and Visible. Both the FUV and NUV (2000–3000 A) channel consists of multiple filters of different bandwidth. The observation in the Visible channel is used to correct the drift caused in the image during the course of observation. Apart from having multiple filters and a wide circular field-of-view of diameter 280 , the UVIT instrument also has around three times better spatial resolution (FWHM of PSF  1.400 ) than the Galaxy Evolution Explorer (GALEX Martin et al. 2005). Such a unique combination has provided a great advantage to study external galaxies using UVIT. In this study, we have used F148W band FUV and N242W band NUV imaging data for the galaxy WLM (Fig. 1), whereas for IC 2574, we have used the data in F148W band (Fig. 2). Each observation was performed with multiple orbits of the AstroSat satellite. The raw data acquired from the UVIT observation were processed with the help of a customised software CCDLAB to produce science ready images (Postma & Leahy 2017). During this conversion, the images are drift-corrected, flat-fielded, aligned, and combined to produce the final deep images. We have also corrected the data for intrinsic distortion and fixed pattern noise

Figure 1. UVIT colour composite image of the galaxy WLM. The FUV F148W and the NUV N242W bands are shown in blue and yellow colour respectively.

Figure 2. UVIT F148W band image of the galaxy IC 2574.

of the detector using the calibration files (Girish et al. 2017; Postma et al. 2011). The final images have 4096  4096 pixel dimension with 1 pixel corresponding to 0.400 . At the distance of the galaxy WLM and IC 2574, this single pixel corresponds to  2 pc

50

Page 4 of 11

J. Astrophys. Astr.

(2021) 42:50

Table 3. Details of UVIT bands and observations. Bandpass ˚ (A)

ZP magnitude (AB)

Unit conversion ˚ (erg/s/cm2/A)

˚ MkðAÞ

F148W

1250–1750

18.016

3.09 910-15

500

N242W

2000–3000

19.81

2.22 9 10-16

785

Filter

Galaxy WLM IC 2574 WLM

Exposure time (s) 2634 10375 2824

and  7.6 pc, respectively. We have listed the exposure time of the UVIT observations and the filter combination in Table 3. The calibration measurements are adopted from Tandon et al. (2017).

3. Analysis 3.1 Identification of star-forming clumps As the young massive stars emit a copious amount of radiation in the FUV wavelength, we used UVIT F148W band images to locate the young-star forming regions in both the galaxies. The size of UVIT PSF is not sufficient to resolve individual stars in these galaxies. The point-like objects or the extended clumps detected with UVIT are actually a group of clusters, associations, or a combination of several such groups. We used F148W band FUV images and employed astrodendro1 package to identify the FUV bright star-forming clumps in both WLM and IC 2574. The astrodendro package helps to identify structures (i.e., intensity peaks) in the intensity map for a given threshold flux and a minimum clump size in pixel unit (Fig. 3). The algorithm identifies individual intensity peaks above the adopted threshold in the image and builds a structure tree (i.e., dendrogram) from the higher to lower flux levels. An intensity peak identified above the threshold is considered as a structure only if the peak height from the local minima is more than the value of minimum delta (another input parameter which we considered as 3 times the average background) and the size is greater than the defined minimum clump size. For a given set of input parameters, the code identifies both child and parent structure (in the form of a structure tree) and provides their position, area, flux, etc. The child structures are individual intensity peaks that could not be resolved further with UVIT, whereas the parent structures are defined as those that contain multiple child structures inside them. Based on the shape and intensity profile, 1

http://www.dendrograms.org/.

Figure 3. A specific region of the galaxy WLM is shown to display the detection of star-forming clumps (child structure – brown contoured region) by astrodendro package.

astrodendro also fits an ellipse for each individual structure and provides the major and minor axes of the fitted ellipse along with the position angle (PA) of its major axis. These parameters are important measures for the characterisation of the identified star-forming clumps. We fixed the minimum number of pixel for clumpidentification as 10. This is so chosen that the size of the smallest clump will be similar to the FWHM of the PSF (which is  1.400  3.4 pixels). At the distance of the galaxy WLM and IC 2574, this limiting size corresponds to  7 and  26 pc, respectively. To fix the threshold flux, we first examined the average background in each galaxy. We found the average FUV ˚ for WLM background flux (log[flux(erg/s/cm2 =A)]) and IC 2574 is  18:54 and 18:96, respectively. For selecting the threshold flux, we started with a flux value of 5 times the average background for each galaxy and increased it with 0.2 intervals in the logarithmic scale up to the value at which we detect less than 10 child structures. We identified both parent and child structures for these varying threshold fluxes and plotted the numbers in Fig. 4. For both the galaxies,

J. Astrophys. Astr.

(2021) 42:50

Page 5 of 11

50

the threshold flux for studying the star-forming clumps in IC 2574. A comparison between both the plots in Fig. 4 signifies that the hierarchical splitting of star-forming regions for varying flux levels is more prominent in the galaxy IC 2574 than WLM. This may be due to the fact that IC 2574 is a bigger galaxy, leading to the formation of bigger structures, which then fragments into smaller clumps, whereas there are only moderately big structures in WLM. To highlight this more, we have selected one parent structure (among the brighter and larger ones) from each of the galaxies and shown them in Fig. 5 along with their dendrograms. The structure trees show the nature of hierarchical splitting from higher to lower flux levels above the selected thresholds in each galaxy. We noticed more substructures at different flux levels for the region in IC 2574 than WLM.

3.2 Properties of the clumps

Figure 4. Number of child and parent structures identified in WLM (top panel) and IC 2574 (bottom panel) for varying threshold flux. The black lines show the ratio of child and parent structures. The vertical green dashed lines signify the value of threshold flux selected for each galaxy.

we noticed the obvious trend of detecting less number of structures with increasing threshold flux. But the overall behaviour is not exactly similar in both the galaxies. For the galaxy WLM, the number of parent and child structures identified for varying threshold flux is comparable (Fig. 4 – top). The black line that shows the ratio of child to parent structures highlights the same. We noticed a slight jump in the black line at ˚ ¼ 17:44). This flux value (log[flux(erg/s/cm2 =AÞ (which is  12 times the average background) is selected for studying the star-forming clumps in WLM. In the case of IC 2574 (Fig. 4 – bottom), we noticed that with decreasing threshold flux, the ratio of child to parent structure gradually increases initially and then becomes relatively flat and further shows a ˚ ¼ 17:87 small jump at log[flux(erg/s/cm2 =AÞ (  12 times the background flux). We selected this as

As our primary aim is to probe the nature of starforming clumps up to smaller scales, we considered only the child structures, that are individual clump identified in the UVIT images for our further analysis. The number of child structures identified for the selected threshold flux in WLM and IC 2574 is 180 and 782, respectively. The astrodendro package provides several physical quantities for the identified structures. We used those parameters to estimate the size, ellipticity, and position angle of the identified clumps. Using the position, orientation, and size of the clumps, we also performed custom aperture photometry with photutils python package to estimate the flux of these clumps. 3.2.1 Size. The clumps identified with astrodendro are mostly irregular in shape. We have shown a specific region of the galaxy WLM to highlight this in Figure 4. To estimate the size of these clumps, we considered the area of each structure, provided by astrodendro, and equated it with the area of a circle with diameter d. The derived value of d is considered as the size of that clump. In Figure 6, we have shown the histogram of the clump-size for both the galaxies. The FUV-bright clumps identified in WLM found to have sizes mostly in the range  7–30 pc, whereas IC 2574 hosts clumps mostly between  26–150 pc. We detected five larger clumps of size between 150–250 pc in IC 2574. The lower limit of the clump-size at the distance of each galaxy is limited by the UVIT resolution. To compare the nature of the histograms,

50

Page 6 of 11

J. Astrophys. Astr.

(2021) 42:50

Figure 5. Two selected star-forming regions (parent structure) from WLM (top panel) and IC 2574 (bottom panel) are shown along with their dendrograms which represent the overall structure tree. The brown contours show the boundary of the parent structures which host multiple child structure inside. Each branch of the dendrograms represents one child structure. The y-axis of the dendrogram shows the flux level in counts per second.

we estimated standard deviation (r) for both the distributions and found it to be 5.1 pc and 20.9 pc respectively for WLM and IC 2574. This highlights that the clump-size distribution in IC 2574 is broader than that in WLM. The smaller clumps dominate in number when compared to larger clumps in both the galaxies. It is also noticed that the galaxy WLM has produced relatively smaller clumps compared to IC 2574. Though, it is possible that the larger clumps identified in IC 2574, located around 4 times farther than WLM, are actually a combination of smaller clumps that could not be resolved further by UVIT. As the two galaxies are located at different distances, the difference found in the overall range of estimated clump-size can be biased. To verify this, we degraded the resolution of WLM by placing it at the distance of IC 2574. We carried out the same analysis with astrodendro and identified clumps in the degraded WLM image. The detected clumps have a size mostly in the range 26–60 pc. Only a few clumps have a size between 60–90 pc. This overall picture is not a match to what we noticed in IC 2574. This signifies there is an intrinsic difference in the overall distribution of clump-size of the two galaxies and hence the difference found in the measured range is not purely a distance effect.

Figure 6. The histogram for size (in pc) of identified starforming clumps. Top panel – WLM, bottom panel – IC 2574.

3.2.2 Ellipticity. We have estimated the ellipticity of the identified clumps (Figures 7 and 8). The astrodendro fits an ellipse inside the irregular-shaped boundary of each clump in the position-intensity plane and provides the length of major (a) and minor (b) axes of the fitted ellipse along with the position angle in the observed reference frame. We used the a, b values to estimate ellipticity of the clumps using qffiffiffiffiffiffiffiffiffiffiffiffi 2  ¼ 1  ba2 . A larger value of  will signify a more elongated shape of the clump, whereas a smaller value will denote a more circular shape. In the top-left panel of Figures 7 and 8, we have shown the distribution of measured ellipticity for the clumps detected in WLM and IC 2574, respectively. We noticed that the majority of the star-forming clumps identified in both the galaxies are elongated in shape. The number of near-circular clumps is comparatively less in both the galaxies. The histogram for the galaxy WLM

J. Astrophys. Astr.

(2021) 42:50

Page 7 of 11

50

Figure 7. The properties of the star-forming clumps identified in the galaxy WLM. Top-left: Histogram of the ellipticity of identified clumps. Top-right: Histogram of the position angle of identified clumps. Bottom-left: The size and the ellipticity of the clumps. Bottom-right: Observed FUV magnitude of the clumps and the corresponding error.

Figure 8. The properties of the star-forming clumps identified in the galaxy IC 2574. Each figure denotes the same as mentioned in Fig. 7.

peaks at   0.7, whereas for IC 2574 the ellipticity value peaks at   0.8. We have shown the size and ellipticity of the identified clumps of both galaxies in Figures 7 and 8(bottom-left). For both WLM and IC 2574, we noticed that most elliptic as well as circular structures are smaller in size. The larger clumps identified in

WLM and IC 2574 found to have moderate ellipticity between  0.5–0.8. 3.2.3 Position angle. The estimated PA of the clumps are shown in top-right panel of Figures 7 and 8. The astrodendro measures PA with respect to the increasing x-axis in the observed frame. We

50

Page 8 of 11

converted these to the WCS frame and estimated position angle with respect to the west direction in a counter clock-wise direction. For example, a position angle in the range 0 \ PA \ 90 will mean the major axis of the clump is oriented along south-east to northwest direction, whereas a value between 90 \ PA \ 180 will signify the clump to be oriented along southwest to north-east. The results show that the majority of the clumps detected in the galaxy WLM have a specific mode in their orientation. This is clear from the distinct peak (around PA  110) noticed in the distribution of PA in Fig. 7(top-right). On the other hand, the distribution is more uniform for the galaxy IC 2574 with a moderate peak around PA  125. 3.2.4 Flux. In order to characterise the star-forming clumps in terms of the FUV flux, we have performed custom aperture photometry using photutils package. We considered the position, major and minor axes (a, b), and the position angle of the fitted ellipse for each clump from astrodendro output. To measure the flux of a clump, we put an elliptic aperture on the clump center with major axis length as 2a and minor axis length as 2b with the same position angle. This is done to cover the identified irregular structure of each clump entirely. The measured flux is further corrected for the background and then converted to magnitude using the zero-points given in Table 3. The apparent magnitudes of the identified clumps for galaxies WLM and IC 2574 are shown in the bottom-right of Figures 7 and 8, respectively. The clumps identified in WLM have a magnitude range between  18–22.5 mag, whereas in IC 2574 it ranges from  16–24 mag. We have also shown error for the measured magnitudes in the same figure.

3.3 Radial distribution The (FUV–NUV) colour of a young star-forming clump is often used to trace the age of the stellar populations it contains. Simulation from the simple stellar population (SSP) models suggests that the UV colour becomes redder with increasing age of the SSP, and it is most sensitive up to a few hundred Myr (Goddard et al. 2010; Mondal et al. 2019b). The galaxy WLM has been observed in both F148W and N242W UVIT filters. In Section 3.2, we have estimated the FUV magnitudes of the identified starforming clumps. We implemented the same positions and apertures, as derived from the F148W band image, on the N242W band image and estimated the

J. Astrophys. Astr.

(2021) 42:50

Figure 9. The (F148W–N242W) colour and the galactocentric distance of the clumps identified in the galaxy WLM.

Figure 10. The (F148W–N242W) colour and the size of the clumps identified in WLM.

NUV magnitudes of the clumps. The NUV magnitudes are also corrected for background emission. Using these magnitudes, we calculated (F148W– N242W) colour for each clump. We have also estimated the galactocentric distance to each identified clump following the steps discussed in Mondal et al. (2019a). In Fig. 9, we have shown the (F148W– N242W) colour and the galactocentric distance (in kpc) of the clumps identified in the galaxy WLM. The clumps with bluer colour are seen between radii 0.4– 0.8 kpc. These clumps are mostly located in the regions R1, R2, and R3, as shown in Mondal et al. (2018). The clumps identified in the central and the outer part of the galaxy are relatively redder in colour. We have also shown the (F148W–N242W) colour and the size of the detected clumps in Fig. 10. We found

J. Astrophys. Astr.

(2021) 42:50

that all the bluer clumps with (F148W–N242W) \  0:5 are smaller in size (\10 pc). We also found a few redder clumps with (F148W–N242W) [0.5) to be smaller than 10 pc in size. We note here that the photometric error in the (F148W–N242W) colour for clumps with extreme blue and red colours (which are smaller in size) is slightly higher (ranges between 0.2–0.4 mag). We could not study the (FUV–NUV) colour of the clumps in IC 2574, as the UVIT NUV data of the galaxy was not there.

4. Results and discussions The key aim of this study is to identify young starforming regions in two nearby dwarf irregular galaxies and understand their properties. Despite having the same morphological class, the galaxy WLM and IC 2574 have significantly different characteristics. IC 2574 is around  5 times larger and  30 times massive than WLM. Though both the galaxies have low metallicity, IC 2574 is relatively metal-rich than WLM. Such contrast in the physical characteristics has motivated us to study the properties of star-forming clumps in each system and further compare them. To identify star-forming clumps, we employed astrodendro python package. We varied the threshold flux in each galaxy and counted the number of child and parent structures for each different flux value. For the lower values of threshold flux, we found that the number of identified child structures is around 2 times higher than the parent structures in IC 2574, whereas for WLM it is around 1.5 times. This signifies the galaxy IC 2574 has more sub-structures at the lower flux levels compared to WLM. In other words, it suggests that the young star-forming regions in IC 2574 are more clumpy in nature. On the other hand, this could also be due to the observational bias as IC 2574 was imaged with 4 times deeper exposure than WLM and hence could detect many fainter clumps. We also note here that due to the deeper exposure, we expect to have less artefact in the identified clumps in IC 2574 than in WLM. With the selected threshold flux (i.e.,  12 times the average background in each galaxy), we identified around 4 times more number of clumps in IC 2574 than WLM. This is mostly because of the larger size of the galaxy IC 2574. The difference in the exposure times can also be an add-on factor for such a contrast in the identified number. We also performed the same identification algorithm with the same thresholds and minimum pixel number on the galaxy images subtracted for smooth galaxy

Page 9 of 11

50

background (produced using sep python package (Bertin and Arnouts 1996; Barbary 2016)) and identified 164 and 514 clumps, respectively in WLM and IC 2574. We searched for the same clumps as detected earlier and found that they have almost the same sizes with slightly less flux value ( which is mostly less than  20% of the earlier value). The identified clumps have sizes mostly between  7–30 pc in WLM and  26–150 pc in IC 2574. We could see that the bigger galaxy IC 2574 has clumps of larger size. The lower limit of the clump-size in each galaxy is fixed by the UVIT PSF and the distance to that galaxy. The overall size-range of the detected clumps is also sensitive to the distance of the galaxy (Bastian et al. 2007; Garcia et al. 2010). To confirm that the difference between the range of detected clump-size for two galaxies is not a result of the difference in their distances, we performed the same analysis on a smoothed image of WLM by pushing it at the distance of IC 2574. With this, we found the identified clumps to be smaller than 60 pc with a few between 60–90 pc. Therefore, the difference in the clump-size of two galaxies is mostly an intrinsic property and not an artefact. As FUV emission directly traces massive stellar populations, it is more likely that the star-forming clumps we identified in both the galaxies are OB associations. Several observations have targeted the OB associations in the Milky Way as well as in nearby spiral and dwarf galaxies and reported an average size between 15–100 pc (Mel’Nik & Efremov 1995; Ivanov 1996; Bresolin et al. 1996, 1998; Bastian et al. 2007; Garcia et al. 2010). The overall range of clump-size found in our study supports the earlier findings. Pellerin et al. (2012) used HST optical observations for a limited region in the northern part of IC 2574 and detected 75 young stellar groups (age  10 Myr) with sizes between 10–120 pc, which matches well with our results. Although, it is possible that the larger clumps we detected are actually a combination of multiple smaller clumps that are not resolved by UVIT. The detected FUV clumps in WLM also show a good spatial correlation with the FUV bright stars as identified by HST (Bianchi et al. 2012b). In both the galaxies, we found the smaller clumps to dominate in numbers, which also have been reported earlier (Garcia et al. 2010). As the identified star-forming clumps have an irregular shape, we used astrodendro to fit ellipse for each of the structures and estimated their ellipticity. From the measured ellipticity values, we found the majority of the clumps to be elongated and very less

50

Page 10 of 11

have near-circular shape. This highlights the structure of young star-forming regions in these galaxies. The stellar groups identified by Pellerin et al. (2012) in IC 2574 were found to have  in the range 0–0.95. One possible reason for the elongated shape can be imaging in the FUV band. FUV emission mostly picks massive stars, which may not be distributed uniformly in a stellar group. Therefore, the shape of the clumps retrieved from the FUV imaging may not be the actual shape of the stellar group we are detecting. This also may be the reason why we see less number of nearcircular clumps. The other important point to note here is the effect of galaxy inclination. Both the galaxies have relatively higher inclination angle (Tables 1, 2). For intrinsically non-spherical clumps, the inclination of the galaxy will have a clear effect on their shape, as seen in the sky plane. Dwarf galaxies are systems that lack strong shear force and the presence of density waves. Such ordered force has a significant impact on the properties of molecular clouds and hence the star-forming clumps (Vogel et al. 1988; Miyamoto et al. 2014). The shape and orientation of the clumps can also be influenced by such organised force (Pellerin et al. 2012). Here, we estimated the major axis PA of identified starforming clumps to see whether there exists any specific trend in their orientation. Our analysis shows that the clumps detected in IC 2574 do not show any distinct mode in their orientation, whereas in WLM we noticed the majority of the clumps are aligned along south-west to north-east direction. Earlier studies by Kepley et al. (2007) and Mondal et al. (2018) have shown the possibility of propagating star formation in the hook-like H I structure around the centre in WLM. The specific mode of orientation of the star-forming clumps may have a connection with this. It may also be possible that the elongation is arising due to an underlying magnetic field or a filamentary structure in the molecular clouds. The youngest clumps (bluer than –0.5 mag) in WLM are identified between radii 0.4 kpc to 0.8 kpc. As WLM has more star formation in the southern part, these clumps are located in the southern half of the galaxy, specifically in regions R1, R2, and R3 as defined in Mondal et al. (2018). Both the central and the outer parts of the galaxy host lesser clumps and they are mostly redder in colour. We also found that all the younger clumps with (F148W–N242W) \0.5 are smaller than 10 pc in size. A few redder clumps of size \10 pc are also detected. We note here that the smaller clumps have a relatively larger

J. Astrophys. Astr.

(2021) 42:50

photometric error in their measured magnitude. Therefore, the clumps with extreme colours will have a larger error (  0.2–0.4) in their colour. We are unable to carry out a similar exercise for IC 2574 as NUV data is not available.

5. Summary The key results of the study are summarized below: (1) We studied the characteristics of FUV-bright starforming clumps in two nearby dwarf irregular galaxies WLM and IC 2574 using UVIT imaging observations. (2) The identified clumps have a size between  7–30 pc in WLM and  26–150 pc in IC 2574. The average size of the clumps is larger in the galaxy IC 2574, which is bigger and massive than WLM. (3) We found that the hierarchical splitting of starforming regions is more prominent in IC 2574 than WLM. (4) The young star-forming clumps identified in both the galaxies are mostly observed to be elongated in shape. (5) We did not find any specific orientation of the clump major axis in IC 2574, whereas in WLM the majority of the clumps are oriented along southwest to north-east direction. (6) The youngest star-forming clumps in WLM are detected between radii 0.4 kpc to 0.8 kpc. Both the central and outer parts of the galaxy are relatively less active in recent times.

Acknowledgements This publication uses the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). This publication uses UVIT data processed by the payload operations centre at IIA. The UVIT is built in collaboration between IIA, IUCAA, TIFR, ISRO, and CSA. Indian Institutions and the Canadian Space Agency have contributed to the work presented in this paper. Several groups from ISAC (ISRO), Bengaluru, and IISU (ISRO), Trivandrum have contributed to the design, fabrication, and testing of the payload. The Mission Group (ISAC) and ISTRAC (ISAC) continue to provide support in making observations with, and reception and initial processing of the data. We gratefully thank all the

J. Astrophys. Astr.

(2021) 42:50

individuals involved in the various teams for providing their support to the project from the early stages of the design to launch and observations with it in the orbit. This research made use of Matplotlib (Hunter 2007), Astropy (Astropy Collaboration et al. 2013, 2018), Astrodendro (http://www.dendrograms. org/), community-developed core Python packages for Astronomy. Finally, we thank the referees for the valuable suggestions.

References Astropy Collaboration: Robitaille T. P., Tollerud E. J. et al. 2013, A&A, 558, A33 Astropy Collaboration: Price-Whelan A. M., Sip}ocz B. M. et al. 2018, AJ, 156, 123 Barbary K. 2016, J. Open Source Software, 1, 58 Bastian N., Ercolano B., Gieles M. et al. 2007, MNRAS, 379, 1302 Bertin E., Arnouts S. 1996, A&AS, 117, 393 Bianchi L., Efremova B., Hodge P., Kang Y. 2012a, AJ, 144, 142 Bianchi L., Efremova B., Hodge P., Massey P., Olsen K. A. G. 2012b, AJ, 143, 74 Bianchi L., Kang Y., Hodge P., Dalcanton J., Williams B. 2014, Adv. Space Res., 53, 928 Bresolin F., Kennicutt Jr. R. C., Stetson P. B. 1996, AJ, 112, 1009 Bresolin F., Kennicutt Jr. R. C., Ferrarese L. et al. 1998, AJ, 116, 119 Calzetti D., Lee J. C., Sabbi E. et al. 2015, AJ, 149, 51 Cannon J. M., Walter F., Bendo G. J. et al. 2005, ApJL, 630, L37 Cignoni M., Sacchi E., Aloisi A. et al. 2018, ApJ, 856, 62 Cignoni M., Sacchi E., Tosi M. et al. 2019, ApJ, 887, 112 Dalcanton J. J., Williams B. F., Seth A. C. et al. 2009, ApJS, 183, 67 de Vaucouleurs G., de Vaucouleurs A., Corwin Herold G. J. et al. 1991, Third Reference Catalogue of Bright Galaxies Elmegreen B. G., Efremov Y., Pudritz R. E., Zinnecker H. 2000, Protostars and Planets IV, 179 Gallouet L., Heidmann N., Dampierre F. 1975, A&AS, 19, 1 Garcia M., Herrero A., Castro N., Corral L., Rosenberg A. 2010, A&A, 523, A23 Gerola H., Seiden P. E., Schulman L. S. 1980, ApJ, 242, 517 Gil de Paz A., Boissier S., Madore B. F. et al. 2007, ApJS, 173, 185 Girish V., Tandon S. N., Sriram S., Kumar A., Postma J. 2017, Experimental Astron., 43, 59 Goddard Q. E., Kennicutt R. C., Ryan-Weber E. V. 2010, MNRAS, 405, 2791

Page 11 of 11

50

Grasha K., Elmegreen B. G., Calzetti D. et al. 2017, ApJ, 842, 25 Hunter D. 1997, PASP, 109, 937 Hunter J. D. 2007, Computing in Science & Engineering, 9, 90 Ivanov G. R. 1996, A&A, 305, 708 Kang Y., Bianchi L., Rey S.-C. 2009, ApJ, 703, 614 Kennicutt R. C., Evans N. J. 2012, ARA&A, 50, 531 Kepley A. A., Wilcots E. M., Hunter D. A., Nordgren T. 2007, AJ, 133, 2242 Kumar A., Ghosh S. K., Hutchings J. et al. 2012, in Proceedings SPIE, Vol. 8443, Space Telescopes and Instrumentation 2012: Ultraviolet to Gamma Ray, 84431N Leaman R., Venn K. A., Brooks A. M. et al. 2012, ApJ, 750, 33 Lee H., Skillman E. D., Cannon J. M. et al. 2006, ApJ, 647, 970 Martin D. C., Fanson J., Schiminovich D. et al. 2005, ApJL, 619, L1 McQuinn K. B. W., Skillman E. D., Dolphin A. E., Mitchell N. P. 2015, ApJ, 808, 109 Melena N. W., Elmegreen B. G., Hunter D. A., Zernow L. 2009, AJ, 138, 1203 Mel’Nik A. M., Efremov Y. N. 1995, Astron. Lett., 21, 10 Miyamoto Y., Nakai N., Kuno N. 2014, PASJ, 66, 36 Mondal C., Subramaniam A., George K. 2018, AJ, 156, 109 Mondal C., Subramaniam A., George K. 2019a, AJ, 158, 229 Mondal C., Subramaniam A., George K. 2019b, J. Astrophys. Astr., 40, 35 Pasquali A., Leroy A., Rix H.-W. et al. 2008, ApJ, 687, 1004 Patton D. R., Torrey P., Ellison S. L., Mendel J. T., Scudder J. M. 2013, MNRAS, 433, L59 Pellerin A., Meyer M. M., Calzetti D., Harris J. 2012, AJ, 144, 182 Postma J., Hutchings J. B., Leahy D. 2011, PASP, 123, 833 Postma J. E., Leahy D. 2017, PASP, 129, 115002 Skrutskie M. F., Cutri R. M., Stiening R. et al. 2006, AJ, 131, 1163 Stierwalt S., Besla G., Patton D. et al. 2015, ApJ, 805, 2 Tandon S. N., Subramaniam A., Girish V. et al. 2017, AJ, 154, 128 Thilker D. A., Bianchi L., Meurer G. et al. 2007, ApJS, 173, 538 Tolstoy E., Hill V., Tosi M. 2009, ARA&A, 47, 371 Urbaneja M. A., Kudritzki R.-P., Bresolin F. et al. 2008, ApJ, 684, 118 Vogel S. N., Kulkarni S. R., Scoville N. Z. 1988, Nature, 334, 402 Weisz D. R., Dalcanton J. J., Williams B. F. et al. 2011, ApJ, 739, 5 Yun M. S. 1999, in Barnes J. E., Sanders D. B., eds, IAU Symposium, Vol. 186, Galaxy Interactions at Low and High Redshift, 81

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:38 https://doi.org/10.1007/s12036-020-09688-x

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

AstroSat observation of 2016 outburst of H 1743-322 SWADESH CHAND1 , V. K. AGRAWAL2, G. C. DEWANGAN3 ,

PRAKASH TRIPATHI3

and PARIJAT THAKUR1,*

1

Department of Pure and Applied Physics, Guru Ghasidas Vishwavidyalaya (A Central University), Bilaspur (C.G.) 495 009, India. 2 Space Astronomy Group, ISITE Campus, ISRO Satellite Centre, Bangalore 560 037, India. 3 Inter-University Centre for Astronomy and Astrophysics, Post Bag 4, Ganeshkhind, Pune 411 007, India. *Corresponding author. E-mail: [email protected]; [email protected] MS received 6 November 2020; accepted 18 December 2020 Abstract. We present the detection of type C quasi-periodic oscillation (QPO) along with upper harmonic at respective frequencies of  0.6 Hz and  1.2 Hz in the single AstroSat observation taken during the 2016 outburst of the low-mass black hole X-ray binary H 1743-322. These frequencies are found to be shifted by  0.4 Hz for the QPO and  0.8 Hz for the upper harmonic with respect to that found in the simultaneous XMM-Newton and NuSTAR observation taken five days later than the AstroSat observation, indicating a certain geometrical change in the system. However, the centroid frequency of the QPO and the upper harmonic do not change with energy, indicating the energy-independent nature. The decreasing trend in the fractional rms of the QPO with energy is consistent with the previous results for this source in the low/hard state. The value of the photon index (C  1:67) also indicates that the source was in the low/hard state during this particular observation. In addition, similar to the XMM-Newton observations during the same outburst, we find a hard lag of  21 ms in the frequency range of  1–5 Hz. The log-linear trend between the averaged time lag and energy indicates the propagation of fluctuations in the mass accretion rate from outer part of the accretion disk to the inner hot regions. Keyword. Black hole physics—binaries: close—X-rays: binaries—X-rays: individual: H 1743-322.

1. Introduction A majority of black hole X-ray binaries (BHXRBs) exhibits transient nature and shows occasional outbursts due to sudden change in the mass accretion rate while spending most of the time in quiescence. The source luminosity may increase up to several orders of magnitude during such outbursts (Tanaka & Shibazaki 1996; Shidatsu et al. 2014; Plant et al. 2015). In the course of a usual outburst, the black hole transients (BHTs) evolve through the low/hard state (LHS) to the high/soft state (HSS) via two intermediate states, viz. the hard and soft intermediate states (HIMS and SIMS; Belloni et al. 2005; Belloni 2010). These states This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

are attributed to certain spectral and timing characteristics, which can be distinguished through the hardness intensity diagram (HID; Belloni et al. 2005; Homan & Belloni 2005; Gierlin´ski & Newton 2006; Fender et al. 2009; Belloni 2010). The X-ray spectrum in the LHS is dominated by the Comptonized emission with a powerlaw index \2 and cutoff energy  100 keV, and the source is associated with strong variability (  30%). On the other hand, the thermal emission from the optically thick and geometrically thin accretion disk dominates the HSS, where the photon index can extend up to 2.5 with a few percent of variability. Low-frequency quasi-periodic oscillations (LFQPOs), ranging from 0.05–30 Hz, are often observed in the BHTs. The exact origin of these LFQPOs are still not clear. However, LFQPOs are categorized into three

38

Page 2 of 9

types: type A, B and C. Among all the three types, the type C QPOs are very common, and appear as strong and narrow variable peaks with strong fractional rootmean-squared (rms) variability (  3–16%) in both the LHS and HIMS. Type B QPOs are typically observed in the intermediate states having rms up to  4%, whereas type A QPOs appear as broad peaks with a few percent of rms (Casella et al. 2005; Motta et al. 2011; Alam et al. 2014). BHTs show variability over the timescale of few seconds to days and the variability relies upon the change in source flux and energy due to different ongoing physical processes. One of the most prominent approach to investigate variability observed in BHTs is the study of time lag between the different energy bands. The measured time lags can be either positive or negative. The positive or hard lag implies that hard photons are delayed relative to the soft ones, and is supposed to be caused by the propagation of mass accretion rate fluctuation in the accretion disk (Page et al. 1981; Miyamoto et al. 1988; Lyubarskii 1997; Nowak et al. 1999a, b; Grinberg et al. 2014; Are´valo & Uttley 2006; De Marco et al. 2013). On the other hand, the coronal X-rays get reflected from the accretion disk and give rise to a time delay between the primary X-ray continuum and the reprocessed emission from the inner accretion disk close to the central source. This time delay is known as the soft lag or the reverberation lag (De Marco et al. 2013; Kara 2014; De Marco & Ponti 2016; De Marco et al. 2017; Kara et al. 2019). H 1743–322, a low-mass black hole transient source discovered with Ariel-V in 1977, shows frequent outburst over the time scale of  200 days (Kaluzienski & Holt 1977; Shidatsu et al. 2012, 2014). The first brightest outburst of this source took place in 2003, and was observed by INTEGRAL (Revnivtsev 2003). RXTE also observed the same outburst, leading to the detection of a pair of high-frequency QPOs (HFQPOs; Homan & Belloni 2005; Remillard et al. 2006). Steiner et al. (2012) estimated the spin of the black hole to be 0:2  0:3 with RXTE observation during the 2003 outburst, and used the radio observation by very Large Array (VLA) telescope of the same outburst to estimate the source distance and inclination angle to be 8:5  0:8 kpc and 75  3 , respectively. Moreover, Sriram et al. (2009) detected QPO in the steep powerlaw state of the source during the 2003 outburst, observed by RXTE. H 1743–322 has also gone through several outbursts until 2008. This outburst in 2008 was found to be a failed one as the source could not reach the HSS over the entire outburst cycle due to abrupt change in

J. Astrophys. Astr. (2021)42:38

the mass accretion rate (Capitanio et al. 2009). A few more outbursts of the source were detected between 2009 and 2013 by several space observatories. Altamirano and Strohmayer (2012) discovered mHz QPO using the RXTE and Chandra observations of 2010 and 2011 outbursts. Using the RXTE observations of the 2010 and 2011 outbursts, Molla et al. (2017) also detected the presense of QPO at  1 Hz, and estimated þ1:65 M . Apart the mass of the black hole to be 11:211:96 from the above, another outburst of the source in 2014 was observed by XMM-Newton, NuSTAR and Swift/ XRT. From the hardness intensity diagram (HID) derived from the Swift/XRT monitoring, Stiele and Yu (2016) found that the 2014 outburst was a failed one as the source stayed in the LHS throughout the full outburst. They also detected QPO (  0.25 Hz) along with upper harmonic (  0.51 Hz) in the commensurate ratio of 1 : 2, and an iron emission line around  6.7 keV in the XMM-Newton observation. The accretion disk, estimated from the relativistic reflection model was found to be truncated during this outburst (Ingram et al. 2017). A soft X-ray lag of  60 ms, arising most probably due to thermal reverberation, was also detected by De Marco and Ponti (2016) in the XMMNewton observations of both the 2014 and 2008 failed outbursts. Furthermore, Chand et al. (2020) analyzed the two consecutive simultaneous XMM-Newton and NuSTAR observations of the 2016 outburst of H 1743322. Unlike the 2008 and 2014 failed outbursts, they found the outburst in the 2016 to be a successful one using the HID derived with the Swift/XRT monitoring of the same outburst. They found the presence of the QPO and its upper harmonic at  1 Hz and  2 Hz, respectively. Moreover, the centroid frequencies of the QPO and its upper harmonic were found to be drifting to higher frequencies over the two epochs. They also detected a fluorescent iron emission line at  6.5 keV and found the accretion disk likely to be truncated. A hard X-ray lag of 0:40  0:15 s and 0:32  0:07 s in the two respective epochs were also found by them. It is worth mentioning here that AstroSat also observed the 2016 outburst of H 1743-322 five days earlier than the above mentioned simultaneous XMM-Newton and NuSTAR observations. Hence, it would be interesting to compare the results from the AsroSat with those obtained from the above mentioned simultaneous XMM-Newton and NuSTAR observations for the 2016 outburst. Broadband coverage of AstroSat can also shed light on the timing properties of the source at higher energy ([30 keV), which was not possible with earlier space observatories. In this paper, we carry out

J. Astrophys. Astr. (2021)42:38

the timing and broadband (0.7–80 keV) spectral study of H 1743-322 using AstroSat observation. The paper is organized as follows. In Section 2, we describe the observation and data reduction. Section 3 presents the analysis and results. Finally, the discussion and concluding remarks are given in Section 4.

Page 3 of 9

38

LAXPC software (Laxpcsoft4). The standard task within Laxpcsoft5 were employed to derive the energy spectrum and light curve. For spectral analysis, we used only the LAXPC20 data due to its low background.

3. Analysis and results 2. Observation and data reduction 3.1 Timing analysis We used the single AstroSat target of opportunity observation (Obs Id: T01 045T01 9000000364) of H 1743-322 taken during its 2016 outburst phase for an effective LAXPC exposure time of  12.6 ks. This observation was taken five days earlier than the simultaneous XMM-Newton and NuSTAR observations of the same outburst. Figure 1 shows the MAXI lightcurve indicating the position of the AstroSat observation along with the simultaneous XMM-Newton and NuSTAR observations during the mentioned outburst period of the source. The Level 2 data of Soft X-Ray Telescope (SXT; Singh et al. 2016, 2017) observations were downloaded from the ISSDC website1. Using the SXT event merger tool2, we obtained an exposure-corrected merged single cleaned event file. To extract the source spectrum, we used the standard available HEASoft 0 tool XSELECT V2.6d and a circular region of 15 from the merged clean event file. For the response matrix file (RMF) and background spectrum, we used the SXT POC team provided ‘‘sxt pc mat g0to12.rmf’’ and the blank sky observation ‘‘SkyBkg comb EL3p5 Cl Rd16p0 v01.pha’’, respectively. By employing the on-axis ARF provided by the SXT POC team, the SXT off-axis auxiliary response file (ARF) was generated with the sxtARFModule tool3, which is suitable for the location of the source on the CCD. The SXT spectrum in the 0.7–6 keV band was used for sepctral analysis. Since the current version of the SXTPIPELINE does not correct for the gain shifts, SXT team recommends gain correction by keeping the slope fixed at unity and the offset to be variable. During the fitting of the SXT spectral data, this same method was applied to modify the gain of the response file. We obtained the Level 2 data of Large Area X-Ray Proportional Counter (LAXPC; Yadav et al. 2016a, b; Agrawal et al. 2017; Antia et al. 2017) using the

We have used Interactive Spectral Interpretation System (ISIS, V.1.6.2–40; Houck & Denicola 2000) for the timing analysis and quoted the errors in 90% confidence level. For the extraction of light curves, both the LAXPC10 and LAXPC20 detectors were used. Power density spectra (PDSs) from the lightcurves of 0.05 s bin were derived with ‘‘POWSPEC’’ task within FTOOLS in the three bands: 3–15, 15–30 and 30–80 keV. The PDSs were normalized as per Leahy et al. (1983) after the subtraction of contribution from the Poisson noise, and then the variability power was converted to the square fractional rms (Belloni & Hasinger 1990). The PDSs derived in all the three bands shows the presence of quasi-periodic oscillation (QPO) at  0.6 Hz. Also the presence of an upper harmonic at  1.2 Hz is found in the 3–15 and 15–30 keV bands. However, it was absent in the 30– 80 keV band, which may be due to poor signal to noise ratio in the data at such high energy. Figure 2 shows the PDS derived in the 3–15 keV band fitted with five Lorenztians required for the QPO, upper harmonic and three band limited noise (BLN) components. In addition to the QPO and its upper harmonic, two additional Lorentzains were employed to describe the two BLN components for the PDS derived in the 15–30 keV band (see Fig. 3). However, only two Lorentzians, required for the QPO and a single BLN component, were employed to describe the PDS in the 30–80 keV band (see Fig. 4). The best fit model parameters are listed in Table 1. It can be seen from Table 1 that the QPO and its upper harmonic are present in the 1 : 2 ratio, and their centroid frequencies do not change with energy. For all the three bands, the quality-factor (Q ¼ mcentroid /FWHM) of the QPOs are found to be similar within the errors. The same is also true for the upper harmonic in the 3– 15 and 15–30 keV bands. The fractional rms amplitude of the QPO in the 15–30 keV band shows a slight

1

https://astrobrowse.issdc.gov.in/astro_archive/archive/Home.jsp. https://www.tifr.res.in/*astrosat_sxt/dataanalysis.html. 3 https://www.tifr.res.in/*astrosat_sxt/dataanalysis.html. 2

4

http://astrosat-ssc.iucaa.in/?q=laxpcData. https://www.tifr.res.in/*astrosat_laxpc/LaxpcSoft.html.

5

38

Page 4 of 9

J. Astrophys. Astr. (2021)42:38

Photons cm s

-2 -1

0.2

0.1

0

57400

57420

57440

57460

57480

Time (MJD)

57500

57520

57540

Figure 1. MAXI lightcurve during the 2016 outburst of H 1743–322. The two blue squares indicate the positions of the two XMM-Newton observations during the same outburst, whereas the AstroSat observation is depicted with the red cross sign.

To study the behavior of the fractional rms of the QPO as a function of energy, we extracted PDSs in different energy bands. These PDSs were fitted with five Lorentzians, where all the parameters except the normalization were kept fixed at the values obtained for the PDS in the 3–15 keV band. Figure 5 depicts that the fractional rms of the QPO shows a decreasing nature with the increase energy. To estimate the time lag as a function of frequency, we used the LAXPC subroutine ‘laxpc_find_freqlag’, which employs the method as described in Vaughan and Nowak (1997) and Nowak et al. (1999a). In this method, time lag between two different time series s(t) and h(t) is calculated as sðf Þ ¼ /ðf Þ=2pf , where /ðf Þ ¼ avg½CðfÞ is the phase of the average cross power spectrum C(f). The cross power spectrum C(f) is defined as Cðf Þ ¼ S ðf ÞHðf Þ, where S(f) and H(f) represent the discrete Fourier transforms of the mentioned time series s(t) and h(t), respectively. As shown in Fig. 6, we found a positive lag of 21:4  2:4 ms in the frequency range  1–5 Hz between the soft photons in the 3–10 keV band and the hard photons in the 20–40 keV band. This indicates that the high energy photons in the 20–40 keV band lag the soft ones in the 3–10 keV band. The coherence between the above mentioned energy bands were found to be nearly equal in the frequency range of 1–5 Hz. This frequency range was used to estimate the averaged time lag as function of energy by considering the same energy bands as used for the derivation of the QPO fractional rms. The averaged time lag estimated in the frequency range 1–5 Hz shows an increasing trend with energy in a log-linear manner (see Fig. 7).

3.2 Broadband spectral analysis Figure 2. Power density spectra in the 3–15 keV band fitted with five Lorentzians.

decrease than that found in the 3–15 keV. However, it decreases marginally in the 30–80 keV band. The fractional rms of the upper harmonic remains similar in the 3–15 and 15–30 keV bands. Moreover, the significance of the QPOs are  30.6r (for 3–15 keV),  26.5r (for 15–30 keV) and  11.7r (for 30–80 keV), which suggests that it is decreasing with increase in energy. In a similar way, the significance of the upper harmonic is  8.7r and  5.2r in the 3–15 keV and 15–30 keV bands, respectively.

Time-averaged broadband X-ray spectral analysis was performed by fitting the SXT (0.7–6 keV) and LAXPC20 (3–80 keV) spectral data simultaneously within ISIS (V.1.6.2-40; Houck & Denicola 2000). All the error bars are quoted in 90% confidence level unless otherwise specified. To account for the calibration uncertainty, we added a 3% systematic uncertainty to SXT and LAXPC20 data. We grouped the LAXPC20 spectral data to a minimum signal to noise (S/N) ratio of 5 and a minimum of 1 channel per bin. For the SXT data, we used a minimum S/N of 10 and a minimum of 3 channels per bin. We fitted the spectral data from both the instruments simultaneously with multi-color disk black body (diskbb;

J. Astrophys. Astr. (2021)42:38

Figure 3. Power density spectra in the 15–30 keV band fitted with four Lorentzians.

Page 5 of 9

38

normalization between the two instruments. The constant factor was fixed at 1 for the LAXPC20 data and kept free to vary for the SXT data. To account for the thermal Comptonization from the hot corona, we used the nthcomp model (Zdziarski et al. 1996; _ Zycki et al. 1999) and tied the seed photon temperature (KTbb ) with the inner disk temperature (KTin ) from the diskbb model. Hence, the model constant*tbabs(diskbb?nthcomp) provides a best fit with v2 /dof ¼ 204:5=248. The best-fit model to the SXT and LAXPC20 spectral data is shown in Fig. 8, whereas all the best-fit model parameters from the broadband X-ray spectral fitting are listed in Table 2. Unlike the XMM-Newton and NuSTAR observations of the same outburst of this source (see Chand et al. 2020), we could not detect the presence of iron emission line from the five days earlier observation by AstroSat. As seen from Table 2, the hydrogen column density is found to be 1:94  0:07. The value of the photon index (C  1:67) indicates that the source was in the LHS during this observation. The electron plasma temperature (KTe ) is found to be [56.9 keV. Moreover, we found the inner disk temperature (KTin ) to be high  1.2 keV.

4. Discussion and concluding remarks

Figure 4. Power density spectra in the 30–80 keV band fitted with two Lorentzians.

Mitsuda et al. 1984) modified by the Galactic absorption model tbabs with the abundances of Wilms et al. (2000) and the cross section given by Verner et al. (1996). A multiplicative constant model was also used to take care of the relative

In this paper, we have carried out a comprehensive spectral and temporal analysis of H 1743-322 with a single AstroSat observation taken during the 2016 outburst of the source. This observation was taken five days earlier than the simultaneous XMM-Newton and NuSTAR ones during the outburst rise phase of the source (see Fig. 1). As similar to the study of the 2016 outburst of this source by Chand et al. (2020) using the above said simultaneous XMM-Newton and NuSTAR observations, QPO along with its upper harmonic in the commensurate ratio of 1 : 2 are detected in both the 3–15 keV and 15–30 keV bands of the AstroSat observation. For the first time, the LAXPC data onboard AstroSat allows us to investigate the properties of PDS of H 1743-322 above 30 keV band. In the 30–80 keV band, we also found the presence of QPO at the same frequency as in the above mentioned two bands. The shape of the PDSs, as well as the values of the quality factor and the fractional rms amplitude of the QPOs in all the above mentioned three energy bands clearly indicate that the observed QPOs fall in the type C category (Casella et al. 2004, 2005; Motta et al. 2011). Moreover, the energy

38

Page 6 of 9

J. Astrophys. Astr. (2021)42:38

Table 1. Best-fit temporal parameters obtained from the PDS derived from LAXPC data. Component mQPO (Hz) WidthQPO (Hz) QQPO rmsQPO (%) mhar (Hz) Widthhar (Hz) Qhar rmshar (%) mbln1 (Hz) Widthbln1 (Hz) Qbln1 rmsbln1 (%) mbln2 (Hz) Widthbln2 (Hz) Qbln2 rmsbln2 (%) mblnðzeroÞ (Hz) WidthblnðzeroÞ (Hz) rmsblnðzeroÞ (%) v2 /dof

3–15 keV

15–30 keV

30–80 keV

0.5960:002 0:075þ0:006 0:005 7.90:6 11.90:3 1.200:01 0.140:03 8:8þ2:1 1:6 4.30:4 0.150:01 0:44þ0:09 0:07 0:35þ0:03 0:04 9.30:6 1:08þ0:12 0:14 1.90:2 0:563þ0:008 0:002 11:8þ0:9 1:1 0 (f) [ 7:9 9:7þ0:9 0:7 215.7/126

0.5980:003 0.0710:006 8:4þ0:7 0:6 10.00:3 1:22þ0:03 0:02 0:23þ0:12 0:07 5:2þ2:5 1:7 4.00:6 0.160:01 0:27þ0:06 0:05 0.600:06 6.10:6 – – – – 0 (f) 3:5þ0:7 0:5 11:9þ0:6 0:7 163.6/129

0.5970:006 0.070:01 8:4þ1:5 1:3 3.20:2 – – – – – – – – – – – – 0 (f) 0:81þ0:25 0:20 3.50:4 174.7/135

f-indicates the fixed parameters.

Figure 5. Evolution of the fractional rms of QPO with energy.

independent nature of the QPO and the upper harmonic indicates that the source was observed in the LHS during the outburst rise period with the luminosity in the X-ray regime capable enough to drive the transition of the source from the hard to soft state (Stiele & Yu (2016)). Similar kind of energy independent nature of type C QPOs along with upper harmonic was reported by Chand et al. (2020) for the

same outburst. However, the centroid frequency of the QPO and its upper harmonic as found by Chand et al. (2020) with the simultaneous XMM-Newton and NuSTAR observations are shifted to higher frequency side with respect to this particular five days earlier AstroSat observation by  0.4 Hz and  0.8 Hz, respectively. This indicates that the centroid frequency of the QPO and its upper harmonic were shifting to the higher frequency side as the time started to increase during the rising phase of the 2016 outburst. Similar kind of increase in the QPO frequency with the increase in time during the 1998 outburst rise phase was observed for the BHT XTE J1550-564 (see Dutta and Chakrabarti 2016). Since the QPOs are associated with the geometry of the corona, the shift in the centroid frequency of the QPOs may indicate certain geometrical change in the corona (see Titarchuk & Fiorito 2004). It is found that the size of the corona decreases when the frequency of the QPO moves up during an outburst rise phase and the vice versa during a declining phase (Chakrabarti et al. 2008, 2009; Debnath et al. 2010; Dutta & Chakrabarti 2016). The overall decreasing trend of fractional rms amplitude of the QPO with energy clearly depicted in Fig. 5 is similar to that found by

J. Astrophys. Astr. (2021)42:38

Figure 6. Frequency dependent time lag between 3–10 keV and 20–40 keV band.

Figure 7. Evolution of the averaged time lag with energy. The reference band is always the 3–5 keV band, corresponding to zero time lag point.

Stiele and Yu (2016) and Chand et al. (2020) for the 2014 and 2016 outbursts, respectively. From broadband spectral analysis (0.7–80 keV) of the simultaneous SXT and LAXPC spectral data, it is found that the X-ray continuum can be described by the multi-colored disk blackbody model diskbb modified by the galactic absorption tbabs, and the thermal Comptonization model nthcomp. The value of the column density (NH ) is found to be 1:94  0:07, which is consistent with those reported by previous workers using XMM-Newton, NuSTAR, Chandra, Suzaku and INTEGRAL observations (Parmar et al. 2003; Corbel et al. 2006; Miller et al. 2006; Shidatsu et al. 2014; Stiele & Yu 2016; Chand et al. 2020). The photon index (C) value of  1.67 indicates that the source was in the LHS during this particular AstroSat observation. The electron temperature (KTe ) in the coronal plasma is found to be[56.9 keV, which

Page 7 of 9

38

is quite higher relative to that reported by Stiele and Yu (2016) during the 2014 outburst. The disk temperature (KTin  1:2 keV) is also found to be higher compared to the 2014 outburst (see Stiele & Yu 2016). Higher value of disk temperature is also reported by Chand et al. (2020) for the 2016 outburst, which may be due to the irradiation of the accretion disk by the high energetic coronal X-rays (Gierlin´ski et al. 2009). We could not detect the presence of the iron line as reported by Chand et al. (2020) with XMM-Newton and NuSTAR observations, which may be due to the limitation of the SXT data up to 6 keV and large statistical uncertainties in the LAXPC data. It is clear from Table 2 that the disk fraction (unabsorbed (Fdisk =Ftotal )) in 0.7–80 keV is found to be  3%, which is consistent with the value reported by Chand et al. (2020). Lower value of disk fraction indicates that the flux from the Comptonizing component has the maximum contribution to the total source flux. As in Fig. 6, a hard lag of 21:4  2:4 ms was detected at the frequency range of  1–5 Hz between 3–10 keV and 20–40 keV bands. However, the lag amplitude is found to be weaker and detected at higher frequency as compared to that obtained by Chand et al. (2020) using the XMM-Newton observations. Higher frequency lag observed in the AstroSat data indicates the short term variability in the source. Morever, it is clear from Figures 2–4 and 6 that the QPO does not contribute significant power in the frequency range, where the time lag is detected. This indicates that the lag is not being triggered by the QPO rather it may arise due to the propagation of fluctuations in the mass accretion rate from outer part of the accretion disk to the inner hot regions (Lyubarskii 1997). This is also confirmed by the log-linear trend of the averaged time lag with energy (see Fig. 7). Similar trend was observed with XMM-Newton up to 10 keV by Chand et al. (2020) but this trend is extended to higher energies up to 60 keV using AstroSat. The absolute log-linear trend between the averaged time lag and energy may also rule out the possibility of hard lag, arising due to thermal Comptonization (Uttley et al. 2011, 2014; De Marco & Ponti 2016). However, a soft X-ray lag was found during the 2008 and 2014 failed outbursts at the Eddington-scaled luminosity of  0.004 in the 3–10 keV band, when the source was in hard state (De Marco et al. 2015; De Marco & Ponti 2016). A similar value of the Eddington-scaled luminosity in the 3–10 keV band was reported by Chand et al. (2020) using the XMM-Newton and NuSTAR observations taken during the 2016 successful outburst. Here, we also

38

Page 8 of 9

J. Astrophys. Astr. (2021)42:38

successful outburst in 2016 may be driven by the physical phenomena other than the source luminosity.

Acknowledgements

Figure 8. The 0.7–80 keV broadband X-ray energy continuum with the best fit model constant *tbabs(diskbb?nthcomp). The black circles represent the SXT spectral data, whereas the blue ones are for the LAXPC spectral data. Table 2. Best-fit spectral parameters for the joint fitting of SXT and LAXPC spectral data in 0:7ndash; 80 keV band. Component

Parameter

TBabs Diskbb

NH KTin (keV) Norm C KTe (keV) norm Ftotal ð 109 Þ Fdisk ð 1010 Þ v2 /dof

NthComp

Value 1.940:07 1:2þ0:1 0:2 3:8þ1:4 1:1 1.670:02 [ 56:9 0:11þ0:02 0:01 4:30þ0:06 0:04 1.30:3 204.5/248

F-unabsorbed flux derived in the 0.7–80 keV band.

derived the Eddington-scaled luminosity for this particular AstroSat observation by assuming the distance to the source and mass of the black hole as in De Marco and Ponti (2016). The estimated Eddingtonscaled luminosity appears to be 0:005  0:002, which is similar to the values obtained by De Marco & Ponti (2016) and Chand et al. (2020). This further confirms the fact that the change in the lag properties between the failed outbursts in 2008 and 2014 and the

We thank the anonymous referee for useful comments that have improved the quality of the paper. The authors acknowledge the financial support of ISRO under AstroSat archival Data utilization program (No: DS-2B-13013(2)/8/2019-Sec.2). This publication uses data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). This work has used the data from SXT and LAXPC instruments onboard AstroSat. LAXPC data were processed by the Payload Operation Center (POC) at TIFR, Mumbai. This work has been performed utilizing the calibration data-bases and auxiliary analysis tools developed, maintained and distributed by the AstroSat-SXT team with members from various institutions in India and abroad, and the SXT POC at the TIFR, Mumbai (https://www.tifr.res.in/*astrosat_sxt/index.html). SXT data were processed and verified by the SXT POC. This research has also used the data from MAXI space telescope provided by RIKEN, JAXA and the MAXI team. VKA thanks GH, SAG; DD, PDMSA and Director, URSC for encouragement and continuous support to carry out this research. P.T. expresses his sincere thanks to the Inter-University Centre for Astronomy and Astrophysics (IUCAA), Pune, India, for granting supports through the IUCAA associateship program. S.C. is also very much grateful to IUCAA, Pune, India, for providing support and local hospitality during his frequent visits.

References Agrawal P. C., Yadav J. S., Antia H. M. et al. 2017, J Astrophys Astr., 38, 30 Alam M. S., Dewangan G. C., Belloni T., Mukherjee D., Jhingan S. 2014, MNRAS, 445, 4259 Altamirano D., Strohmayer T. 2012, ApJL, 754, L23 Antia H. M., Yadav J. S., Agrawal P. C. et al. 2017, ApJS, 231, 10 Are´valo P., Uttley P. 2006, MNRAS, 367, 801 Belloni T., Hasinger G. 1990, A&A, 230, 103 Belloni T., Homan J., Casella P. et al. 2005, A&A, 440, 207 Belloni T. M. 2010, in Belloni T., ed., Lecture Notes in Physics, Springer, Berlin, Volume 794, p. 53 Capitanio F., Belloni T., Del Santo M., Ubertini P. 2009, MNRAS, 398, 1194

J. Astrophys. Astr. (2021)42:38 Casella P., Belloni T., Homan J., Stella L. 2004, A&A, 426, 587 Casella P., Belloni T., Stella L. 2005, ApJ, 629, 403 Chakrabarti S. K., Debnath D., Nandi A. et al. 2008, A&A, 489, L41 Chakrabarti S. K., Dutta B. G., Pal P. S. 2009, MNRAS, 394, 1463C Chand S., Agrawal V. K., Dewangan G. C., Tripathi P., Thakur P. 2020, ApJ, 893, 142 Corbel S., Tomsick J. A., Kaaret P. 2006, ApJ, 636, 971 Debnath D., Chakrabarti S. K., Nandi A. 2010, A&A, 520A, 98D De Marco B., Ponti G., Cappi M. et al. 2013, MNRAS, 431, 2441 De Marco B., Ponti G. 2016, ApJ, 826, 70 De Marco B., Ponti G., Munoz-Darias T., Nandra K. 2015, ApJ, 814, 50 De Marco B., Ponti G., Petrucci P. O. et al. 2017, MNRAS, 471, 1475 Dutta B. G., Chakrabarti S. K. 2016, ApJ, 828, 101 Fender R. P., Homan J., Belloni T. M. 2009, MNRAS, 396, 1370 Gierlin´ski M., Newton J. 2006, MNRAS, 370, 837 Gierlin´ski M., Done C., Page K. 2009, MNRAS, 392, 1106 Grinberg V., Pottschmidt K., Bo¨ck M. et al. 2014, A&A, 565, 1 Homan J., Belloni T. 2005, Ap&SS, 300, 107 Houck J. C., Denicola L. A. 2000, in Manset N., Veillet C., Crabtree D., eds, ASP Conf. Ser. 216, Astronomical Data Analysis Software and Systems IX, ASP, San Francisco, CA, p. 591 Ingram A., van der Klis M., Middleton M., Altamirano D. 2017, MNRAS, 464, 2979 Kaluzienski L. J., Holt S. S. 1977, IAUC, 3099, 3 Kara E. et al., 2014, MNRAS, 445, 56 Kara E., Steiner J. F., Fabian A. C. et al. 2019, Nature, 565, 198 Leahy D. A., Elsner R. F., Weisskopf M. C. 1983, ApJ, 272, 256 Lyubarskii Yu. E. 1997, MNRAS, 292, 679 Miller J. M., Raymond J., Homan J. et al. 2006, ApJ, 646, 394 Mitsuda K., Inoue H., Koyama K. et al. 1984, PASJ, 36, 741 Miyamoto S., Kitamoto S., Mitsuda K., Dotani T. 1988, Nature, 336, 450

Page 9 of 9

38

Molla A. A., Chakrabarti S. K., Debnath D., Mondal S. 2017, ApJ, 834, 88 Motta S., Mun˜oz-Darias T., Casella P., Belloni T., Homan J. 2011, MNRAS, 418, 2292 Nowak M. A., Vaughan B. A., Wilms J., Dove J. B., Begelman M. C. 1999a, ApJ, 510, 874 Nowak M. A., Wilms J., Dove J. B. 1999b, ApJ, 517, 355 Page C. G., Bennetts A. J., Ricketts M. J. 1981, SSRv, 30, 369 Parmar A. N., Kuulkers E., Oosterbroek T. et al. 2003, A&A, 411, L421 Plant D. S., Fender R. P., Ponti G., Munoz-Darias T., Coriat M. 2015, A&A, 573, 120 Remillard R. A., McClintock J. E., Orosz J. A., Levine A. M. 2006, ApJ, 637, 1002 Revnivtsev M. 2003, A&A, 410, 865 Shidatsu M., Negoro H., Nakahira S. et al. 2012, ATel, 4419, 1 Shidatsu M., Ueda Y., Yamada S. et al. 2014, ApJ, 789, 100 Singh K. P., Stewart G. C., Chandra S. et al. 2016, Proc. SPIE, 9905, 99051E Singh K. P., Stewart G. C., Westergaard N. J. et al. 2017, J Astrophys Astr., 38, 29 Sriram K., Agrawal V. K., Rao A. R. 2009, RAA, 9, 901 Steiner F. J., McClintock J. E., Reid M. J. 2012, ApJ, 745, 7 Stiele H., Yu W. 2016, MNRAS, 460, 1946 Tanaka Y., Shibazaki N. 1996, ARA&A, 34, 607 Titarchuk L., Fiorito R. 2004, ApJ, 612, 988 Uttley P., Cackett E. M., Fabian A. C., Kara E., Wilkins D. R. 2014, A&ARv, 22, 72 Uttley P., Wilkinson T., Cassatella P. et al. 2011, MNRAS, 414, 60 Vaughan B. A., Nowak M. A. 1997, ApJ, 474, L43 Verner D. A., Ferland G. J., Korista K. T., Yakovlev D. G. 1996, ApJ, 465, 487 Wilms J., Allen A., McCray R. 2000, ApJ, 542, 914 Yadav J. S., Agrawal P. C., Antia H. M. et al. 2016a, Proc. SPIE, 9905, 99051D Yadav J. S., Misra R., Verdhan Chauhan J. et al. 2016b, ApJ, 833, 27 Zdziarski A. A., Johnson W. N., Magdziarz P. 1996, MNRAS, 283, 193 _Zycki P. T., Done C., Smith D. A. 1999, MNRAS, 309, 561

 Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:81 https://doi.org/10.1007/s12036-021-09719-1

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

Spectro-timing analysis of a highly variable narrow-line Seyfert 1 galaxy NGC 4748 with AstroSat and XMM-Newton MAIN PAL1,* , NEERAJ KUMARI2,3, P. KUSHWAHA4, K. P. SINGH5,

ALOK C. GUPTA4, SACHINDRA NAIK2, G. C. DEWANGAN6, P. TRIPATHI6, RATHIN ADHIKARI1, O. ADEGOKE7 and H. NANDAN8 1

Centre for Theoretical Physics, Jamia Millia Islamia University, New Delhi 110 025, India. Astronomy and Astrophysics Division, Physical Research Laboratory, Ahmedabad 380 009, India. 3 Indian Institute of Technology Gandhinagar, Gandhinagar 382 355, India. 4 Aryabhatta Research Institute of Observational Sciences, Manora Peak, Nainital 263 002, India. 5 Indian Institute of Science Education and Research, Mohali, Manauli P.O. 140 306, India. 6 Inter University Centre for Astronomy and Astrophysics, Pune 411 007, India. 7 Department of Astronomy, University of Geneva, Versoix 1290, Geneva, Switzerland. 8 Department of Physics, Gurukul Kangri University, Haridwar 249 404, India. *Corresponding Author. E-mail: [email protected] 2

MS received 7 November 2020; accepted 12 January 2021 Abstract. We present a detailed timing and spectral study of an extremely variable narrow-line Seyfert 1 galaxy NGC 4748 using observations in the year 2017 and 2014 performed with AstroSat and XMM-Newton, respectively. Both observations show extremely variable soft and hard X-ray emission that are correlated with each other. In the 2014 data set, the source retains its general behaviour of ‘‘softer when brighter’’ while the 2017 observation exhibits a ‘‘harder when brighter’’ nature. Such changing behaviour is rare in AGNs and is usually observed in the black hole binary systems. The ‘‘harder when brighter’’ is confirmed with the anticorrelation between the photon index and the 0.3–10 keV power-law flux. This suggests a possible change in the accretion mode from standard to the advection-dominated flow. Additionally, both the observations show soft X-ray excess below 2 keV over the power-law continuum. This excess was fitted with a single or multiple blackbody component(s). The origin of soft excess during the 2017 observation is likely due to the cool Comptonization as the photon index changes with time. On the other hand, the broad iron line and delayed UV emission during the 2014 observation strongly suggest that X-ray illumination onto the accretion disk and reflection and reprocessing play a significant role in this AGN. Keywords. Accretion—accretion discs—galaxies: active—galaxies: individual: NGC 4748—galaxies: nuclei—X-rays: galaxies.

1. Introduction Radio-quiet Active Galactic Nuclei (AGNs) are a subclass of AGNs that lack the radio jet emission. The AGNs are powered by accretion of material onto the central supermassive black hole (SMBH) through the accretion disk (Hoyle & Fowler 1963; Lynden-Bell 1969). These objects are found to be extremely This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

variable over a wide range of time scales from minutes to years (Gaskell & Klimek 2003; Bon et al. 2016). The variability properties such as amplitude and radiated power vary over different bands. The variability seen in the optical/Ultraviolet (UV) to X-ray bands has been studied extensively in several AGNs (Edelson et al. 1990; Ricci et al. 2011; Grupe et al. 2007). Highly variable X-ray emission is considered to be originated from the compact hot plasma while the UV/optical emission are considered to be originated from the accretion disk. Variations of

81

Page 2 of 13

various physical properties, i.e. luminosity, accretion rate, and derived parameters such as the spectral indices obtained from the analysis of UV/optical and X-ray emission have been useful to understand these bright objects (Shapovalova et al. 2019; Li et al. 2019; Gallo 2006; Pal et al. 2018). Sometimes, variations in the multi-waveband light curves, for example, in the X-ray to UV/optical bands, follow each other. The timing information of these changes suggest a common driving mechanism for their cause of origin. In several objects, the soft photons (i.e., soft X-ray and UV/optical photons) are observed before the hard X-ray photons. This is generally referred as the ‘‘hard delay’’ or ‘‘hard lag’’. The hard lag was observed first time in the binary black hole candidate GX 339-4 (Miyamoto et al. 1991, 1988). After the discovery of hard lag in GX 339-4, similar type of delay was reported in a Seyfert 1 galaxy NGC 7469 by Papadakis et al. (2001). Since then, these lags have been detected in a number of AGNs (Emmanoulopoulos et al. 2011; McHardy et al. 2004; Papadakis et al. 2019). This hard lag is generally interpreted in two ways: the light travel time resulting from single or multiple inverse Compton scattering in the hot plasma (Lobban et al. 2020; Alston et al. 2014) and time taken by the density fluctuations of the accretion flow in the disk towards the centre (Lobban et al. 2018). In general, the time taken in the propagation of density fluctuations is much longer than the light travel time. Contrary, when the soft photons are detected after the hard X-ray photons, this delay is termed as the ‘‘soft lag’’ or the‘‘reverberation lag’’ (Vincentelli et al. 2020; Kara et al. 2014; Zoghbi et al. 2011; Fabian et al. 2009). The correlated variations in the X-ray and the UV/ Optical emission have been seen in a few AGNs. The changes in the UV/optical emission are sometimes observed before the variations seen X-ray emission. This type of process is expected when the UV/optical emission from the accretion disk are inverse-Comptonized in the plasma to give the power-law X-ray continuum. The time delay in X-ray emission is measured as the light crossing time scale or light travel time (Are´valo et al. 2008; Adegoke et al. 2019). However, the fluctuations travel with the sound speed in the matter and thus hard lag has been seen over days to months and years time scales (Marshall et al. 2008; Mehdipour et al. 2011). Sometimes, the variations in X-ray emission lead the changes in the UV/Optical emission. This is expected when X-ray emission from the hot plasma illuminates the accretion

J. Astrophys. Astr.

(2021) 42:81

disk and get absorbed, it is re-radiated at longer wavelengths for example, in the UV/optical bands. This phenomenon is called X-ray reprocessing. This type of phenomenon has been observed in several different types of AGNs (McHardy et al. 2014; Pal et al. 2017; Pal & Naik 2018; Naik et al. 2019; Herna´ndez Santisteban et al. 2020; Pal et al. 2020). Moderately correlated variations in X-ray and UV/ Optical bands have also been reported and explored. There are several cases where the UV/optical emission and X-ray emission are found to be uncorrelated. These erratic flux variations in the UV/optical emission may be associated to the local anisotropic magnetic origin in the disk or the complex absorption/ extinction present along the line of sight or the multiple mechanisms at the same time (Pawar et al. 2017; Nandra et al. 1998; Gaskell 2008; Maoz et al. 2002). In such cases, the understanding of spectral energy distribution (SED) plays a crucial role. Though, the SED in X-ray to UV/optical band is complex due to the mixture of intriguing spectral components. The main components are: (i) power-law continuum obtained from the inverse Comptonization of seed photons from the disk, (ii) the UV/optical continuum originated from the accretion disk, (iii) soft X-ray excess over the power-law continuum below  2 keV. The origin of the soft X-ray excess is not well understood to date. The possible explanation for this component is associated either with the cool Comptonization or the blurred reflection from partially ionized accretion disk (Done et al. 2012; Crummy et al. 2006; Pal et al. 2016; Dewangan et al. 2007). NGC 4748 is a narrow-line Seyfert 1 galaxy (Bentz et al. 2009) located at a redshift of z ¼ 0:014. It harbours one of the lowest mass SMBH (MBH ¼ 6 2:6þ1:6 1  10 M ) among AGNs which makes it interesting and important in terms of its observed variability time scales. This nearby AGN has been found to be highly variable on short and long timescale as seen by XMM-Newton and Swift missions (Vasylenko 2018; Fedorova 2017). Fedorova (2017) used XMM-Newton EPIC-pn data to investigate the short-term variability and a  10 ks periodicity was found in the 68 ks long stare time. Using this periodicity, the upper limit on the mass of the blackhole was estimated to be MBH  6  107 M . Vasylenko (2018) investigated the origin of the soft X-ray excess in the broad-band SED (0.5–500 keV range) using

J. Astrophys. Astr.

(2021) 42:81

multiple mission archival data. Both the cool Comptonization and blurred reflection models were tested and found equally probable. In this work, we explore the variable nature of this AGN as well as the origin of the soft X-ray excess using AstroSat and XMM-Newton data. In the very next section, we describe the data used in the present work and the reduction procedures. In Section 3, we explore both long-term and short-term variability properties. Spectral analysis is presented in Section 4. Finally, we discuss our results in Section 5. In this paper, we used the cosmological parameters H0 ¼ 67:04 km s1 Mpc1 , Xm ¼ 0:3183 and XK ¼ 0:68171 to calculate the distance.

2. Observational data and its reduction We observed (PI: Main Pal) NGC 4748 with the Soft X-ray Telescope (SXT) onboard first Indian multiwavelength space observatory AstroSat (Singh et al. 2014; Agrawal 2006). The SXT consists of shells of conical mirrors that focus the X-ray photons in the 0.3–8 keV band onto a CCD detector. The field-ofview of the SXT is 40 arcmin. The effective area of the telescope is 90 cm2 at 1.5 keV. The energy resolution of the detector is 90 eV at 1.5 keV and 136 eV at 5.9 keV. The detailed description of the SXT instrument is given in Singh et al. (2016, 2017). Along with the SXT data, we also used publicly available archival data on NGC 4748 from the XMMNewton mission. Details of the observations used are listed in Table 1. Below, we describe the data reduction procedure followed in this work.

2.1 AstroSat/SXT In order to analyse the AstroSat/SXT data, we followed the standard analysis procedure as outlined in the AstroSat Science Support Cell webpage (ASSC2). First of all, we obtained Level 1 data from the AstroSat archive at the Indian Space Science Data Centre (ISSDC). We then run the sxtpipeline tool to get clean event files calibrated with the latest calibration database provided by the instrument team. We merged all the event files obtained from each orbit of observation into a single event file. Using the xselect task of FTOOLS, we extracted scientific 1

http://www.kempner.net/cosmic.php. http://astrosat-ssc.iucaa.in/.

2

Page 3 of 13

81

products from the merged event file. The source spectrum was extracted by selecting a circular region of 14 arc-minute radius centred at the source co-ordinates (see Fig. 1). The payload operation centre (POC) performed deeper observations to get good statistics for stable background spectrum of SXT instrument while background light curve extracted from our own observation shows some variations with lesser statistics. We therefore preferred background spectrum provided by the POC team over the background spectrum obtained from selected rectangular box near the source. The ARF was generated using latest merged event file and also corrected for the vignetting effect. To extract the source and background light curves, we checked the background light curve above 7 keV for any solar/proton flare and did not find any sharp change in the background light curve. We searched for a suitable region on the CCD of SXT to find the stable background. Fortunately, the observations of NGC 4748 were carried out in such a way that the image of the source is located at a marginally offcentered position on the CCD. We then estimated the radial profile of the source plus background down to 16.5 arcmin of a circular region. For background light curve, we used a rectangular region of length 27 arcmin and width 8 arcmin away from the source as shown in Fig. 1. The radial profiles of the selected source and background regions are shown in Fig. 2. Clearly, we found a stable background with marginal variations away from the source during the AstroSat observation of NGC 4748. This selected rectangular region was used for the extraction of only background light curves in different energy bands. We extracted light curves of background and source plus background in three energy ranges namely 0.6–2 keV, 2–7 keV and 0.6–7 keV bands using xselect task. We then obtained background subtracted light curves and then computed the hardness ratio for the source.

2.2 XMM-Newton NGC 4748 observation was performed with the European Photon Imaging Camera (EPIC) Stru¨der et al. (2001) and Optical Monitor (OM) (Mason et al. 2001) onboard the XMM-Newton observatory in January 2014. The Science Analysis System (SAS 15.0 ; Gabriel et al. 2004) software package was used for the data reduction. We extracted data from the EPIC-pn and OM using the standard procedure described in the ‘‘ABC Guide’’ of XMM-Newton. We processed the

81

Page 4 of 13

J. Astrophys. Astr.

(2021) 42:81

Table 1. Details of the observations used in the present work. Start and end times are quoted in UTC. Observatory AstroSat XMM-Newton*

Obs.-ID

Start time

End time

GTI (ks)

9000001226 0723100401

15 May 2017; 14:57:15 14 Jan. 2014; 07:35:21

17 May 2017; 14:27:36 15 Jan. 2014; 01:45:03

56.8 43.5

*Optical Monitor (OM) observes filters UVW1, UVM2 and U partly overlapping to X-ray, and X-rays were found to be flared.

count s1 . After removing the flaring duration from the data, we got  43 ks of useful data. We used the single and double events (pattern  4) for EPIC-pn and ignored the events present at the edges and on bad pixels (flag = 0). We selected a circular region of 50 arcsec centred at the source co-ordinates for the source spectrum and light curves. We extracted the background spectrum and light curves from a circular region of the same size located away from the source and free from any other source. The observation was not affected by pile-up effect significantly. We generated the response files such as redistribution matrices and auxiliary files using rmfgen and arfgen tasks, respectively. We grouped the spectra with a minimum of one count per bin with an oversampling factor of three using specgroup SAS tool. For the OM data, we extracted only UVW1 and UVM2 bands in fast mode using omfchain tool.

3. Temporal analysis

Figure 1. Image of NGC 4748 from AstroSat/SXT data showing a circular region with radius of 14 arcmin centered at the source position and a rectangular box selected for source and background regions, respectively, for the extraction of light curves and spectra.

raw data with the latest calibration using the epproc task and obtained the event files. The background light curve above 10 keV showed several solar/proton flares during the observation. We generated the good time intervals after removing the count rates above 1.1

Soft (i.e. below 2 keV) and hard (i.e. above 2 keV) Xray light curves obtained from the AstroSat/SXT and XMM-Newton/EPIC-pn data, binned at 1000 s, are shown in Fig. 3. The top and middle panels in left side show the background subtracted light curves obtained from the SXT data in the 0.6–2 keV and 2–7 keV bands, respectively. The hardness ratio i.e. the ratio between the 2–7 keV and 0.6–2 keV light curves is also shown in the left bottom panel of the figure. Variations based on binning time-interval and during the net exposure can be considered using ratio of the maximum to the minimum count rates/flux (Fmax =Fmin ). On visual inspection, we can see that the short-term variability on 1000 s time scale is by about a factor of three, whereas the long-term variability during the two days observation is found to be striking (by a factor of about thirteen) in the 0.6–2 keV band. In the 2–7 keV band, the changes on 1000 s time scale is found to be about a factor of four while the source is extremely variable (by a factor of forty five) during

J. Astrophys. Astr.

(2021) 42:81

Page 5 of 13

81

Counts/arcsecs 2

0.1

0.01

0

200

400 600 Average size (arcsecs)

800

1000

0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 0.1 0.08 0.06 0.04 0.02 0 0.8

14

0.6-2 keV

10 T3

T2

T1

2-7 keV

2-7 keV/0.6-2 keV

0.2

2-10 keV

0.13 HR

HR

0.4

8

6 1.5 1.4 1.3 1.2 1.1 1 0.9 0.8 0.14

0.6

0

0.3-2 keV

12

Count Rate (c/s)

Count Rate (c/s)

Figure 2. AstroSat/SXT data: Radial profiles for source plus background (green) and background (red) are shown here for selected circular and rectangular regions in Fig. 1, respectively.

0.12 0.11 0.1

0 2.5×10

4

4

7.5×10 10 Time (s)

5

1.5×10

5

2-10 keV/0.3-2 keV 0

10

4

4

2×10 3×10 Time (s)

4

4×10

4

5×10

4

Figure 3. 1000 s binned light curves in the soft and hard X-ray bands (top and middle panels) and hardness ratio (bottom panels) are shown here for left: AstroSat/SXT data (soft X-ray band: 0.6–2 keV, hard X-ray band: 2–7 keV) and right: XMM-Newton Epic-pn data (soft X-ray band: 0.3–2 keV, hard X-ray band: 2–10 keV). The time-bin size for each light curve is 1000 seconds. The vertical lines divide entire SXT light curve into three segments (T1, T2 and T3) which are used for time-resolved spectroscopy analysis.

the observation of about two days long used in this study. Different variability pattern suggests the presence of different spectral components in both the bands. Similarly, the hardness ratio shows large variation within the AstroSat observation and these changes seem interesting. The 2014 observation of NGC 4748 performed with EPIC-pn also showed variation in the soft X-ray (0.3–2 keV) and hard X-ray (2–10 keV) light curves (right panels of Fig. 3). The observed changes on binning time scale (1000 s) are about a factor of two

and the variations within the net exposure are found to be more than a factor of two in the 0.3–2 keV band. In the hard band, we found a similar level of variability. Both the light curves seem correlated to each other. However, the variations in the hardness ratio show a nice pattern but opposite to that observed in the soft and hard X-ray bands. Therefore, the hardness ratio and its relation to the hard X-ray band would be crucial to investigate the cause of their variation. The rapid variations present in each light curve of different energy bands appear to show a similar trend

81

Page 6 of 13

J. Astrophys. Astr.

during both the observations. In order to understand the correlation between the soft and hard X-ray bands, we used the Pearson’s correlation coefficient ‘qxy ’. However, measuring ‘qxy ’ is not enough to relate the variables as the coefficient ‘qxy ’ does not take into account the two-dimensional normal distribution of both the variables. The probability p(qxy ; n) is a useful quantity to determine the significance of correlation. This quantity determines how unlikely a given correlation coefficient qxy would be to yield no association between the variables of a sample. Small values (i.e.  0.05) states that the observed variables are likely correlated. We determined the inter-band correlation coefficient and the probability (p-value) using the Python routine of the above technique. We found the correlation coefficient qxy ¼ 0:36 and probability p ¼ 0:002 for soft and hard X-ray bands for the SXT data (left panel of Fig. 4). For EPIC-pn data, the corresponding values are qxy ¼ 0:92 and p ¼ 3:5  1014 for soft and hard X-ray bands and it is shown in the right panel of Fig. 4. From this analysis, it is clear that there exists some relationship between the soft and hard X-ray bands. The observed hardness ratios are extremely variable during both 2014 and 2017 observations. The variations seen in the hardness ratio, sometimes, follow the hard X-ray band. We estimated the Pearson correlation coefficient qxy between the hardness ratio and hard X-ray band. We found qxy ¼ 0:75 and p ¼ 1:4  1013 for the hardness ratio and hard X-ray band for the SXT data, while for the EPIC-pn data, corresponding values are qxy ¼ 0:55 and p ¼

0.48

=

4. Spectral analysis 4.1 Long term variability: Average spectrum of 2014 and 2017 observations We employed C-statistic for finding the best fit model and to estimate 1  r range for each parameter. We used the SXT data in the 0.6–6 keV range for spectral analysis. Below 0.6 keV, the response of the detector is uncertain, whereas above 6 keV, the data is dominated by background. In the beginning, the powerlaw continuum model powerlaw was used to fit the data in 2–6 keV band. We then modified the model for the Galactic absorption by using tbabs model and fixed it to the Galactic column density NH ¼ 3:6  1020 cm2 (HI4PI Collaboration et al. 2016). The bestfit power-law photon index in the 2–6 keV band was found to be C ¼ 2:08  0:15 with the fit statistics C=dof ¼ 412:8=387, where dof is the degree of freedom. Then the fitted band was extrapolated down to 0.6

1.19 67

χ2ν dof

14 13

0.36 0.30 0.24 0.18 0.12

=

2.9 32

12 11 10 9 8 7

0.06 0.00

7:9  104 , respectively. The correlations are shown in the left and right panels of Fig. 5 for SXT and EPIC-pn, respectively. Thus, highly variable soft and hard X-ray bands appear to be strongly correlated, however the hardness ratio with hard X-ray band is negatively correlated for EPIC-pn data (right panel of Fig. 5). The linear fit to the observed correlation also supports that the 2017 observation is positively correlated while the 2014 data exhibits negative correlation. To understand this contrary behaviour, we performed a systematic spectral analysis.

Soft X-ray (c s−1)

Soft X-ray (c s−1)

0.42

χ2ν dof

(2021) 42:81

y = (0.014 ± 0.013) + (4.5 ± 0.4)x 0.00

0.05

0.10 −1

Hard X-ray (c s )

0.15

y = (−2.8 ± 0.9) + (11.0 ± 0.8)x

6 0.8

0.9

1.0

1.1

1.2

1.3

Hard X-ray (c s−1)

1.4

1.5

1.6

Figure 4. The correlation between the soft and hard X-ray light curves are presented (left panel: SXT data and right panel: EPIC-pn data). The red line in each panel shows linear-fit function.

J. Astrophys. Astr.

(2021) 42:81

Page 7 of 13 χ2ν dof

HR (2-7 keV/0.6-2 keV)

1.2 1.0

=

0.6 0.4 0.2 0.0

y = (0.039 ± 0.015) + (4.3 ± 0.4)x −0.025 0.000

HR (2-10 keV/0.3-2 keV)

=

0.025

0.050

0.075

0.100

Hard X-ray (c s−1)

0.125

0.150

0.175

χ2ν dof

0.144

4.0 32

HR (2-10 keV/0.3-2 keV)

χ2ν dof

0.138

0.3 67

0.8

−0.2

0.144

81

0.132 0.126 0.120 0.114 0.108 0.102

0.138

=

3.0 32

y = 0.144(±0.007) − 0.0024(±0.0006)x

0.132 0.126 0.120 0.114 0.108 0.102 0.096

0.096

y = (0.13 ± 0.01) + (−0.014 ± 0.008)x

0.090 0.8

0.9

1.0

1.1

1.2

1.3 −1

1.4

1.5

1.6

Hard X-ray (c s )

0.090 8

10

12

X-ray (c s−1)

14

16

Figure 5. Correlation between hard X-ray band and hardness ratio for SXT data (top panel) and EPIC-pn data (bottomleft panel). Bottom-right: correlation between hardness ratio and X-ray band (0.3–10 keV) of EPIC-pn data. The linear best-fit equation is shown with red colour in each panels of the figure.

keV and a marginal soft excess was found over the power-law continuum (see left panel of Fig. 6). To model this soft excess, a redshifted blackbody model zbbody was used. The best-fit blackbody temperature was found to be kTbb  120 eV. After freeing the index parameter for the best-fit model tbabs (powerlaw ?zbbody), the fit statistic resulted in C=dof ¼ 596:4=526. The best-fit model, data, and residuals are shown in the left panel of Fig. 7 and the best-fit parameters are listed in Table 2. We followed similar procedure to fit the EPIC-pn data. First, we fitted the 2–5 keV band by absorbed power-law model and found the power-law photon index to be 2:21  0:03 which is consistent with the photon index obtained from the SXT data (within errors). We then extrapolated down to 0.3 keV and also up to 10 keV. This shows the presence of a very strong soft excess below 2 keV and a broad emission line near 6 keV as shown in the right panel of Fig. 6. The observed soft excess was fitted with two blackbody components and the broad emission line was

modelled with a Gaussian function. The broad emission line was centred at  6.9 keV and the width of the line was found to be greater than 0.8 keV. The model tbabs (blackbody1?blackbody2? powerlaw) without any significant residuals resulted in C=dof ¼ 189:1=171. The best-fit model, data and residuals are shown in the right panel of Fig. 7 and the best-fit parameters are listed in Table 2. As a strong soft X-ray excess is detected in the EPIC-pn data, the origin of this component would be important to investigate. To understand this, we plotted UVW1 and UVM2 data along with the Xray data in 0.3–2 keV and 2–10 keV ranges, as shown in Fig. 8. Here, we used a time binsize of 500 s. It should however be noted that large time gaps in the data sets shown in Fig. 8 prohibit quantitative estimation of any time lag between Xrays and UV emission. Nonetheless, the figure hints that the UV data sets appear to be delayed by  10 ks with respect to the X-ray emission. Such delay is

Page 8 of 13

J. Astrophys. Astr.

(2021) 42:81

1.5 1

1

Data/Model

Data/Model

1.5

2

81

1

2 Energy (keV)

5

0.5

1

2 Energy (keV)

5

10

Figure 6. Left: Absorbed powerlaw model fitted in the 2–6 keV range and then extrapolated down to 0.6 keV for SXT data. A marginal excess is seen below 1 keV. Right: Absorbed powerlaw model in the 2–5 keV band and then extrapolated down to 0.3 keV and up to 10 keV bands for EPIC-pn data. A strong soft X-ray excess below 2 keV and a broad emission line near 6 keV are found.

expected in the X-ray reprocessing scenario and thus favours the blurred reflection for the origin of the soft excess. 4.2 Short term variability: Time-resolved spectroscopy A clear positive trend in the hardness ratio and hard X-ray light curve from the SXT data is the first time finding in this AGN. Normally, this type of AGNs show negative correlation (Constantin et al. 2009). The observed positive correlation between the hard Xray band and the hardness ratio is unique. This can be verified if we find some variations for power-law

photon index with its flux. This can be done by using the time resolved spectroscopy technique. For this, we divided the entire SXT exposure into three parts with exposures of  0–50 ks,  50–100 ks and  100– 170.5 ks (see vertical lines in the left graph of Fig. 3). We then obtained the source spectra for all the three time windows and used the background spectrum and response matrices provided by the POC team. The effective area files (ARF) generated for the spectral fitting were applied here. We jointly fitted all three spectra using the best fit model obtained from the spectral modeling. We derived the parameters and unabsorbed fluxes in various bands as listed in Table 2. From Table 2, one can find a relation between power-law photon index (C) and power-law

1

2 Energy (keV)

5

keV (Photons cm−2 s−1 keV−1)

1

10−4 1.5

0.01

Data/Model

keV (Photons cm−2 s−1 keV−1)

10−3

ratio

0.01

1.2

10−3

10−4 1.4

1 0.8 0.5

1

2 Energy (keV)

5

10

Figure 7. Best-fit model (solid line) consisting of an absorbed power-law (dotted line) and blackbody(s) (dotted curve) components, data and residuals (data/model) for AstroSat/SXT data in the 0.6–6 keV (left panel) and XMM-Newton/EPICpn data in the 0.3–10 keV band (right panel).

J. Astrophys. Astr.

(2021) 42:81

Page 9 of 13

81

Table 2. Best-fit parameters of NGC 4748 obtained from the spectral analysis of AstroSat/SXT and XMM-Newton EPICpn data. Model Tbabs Power-law

Blackbody-1

Blackbody-2

Gaussian line

Stats

Model comp.

SXT

EPIC-pn

T1

T2

T3

NH (1020 cm2 ) C Norm. ð103 Þ F0:32 keV F0:310 keV kTbb (in eV) Norm. ð105 Þ F0:32 keV kTbb (in eV) Norm. ð105 Þ F0:32 keV E (keV) Width (keV) Norm. ð105 Þ C=dof

3.6 (fixed) 2:07þ0:06 0:07 2:81  0:13 8:7  0:6 15:24  0:36 120:6þ10:1 10:3 6:0þ1:2 1:0 3:5  0:5 – – – – – – 596.4/526

3.6 (fixed) 2:15þ0:02 0:03 4:04þ0:08 0:13 12:8þ0:6 0:4 21:1þ0:45 0:30 83:5  2:5 10:3þ0:8 0:5 3:95þ0:37 0:30 178:8þ11:2 9:2 3:0þ0:4 0:3 2:15  0:20 6:88þ0:13 0:15  0:77 5:2þ0:6 1:3 190.5/171

3.6 (fixed) 1:96þ0:10 0:10 3:5  0:3 10:5  1:2 20:1  0:8 133:9þ13:7 14:2 8:2þ1:8 1:6 5:1  0:9 – – – – – – –

3.6 (fixed) 1:91þ0:08 0:09 3:3þ0:2 0:2 9:9  0:8 19:8  0:6 113:4  13:0 10:4þ3:7 2:4 5:7  1:1 – – – – – – 1282.7/1295

3.6 (fixed) 2:67þ0:10 0:10 1:55þ0:06 0:06 6:0  0:4 7:5  0:3 – – – – – – – – – –

Results from the time-resolved spectroscopy analysis of AstroSat/SXT are also listed. Flux is measured in 1012 ergs1 cm2 . ‘‘T’’ represents the time segment of long SXT light curve, which is used to extract the spectral parameters.

Count rate (counts s-1)

Short term variability 22 21 20 19 5.25 5

UVW1

UVM2

4.5 1.4 1.2 1 0.8 12 8 4 0

2-10 keV

0.3-2 keV

104

2×104

3×104 4×104 Time (s)

5×104

6×104

7×104

Figure 8. 500 s binned light curves in (i) UVW1 and UVM2 bands obtained from Optical Monitor, and (ii) soft X-ray and hard X-ray bands extracted from EPIC-pn observations of NGC 4748. The straight lines hint the echoes of X-ray reprocessing process in the UV bands.

flux in the 0.3–10 keV band. It is clear that the powerlaw photon index increases with the decrease in power-law continuum flux (see bottom panel of Fig. 9). The best-fit model, data and residuals are shown in the top graph of Fig. 9.

5. Summary and discussion We studied available long exposure X-ray observations (only 2) of NGC 4748 carried out in 2017 with AstroSat and in 2014 with XMM-Newton. Data

Data/Model

keV (Photons cm−2 s−1 keV−1)

81

Page 10 of 13

J. Astrophys. Astr.

0.01

•

10−3

10−4 2 1.5 1 0.5

•

0

1

2 Energy (keV)

5

•

2.70

Photon Index (Γ)

2.55 2.40

•

2.25 2.10 1.95 1.80 8

10

12

14

16

18

X-ray flux (10−12 ergs s−1 cm−2)

20

Figure 9. Top: Spectral fitting of data during all three segments of SXT exposures. The black, red and green colours represent the data and best-fit models for each segment. Dotted straight and sloppy lines are the power-law continuum and dotted curves show a blackbody model used to fit the soft X-ray excess. Bottom: An anti-correlation between photon index and the 0.3–10 keV power-law continuum flux is shown.

obtained from both the observations show extreme variations in the soft X-ray as well as the hard X-ray bands. Along with large changes in the soft and hard X-ray bands, we found a soft X-ray excess in both the observations. SXT and EPIC-pn spectra were fitted by using phenomenological models such as power-law continuum and thermal blackbody emission. We found the following results from our analysis: • A correlated variability of the soft and hard X-ray bands is seen. • The amplitude of variations in the hard X-ray band of SXT data is surprisingly higher, almost by a factor of forty-five which has never seen in this AGN. On the other hand, flux variations in the soft band in the SXT data varies by a factor of three and thirteen on short and long time

(2021) 42:81

scales, respectively. For EPIC-pn data, both the soft and hard X-ray flux show similar variations (by a factor of two). The changes seen in the soft and hard X-ray flux are positively correlated for both while positive and negative correlation between hard X-ray flux and hardness ratio were found for the 2017 and 2014 observations, respectively. The timeresolved spectroscopy confirms a ‘‘harder when brighter’’ nature observed in 2017 data set. Both the observations show the presence of soft X-ray excess below  2 keV over the powerlaw continuum. The observed soft excess in 2017 is well described by a single blackbody while the 2014 observation requires two blackbody components to fit the soft X-ray excess. Average power-law continuum slopes are consistent (within error-bars) for both the observations, suggesting that the Comptonization may not be a dominant process during both the observations. However, a variation of photon index within the 2017 observation can be seen (see Fig. 9). On the other hand, the 2014 data shows the presence of broad iron-Ka emission line and hints a delayed UV emission. These findings do not favour a common origin of soft excess for both the data set.

The observed variations in the soft and hard X-ray bands and their correlations are interesting to understand the characteristics of NGC 4748. The variations seen in 2017 data set is different in both the soft X-ray and hard X-ray bands while the 2014 data shows similar changes in both the bands. The significant correlations in soft and hard X-ray bands suggest a relationship between these bands. Further, the positive offset in 2017 data suggests the presence of a slowly variable component in the soft X-ray emission while negative offset in the 2014 data set (loosely speaking) may infer a possible absorption in the soft band (see Fig. 4). However, we do not see any absorption signature in the soft X-ray band (see the residuals in the right panel of Fig. 7). The negative offset may be due to the presence of slowly variable spectral component such as broad Fe-K line along with a highly variable power-law continuum. Fabian et al. (2002) studied a Seyfert 1 galaxy MCG 6-30-15 using about four days continuous XMM-Newton observation. They found that the skewed broad Fe-K emission line near 6 keV showed a minimum variability compared to any other spectral components in the X-ray band. As we

J. Astrophys. Astr.

(2021) 42:81

detected a very broad iron line near 6 keV whose width is found to be greater than 0.8 keV, the negative offset is possibly caused due to a slowly varying broad iron emission line. Such a low variable broad iron emission line is to be originated by the fluorescence phenomena followed by photo-absorption of X-ray power-law emission in the partially ionized accretion disk near the supermassive black hole (Miniutti & Fabian 2004; Fabian et al. 2002). Power-law continuum is believed to be the dominant spectral component in the energy band above 2 keV. Sometimes, the variations seen in hardness ratio depict the presence of the spectral features, i.e., neutral absorption along the line-of-sight or the changes in the geometry/size of the hot corona or the changes in the accretion flow state. The observed hardness ratio is highly variable and the correlation with hard X-ray flux appears to be strong in the 2017 data set while 2014 data shows moderate anti-correlation between the hardness ratio and hard X-ray band flux (see Fig. 5). The anti-correlation appears stronger when we used entire X-ray band with hardness ratio as shown in the bottom right panel of Fig. 5. Thus, the 2017 data set shows a ‘‘harder when brighter’’ behaviour and the 2014 data set exhibits the ‘‘softer when brighter’’ state of this AGN. From the Table 2, the power-law flux in the 0.3–10 keV band varies  30% from 2014 to 2017 which appears to be a significant change likely associated with a change in the accretion rate. This type of a behaviour of NGC 4748 is interesting as this AGN has gone through likely a state transition from high flux state to low flux state during 2014-2017. Such state transition is analogous to the black hole binary systems (Wu & Gu 2008). Wu & Gu (2008) investigated the spectral evolution of six X-ray binary systems and found a critical transition point at Eddington accretion rate 0.01 and photon index C  1:5. These accretion states are normally known as low/hard and high/soft about this transition point and spectral states are completely different around such transition point. Typically in X-ray band, the low/hard state is described by a power-law continuum and the high/soft state is dominated by the thermal disk component along with power-law emission (Remillard & McClintock 2006). Similarly, Gu and Cao (2009) found a significant anti-correlation between the photon index and Eddington ratio for a number of low luminosity AGNs which was opposite to that seen in most of the AGNs. Recently, a similar study was performed using Swift data by Connolly et al. (2016) and they found a

Page 11 of 13

81

‘‘harder when brighter’’ trend in the low luminosity AGNs also. In the low luminosity AGNs, this is expected due to the radiatively inefficient accretion flow such as the advection dominated accretion flow (Esin et al. 1997). Shemmer et al. (2006) studied a sample of radio-quiet AGNs and found that the powerlaw photon index depends primarily on the accretion rate expected in the geometrically thin and optically thick standard disk. Thus, normally radio-quiet AGNs show the positive correlation between C and Eddington ratio being in ‘‘softer when brighter’’ nature. Here, NGC 4748, being a radio quiet AGN, showed both ‘‘harder when brighter’’ and ‘‘softer when brighter’’ nature which seems rare in AGNs. Therefore, this AGN exhibits two states within a three year span likely with two different spectral states. In the ‘‘harder when brighter’’ state in 2017 data, the observed soft X-ray excess was described by a single blackbody with inner disk temperature of about 120 eV. This huge temperature is not acceptable theoretically and can be achieved by a physical phenomenon such as Comptonization process. Done et al. (2012) described the soft X-ray excess as a result of cool Comptonization process. In this process, the seed photons from the disk are inverse Compton scattered in the optically thick (optical depth [1) and cool plasma of about 0.2 keV. This explanation is supported as we obtained a varying photon index during the 2017 observation. To understand this behaviour in more detail, we have proposed new long AstroSat observation which will be performed simultaneously in UV and X-ray from the onboard UVIT and SXT instruments. The strong soft X-ray excess observed in the 2014 data, on the other hand, may be associated with a different mechanism as it belongs to the ‘‘softer when brighter’’ state of the AGN, a state similar to high/soft behavior in high accretion flow regime. The soft excess was modelled by two blackbody components with temperatures of  80 eV and  180 eV. To investigate its origin in the 2014 data, we found possibly a delayed UV emission of about 10 ks in the UVW1 (  2910 Angstrom) and UVM2 (  2310 Angstrom) bands (see Fig. 8). This delayed emission could be due to the X-ray reprocessing in the accretion disk. To confirm this, we used temperature profile obtained from the standard accretion disk (Shakura & Sunyaev 1973) to estimate the light travel time for X-ray emission. Using its known mass 2:6  106 from literature (Bentz et al. 2009) and

81

Page 12 of 13

acceptable Eddington ratio (  0.01), we obtained a time delay of  10 ks for UVW1 band. This is consistent with the time-delay observed in UVW1 band within an observational cadence. The trend and dips seen in the X-ray bands near 36 ks seem to appear in the UVM2 band near 45 ks shown in Fig. 8. These trend and dips support the scenario for the X-ray reprocessing on the outer disk. In addition, a broad Fe-K emission line seen around 6 keV suggests that this is originated from the vicinity of the SMBH and is blurred likely due to relativistic effects. The presence of strong and broad iron line near 6 keV and the delayed UV emission suggests that the origin of the soft X-ray excess is likely due to blurred reflection as a result of the strong light bending close to SMBH. Acknowledgements We are thankful to the anonymous referee for his/her useful comments which have improved the manuscript significantly. MP is thankful for financial support from the UGC, India through the Dr. D. S. Kothari PostDoctoral Fellowship (DSKPDF) program (Grant No. BSR/2017-2018/PH/0111). PK acknowledges support from the Aryabhatta Post-Doctoral Fellowship (APDF) grant (AO/A-PDF/770). This publication uses the data from AstroSat mission of the ISRO, archived at the Indian Space Science Data Centre (ISSDC). We thank members of the SXT team for their contribution to the development of the instrument and analysis software. We also acknowledge the contributions of the AstroSat project team at ISAC and IUCAA. This work has been performed utilising the calibration data-bases and auxiliary analysis tools developed, maintained and distributed by AstroSat/SXT team with members from various institutions in India and abroad. This research has also used data from the XMM-Newton, operated by the European Space Agency.

References Adegoke O., Dewangan G. C., Pawar P., Pal M. 2019, ApJL, 870, L13 Agrawal P. 2006, Advances in Space Research, 38, 2989, Spectra and Timing of Compact X-ray Binaries Alston W. N., Done C., Vaughan S. 2014, MNRAS, 439, 1548 Are´valo P., Uttley P., Kaspi S. et al. 2008, MNRAS, 389, 1479

J. Astrophys. Astr.

(2021) 42:81

Bentz M. C., Walsh J. L., Barth A. J. et al. 2009, ApJ, 705, 199 Bon E., Zucker S., Netzer H. et al. 2016, ApJS, 225, 29 Connolly S. D., McHardy I. M., Skipper C. J., Emmanoulopoulos D. 2016, MNRAS, 459, 3963 Constantin A., Green P., Aldcroft T. et al. 2009, ApJ, 705, 1336 Crummy J., Fabian A. C., Gallo L., Ross R. R. 2006, MNRAS, 365, 1067 Dewangan G. C., Griffiths R. E., Dasgupta S., Rao A. R. 2007, Astrophys. J., 671, 1284 Done C., Davis S. W., Jin C., Blaes O., Ward M. 2012, MNRAS, 420, 1848 Edelson R. A., Krolik J. H., Pike G. F. 1990, ApJ, 359, 86 Emmanoulopoulos D., McHardy I. M., Papadakis I. E. 2011, MNRAS, 416, L94 Esin A. A., McClintock J. E., Narayan R. 1997, ApJ, 489, 865 Fabian A. C., Vaughan S., Nandra K. et al. 2002, MNRAS, 335, L1 Fabian A. C., Zoghbi A., Ross R. R. et al. 2009, Nature, 459, 540 Fedorova E. 2017, Bulletin of Taras Shevchenko National University of Kyiv Astronomy, 56, 18 Gabriel C., Denby M., Fyfe D. J. et al. 2004, in Ochsenbein F., Allen M. G., Egret D., eds, Astronomical Society of the Pacific Conference Series, vol. 314, Astronomical Data Analysis Software and Systems (ADASS) XIII, 759 Gallo L. C. 2006, MNRAS, 368, 479 Gaskell C. M. 2008, in Revista Mexicana de Astronomia y Astrofisica Conference Series, vol. 32, Revista Mexicana de Astronomia y Astrofisica Conference Series, p. 1–11 Gaskell C. M., Klimek E. S. 2003, Astron. Astrophys. Trans., 22, 661 Grupe D., Komossa S., Gallo L. C. 2007, ApJL, 668, L111 Gu M., Cao X. 2009, MNRAS, 399, 349 Herna´ndez-Santisteban J. V., Edelson R., Horne K. et al. 2020, MNRAS, 498, 5399 HI4PI Collaboration: Ben Bekhti N., Flo¨er L. et al. 2016, A&A, 594, A116 Hoyle F., Fowler W. A. 1963, Nature, 197, 533 Kara E., Cackett E. M., Fabian A. C., Reynolds C., Uttley P. 2014, MNRAS, 439, L26 Li Y.-R., Wang J.-M., Zhang Z.-X. et al. 2019, ApJS, 241, 33 Lobban A. P., Vaughan S., Pounds K., Reeves J. N. 2018, MNRAS, 476, 225 Lobban A. P., Turner T. J., Reeves J. N., Braito V., Miller L. 2020, MNRAS, 494, 5056 Lynden-Bell D. 1969, Nature, 223, 690 Maoz D., Markowitz A., Edelson R., Nandra K. 2002, Astron. J., 124, 1988 Marshall K., Ryle W. T., Miller H. R. 2008, ApJ, 677, 880 Mason K. O., Breeveld A., Much R. et al. 2001, A&A, 365, L36 McHardy I. M., Papadakis I. E., Uttley P., Page M. J., Mason K. O. 2004, MNRAS, 348, 783

J. Astrophys. Astr.

(2021) 42:81

McHardy I. M., Cameron D. T., Dwelly T. et al. 2014, MNRAS, 444, 1469 Mehdipour M., Branduardi-Raymont G., Kaastra J. S. et al. 2011, A&A, 534, A39 Miniutti G., Fabian A. C. 2004, MNRAS, 349, 1435 Miyamoto S., Kitamoto S., Mitsuda K., Dotani T. 1988, Nature, 336, 450 Miyamoto S., Kimura K., Kitamoto S., Dotani T., Ebisawa K. 1991, ApJ, 383, 784 Naik S., Pal M., Kumari N. 2019, Bulletin de la Societe Royale des Sciences de Liege, 88, 143 Nandra K., Clavel J., Edelson R. A. et al. 1998, ApJ, 505, 594 Pal M., Naik S. 2018, MNRAS, 474, 5351 Pal M., Dewangan G. C., Misra R., Pawar P. K. 2016, MNRAS, 457, 875 Pal M., Dewangan G. C., Connolly S. D., Misra R. 2017, MNRAS, 466, 1777 Pal M., Dewangan G. C., Kembhavi A. K., Misra R., Naik S. 2018, MNRAS, 473, 3584 Pal M., Kushwaha P., Dewangan G. C., Pawar P. K. 2020, ApJ, 890, 47 Papadakis I. E., Nandra K., Kazanas D. 2001, ApJL, 554, L133 Papadakis I. E., Rigas A., Markowitz A., McHardy I. M. 2019, MNRAS, 485, 1454 Pawar P. K., Dewangan G. C., Papadakis I. E. et al. 2017, MNRAS, 472, 2823 Remillard R. A., McClintock J. E. 2006, ARA&A, 44, 49

Page 13 of 13

81

Ricci C., Paltani S., Soldi S., Courvoisier T. J. L. 2011, in Fast X-ray Timing and Spectroscopy at Extreme Count Rates (HTRS 2011), 67 Shakura N. I., Sunyaev R. A. 1973, A&A, 500, 33 Shapovalova A. I., Popovic´ L. Cˇ. et al. 2019, MNRAS, 485, 4790 Shemmer O., Brandt W. N., Netzer H., Maiolino R., Kaspi S. 2006, ApJL, 646, L29 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, in Takahashi T., den Herder J.-W. A., Bautz M., Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, vol. 9144, Space Telescopes and Instrumentation 2014: Ultraviolet to Gamma Ray, 91441S Singh K. P., Stewart G. C., Chandra S. et al. 2016, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, vol. 9905, Space Telescopes and Instrumentation 2016: Ultraviolet to Gamma Ray, 99051E Singh K. P., Stewart G. C., Westergaard N. J. et al. 2017, J. Astrophys. Astr., 38, 29 Stru¨der L., Briel U., Dennerl K. et al. 2001, A&A, 365, L18 Vasylenko A. A. 2018, ApSS, 363, 228 Vincentelli F. M., Mastroserio G., McHardy I., Ingram A., Pahari M. 2020, MNRAS, 492, 1135 Wu Q., Gu M. 2008, ApJ, 682, 212 Zoghbi A., Uttley P., Fabian A. C. 2011, MNRAS, 412, 59

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:51 https://doi.org/10.1007/s12036-021-09704-8

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

AstroSat view of the NLS1 galaxy Mrk 335 SAVITHRI H. EZHIKODE* , GULAB C. DEWANGAN and RANJEEV MISRA Inter-University Centre for Astronomy & Astrophysics, Post Bag 4, Ganeshkhind, Pune 411 007, India. *Corresponding Author. E-mail: [email protected] MS received 7 November 2020; accepted 24 December 2020 Abstract. We present the results from the multi-wavelength monitoring observations of the Narrow-Line Seyfert 1 galaxy Mrk 335 with AstroSat. We analysed both the X-ray (SXT and LAXPC) and UV (UVIT) data of the source at two epochs, separated by  18 days. The source was in a low flux state during the observations, and the X-ray spectra were found to be harder than usual. The presence of soft X-ray excess was identified in the observations, and the broadband X-ray continuum was modelled with power-law and black body (modified by intrinsic absorption) and a distant neutral reflection component. We did not find any variability in the X-ray spectral shape or the flux over this period. However, the UV flux is found to be variable between the observations. The obtained results from the X-ray analysis point to a scenario where the primary emission is suppressed and the component due to distant reflection dominates the observed spectrum Keywords. Active galactic nuclei—narrow-line Seyfert 1—Mrk 335—X-ray—ultra-violet.

1. Introduction Narrow-line Seyfert 1 galaxies (NLS1s) are a peculiar class of active galactic nuclei (AGN) with some extreme properties. They show strong Fe-II emission in the optical band along with narrow permitted lines and weak [OIII] emission (Osterbrock & Pogge 1985; Goodrich 1989). In the X-ray band, they are generally characterised by enhanced spectral and flux variability and the presence of soft excess emission (e.g. Boller et al. 1996). Mrk 335 (RA = 00 h 06 m 19.5 s, DEC = ?20 d 12 m 11 s) is an NLS1 galaxy at redshift = 0.026. The source is well known for showing dramatic fluctuations between high and low flux states in the X-ray band. Mrk 335 was observed with earlier observatories like Uhuru, Einstein, EXOSAT, Ginga, ROSAT, ASCA, and BeppoSAX when it was an X-ray bright source in the sky (e.g. Tananbaum et al. 1978; Halpern 1982). Its intensity dropped from the brightest stage to the very low flux state in 2007 (Grupe et al. This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

2007). Later, the source remained mostly in a low flux state though it has been reported to showing episodes of X-ray flaring activities (e.g. Gallo et al. 2018). Mrk 335 has been extensively studied in the optical/UV and X-rays (Grupe et al. 2007, 2008, 2012; Longinotti 2013; Gallo 2013; Parker 2014; Komossa et al. 2014; Chainakun & Young 2015; Keek & Ballantyne 2016; Sarma et al. 2015; Gallo 2015; Wilkins 2015). The X-ray spectra obtained with Ginga, ASCA, BeppoSAX, XMM-Newton and NuSTAR showed evidences for reflection and warm absorption features in the source (Nandra & Pounds 1994; George et al. 2000; Leighly 1999a; Ballantyne et al. 2001; Parker 2014; Longinotti 2013, 2019; Ezhikode 2020). Various attempts to model the X-ray spectra of Mrk 335 in the past suggested the possibility of changes in the geometry of the corona leading to the state changes in the source (e.g. Gallo 2013, 2015; Wilkins 2015; Gallo 2018). Signatures of soft excess has been persistently seen in the X-ray spectra of the source (e.g., Bianchi 2001; Grupe et al. 2001, 2008; Chainakun & Young 2015; Gallo 2015). The source is also known to show considerable variability in the optical/UV band which was found to

51

Page 2 of 12

J. Astrophys. Astr. (2021)42:51

be correlated and uncorrelated with the X-ray variability at various epochs (Buisson et al. 2017; Gallo et al. 2018). The long-term monitoring of Mrk 335 with Swift in optical–UV–X-ray bands and the observations with other telescopes revealed the properties of the source at different phases of its variability (Buisson et al. 2017; Tripathi et al. 2020) AstroSat (Singh et al. 2014; Agrawal 2017) monitored Mrk 335 at two epochs in 2017. We present the results from these multi-wavelength observations in X-ray and UV bands. We studied the X-ray spectral features by modelling the soft and hard X-ray spectra with different models. The details of observations and data processing are given in Section 2. The analysis of X-ray and UV data are described in Sections 3, 4 and 5. In Section 6, we discuss the results of the study. The summary and discussion of the work are given in Section 7.

2. Observations Mrk 335 was observed simultaneously in the X-ray and UV bands with Soft X-ray Telescope (SXT: Singh et al. 2017), Large Area X-ray Proportional Counter (LAXPC: Yadav 2016; Antia 2017), Cadmium–Zinc–Telluride Imager (CZTI: Rao et al. 2017; Bhalerao 2017; Vadawale 2016), and Ultra Violet Imaging Telescope (UVIT: Tandon 2017a, b) onboard AstroSat on October 31, 2017 (Obs 1) and November 18, 2017 (Obs 2). Here, we use the data from SXT, LAXPC, and UVIT observations. The details of these observations are given in Table 1 and Table 2, in the subsequent sections. The data used for the study are available at the Astrobrowse archive handled by Indian Space Science Data Centre (ISSDC).

3. X-ray analysis 3.1 Data reduction The Level-2 data products for SXT and LAXPC observations were obtained from the Level-1 data using the processing pipelines. The SXT observations were performed in the photon counting (PC) mode. We used SXTPIPELINE 1.4B (Release Date: 2019-01-04) for reducing the Level-1 SXT data. The pipeline produced cleaned event list for each orbit. These event lists were then merged using the SXTEVTMERGER tool in Julia. The merged event list in each observation was used to create high-level science products using XSELECT. We used the software LAXPCSOFT for processing LAXPC data. From the Level-2 event file and the GTI file created, we generated the light curves and spectra with the various tasks in the tool. Since the exposure time for the observations are less than that necessary for obtaining a good signal to noise data, and the source was in a low flux state, the data quality is found to be poor.

3.2 Light curves We created the light curves in both soft and hard X-ray bands. SXT light curves for the two observations were generated in the 0.7-7 keV band, from a region of 16 arcmin radius circle, for different time bins using XSELECT. We also generated LAXPC light curves in the energy range of 4–20 keV for various bin sizes. Since the source is very faint in the hard X-ray band, we used the specific LAXPCSOFT code for faint source background for light curve creation. Figure 1 shows the 0.7–7 keV SXT and 4–20 keV LAXPC 20 light curves for the two observations, created for a time bin of 500 s. We checked the

Table 1. The details of AstroSat observations of Mrk 335. Observation Number Obs 1 Obs 2

Exposure (ks)

Count rate (counts/s)

ID

Date

SXT

LXP 20

SXT (102 )

LXP 20

9000001654 9000001700

31/10/2017 18/11/2017

*14 *16

*17 *24

3:60  0:33 4:14  0:32

1:14  0:07 1:36  0:07

The quoted count rates are background subtracted values in the 0.3–8 keV band for SXT and in the 4–20 keV band for LAXPC 20 (LXP 20).

J. Astrophys. Astr. (2021)42:51

Page 3 of 12

51

Table 2. The details of UVIT observations at the two epochs. Exposure (s)

Count rate (counts/s)

Band

Filter

Wavelength ˚ (A)

Width ˚ (A)

Obs 1

Obs 2

Obs 1

Obs 2

NUV

N242W N245M N263M N279N

2418.0 2447.0 2632.0 2792.0

785.0 280.0 275.0 90.0

1881.957 667.5162 2523.842 563.2253

1949.968 2410.008 1361.323 688.3519

27.30 ± 0.12 15.35 ± 0.15 13.35 ± 0.07 3.48 ± 0.08

27.81 ± 0.12 16.16 ± 0.08 13.97 ± 0.10 3.37 ± 0.07

FUV

F148Wa F154W F169M F172M

1485.0 1541.0 1608.0 1717.0

500.0 380.0 290.0 125.0

1874.5 1025.182 1876.111 564.0201

816.4949 761.7113

10.70 ± 0.08 8.64 ± 0.09 6.76 ± 0.06 2.24 ± 0.06

11.67 ± 0.12 9.54 ± 0.11

The last two columns show the measured count rate (background subtracted) from a circular regions of 30 subpixel radius.

variability of the light curves using the ftool LCSTATS and found no variability in SXT. Though the LAXPC light curves showed significant variability in terms of the fractional RMS variability amplitude, a similar variability pattern was observed in the background light curve. Hence, the variability seen in the net LAXPC light curves in Fig. 1 is not intrinsic to the source.

3.3 Spectral analysis The SXT source spectra were extracted from circular regions of 16 arcmin radius, whereas the blank sky spectrum was used for the background. We used the background spectrum and the response files provided by the SXT-POC team. The Ancillary Response Function (ARF) file corrected for vignetting, PSF and exposure was generated using the latest module released on 2019 July 18, SXTEEFMODULE_V02. The rmf file for grade 0–12 was used for the analysis. The spectra were also grouped so that we can apply v2 statistic. As the SXT response is not well characterised below 0.7 keV, the region was ignored in the analysis. The LAXPC background spectra were created with the faint source code mentioned above. Since the background is more stable for LAXPC 20, we used only LAXPC 20 spectra for the analysis. Here, we ignored below 4 keV and above 20 keV as the regions were dominated by background. The SXT and LAXPC spectra at the two epochs are shown in Fig. 2 and Fig. 3. Both the SXT and LAXPC 20 spectra were analysed simultaneously to characterise the broadband X-ray continuum of the source. The spectral analysis

was done using XSPEC version 12.9. To account for the shift in SXT response, gain command in XSPEC was used with offset parameter fixed at 0.02. The model CONSTANT was used to take care of the cross normalisation between SXT and LAXPC 20. Also, a systematic error of 3% was applied while fitting. The X-ray spectral analysis was started by jointly fitting the SXT and LAXPC 20 spectra in the hard X-ray band (2-20 keV) with an absorbed power-law (tbabs  powerlaw) model. The Galactic column density (NH ) for the tbabs component was fixed at 3:56  1020 cm2 which we obtained from the LAB survey (Kalberla et al. 2005). The fit yielded a photon index (C) of less than 1 for both the observations. To check the presence of intrinsic absorption, we added a ztbabs component. However, the fit did not improve, and in the intrinsic equivalent hydrogen column density (NHInt ) was not constrained. Since such flat hard X-ray spectra could be the result of intrinsic absorption and the presence of distant reflection, we also included one xillver (Garcia & Kallman 2010; Garcia et al. 2013) component. Only the normalisation (Nxl ) and reflection fraction (frefl ) parameters of xillver were allowed to vary during the fit. The photon index of xillver component was tied to the slope of powerlaw model. Inclination (i), high-energy cut-off (Ecut ) and the iron abundance (AFe ) were fixed at 30 , 300 keV and 1 (in solar abundance), respectively. The ionisation parameter (n) was set to the minimum value, log n ¼ 0, to account for the reflection from neutral material. The new model yielded a marginally better fit for both the observations with Dv2   4:9 for Obs 1 and Dv2   8:5 for Obs 2 for a change in degrees of freedom (dof) of 2. The photon index also slightly increased with the addition of xillver component.

51

Page 4 of 12

J. Astrophys. Astr. (2021)42:51 Obs 1 500.0

s

4 0.05 0.1 0.15 0.2

SXT (0.7−7 keV)

1

2

3

LXP 20 (4−20 keV)

0

Count Rate (counts/s)

Count Rate (counts/s)

Bin time:

104

0

2×104 Time (s)

Start Time 18057 9:28:50:850

3×104

Stop Time 18057 18:13:50:850

Obs 2 0.3

500.0

s

40

0.1

0.2

SXT (0.7−7 keV)

1

2

3

LXP 20 (4−20 keV)

0

Count Rate (counts/s)

Count Rate (counts/s)

Bin time:

0

104

2×104

3×104 Time (s)

Start Time 18075 2:52:37:827

4×104

5×104

Stop Time 18075 16:20:57:827

Figure 1. X-ray light curves of Mrk 335 for the two observations (Obs 1 and Obs 2) binned for 500 s. SXT (black) and LAXPC 20 (LXP 20: red) light curves are extracted from 0.7–7 keV and 4–20 keV bands, respectively.

Further, we noticed the energy range below 2 keV and found that the spectrum rises above the current model. This is a clear indication of the soft excess emission. Therefore, we added bbody to model the soft X-ray excess and the fit provided a blackbody temperature (kTbb ) of  0.1 keV. All parameters are well constrained, except for the xillver normalisation and reflection fraction. However, the model tbabsðztbabs  ðpowerlaw þ bbodyÞ þ xillverÞ was preferable than the one without either xillver or bbody component. Another possible reason behind the observed hard spectrum could be the presence of partial covering absorption. Therefore, we also tried fitting the spectra

with zpcfabs and zxipcf models. However, the fits provided poor statistic and poorly constrained parameters.

4. UVIT analysis The source was observed with UVIT at both NUV and FUV wavelengths. Four filters (in PC mode) were used for both NUV and FUV observations at the first epoch. In the second observation, only two FUV filters were used as the instrument stopped working during that time. The filter information and other details of the exposures are given in Tabel 2. The Level-2 data,

J. Astrophys. Astr. (2021)42:51

Page 5 of 12

51

SXT

10−3

0.01

LXP 20

10−4

normalized counts s−1 keV−1

0.1

Obs 1

1

2

5 Energy (keV)

10

20

SXT

10−3

0.01

LXP 20

10−4

normalized counts s−1 keV−1

0.1

Obs 2

1

2

5 Energy (keV)

10

20

Figure 2. SXT (0.7–7 keV: black) and LAXPC 20 (4–20 keV: red) spectra at the two epochs, Obs 1 and Obs 2.

already processed with the latest pipeline UVIT LEVEL-2 PIPELINE (UL2P) version 6.3 by the POC were available at the archive. We used these data sets for further analysis. We carried out the photometry on the combined image in each filter. The combined images are obtained from the Level-2 data created using aspect correction done with VIS or NUV data. To do photometry, we chose a circular region of radius 30 subpixels (  12.500 ) centred around the source in each filter. This would give enclosed energy of around 97% in both NUV and FUV filters (Tandon et al. 2020). The observed count rates are given in Table 2. Most of these observations suffer

from saturation. Hence, we followed the procedure described in Tandon et al. (2017, 2020) to correct for the effect. We note that the NUV N242W filter records a total count rate of  30 counts/s where the abovementioned saturation correction is not valid. Hence, we do not use the data from N242W filter for further analysis. The background regions were selected from different circles with radii of 60 subpixels. The average background count rates were then subtracted from the saturation corrected values. We obtained the AB magnitude (mAB ) from the count rates and the zero-point (ZP) magnitudes and calculated the corresponding flux density Fk (Tandon et al. 2017, 2020) in each filter. We

51

Page 6 of 12

J. Astrophys. Astr. (2021)42:51 0.01

Obs 1

10−3 10−4

xillver bbody

10−5

powerlaw

10−6

2 1 0 1

0.01

2

5 Energy (keV)

10

2

5 Energy (keV)

10

Obs 2

10−3 10−4 10−5 10−6

2 1 0 1

Figure 3. The spectral fitting plots for the model tbabsðztbabs  ðpowerlaw þ bbodyÞ þ xillverÞ in the 0.7–20 keV for the two observations. The unfolded spectra and the model are shown in the upper panels and the ratio of data to model are shown in the lower panels. The solid lines in the upper panels represent the overall model whereas the dotted lines denote the individual model components powerlaw, bbody and xillver. The SXT data and model are shown in black colour while those of LAXPC are shown in red colour.

have also applied the Galactic extinction correction for the estimated Fk values using Cardelli et al. (1989) relation for RV ¼ 3:1 and AV ¼ 0:118 (Schlegel et al. 1998).

5. Spectral energy distribution A comprehensive modelling of the UV to X-ray spectral energy distribution (SED) of AGN can unveil the geometry of the central emitting regions and physics related to the variability mechanisms. We

have UV observations of Mrk 335 with various filters in NUV and FUV bands (see Table 2). But modelling the accretion disc emission from these photometric data is complicated as many components, like host galaxy and emission lines, can contribute to the observed flux at these wavelengths. Moreover, most of these observations are affected by saturation. Though we corrected the source count rates for the saturation effects (as mentioned in the previous section), there could still be uncertainties associated with the flux estimation. Also, it is difficult to derive the source flux free from the host galaxy contamination. To account

J. Astrophys. Astr. (2021)42:51

Page 7 of 12

51

Obs 1

0.01

FUV

LXP 20 SXT

10−3 10−4 10−5 10−6

2 1 0 0.01

0.1

1

10

Energy (keV)

Figure 4. Broadband FUV–X-ray SED of Mrk 335 with SXT, LAXPC 20 and UVIT (FUV filters: F148Wa, F154W, F169M and F172M) data from Obs 1. The data are modelled with optxagnf and xillver, modified by both Galactic and intrinsic absorption and reddening.

for these uncertainties, we added a systematic of 5–10% and fitted the UV–X-ray SED. The UV spectra were created by converting the flux (corrected for the Galactic reddening) in each filter (given in Table 5) using the task FLX2XSP. We used the XSPEC model optxagnf (Done et al. 2012) to fit the broadband SED of the source. The model can describe the emissions form accretion disc together with the soft and hard X-ray components. Here, we show the example of SED fitting for Obs 1 since there are more FUV data points for this observation. We began with the analysis of X-ray spectra by replacing powerlaw?bbody components with optxagnf in the best-fit model tbabsðztbabs ðpowerlaw þ bbodyÞ þ xillverÞ. Further, we added the FUV spectra for the filters F148Wa, F154W, F169M & F172M, and included the model zreddedn to correct for intrinsic reddening. The parameter E(B-V) for zredden was obtained from the intrinsic column density N Int H using the relation given by Bessell (1991). The parameters of xillver for the X-ray part were fixed at the best-fit values, whereas the component was not used for UV spectra. The cross normalisation constant for both the NUV and FUV spectral groups were tied to that of the SXT spectrum. We notice that fitting the FUV–X-ray SED resulted in a reasonable v2 of 91.78 for 71 degrees of freedom when a systematic error of 5% was applied. The fit provided a hard X-ray photon 21 2 index of  1.1 and intrinsic N Int H of 2:5  10 cm

(fixed). The other parameters obtained from the fit are Eddington ratio  1, coronal radius Rcor  19Rg and the fraction of power below Rcor emitted as the hard X-ray component fpl  0:9. The temperature and optical depth of the soft X-ray component are  0.1 keV,  79.4, respectively. The SED plot for the fit is shown in Fig. 4. When the NUV spectra were added, the fit worsened. The fit seemed to be improving when the systematic error was increased up to 10%. From FUV–X-ray SED analysis, we see that the accretion disc is truncated at a radius of about 19Rg , below which the energy is dissipated as Comptonised emission. However, this is a preliminary analysis and the errors on parameters are not obtained. A detailed and systematic study of the multi-wavelength SED of Mrk 335 with AstroSat data will be done later.

6. Results As expalined in Section 3.3, we tried fitting the X-ray spectra with different models. The best-fit parameters for these models are given Table 3. Figure 3 shows the various spectral fitting plots for Obs 1 and Obs 2. We found that tbabs (ztbabs  (powerlaw ? bbody) ? xillver) better fits the data in the 0.7–20 keV band than the other models. We estimated the unabsorbed flux in different energy bands using the cflux convolution

51

Page 8 of 12

J. Astrophys. Astr. (2021)42:51

Table 3. Best-fit parameters for the spectral fits in the 0.7–20 keV band for AstroSat (SXT and LAXPC 20) observations taken on 31/10/2017 (Obs1) and 18/11/2017 (Obs2). The xillver parameters that kept fixed while fitting are i ¼ 30 , AFe ¼ 1, log n ¼ 0 and Ecut ¼ 300 keV Energy range 2–20 keV

Model

Parameter

Obs 1

Obs 2

tbabs  powerlaw

C Npl (104 ) Constant

0.31þ0:28 0:32 1.64þ0:79 0:60 0.50þ0:31 0:20 28.74/39 \3:7 0.35þ0:31 0:33 1.88þ1:74 0:78 0.49þ0:30 0:19 28.56/38 \4:8 0.85þ0:41 0:26 3.91þ2:59 1:39 [ 0:007 [ 158:7 0.55þ0:31 0:19 23.66/36 \0:3 0.83þ0:41 0:28 2.56þ0:71 0:28 [ 0:003 [ 0:3 0.79þ0:24 0:30 47.57/68 0.89þ0:98 0:70 0.85þ0:39 0:26 3.47þ1:50 0:99 0.08þ0:04 0:02 4.23þ39:04 4:17 0.06þ999:92 0:02 [ 8:0 0.60þ0:29 0:19 39.46/66 \1:2 0.34þ0:29 0:32 1.81þ1:10 0:73 0.10þ0:08 0:04 1.40þ42:11 1:36 0.49þ0:29 0:20 43.76/68

0.66þ0:22 0:23 2.69þ0:96 0:77 0.87þ0:40 0:28 31.68/38 \4:1 0.74þ0:24 0:26 3.73þ2:70 1:58 0.79þ0:37 0:24 30.28/37 2.43þ2:94 2:12 1.27þ0:39 0:37 6.89þ4:61 2:96 [0.7 [1.7 0.82þ0:37 0:24 21.83/35 \0:9 1.05þ0:45 0:32 3.74þ2:07 1:20 0 – 0(?) [ 2:5 1.03þ0:48 0:30 45.28/66 0.86þ0:78 0:64 1.18þ0:40 0:36 5.04þ2:68 1:87 0.07þ0:03 0:06 6.66þ143:85 6:60 5.16þ994:84 4:29 [ 1:6 0.95þ0:41 0:28 39.37/64 \1:1 0.68þ0:23 0:12 2.92þ1:43 0:05 0.08þ0:08 0:07 0.43þ14:03 0:43 0.85þ0:39 0:27 46.21/66

v2 /dof tbabsztbabs  powerlaw

NHInt (1022 cm2 ) C Npl (104 ) Constant

v2 /dof tbabsðztbabs  powerlaw þ xillverÞ

0.7–20 keV

NHInt (1022 cm2 ) C Npl (104 ) frefl Nxl (105 ) Constant

v2 /dof tbabsðztbabs  powerlaw þ xillverÞ

NHInt (1022 cm2 ) C Npl (104 ) frefl Nxl (105 ) Constant

v2 /dof tbabsðztbabs  ðpowerlaw þ bbodyÞ þ xillverÞ

v2 /dof tbabs  ztbabs  ðpowerlaw þ bbodyÞ

v2 /dof

model. The obtained flux values are provided in Table 4. Both observations are found to be intrinsically 21 cm2 . The source does absorbed with N Int H  9  10 not show significant X-ray spectral or flux variability

NHInt (1022 cm2 ) C Npl (104 ) kTbb (keV) Nbb (103 ) frefl Nxl (104 ) Constant NHInt (1022 cm2 ) C Npl (104 ) kTbb Nbb (103 ) Constant

between the observations. The X-ray spectra seem to be harder in both the observations with C  0:9ð1:2Þ (consistent within error bars) for Obs 1(Obs 2) while the reflection parameters are not properly constrained. The total flux in the 2–20 keV is roughly

J. Astrophys. Astr. (2021)42:51

Page 9 of 12

2  1011 erg cm2 s1 , and in the 0.7–2 keV band it is around 6  1013 erg cm2 s1 . Unlike the X-ray observations, UV emission from the source shows variability in both FUV and NUV bands. The net count rate and flux in each filter are mentioned in Table 5. The flux variability (except for the FUV filters F169M, F172M and the NUV filter N245M) are shown in Fig. 5. In order to check if the variability is an instrument artefact, we obtained the light curves of a star in both NUV and FUV images (since the stars were too faint in NUV N279, F154W and F148Wa filters we did not obtain the count rate for those exposures). The net count rates for the star seem to be non-variable in the NUV band showing that the variability shown by the source is real.

6.1 Comparison with other observations Mrk 335 has been observed at optical, UV and X-ray wavelengths with various missions. Here, we give a brief summary of the analysis of some of these data Table 4. The total unabsorbed X-ray flux for the model tbabsðztbabs  ðpowerlaw þ bbodyÞ þ xillverÞ from the two AstroSat observations and the near-XRT observation. Energy range (keV) 0.7–20 0.7–2 2–20 2–10 0.3–2 0.3–10

Flux (1011 erg cm2 s1 ) Obs 1

Obs 2

2.14þ0:86 0:63

1.66þ0:59 0:45

0.060:01 2.09þ0:86 0:62 0.64þ0:18 0:16 0.07þ0:07 0:02 0.71þ0:17 0:15

0.060:01 1.61þ0:59 0:45 0.59þ0:15 0:14 0.07þ0:25 0:02 0.66þ0:09 0:13

XRT

0.32þ0:21 0:14 0.16þ0:06 0:05 0.49þ0:21 0:15

51

and compare those with the results from our AstroSat observations. Swift has been monitoring Mrk 335 for years in X-rays and optical/UV. We analysed one Swift observation close to AstroSat observations as there are no observations strictly simultaneous with that of AstroSat. We retrieved XRT and UVOT data taken on November 03, 2017 (almost three days after Obs 1 and two weeks before Obs 2). This near-simultaneous Swift observation (ID: 00033420140) has an exposure time of only  800 s. The XRT observation was made in PC mode. We reduced the data with XRTPIPELINE and extracted the spectrum and light curve from a circular region of radius 30 pixels. The background region of 50 pixels radius circle was also selected from the same image. We generated the XRT light curve in the 0.3–10 keV band and did not find any variability. The net count rate of the XRT observation is around 0.07 counts s1 (0.3–10 keV). Since the data quality is not good, we grouped the spectrum for minimum 5 counts per bin and used cstat while fitting. Modelling the 0.3–10 keV spectrum with TBabs  zTBabs  powerlaw gave a photon index of about 1.8, softer than that obtained for the fit of combined SXT and LAXPC spectra in the 0.7–20 keV band. The corresponding fitstatic is cstat/dof = 14.40/9. When a xillver component was added the fit-static reduced to 1.46/7 with a steeper unconstrained photon index. Other parameters 22 cm2 , Npl ¼ 7:59þ9:68 are NHInt ¼ 11:7515:94 6:37  10 5:61  þ2:78 3 6 10 , Nxl ¼ 6:762:45  10 and frefl was not constrained. Adding a bbody, with kTbb fixed at 0.1 keV did not change the fit any more. However, the spectrum remained steeper with C [ 2:3. We also fitted the spectrum with TBabs  zTBabsðbbodyþ powerlawÞ model which resulted in a cstat/dof of 1.35/8 with parameter values, NHInt \0:4  1022 cm2 , þ0:04 4 C\1:2, Npl ¼ 1:04þ1:71 0:61  10 , kTbb ¼ 0:140:06 keV

Table 5. Results from the analysis of UVIT data. The saturation corrected net count rates with the corresponding magnitude and flux density (Fk ) for each filter are given in the last tree columns. Fk values are corrected for the Galactic reddening.

Filter

Zero point (magnitude)

Count rate (counts/s) Obs 1

Obs 2

N245M 18.452?/–0.005 21:84  0:18 23:51  0:1 N263M 18.146?/–0.010 18:14  0:08 19:3  0:12 N279N 16.416?/–0.010 3:75  0:08 3:62  0:07 F148Wa 17:994  0:010 13:24  0:08 14:78  0:13 F154W 17:771  0:010 10:2  0:1 11:5  0:12 F169M 17:410  0:010 7:67  0:06 F172M 16:274  0:020 2:34  0:06

mAB (magnitude) Obs 1

Obs 2

15.104?/–0.010 14.999?/–0.011 14.982?/–0.026 15.189?/–0.012 15.250?/–0.015 15:198  0:013 15:350  0:040

15.024?/–0.007 14.932?/–0.012 15.020?/–0.024 15.070?/–0.014 15.119?/–0.015

Fk ð1014 erg cm2 s1 Þ Obs 1

Obs 2

2:15  0:02 2:32  0:01 1:98  0:02 2:1  0:02 1:76  0:04 1:7  0:04 5:55  0:06 6:2  0:08 4:84  0:07 5:46  0:08 4:63  0:06 3:51  0:12

51

Page 10 of 12

J. Astrophys. Astr. (2021)42:51

Figure 5. Plot showing the variability in FUV (F148Wa and F154W) and NUV (N245M, N263M and N279M) emission from Mrk 335 between the two epochs. 3 and Nbb ¼ 0:02þ0:11 0:01  10 . Here, the black body normalisation is lower than the results from AstroSat whereas the temperature remains similar. In the optical/UV band, the observation was made with only uvw2 filter. We estimated the flux of the source using the task UVOTPRODUCTS. For this, we selected a source region of 500 radius circle and background circles of larger radius at different regions from the observed images. The background subtracted flux for uvw2 filter is 2:32  0:05  1014 erg cm2 s1 A1 (not corrected for Galactic reddening). The X-ray spectrum of Mrk 335 is complex to be modelled with low-quality data from AstroSat and Swift. We notice that when the X-ray spectra from XRT and AstroSat (SXT & LAXPC 20) observations were fitted in the same energy range of 0.7–10 keV with simple models like (tbabs  zTBabs powerlaw), the parameters agree well within error bars. The discrepancy arises when we fit the broadband X-ray spectrum, including LAXPC data. We also carried out a preliminary analysis of one of the XMM-Newton observations of Mrk 335 taken in 2019 January. The EPIC-pn spectrum (net exposure  66 ks) showed soft excess emission and a broad iron emission line. We fitted the spectrum in the 0.3–10 keV band with an absorbed (Galactic and intrinsic) power-law, black body and a redshifted broad Gaussian component. The fit yielded a photon

index of  0.7 and NHInt \0.19 9 1022 cm-2. When a xillver component was added, the fit improved significantly, and the photon index increased to around 1.4 while NHInt remained unconstrained. The X-ray (2–10 keV) flux from the observation is found to be decreased roughly by a factor of 5–6 as compared to the AstroSat observations, but C is consistent within error bars.

7. Discussion and summary We found harder X-ray spectra for both AstroSat observations. A similar photon index was obtained when the XRT spectrum was fitted with an absorbed power-law and blackbody model (0.3–10 keV), that is consistent with the result obtained by Tripathi et al. (2020). However, when a reflection component was added in the model, the primary power-law appeared to be softer in the XRT observation. The significance of reflection in the source was noticed in the previous studies as well. For example, Parker (2014) studied the XMM-Newton, Swift and NuSTAR spectra taken in 2018–2019 when the source was showing an extremely low flux level in X-rays. By modelling the broadband continuum in detail, they found that the hard X-ray spectrum is dominated by distant reflection and the soft part by photoionised emission lines. They also observed

Page 11 of 12

51

1.5

Confidence contours: Chi−Squared cross = 39.373; Levels = 41.673 43.983 48.583

1

+

0.5

Parameter: PhoIndex

2

J. Astrophys. Astr. (2021)42:51

0

1

2 Parameter: nH (1022)

3

4

Figure 6. The confidence (one, two and three sigma) contour plot of the parameters NHInt and C (for Obs 2).

steep X-ray spectra C  2 and a significant blackbody component. Earlier observations of Mrk 335 reported partial covering absorption and relativistic reflection in the source (Longinotti 2019; Parker 2014) that we could not parameterise with the AstroSat data. In a previous study on the correlation between the reflection fraction and photon index in AGN (Ezhikode 2020), we analysed the NuSTAR spectrum of Mrk 335 observed in 2013 June. The spectrum showed the presence of broad and narrow emission lines, and we fitted the 3–79 keV spectrum using the models relxill and xillver. The spectrum was steep with gamma around 2.2, and we did not see any significant intrinsic absorption. A larger X-ray photon index of CJ1:5 is typically observed in NLS1s. Here, we observe a different behaviour even after including the neutral reflection model (though the slope of the second observation is marginally within this range). We also note that the Xray spectrum gets steeper when the intrinsic obscuration is fixed at larger values, although the fit worsens. We obtained the confidence contour plot (see Fig. 6) for photon index and intrinsic absorption. It is clear from the plot that the data could not constrain C well and a higher index similar to those found in other AGN is not ruled out. The X-ray emission from the source may be obscured intrinsically, and hence the distant reflection may be dominating the observed spectra. During AstroSat observations, Mrk 335 was in a lowflux state in the UV band as well. Our observations, separated by almost 18 days, show variability in both NUV fand FUV emissions. However, no significant variability was found between the two X-ray observations. Variable

UV emission using Swift UVOT observations was detected by Grupe et al. (2008) on time-scales of days to weeks. They found the UV variability to be following the XRT light curve, suggesting the possibility of the same mechanism triggering both the emissions. Considering the time-scale of variability in our observations, X-ray reprocessing could be the origin of the observed UV variability in the source. However, a similar variability is not observed in X-ray emissions, and we do not have enough monitoring observations to confirm this. The obscuration of X-rays by clouds could be another possibility of the observed UV variability that is unrelated to X-ray emission. Detailed modelling of the broadband SED is required to explain the scenario. Owing to the low signal-to-noise of the X-ray spectra and uncertainties in the UV flux measurements, a proper modelling of the UV-X-ray SED is difficult. Hence, we do not make a definitive statement regarding the UV variability in the source. A more detailed characterisation of broadband X-ray continuum emissions may be carried out with future better observations with AstroSat. With simultaneous filter and grating observations with UVIT, we can also study the nature of variability in the accretion disc emission in depth.

Acknowledgements We would like to acknowledge the anonymous referee for the helpful comments and suggestions. We thank Prof. Shyam Tandon and Mr. Prajwel Joseph for the

51

Page 12 of 12

useful discussions on UVIT data analysis. This publication uses the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). This work has used the data from the Soft X-ray Telescope (SXT) developed at TIFR, Mumbai, and the SXT POC at TIFR is thanked for verifying and releasing the data via the ISSDC data archive and providing the necessary software tools. We thank the UVIT POC at IIA, Bangalore for the data and their support. This research has made use of data, software and/or web tools obtained from the High Energy Astrophysics Science Archive Research Center (HEASARC), a service of the Astrophysics Science Division at NASA/GSFC and of the Smithsonian Astrophysical Observatory’s High Energy Astrophysics Division.

References Antia H. M. et al. 2017, ApJS, 231, 10 Agrawal P. C. 2017, J. Astrophys. Astr., 38, 27 Ballantyne D. R., Iwasawa K., Fabian A. C. 2001, MNRAS, 323, 506 Bessell M. S. 1991, A&A, 242, L17 Bianchi, S. et al. 2001, A&A, 376, 77 Boller T., Brandt W. N., Fink H. 1996, A&A, 305, 53 Bhalerao V. et al. 2017, J. Astrophys. Astr., 38, 31 Buisson D. J. K., Lohfink A. M., Alston W. N., Fabian A. C. 2017, MNRAS, 464, 3194 Cardelli J. A., Clayton G. C., Mathis J. S. 1989, ApJ, 345, 245 Chainakun P., Young A. 2015, ebha.conf, 76 Done C., Davis S. W., Jin C. et al. 2012, MNRAS, 420, 1848 Ezhikode S. H. et al. 2020, MNRAS, 495, 3373 Gallo L. C. et al. 2013, MNRAS, 428, 1191 Gallo L. C. et al. 2015, MNRAS, 446, 633 Gallo L. Blue D. M., Grupe D. et al. 2018, MNRAS, 478, 2557 Gallo L. 2018, rnls.conf, 34 Garcia J., Kallman T. R. 2010, ApJ, 718, 695. Garcia J., Dauser T., Reynolds C. S., Kallman T. R., McClintock J. E., Wilms J., Eikmann W. 2013, ApJ, 768, 146. George I. M., Turner T. J., Yaqoob T., Netzer H., Laor A., Mushotzky R. F., Nandra K., Takahashi T. 2000, ApJ, 531, 52 Goodrich R. W. 1989, ApJ, 342, 224

J. Astrophys. Astr. (2021)42:51 Grupe D., Thomas H.-C., Beuermann K. 2001, A&A, 367, 470 Grupe D., Komossa S., Gallo L. 2007, ApJ, 668, L111 Grupe D., Komossa S., Gallo L. C., Fabian A. C., Larsson J., Pradhan A. K., Xu D., Miniutti G. 2008, ApJ, 681, 982 Grupe D., Komossa S., Gallo L. C., Longinotti A. L., Fabian A. C., Pradhan A. K., Gruberbauer M., Xu D. 2012, ApJS, 199, 28 Halpern J. P. 1982, Ph.D. thesis, Harvard Univ. Kalberla P. M. W., Burton W. B., Hartmann D., Arnal E. M., Bajaja E., Morras R., Po¨ppel W. G. L. 2005, A&A, 440, 775 Keek L., Ballantyne D. R. 2016, MNRAS, 456, 2722 Komossa S., Grupe D., Saxton R., Gallo L. 2014, Proceedings of Swift: 10 Years of Discovery (SWIFT 10), id. 143 Komossa S., Grupe D., Gallo L. C., Poulos P., Blue D., Kara E., Kriss G., Longinotti A. L., Parker M. L., Wilkins D. 2020, A&A 643, L7 Leighly K. M. 1999a, ApJS, 125, 297 Longinotti A. L. et al. 2013, ApJ, 766, 104 Longinotti A. L. et al. 2019, ApJ, 875, 150 Nandra K., Pounds K. A. 1994, MNRAS, 268, 405 Osterbrock D. E., Pogge R. W. 1985, ApJ, 297, 166 Parker M. L. et al. 2014, MNRAS, 443, 1723 Rao A. R., Bhattacharya D., Bhalerao V. B., Vadawale S. V., Sreekumar S. 2017, Curr. Sci., 113, 595 Singh K. P., Tandon S. N., Agrawal P. C. et al 2014, Proc. SPIE, 9144, 91441S Singh K. P., Stewart G. C., Westergaard N. J. et al. 2017, J. Astrophys. Astr., 38, 29 Sarma R., Tripathi S., Misra R., Dewangan G., Pathak A., Sarma J. K. 2015, MNRAS, 448, 1541 Schlegel D. J., Finkbeiner D. P., Davis M. 1998, ApJ, 500, 525 Tananbaum H., Peters G., Forman W., Giacconi R., Jones C., Avni Y. 1978, ApJ, 223, 74 Tandon S. N. et al. 2017a, J. Astrophys. Astr., 38, 28 Tandon S. N. et al. 2017b, AJ, 154, 128 Tripathi S., McGrath K. M., Gallo L. C. et al. 2020, MNRAS, 499, 1266 Vadawale S. V. et al. 2016, in Space Telescopes and Instrumentation 2016: Ultraviolet to Gamma Ray, p. 99051G Wilkins D. R et al. 2015, MNRAS, 454, 4440 Yadav J. S. et al. 2016, Large Area X-ray Proportional Counter (LAXPC) Instrument Onboard ASTROSAT, p. 99051D, https://doi.org/10.1117/12.2231857 Yadav J. S. et al. 2016, Proc. SPIE, 9905, 99051D

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:93 https://doi.org/10.1007/s12036-021-09763-x

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

The AstroSat mass model: Imaging and flux studies of off-axis sources with CZTI SUJAY MATE1,2 , TANMOY CHATTOPADHYAY3,4 , VARUN BHALERAO1,* ,

E. AARTHY5 , ARVIND BALASUBRAMANIAN1,6 , DIPANKAR BHATTACHARYA7 , SOUMYA GUPTA7 , KRISHNAN KUTTY8, N. P. S. MITHUN5 , SOURAV PALIT1 , A. R. RAO8 , DIVITA SARAOGI1 , SANTOSH VADAWALE5 and AJAY VIBHUTE7 1

Indian Institute of Technology Bombay, Mumbai 400 076, India. IRAP, CNES, CNRS, UPS, Universite´ de Toulouse, Toulouse, France. 3 Department of Physics, Stanford University, 382 Via Pueblo Mall, Stanford, CA 94305, USA. 4 Kavli Institute of Astrophysics and Cosmology, 452 Lomita Mall, Stanford, CA 94305, USA. 5 Physical Research Laboratory, Ahmedabad 380 009, India. 6 Department of Physics and Astronomy, Texas Tech University, Box 1051, Lubbock, TX 79409-1051, USA. 7 The Inter-University Centre for Astronomy and Astrophysics, Pune 411 007, India. 8 Tata Institute of Fundamental Research, Mumbai 400 005, India. *Corresponding Author. E-mail: [email protected] 2

MS received 3 November 2020; accepted 16 April 2021 Abstract. The Cadmium Zinc Telluride Imager (CZTI) on AstroSat is a hard X-ray coded-aperture mask instrument with a primary field-of-view of 4:6  4:6 (FWHM). The instrument collimators become increasingly transparent at energies above  100 keV, making CZTI sensitive to radiation from the entire sky. While this has enabled CZTI to detect a large number of off-axis transient sources, calculating the source flux or spectrum requires knowledge of the direction and energy dependent attenuation of the radiation incident upon the detector. Here, we present a GEANT4-based mass model of CZTI and AstroSat that can be used to simulate the satellite response to the incident radiation, and to calculate an effective ‘‘response file’’ for converting the source counts into fluxes and spectra. We provide details of the geometry and interaction physics, and validate the model by comparing the simulations of imaging and flux studies with observations. Spectroscopic validation of the mass model is discussed in a companion paper, Chattopadhyay et al. (J. Astrophys. Astr., vol. 42 (2021) https://doi.org/ 10.1007/s12036-021-09718-2). Keywords. Gamma-ray bursts—AstroSat—Cadmium Zinc Telluride Imager—mass model simulations.

1. Introduction The Cadmium Zinc Telluride Imager on board AstroSat is a hard X-ray (20–200 keV) coded aperture mask instrument with a 4:6  4:6 field-of-view (Singh et al. 2014; Bhalerao et al. 2017b). The primary objectives of the instrument are spectroscopy, imaging, and timing studies of hard X-ray sources. At energies above  100 keV, the instrument collimators This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

become increasingly transparent to the radiation from off-axis directions (Rao et al. 2017). Thus, the CZT detectors are sensitive to the entire sky up to  200 keV. This sensitivity is extended to  650 keV by the CsI anti-coincidence ‘‘veto’’ detectors installed in the instrument. This all-sky sensitivity has been leveraged for detection of transient sources like Gamma Ray Bursts (GRBs). CZTI detected its first GRB, GRB 151006A, on the very first day it was switched on (Bhalerao et al. 2015; Rao et al. 2016). In the five years since, CZTI

93

Page 2 of 26

has detected more than 400 GRBs, 88 of which are being reported for the first time in Sharma et al. (2021). The sensitivity of CZTI is comparable to several other GRB missions (Bhalerao et al. 2017a), which has led to many significant results. The detection of GRB 170105A, which was missed by all major missions, conclusively proved that an ‘‘orphan afterglow’’ discovered in optical was not related to the binary black hole merger GW170104 (Marcinkowski et al. 2017; Sharma et al. 2017; Bhalerao et al. 2017). When CZTI did not detect GW170817, we inferred that the source was occulted by the Earth—narrowing down the source localisation by a factor of two (Kasliwal et al. 2017). The off-axis sensitivity of CZTI has also been leveraged for Earth-occultation studies of sources (Singhal et al. 2021). Compton scattering within the CZT detectors is sensitive to polarisation of incoming photons, and has been successfully used to measure polarisation of GRBs (Chand et al. 2019; Chattopadhyay et al. 2019) and the Crab pulsar (Vadawale et al. 2018). While CZTI has had great success in detecting offaxis transients, the interpretation of detected signals requires a detailed modelling of the instrument and satellite. Incoming X-ray photons interact with various satellite elements undergoing absorption, coherent and incoherent scattering, etc. Photons can get absorbed and re-emitted as fluorescence lines. Such interactions modify the energy, direction, and position of interaction of incoming radiation. These effects are strongly direction-dependent, based on the mass distribution of various materials in the satellite. In this paper, we present the AstroSat mass model: a numerical simulation of such interactions to calculate the observed spatial and energy distribution of photons for any given astrophysical source. The basic concepts of this mass model were introduced in Chattopadhyay et al. (2019). In this paper, we give the full details of the mass model and give results on imaging and flux studies. A companion paper (Chattopadhyay et al. 2021) discusses details of sub-MeV spectroscopy using the mass model, and compares the results with various other sub-MeV spectroscopic methods. Numerical simulations using the mass model are utilised for three key calculations. First, these simulations are used for mapping the observed count rates and spectra to a source spectrum. We discuss examples with count rates in this paper and spectra are discussed in Chattopadhyay et al. (2021). The same technique is used to calculate flux upper limits for non-detections, based on an assumed source spectrum

J. Astrophys. Astr.

(2021) 42:93

(Sharma et al. 2021). Secondly, mass model simulations form a key part of the measurement of polarisation of astrophysical sources (Chattopadhyay et al. 2014; Vadawale et al. 2015). Thirdly, the observed distribution of photons on the detectors (Detector Plane Histogram, DPH) is strongly dependent on the incident direction of the source photons. Crudely speaking, different satellite elements cast unique shadows on the detector plane, allowing us to use the entire satellite as a mask for locating source positions. A basic ray-tracing version of this concept was used to localise GRB 170105A (Bhalerao et al. 2017a), and the more refined mass model is expected to improve such localisations. This paper is organised as follows: in Section 2, we describe the framework used for the numerical simulations. A detailed discussion of the model components and simplifying approximations is presented in Section 3, it is followed by a discussion of the model physics in Section 4. In Section 5, we discuss the efficacy of the model by comparing simulations to observations. We conclude by discussing future work in Section 6.

2. Numerical modelling As discussed above, our goal is to create a numerical model that can simulate the interaction of incoming radiation with the satellite, to yield the final energies (spectrum) and positions (DPH) of photons incident on the detector. The first step is to create a detailed digital representation of the the satellite. Then, we need a software to simulate the interaction and passage of radiation through the satellite. These interactions are probabilistic in nature: multiple photons entering the satellite at exactly the same point from the same direction may all undergo different interactions with different satellite elements. The final spectrum and DPH can hence be interpreted only in the average sense, and this caveat underscores all comparisons with observations. AstroSat, like any other satellite, is a complex structure and modelling all interactions is a non-trivial task. We tackle this problem with the aid of the GEANT4 toolkit for particle, photons and matter interactions.1 GEANT4 has wide range of applications in high energy physics, space sciences, and medical science (Agostinelli et al. 2003; Allison et al. 2006, 2016). It is an easy to use open source 1

http://geant4.web.cern.ch/geant4/.

J. Astrophys. Astr.

(2021) 42:93

simulation toolkit which provides all the necessary building blocks to simulate complex particle matter interactions. GEANT4 features pre-defined geometry classes and a large materials database for easy construction of elaborate geometrical structures. In addition, computer-aided design (CAD) models can also be imported into the code. The toolkit supports a wide range of physical processes for photons and particles, and allows tracking and extraction of photon properties at any stage of simulation. Lastly, it is a highly scalable toolkit making it easy to develop and test the mass model simulation on personal computers, and run it on high performance computing clusters for speed and performance. Thanks to these features, we selected GEANT4 for creating the AstroSat mass model. We now discuss the satellite geometry construction in Section 3, followed by the interaction physics in Section 4.

3. Geometry constructions The basic structure of AstroSat is a cuboid of dimensions  2  1:8  1:8 m that supports all instruments (Fig. 1). CZTI (Section 3.1) is mounted on the ‘‘top’’ deck of this cuboid, along with the Soft X-ray Telescope (SXT, Section 3.3) and all three units of the Large Area X-ray Proportional Counter (LAXPC, Section 3.2). The Ultraviolet Imaging Telescope (UVIT, Section 3.4) is mounted on the

Page 3 of 26

93

bottom deck and penetrates the entire satellite body. The Scanning Sky Monitor (SSM, Section 3.5) is mounted on a rotating platform on one of the sides of the cuboid. Various other satellite bus components are also contained in the cuboid. The geometry and materials of AstroSat are simulated by creating a custom detector construction class derived from the standard GEANT4 G4VUserDetectorConstruction. The geometrical shapes of the model are defined either using the built-in GEANT4 geometry shapes (G4Tubes, G4Box, G4Sphere etc.) or by importing CAD files (in case of complex geometry) in .stl format using the CADmesh2 library (Poole et al. 2011). The material properties, both for pure and composite materials, are defined using the G4Material class. GEANT4 derives these properties from the NIST3 database. Several small-scale structural intricacies of AstroSat are unimportant for the key goals of our numerical simulations, but add a significant computational overhead in the simulations. Hence, we make certain simplifying assumptions while modelling them in GEANT4. In several cases, a component is replaced with a uniform block of the same size, mass, and average composition. Some other small components like Titanium screws have been completely ignored in the mass model. Thus, the total simulated mass is comparable to the actual satellite mass (Table 1). We now discuss the specifics about the model and approximations used in each instrument:

3.1 Cadmium Zinc Telluride Imager The collimators, housing, and other structural elements of CZTI are closest to the detectors, and hence have the most influence on the spatial and energy redistribution of incident photons. Hence the CZTI geometry is modelled as accurately as possible (Fig. 2). The overall construction of CZTI is discussed in Bhalerao et al. (2017a). Some elements of the geometry are coded using the GEANT4 geometry classes while the complex structure like coded masks (both top and side), heat pipes, detector and collimator housing, etc. are imported from the CAD geometry file. The coded masks are pure tantalum while the collimators are 1 mm aluminium with 0.07 mm tantalum pasted on one side.

Figure 1. Representative CAD model of AstroSat with solar panels folded.

2

https://github.com/christopherpoole/CADMesh. http://physics.nist.gov/PhysRefData/Star/Text/method.html.

3

93

Page 4 of 26

J. Astrophys. Astr.

(2021) 42:93

Table 1. Comparison of simulated and actual masses of key components of AstroSat. Instrument

Simulated mass (kg)

Actual mass (kg)

41.87 44.03 212.58 350.87 71.53 657.22 1378.10

50.29 57.64 202.06 389.10 71.53 668.47 1439.09

CZTI SXT UVIT LAXPC SSM Satellite bus and electronics Total

For simplicity, each CZT module has been modelled to be 40  40  5 mm instead of 39.06  39.06  5 mm. This change increases the effect area of each module by about 5%. Therefore each pixel size in simulation is 2.5 mm instead of the actual sizes of 2.46 mm for central pixels and 2.31 mm for edge pixels. The detector composition is 43% cadmium, 2.8% zinc and 54.2% tellurium. To model the electronics cards of CZTI, a PCB material is defined using G4Material class by assuming elemental composition of a regular PCB. The dimensional accuracy of cards is within 5%. The veto detector is defined using GEANT4 geometry class as a continuous slab of CsI(Tl) with dimensions 160  160  20 mm. The composition of CsI(Tl) is 51% Cs, 48% I, 1% Tl. The veto casing is imported using the CAD model. The Photo-Multiplier Tube (PMT) inside the casing is not modelled. The veto electronics cards are defined similarly as in case of CZTI.

3.2 Large Area X-ray Proportional Counter The mass model of LAXPC is taken from Antia et al. (2013) and the geometry is defined completely by GEANT4 geometry classes (Fig. 3(b)). A particularly complex system is the collimator, comprising of uniformly spaced parallel slats. However, the spacing between the slats (  cm) is much smaller than the distance between the collimators and the CZT detectors ([several tens of cm) which are the focus of our work. Hence, we can safely approximate the collimators of LAXPC as box of effective mass and composition to reduce the simulation time and for simplicity of incorporating the model in our code. All the other elements, except gas pump and electronic cards, are modelled accurately. The three units do not have the same gas: LAXPC 10 and 20 have 90%

xenon and 10% methane, while LAXPC 30 has 84% xenon, 9.4% methane and 6.2% argon, all at 1520 torr pressure (Bhattacharya et al. 2016). This gives density of 1:07  102 g cm3 for LAXPC 10 and 20 and 1:22  102 g cm3 for LAXPC 30.

3.3 Soft X-ray Telescope The geometry of SXT is completely imported as a CAD model, with some approximations for compositions of geometric parts (Fig. 3(a)). The focusing tube is made of carbon Fibre Reinforced Polymer (CFRP), which we define as a new material with low density carbon as the only element. The optics housing is pure aluminium. The actual optics consist of concentric shells of aluminium coated with gold, but for simplicity we model them as a solid body composed of a gold–aluminium alloy with the same fractional abundances and total mass. Similarly in case of the camera assembly, the composition is kept as an alloy of aluminium, nickel and gold with effective fraction and mass to incorporate the effect of nickel and gold coating as in real case.

3.4 Ultraviolet Imaging Telescope Bulk of the UVIT geometry is imported from a CAD model (Fig. 3(d)). A notable exception is the highly complex geometry of the camera assembly, which we model with a GEANT4 geometry class as two aluminium cylinders with the same effective mass as the camera. The focusing tubes are made of two parts: aluminium for parts above the top deck of the satellite, and invar for the part between the top and bottom deck. The satellite adapter connecting the tubes and central cylinder between decks is of pure titanium. The thermal blanket covering camera assembly is

J. Astrophys. Astr.

(2021) 42:93

Page 5 of 26

93

the coded mask is aluminium and the gas composition is 25% xenon and 75% P10 (90% argon ? 10% methane) at 800 torr pressure. The average density of the gas is 2:9  103 g cm3 (Bhattacharya et al. 2016). SSM is mounted on a rotating platform, and it changes the orientation every ten minutes in routine AstroSat observations. However, we do not model this platform and keep the orientation of SSM fixed to the configuration at the time of the launch.

3.6 Satellite body

Figure 2. Panel (a): Rendering of the CZTI mass model. The green part houses the CZT and Veto detectors. The quadrants are labeled to show their relative orientation with respect to the radiator plate. Panel (b): Inner view of the green section, showing the four arrays of CZT detectors (purple) and the four Veto detectors (yellow).

made of an aluminium and titanium alloy, where the titanium fraction approximates the fasteners used in the assembly.

3.5 Scanning Sky Monitor The SSM geometry is also imported from the CAD model (Fig. 3(c)). The body composition, along with

Various parts of the satellite structure (satellite bus, support structures, electronics, etc.) significantly scatter incident photons. This effect is most prominent for GRBs shining onto CZTI from under the satellite (Z direction). This warrants a careful modelling of the satellite supporting structures as well as auxiliary electronics. This modelling is simplified by noting that if the satellite components are not too close to the CZTI detector plane, then small-scale structural details do not matter. For such components, only the effective mass, composition and geometry is modelled accurately. This is done especially in case of electronic boxes on inside of the side plates of the satellite body. Instead of modelling each box, effective mass is split into separate boxes and they are placed uniformly on the side panels from inside. Other structures inside of the satellite such as fuel tanks, vertical support slats, and UVIT connector cylinder are modelled accurately with appropriate GEANT4 geometry classes. Solar panels are approximated as a single sheet instead of two. The rotation with the orbit is not accounted while simulating and the orientation of panels is always vertical. Most of the satellite body is aluminium honeycomb (lower density aluminium). The electronic boxes are PCB plus aluminium composite. Figure 4 shows rendering of complete AstroSat geometry and top view into the inside of the satellite.

4. Physics and tracking To define the particles and physics processes for the simulation, we employ user defined physics list derived from the G4VUserPhysicsList class in GEANT4. As our interest is to obtain the response of CZTI for photon energies less than a few MeV, physics processes involving low energy X-ray photons

93

Page 6 of 26

J. Astrophys. Astr.

(2021) 42:93

Figure 3. (a)–(d) GEANT4 rendering of major AstroSat instruments.

Figure 4. (a), (b) GEANT4 rendering of the satellite assembly. SSM and the solar panels are always assumed to be fixed at the orientation shown in this figure.

and electrons (secondary particles generated by photon interactions) are included in the physics list. In particular the following processes are used: G4LivermorePhotoElectricModel, G4Live

rmorePolarizedComptonModel, G4Liver morePolarizedRayleighModel, G4Liver moreIonisationModel, G4LivermoreBrems strahlungModel and G4eMultiple Scat

J. Astrophys. Astr.

(2021) 42:93

tering. Note that for Compton and Rayleigh scattering, models for polarised photons are added, as the mass model is also used to study the polarization characteristics of off-axis sources like GRBs (Chattopadhyay et al. 2019). A GEANT4 simulation ‘‘Run’’ includes generation of multiple seed photons and tracking their interaction with the volumes to obtain the required information. In AstroSat mass model simulations, CZT and veto CsI detectors are active volumes, for which the details of interaction and energy depositions are recorded for further analysis. The interactions of a given seed photon and the secondaries produced are part of an event of the Run. Energy depositions and interaction positions in the CZT and CsI detector volumes for each step of an event are accumulated. Here, the processing diverges for two different modes that we use in simulations. The first mode is the ‘‘response mode’’, where our interest is to calculate the effective response of the satellite for incident photons, broadly equivalent to a combination of the ancillary response file (ARF) and the redistribution matrix file (RMF), but not including the detector effects like energy resolution. In this mode, at the end of an event, for interactions in the CZT detectors, the pixel numbers are computed from the recorded interaction positions and total energy deposited in each pixel is calculated by adding up the energy depositions from all the interactions within a pixel. We only store events where the total energy is deposited in one pixel (single event) and discard the events where the total energy is shared in multiple pixels (multiple events). The simulation output is saved as the total number of single events as a function of energy for each CZT pixel. For veto interactions, the total energy deposited is calculated and the net spectrum (number of events as a function of energy) is saved for each of the four veto detectors. The second mode is a ‘‘polarisation mode’’, where it is important to process the photon interactions in further detail to identify Compton scattering double-pixel events for polarisation analysis. In this mode, instead of recording just the total energy deposit in a pixel, details of all the interactions (like xyz positions, energy deposition in each interaction, type of interaction etc.) for each incident photon (and its secondaries) within the detector volume are written out to a file. This file is then post-processed outside GEANT4for polarization analysis.

Page 7 of 26

93

5. Validation Validation of the mass model is crucial to understand its strengths and limitations. Like all satellites, the instruments of AstroSat were built separately and integrated together only at the final stage, leaving no opportunity for ground calibration with off-axis radioactive sources. Instead, we rely on astrophysical sources for validating the mass model. Since a large number of sources spread over the sky are contributing to the DPH and spectra, we cannot decompose the data into contributions from individual sources for such analysis. Instead, we rely on GRBs for the validation: we can estimate the DPH and spectrum of the GRB by subtracting the background contribution from all other sources as measured from time intervals just before and after the GRB. We then simulate the same GRBs by using incident spectra and fluxes from literature, and compare the simulation outputs with observations. We note an important caveat here: GRB photons are scattered by the Earth’s atmosphere, and some of these scattered photons reach the satellite, and this ‘‘albedo’’ flux can be as high has 30% of the incident flux (Palit et al. 2021). The observed magnitude of the effect depends on the direction-dependent sensitivity of the satellite, the spectrum of the GRB, and relative positions of the source, Earth, and the satellite. In extreme cases where the source is in a low sensitivity part of the sky while the Earth is in a high-sensitivity direction, the detected albedo component may even be higher than the direct GRB flux. The joint modelling of the Earth albedo and the satellite response will be taken up in a future work, here we simply note that the albedo effect may be responsible for some discrepancy in results.

5.1 GRB selection The primary criterion for selecting GRBs for validation was having a large number of photons ([5000) detected by CZTI.4 We took GRB positions reported by other missions and calculated the coordinates of each GRB in the CZTI coordinate system (Fig. 2). The angle h is defined as the angle from the Z axis, while / is the azimuthal angle measured from the X axis in the usual right-handed sense. We selected a 4

Note: This criteria forces our sample to consist of long GRBs only. The shortest duration GRB in the sample is  30 s and

the longest one is  200 s.

93

Page 8 of 26

J. Astrophys. Astr.

(2021) 42:93

Figure 5. GRB directions in CZTI reference frame. The mollweide projection is centred on the CZTI bore sight, such that h ¼ 0 ; / ¼ 0 (Z axis) is at the centre of the figure. The concentric dotted circles show constant h in steps of 15 , while the dotted lines originating from the centre are lines of constant /, in steps of 30 . The X axis (/ ¼ 0) points towards the bottom, and the Y axis points towards the right. Colours denote the total number of counts detected in CZTI.

subset of GRBs that were well-spread out in this h  / coordinate system (Fig. 5, Table 2). Finally, to validate our simulations against the observed data, we needed knowledge of the flux and spectral parameters of the bursts. We observed that incident photons with energies as high as a few MeV can get down-scattered to the 20–200 keV range of interest. However, the number of such photons in typical GRBs is very low. Based on convergence studies, we concluded that considering the incident spectrum up to 2 MeV was sufficient for our simulations. Hence, we preferred using GRBs where the spectra had been modelled to high energies, which led us to select GRBs detected by either Fermi or KonusWind missions. We use the Band model parameters (Band et al. 1993) reported in the literature to simulate the source spectrum. The normalisation is calculated using the total fluence of the burst, circumventing the problem of variation of count rates over time. The final list of selected GRBs along with references for spectral parameters is given in the Table 2.

5.2 Simulations Instead of implementing the Band function in GEANT4, we opted to simulate monochromatic photon beams at various energies, and synthesise the final output by numerically integrating over the outputs of these simulations. An advantage of this method is the possibility of reusing the same simulation outputs for any other incident spectral model, by simply changing

the weights in the final co-addition. We choose an energy grid based on the energy resolution of CZTI. The simulations start at 20 keV, with 5 keV steps till 500 keV. The step size increases to 10 keV in the 500 keV–1 MeV range, followed by 20 keV steps in the 1–2 MeV range. For each energy, we simulate 6.97 million incident photons shining on the entire satellite by using a circular source plane of 200 cm radius, generating a photon flux of 55.46 photons cm2 . Both observed data and simulations show significant flux in tantalum fluorescence lines in the 55–65 keV region (Ka1 ¼ 57:5 keV, Ka2 ¼ 56:3 keV, Kb ¼ 65:2 keV). These fluorescence lines give almost no information about the energy or direction of incident photons. Hence, we ignore the 50–70 keV band in our analysis. The satellite structure is highly absorbing at energies below 50 keV, limiting the utility of that energy range as well. Thanks to these cuts, all analyses discussed in this section are for the CZT data in the 70–200 keV band. Veto detectors are non-imaging detectors. These detectors have poorer spectral resolution and underwent limited ground calibration. Hence, we do not discuss veto data in this paper.

5.3 Comparing observations and simulations Here, we show the comparison between our observed and simulated data for a few selected GRBs. Details for the remainder of the sample are given in Appendix B.

3:36  1:47 4:67  9:50

GRB180314A 48.93 106.48 0:40  0:07

GRB170228A 51.06 109.42 0:79  0:04 -

110:07  2:65

2:92  0:11

GRB180427A 42.17 250.99 0:51  0:04

-

747:34  63:21

102:95  4:48

639:83  49:97 -

-

0.00477 0.06407 -

0:870:12 0:1

3:10:17 0:24 \  3:08 \  1:98 2:670:27 0:6

0.12519 0:680:13 0:12 0.07338 0:610:17 0:16 0.00637 1:03þ0:22 0:18 1:190:07 0:08 -

2:43

0.00959 0:680:13 0:12

2:5??

19018 17

1017657 317

11166

11644

30130 25

57155 47

27525 30

21431 23

2:450:01 0:01 2:440:19 0:24

Epy

b

Konus-Wind

0.00774 Tsvetkova et al. (2016a)w 0.02210 Svinkin et al. (2018)w, Moss et al. (2018){ 0.03144 Frederiks et al. (2017b)w, Hurley et al. (2017){ 0.11279 Tsvetkova et al. (2018b)w, Bissaldi et al. (2018){ 0.13346 Kozlova et al. (2018)w, Hurley et al. (2018){ 0.03988 Tsvetkova et al. (2018a)w 0.00490 Tsvetkova et al. (2017)w 0.03198 Frederiks et al. (2019b)w, Hurley et al. (2019){

Nz

GCN reference (in addition to Fermi burst catalog)

(2021) 42:93

GRB190117A 51.27 178.51

a

0.01752 0:930:15 0:14

Nz

1094:61  46:52 0.01415 0:71þ0:09 0:09

325:23  77:83 125:88  2:46 -

239:66  13:64

Epy

Band parameters

GRB180605A 33.9 277.37 0:66  0:04 3:13  66:78 GRB180914A 40.68 215.76 -

99.79

5:40  8:87

0.65

GRB170527A 30.19

GRB160325A

1:16  0:00

b

1:81  0:12 2:63  0:08 -

a

GRB170726A 5.46 147.88 1:19  0:06 GRB180120A 15.88 206.27 1:09  0:01 GRB180809B 26.46 198.83 -

/ 2:32  0:15

h

Fermi GBM

159.48 0:74  0:04

GRB name

Co-ordinates (deg)

CZTI

Table 2. Spectral properties of all GRBs used in the validation analysis.

J. Astrophys. Astr. Page 9 of 26 93

2:650:31 1:06 \  3:05

0:50:21 0:19

0.03341 1:050:1 0:09

13876 \  3:3

0.05564 1:420:05 0:05 88:39  3:53

0.00804 Svinkin et al. (2017)w, Melandri et al. (2017){ Frederiks et al. (2017d)w, D’Avanzo et al. (2017){ 0.04754 Frederiks et al. (2018a)w, Troja et al. (2018){ 0.10023 Tsvetkova et al. (2019)w, von Kienlin (2019) 0.07075 Svinkin et al. (2016b)w, Svinkin et al. (2016a){ 0.08483 Frederiks et al. (2018b)w, Veres et al. (2018) Sbarufatti et al. (2017){ 0.05074 Frederiks et al. (2017a)w 0.07566 Kozlova et al. (2017)

Nz

GCN reference (in addition to Fermi burst catalog)

J. Astrophys. Astr.

GRB170921B 136.68 302.73 1:28  0:25 2:41  0:07

1114180 156

9776

2:490:31 0:9

2:570:12 0:18

GRB170115B 116.27 132.60 0:88  0:01 2:47  0:36 1563:00  106:00 0.01406 0:860:06 0:06

-

1:480:12 0:1

63836 33

12576

-

0.02285 0.00984

0.02355 0.28219

0.03595 0.11191 0:930:03 0:03

-

\  3:5

17647 35

4x

1:210:2 0:35

-

30650 39

7012 12

\  2:4

-

1:20:48 0:38

Epy

b

a

Nz

Konus-Wind

Page 10 of 26

-

307:221  16:55 413:86  35:41

GRB160106A 106.12 255.69 0:62  0:04 2:14  0:07 GRB170121B 115.98 200.58 1:03  0:02 2:68  1:61

-

571:49  21:48 79:20  1:40

97.25 97.66

GRB170607B GRB180728A

-

-

169.64 0:75  0:02 2:71  0:23 146.24 1:54  0:01 2:46  0:02

81.22 91.72

GRB170511A GRB160530A

290.53 1:04  0:03

-

-

-

Epy

86:87  3:93 181:77  2:77

77.14

GRB190519A

188.27

-

-

b

263.31 1:04  0:03 2:48  0:10 67.94 0:68  0:01 2:50  0:04

73.53

GRB180325A

216.13

-

a

108:00  3:00

66.01

GRB171027A

296.33

/

Fermi GBM

Band parameters

3:1  0:30

51.86

h

GRB170822A

GRB name

Co-ordinates (deg)

CZTI

Table 2. Continued.

93 (2021) 42:93

h

/

b

Epy

Nz a

b

Konus-Wind Epy

Nz

GCN reference (in addition to Fermi burst catalog)

4. Konus-Wind spectral parameters are taken from the GCNs marked with H . 5. For GRB171027A the GCN uses cut-off power-law (CPL) to fit the spectrum, however since we only use Band function, we use b value (marked by x) which reasonably approximates the CPL.

(2021) 42:93

3. Fermi GBM spectral parameters are also taken from the GBM burst catalogue (von Kienlin et al. 2020) except for GRB190519A, GRB180728A and GRB190530A where the reference are taken from the GCNs marked with  .

1. Lack of value imply no detection by the respective instrument; y Peak energy in keV; z Band function norm at 100 keV in ph/cm2 /s/keV. 2. CZTI coordinates are computed by converting the GRB RA/Dec to the CZTI frame. The RA/Dec are taken from GBM burst catalogue (https://heasarc.gsfc.nasa. gov/W3Browse/fermi/fermigbrst.html) (von Kienlin et al. 2020) except for GRB180809B, GRB170527A, GRB180914A, GRB180427A, GRB190117A, GRB170822A, GRB171027A, GRB180325A, GRB160530A, GRB170121B, GRB160607A where they are taken from the GCNs cited with { .

GRB170614A 137.69 340.67 GRB160607A 138.85 315.77

1:12  0:04 -

a

Fermi GBM

Band parameters

7:17  309:34 348:96  28:35 0.01749 1:47 0:26 25 0.02180 Tsvetkova et al. 0:680:43 2:510:35 17642 (2016b)w, Ukwatta et al. (2016){ 0:06 GRB171010A 142.51 242.02 1:09  0:005 2:19  0:01 137:66  1:42 0.11804 1:090:04 17176 0.11693 Frederiks et al. 0:04 2:420:07 (2017c)w 0:14 GRB160720A 143.38 73.06 0:92  0:17 1:95  0:28 153:84  0:00 0.00939 0:870:04 22798 0.01888 Svinkin et al. (2016c)w 0:04 2:870:18 0:36 GRB190530A 154.50 80.31 1:00  0:01 3:64  0:12 900:00  10:00 0.19478 1:030:02 0.08921 Frederiks et al. 84844 0:02 3:030:22 33 (2019a)w, Bissaldi & Meegan (2019) 47 GRB160821A 156.18 59.27 1:05  0:00 2:30  0:02 940:62  16:02 0.03263 10:02 0.02021 Kozlova et al. (2016)w 1:990:04 0:02 0:04 71044

GRB name

Co-ordinates (deg)

CZTI

Table 2. Continued.

J. Astrophys. Astr. Page 11 of 26 93

93

Page 12 of 26

J. Astrophys. Astr.

(2021) 42:93

Figure 6. (a)–(d) Observed and simulated data for four GRBs above the detector plane. The left panels show observed, background-subtracted DPHs, while middle panels show simulated DPHs. These are binned in 4  4 pixel bins, and grey lines denote boundaries of detector modules. The DPHs are oriented such that the CZTI radiator plate is along the left side, and / ¼ 0 is along the þX axis. In this orientation, quadrants are oriented in a clockwise manner with quadrant A at the top left adjacent to the radiator plate. For instance, in case of GRB170527A (6d), quadrant C data was unusable as explained in the text. Right panels show a scatter plot of module-wise observed versus expected counts, along with fits discussed in Section 5.4.

Figures 6 and 7 show the comparison between observations and our mass model simulations for eight GRBs. The left and middle panels show the detector

plane histograms (DPH) of the background-subtracted observed data and the mass model simulation respectively. We take a list of dead and disabled pixels

J. Astrophys. Astr.

(2021) 42:93

Page 13 of 26

93

Figure 7. (a)–(d) Observed and simulated data for two GRBs incident at oblique angles, and two GRBs below the detector plane. Details are as in Fig. 6.

from the CZTI pipeline, and set those pixels to zero in the simulations as well. The complete CZTI detector plane is a 256  256 pixel grid. In these figures, we bin the DPH in 4  4 pixel bins to visually suppress the inter-pixel Poisson variations. We plot the DPH as a single image, without rendering the inter-detector or inter-quadrant spacing seen in Fig. 2(b). Grey lines denote the boundaries of individual detector modules.

Some pixels in the source DPH can be negative as a result of background subtraction and Poisson noise. For visual clarity, the minimum value of the source DPH colour bar is set to zero, such that pixels with  0 are rendered as purple. This problem does not affect the simulated DPHs, and we set the lower bound to zero for all of them. In all DPHs (observed and simulated), the upper limit of the colour bar is set

93

Page 14 of 26

J. Astrophys. Astr.

to the 99th percentile of the DPH counts to prevent a few bright pixels from skewing the rendering. The right panels show comparisons of the observed and simulated counts in each detector module, and are discussed in detail in Section 5.4. We caution the readers about two effects here. First, there are unknown contributions from Earth albedo to these DPHs, but they are not discussed in this paper. Second, the background subtraction process leaves residual Poisson noise in the source DPH which is not present in the simulation. This is discussed in Appendix A. Let us first discuss the qualitative comparisons of the observed and simulated DPHs in Figures 6 and 7. Figure 6 shows the comparisons for GRBs incident from well above the detector plane (h.60 ). We see that the simulations reproduce observations quite well in most cases. We see

(2021) 42:93

excellent agreement in the brightness patterns in GRB 180809B (Fig. 6(a)), including not just the vertical and horizontal bright bands, but also vertical dark lines within the bright band that are seen in the observed as well as in the simulated data. We note with satisfaction that for GRB 180314A and GRB 170228A which are separated by just 3 , both observed and simulated DPHs are similar (Fig. 6(b), (c)). At the same time, some features stand out. For instance, Quadrant C was saturated with noise at the time of GRB 170527A, rendering parts of the data unusable. This decreases the overall counts in that quadrant (Fig. 6(d)), an effect which is not replicated in the simulation. The two bright spots at the top of the observed DPH seem to be related to the scattering from the alpha detector holders, but we could not replicate this feature in simulations.

Table 3. Best-fit scaling parameters and reduced v2 values for detector-by-detector comparisons for all GRBs. GRB name GRB160106A GRB160325A GRB160530A GRB160607A GRB160720A GRB160821A GRB170115B GRB170121B GRB170228A GRB170511A GRB170527A GRB170607B GRB170614A GRB170726A GRB170822A GRB170921B GRB171010A GRB171027A GRB180120A GRB180314A GRB180325A GRB180427A GRB180605A GRB180728A GRB180809B GRB180914A GRB190117A GRB190519A GRB190530A 

h

/

Spectral parameter source

m

c

v2

v2f

106.12 0.65 91.72 138.85 143.38 156.18 116.27 115.98 51.06 81.22 30.19 97.25 137.69 5.46 51.86 136.68 142.51 66.01 15.88 48.93 73.53 42.17 33.90 97.66 26.46 40.68 51.27 77.14 154.50

255.69 159.48 67.94 315.77 73.06 59.27 132.60 200.58 109.42 263.31 99.79 169.64 340.67 147.88 296.33 302.73 242.02 216.13 206.27 106.48 188.27 250.99 277.37 146.24 198.83 215.76 178.51 290.53 80.31

Fermi Fermi Fermi Konus-Wind Fermi Fermi Fermi Fermi Fermi Fermi Fermi Fermi Fermi Fermi Konus-Wind Fermi Fermi Konus-Wind Fermi Fermi Konus-Wind Fermi Fermi Fermi Konus-Wind Konus-Wind Konus-Wind Fermi Fermi

0.48 0.16 0.12 1.32 0.24 1.22 0.40 0.41 0.90 0.83 0.87 0.64 0.20 0.30 0.72 0.46 0.52 1.16 0.20 0.45 0.38 0.75 0.41 0.19 21.43 0.10 0.54 0.81 0.13

6.46 70.55 48.93 30.81 31.88 222.51 21.37 23.13 15.36 15.62 82.5 2.78 16.89 47.69 10.30 10.58 313.76 14.12 21.36 10.74 5.78 20.72 20.67 -4.91 99.22 42.54 23.68 8.56 115.62

767.87 15132.10 243.43 421.51 372.18 6078.98 454.12 241.40 151.92 132.19 1129.58 414.50 7147.74 274.68 78.83 291.65 3897.44 103.07 3955.80 339.23 968.52 726.79 485.38 8250.68 9838.36 39249.08 519.76 132.49 6377.13

151.07 172.43 88.49 105.95 126.78 454.69 120.01 106.95 113.44 92.92 736.28 110.57 311.75 72.85 73.91 122.17 2164.37 61.32 182.13 131.16 57.03 645.48 129.84 102.09 724.93 342.96 145.84 128.02 1104.98

Quadrant C was extremely noisy at the time of GRB 170527A, and was excluded from the fit.

J. Astrophys. Astr.

(2021) 42:93

Page 15 of 26

93

detector plane shine on the 5 mm vertical sides of these pixels, thus giving higher counts. Our simulations slightly overestimate this effect, and also create a bright row at the bottom of the DPH. Observed and simulated DPHs for GRBs incident from below the detector plane (hJ120 ) show some peculiar features: in particular, some modules may end up being much brighter in the simulation than in observed data. Inspection of DPHs for GRB 171010A shows overall good agreement in the features: the presence of bright edges in the left and bottom of the figure, as well as a plus-shaped fainter feature running across the centre (Fig. 7(c)). However, we find that the left edge is predicted to be significantly brighter in the simulations as compared to data. GRB 160720A shows a similar problem with two bright modules in the upper left corner of the simulated DPH (Fig. 7(d)). If we ignore these two modules, the features match better: the fifth and sixth detector modules of the top row are a bit brighter, and there are fainter ‘‘blobs’’ just below the centre, and the middle of the left edge. The effects of excluding such detectors from the fit are discussed in Section 5.4.

5.4 Comparison of count rates

Figure 8. (a), (b) The original scaling relationships (dotted black line) for GRB 171010A and GRB 160720A are heavily skewed by simulated count rates of a few bright modules. Excluding these modules (red symbols) from the fit drastically improves the quality of the fit (solid black line). The dashed orange line denotes the ideal condition where observed counts are equal to predicted counts.

Figure 7 shows GRBs with incidence directions closer to or below the detector plane. GRBs incident at oblique angles show fewer features in the DPH. For instance, GRB 180325A (Fig. 7(a)) simply shows a broad gradient from upper left to lower right direction in both observations and simulations. The observed horizontal row of bright pixels in GRB 170511A (Fig. 7(b)) is the last row of pixels in quadrant B, and there is a large physical gap between these and the next row that has not been shown in the DPH. GRB photons coming at an angle of just 8:8 from the

Next, we undertake a quantitative analysis by comparing the observed and simulated counts in each detector (right panels in Figures 6 and 7). We define the v2 metric for this comparison as: X ðNsim  Nobs Þ2 v2 ¼ : ð1Þ r2sim þ r2obs The terms in this equation warrant some explanation. The observed count rate (Nobs ) is calculated as the difference between the number of counts in the GRB time window minus the background counts rate. We assume that the count rate is Poisson, and calculate r2obs by adding the errors in quadrature. Each simulation of a monochromatic beam with energy E gives N(E) detected photons. As discussed in Section 5.2, we calculate weights w(E) based on the flux and spectrum of the GRB, and obtain the simulated counts by numerically integrating over E. To calculate the total error, we assume that the uncertainty in N(E) is pffiffiffiffiffiffiffiffiffiffiffi rðEÞ ¼ NðEÞ. The final uncertainty is calculated by applying the same weights adding these uncertainties P P in quadrature: r2sim ¼ wðEÞr2 ðEÞ ¼ wðEÞNðEÞ. We note that this uncertainty calculation is not exact,

93

Page 16 of 26

J. Astrophys. Astr.

(2021) 42:93

Table 4. Comparing the 70–200 keV fluence values from Fermi and Konus-Wind parameters, in units of erg cm2 . GRB name GRB160325A GRB160530A GRB160720A GRB160821A GRB170115B GRB170228A GRB170527A GRB170921B GRB171010A GRB180314A GRB180427A GRB180728A GRB190519A GRB190530A

Fermi 6.11 3.43 8.17 1.04 1.17 4.28 1.80 2.02 2.06 7.53 2.07 1.61 1.17 6.99

9 9 9 9 9 9 9 9 9 9 9 9 9 9

-6

10 10-5 10-7 10-4 10-5 10-5 10-5-5 10-5 10-4 10-6 10-5 10-5 10-5 10-5

Konus-Wind 3.81 4.43 4.54 4.31 4.77 1.83 2.96 1.88 1.76 5.95 2.12 1.54 1.21 7.00

9 9 9 9 9 9 9 9 9 9 9 9 9 9

-6

10 10-5 10-5 10-4 10-5 10-5 10-5 10-5 10-4 10-6 10-5 10-5 10-5 10-5

Ratio (KW/Fermi)

m

0.62 1.29 55.6 4.11 4.09 3.79 1.65 9.28 0.85 0.79 1.02 0.96 1.03 1.00

0.16 0.12 0.24 1.22 0.40 0.90 0.95 0.46 0.52 0.45 0.75 0.19 0.81 1.13

We used Fermi spectral parameters for simulations of all these GRBs. The last two columns give the ratio of the the Fermi fluence to Konus-Wind fluence, and the multiplication factor required for matching Fermi-derived count rates to observed CZTI data.

but it serves as a sufficient approximation for our purposes. To quantify the visual comparisons in Section 5.3, we now focus on the plots showing observed versus simulated counts (right panels). The dashed orange lines are unity-slope lines, for cases when the observed counts are equal to the simulated counts. GRB 170228A (Fig. 6(c)), GRB 170527A (Fig. 6(d)), GRB 170511A (Fig. 7(b)) are some of the GRBs where the points cluster around these lines, showing a good agreement between observations and simulations. Any offsets from such a line would indicate improper background subtraction. In other cases like GRB 180809B and GRB 180325A (Figures 6(a), 7(a)) we see a strong linear correlation between the points, but with a different slope. We fit a straight line to this plot, and evaluate a v2f value for this fit. Since rsim robs , we ignore rsim in this calculation. This fit gives us a scaling parameter (the slope m) and an intercept (c), which are listed for all simulated GRBs in Table 3. The intercept is interpreted as a residual due to improper background subtraction, which may arise from the orbital variation in background (Bhalerao et al. 2017b). We note that the GRBs with the highest values of c are those with large h values. The intercept here is indicative of an overall poorer match between our simulations and observations. Some GRBs have a rather large v2f value, indicative of a poor fit despite scaling (Table 3). Two

examples of this were discussed in Section 5.3: GRB 171010A and GRB 160720A (Figures 7(c), (d)). The visual inspection of DPHs shows that counts in some modules have been over-predicted in both cases. For GRB 171010A if we ignore the six bright modules on the left side from the fit, we get a much better scaling relation (m0 ¼ 1:00, c0 ¼ 191) and the v2f value decreases from 2164 for 62 degrees of freedom to 952 for 56 degrees of freedom (Fig. 8(a)). A similar scenario is seen for GRB 160720A (Fig. 8(b)), where the exclusion of the upper two modules results in the v2f decreasing from 127 for 62 degrees of freedom to 79 for 60 degrees of freedom. The scaling factor changes from m ¼ 0:60  0:16 to m0 ¼ 4:1  0:5. We carefully examined the simulated data for such cases and found abnormalities in data quality. We infer that these abnormally bright patches, which strongly depend on incident direction, are a result of limitations of our mass model. Such artefacts could arise from missing components in the mass model, and can be considered as systematic errors. However, we need to characterise them and create objective criteria for excluding certain areas of the DPH for radiation incident from a given direction. In other cases, we see that the slope of our best-fit is different from unity: for instance, m ¼ 0:45 for GRB 180314A (Fig. 6(b)), and m ¼ 0:38 for GRB 180325A (Fig. 7(a)). While the bright spots seen for some simulations indicate there although there are minor

J. Astrophys. Astr.

(2021) 42:93

discrepancies in the mass model, the linear correlation demonstrates the validity of the placement of various components in the mass model and their relative opacities to the incoming radiation for these GRB directions. While detailed spectral modelling required to measure CZTI flux is beyond the scope of this paper, we can use our scaling parameter m as a proxy for ratio of fluence measured by CZTI to that of the reference spectral model. We note that no strong correlation is seen between the scaling parameter and the h or / coordinates of the GRB. It is then likely that these slopes are indicative of the uncertainty in the source properties itself. These may be measurement uncertainties, or uncertainties in extrapolating from the energy range of the detecting instrument to the CZTI energy range. To illustrate this case, we consider the twelve GRBs from our sample where spectral parameters are known from both Fermi and Konus-Wind data. On comparing the incident flux in the 70–200 keV range using both spectral models (Table 4), we find that while the inferred fluence values are in good agreement for about half the GRBs, in extreme cases they disagree by up to a factors of 9.3 (GRB 170921B) or even 55 (GRB 160720A). These discrepancies may arise from differences in the spectral parameters due to the different energy ranges of the instruments. The energy range and sensitivity differences even lead to different T90 values for the two instruments, and thus some differences in the fluence values calculated by different instruments are unsurprising. Extending the same argument, even for a perfect CZTI Mass Model and simulation, we may expect disagreements of a similar magnitude when compared Fermi or KonusWind. We see that while the flux estimates from Konus-Wind are often higher than those from Fermi spectral parameters, the range of values of m is similar to that of the ratio of Konus-Wind and Fermi fluence values.

5.5 Spectral comparisons Besides the count rate comparisons as demonstrated in the previous section, we also attempted to validate the mass model by performing broad band spectroscopy for a number of GRBs using Fermi, Swift and AstroSat-CZTI data, where the spectral response file for CZTI is generated from the GEANT4 simulations of full AstroSat mass model. Using an older version of this mass model with minor differences, a satisfactory agreement in spectral parameters between Fermi and

Page 17 of 26

93

CZTI was reported in Chattopadhyay et al. (2019) for GRB 160821A. Chattopadhyay et al. (2021) used this latest version of the mass model to expand the work to eleven bright GRBs detected in the first year of CZTI operation and showcases capability of CZTI as a subMeV spectrometer, and thereby validating this AstroSat-CZTI mass model further.

6. Discussion The mass model presents a substantial development over the raytrace codes used in Bhalerao et al. (2017b): extending our understanding of CZTI sensitivity from  30% of the sky to the entire visible sky. The all-sky median effective areas are 32.8 cm2 , 74.6 cm2 , and 68.2 cm2 at 60 keV, 120 keV and 180 keV respectively. Bhalerao et al. (2017b) considered only 30% of the sky and calculated the median effective area at 180 keV to be 190 cm2 . Based on the more detailed physics of the mass model, this number decreases to 115 cm2 for the same part of the sky. The conversion of effective area to sensitivity depends on the source spectrum, duration, as well as detector noise properties. Sharma et al. (2021) reported that the typical minimum detectable count rate for CZTI is 284 counts s1 for a 1 s burst, and 42 counts s1 (total 420 counts) for a burst with a 10 s duration. These count rates can be converted into directiondependent sensitivity by assuming a source spectrum. As a proxy for an all-sky calculation, we evaluate the sensitivity in the directions of the 29 GRBs, where we have conducted mass model simulations. For short GRBs, we consider a spectrum defined by a Band function with a ¼ 0:5, b ¼ 2:25, and Epeak ¼ 800 keV (Wanderman & Piran 2015). The light curve is assumed to comprise of a top-hat pulse with a width of 1 s. For such a burst, the CZTI fluence sensitivity ranges from  8  108 erg cm2 to  2  106 erg cm2 , with a median value of 4  107 erg cm2 . For long GRBs, we consider a spectrum defined by a Band function with a ¼ 1, b ¼ 2:25, and Epeak ¼ 511 keV (Wanderman & Piran 2010) and a 10 s top-hat pulse to get fluence sensitivities in the range from 1  107 erg cm2 to 4  106 erg cm2 with a median value of 7  107 erg cm2 . The range of these values is comparable to Fermi GBM. Detailed analyses of GRBs detected by CZTI and quantification of the sensitivity will be addressed in a later work.

93

Page 18 of 26

We discuss a few observations noted during our studies. For instance, we note that the DPH is not sensitive to the small changes in the input spectrum. Furthermore, Ta fluorescence lines dominate the observed spectrum in the 50–70 keV energy range making it difficult to ascertain the source spectrum in this energy range. Finally, for very accurate analysis of GRBs coming from the SSM direction exact position the SSM should be taken into account as the orientation of SSM may affect the observed counts distribution. However, such GRBs would have a high h, and we have found that such cases show higher disagreements between observations and simulations, independent of /. An older version of AstroSat CZTI mass model with minor differences has been successfully used to study the polarisation of prompt emission in several GRBs (Rao et al. 2016; Chattopadhyay et al. 2019; Chand et al. 2019). Singhal et al. (2021) used the same older version to measure fluxes of bright sources by the earth occultation technique. (Chattopadhyay et al. 2021) demonstrated the validity of the mass model simulations for spectroscopic analyses, using this current mass model version. Simulations of GRBs using the AstroSat CZTI mass model show satisfactory match with observed data. Here, we have demonstrated the good correspondence between observed and simulated count rates and DPHs. Our simulations with earlier versions of the mass model show that DPHs can show measurable differences over scales of  10 . This can be leveraged for localising GRBs. To aid this work, we will calculate the mass model response over the entire sky using the vertices of a level 4 HTM grid, with a nominal spacing of 5:6 . Preliminary testing shows that by comparing simulated DPHs for various points on this grid with the observed DPH, we can localise GRBs to .20 on the sky. Accurate localisation techniques will also have to correctly account for the effect of Earth albedo on the observed distribution of source photons on the detector plane. The same grid will also be utilised for future studies of counterparts to fast radio bursts and gravitational wave sources.

Acknowledgements CZT–Imager is built by a consortium of Institutes across India. The Tata Institute of Fundamental Research, Mumbai, led the effort with instrument

J. Astrophys. Astr.

(2021) 42:93

design and development. Vikram Sarabhai Space Centre, Thiruvananthapuram provided the electronic design, assembly and testing. ISRO Satellite Centre (ISAC), Bengaluru provided the mechanical design, quality consultation and project management. The Inter University Centre for Astronomy and Astrophysics (IUCAA), Pune did the Coded Mask design, instrument calibration, and Payload Operation Centre. Space Application Centre (SAC) at Ahmedabad provided the analysis software. Physical Research Laboratory (PRL) Ahmedabad, provided the polarisation detection algorithm and ground calibration. A vast number of industries participated in the fabrication and the University sector pitched in by participating in the test and evaluation of the payload. The Indian Space Research Organisation funded, managed and facilitated the project. We thank the satellite and instrument teams for sharing their design details, and engineers at ISAC, ISITE, ISRO for providing CAD files that could be used in the mass model. In particular, we thank Prof. Shyam Tandon (IUCAA, Pune), Prof. H.M. Anitia (TIFR, Mumbai), Mr. Harshit Shah (TIFR, Mumbai), and Mr. Nagabhushana S. (IIA, Bangalore) for their assistance. We thank Mr. Dhanraj Borgaonkar (IUCAA, Pune) for the help in profiling the mass model. We thank Gaurav Waratkar (IIT Bombay) and Vedant Shenoy (IIT Bombay) for assisting in the data analysis. We acknowledge the use of Vikram-100 HPC at the Physical Research Laboratory (PRL), Ahmedabad and Pegasus HPC at the Inter University Centre for Astronomy and Astrophysics (IUCAA), Pune. This work utilised various software including Python, AstroPy (Robitaille et al. 2013), NumPy (van der Walt et al 2011), Matplotlib (Hunter 2007), IDL Astrolib (Landsman 1993), FTOOLS (Blackburn 1995), C, and C??.

Appendices Appendix A: Visual impact of residual Poisson noise For a few GRBs, we see that the scatter plots of module-wise observed versus simulated counts show a good correlation, but the DPHs are visually discrepant. A key factor in this is Poisson noise present in the observed data. We illustrate this with the example of GRB 160607A. In Fig. A1, the upper left and right panels show the observed and simulated DPH for the GRB. These DPHs appear quite distinct. However, Fig. A2 shows that there is a modest correlation

J. Astrophys. Astr.

(2021) 42:93

Page 19 of 26

93

Figure A2. The observed and simulated module-wise count rates of GRB 160607 show a reasonable correlation. The best-fit line gives a scaling factor of 1.3. Figure A1. Upper left: Background-subtracted source DPH for GRB 160607A. Lower left: Simulated DPH including a residual Poisson noise component. For both these panels, the colour bar ranges from first to 99th percentile values. Upper right: Simulated mass model DPH, with the maximum set at 99th percentile value. Lower right: Simulated Poisson background noise, with the mean value subtracted. The range of the colour bar is the actual range for this residual Poisson noise.

between the observed and simulated module-wise count rates. In particular, the modules with high simulated counts (J25) also have high observed counts (J40). The key cause of this visual discrepancy is residual Poisson noise. For all of our GRBs, source DPH is obtained by creating a DPH for the GRB interval and subtracting a background DPH estimated from pre– and post–GRB intervals. These intervals are selected to be much longer than the GRB, and the count rates are scaled to the GRB duration to suppress uncertainty in the estimate of the background. However, this removes only the mean background from the actual GRB data, leaving residual noise. To demonstrate this, we create a simulated background DPH using a Poisson distribution with the expected background count rate. We then subtract the mean from this DPH to leave only the Poisson-induced noise (Fig. A1, lower right panel). Adding this residual noise component to the simulation results in the DPH shown in the lower left panel: the prominent contrast from the simulation has been lost due to the high variations added by the residual noise.

Appendix B: Comparisons for the full GRB sample In Sections 5.3 and 5.4 we have discussed the comparisons between the observed and simulated data for eight selected GRBs. Here, we discuss twenty more GRBs that were studied in this work. The observed and simulated DPHs of the on-axis GRB 160325A show good agreement, including vertical and horizontal patterns caused by shadows of the collimators (Fig. B1(a)). The DPHs for GRB 170726A, located just outside the primary field of view, is more discrepant: but in reasonable agreement in terms of counts per detector (Fig. B1(b)). In Section 5.3 we had pointed out that the observed DPH of GRB 170527A shows two bright spots caused due to scattering from the alpha detector holders (Fig. 6(d)), an effect we could not replicate in our simulations. A similar effect is seen at the bottom of the observed DPH for GRB 171010A which is incident from / ¼ 242 (Fig. 7(c)), GRB 180605A with / ¼ 277 (Fig. B1(d)), GRB 180914A with / ¼ 217 (Fig. B2(a)), and GRB 180427A with / ¼ 251 (Fig. B2(b)). Figures B3 and B4 compare the observed and simulated DPHs for GRBs incident at oblique angles, 60 \h  120 . The observed DPHs are relatively featureless here, owing to the oblique angle of incidence. In Section 5.3, we discussed GRB 170511A (Fig. 7(b)) as an example of how the edge pixels of a quadrant can get significantly higher counts in such oblique cases: an effect slightly overestimated in our simulations. This effect is also seen at the bottom edges of GRB 160106A (Fig. B4(b)), and the top edge

93

Page 20 of 26

J. Astrophys. Astr.

(2021) 42:93

Figure B1. (a)–(d) Observed and simulated DPHs for GRBs incident from above the detector plane, h  60 . Details are as in Fig. 6.

J. Astrophys. Astr.

(2021) 42:93

Page 21 of 26

93

Figure B2. (a)–(d) Observed and simulated DPHs for GRBs incident from above the detector plane, h  60 . Details are as in Fig. 6.

93

Page 22 of 26

J. Astrophys. Astr.

(2021) 42:93

Figure B3. (a)–(d) Observed and simulated DPHs for GRBs incident at oblique angles, 60 \h  120 . Details are as in Fig. 6.

J. Astrophys. Astr.

(2021) 42:93

Page 23 of 26

93

Figure B4. (a)–(d) Observed and simulated DPHs for GRBs incident at oblique angles, 60 \h  120 . Details are as in Fig. 6.

93

Page 24 of 26

J. Astrophys. Astr.

(2021) 42:93

Figure B5. (a)–(d) Observed and simulated DPHs for GRBs incident from below the detector plane, h [ 120 . Details are as in Fig. 6.

J. Astrophys. Astr.

(2021) 42:93

of GRB 160530A (Fig. B3(c)). Even the visually discrepant GRB 171027A (Fig. B3(a)) shows a good correlation in the module-wise count rates, with slope close to unity. Such visual discrepancy is discussed in Appendix A. Figure B5 shows comparisons for GRBs incident from below the focal plane (h [ 120 ). Here we see greater discrepancies between observations and simulations. This is an unsurprising effect, as the satellite body has not been modelled very accurately. As discussed in Section 5.3, a common discrepancy seen here is the presence of ‘‘hotspots’’ in the simulation: a certain region of the DPH is disproportionately brighter than the rest. Such an effect is seen in the top right modules of GRB 170614A (Fig. B5(b)) and two detector modules along the top row of GRB 190530A (Fig. B5(c)). There is still broad agreement in the observations and simulations: the bright lower edge, right side, and middle ‘‘spike’’ in the simulations of GRB 170921B can be discerned in the observed data (Fig. B5(a)). Overall, it appears as if the observed DPHs are ‘‘blurred’’ versions of the simulations, the sharper simulated DPHs are likely an artefact of our choice of clumping several satellite components into compact boxes and sheets.

References Agostinelli S., Allison J., Amako K. et al. 2003, Nuclear Instrum. Methods Phys. Res. A, 506, 250 Allison J., Amako K., Apostolakis J. et al. 2006, IEEE Trans. Nucl. Sci., 53, 270 Allison J., Amako K., Apostolakis J. et al. 2016, Nuclear Instruments and Methods in Physics Research A, 835, 186 Antia H. M., Chitinus V. R., Katoch T. V. et al. 2013, GEANT4 simulation of LAXPC background, Tech. Report, Tata Institute of Fundamental Research (TIFR) Band D., Matteson J., Ford L. et al. 1993, APJ, 413, 281 Bhalerao V., Bhattacharya D., Rao A. R., Vadawale S. 2015, GRB Coordinates Network, 18422 Bhalerao V., Bhattacharya D., Rao A. R., Vadawale S. 2017, GRB Coordinates Network, 20412 Bhalerao V., Kasliwal M., Bhattacharya D. et al. 2017a, Astrophys. J., 845, https://doi.org/10.3847/1538-4357/ aa81d2 Bhalerao V., Bhattacharya D., Vibhute A. et al. 2017b, J. Astrophys. Astr., 38, 31 Bhattacharya D., Dewangan G. C., Antia H. M. et al. 2016, AstroSat Handbook, https://www.iucaa.in/*astrosat/ AstroSat_handbook.pdf Bissaldi E., Kocevski D., Omodei N. 2018, GRB Coordinates Network, 23225, 1

Page 25 of 26

93

Bissaldi E., Meegan C. 2019, GRB Coordinates Network, 24692, 1 Blackburn J. K. 1995, Astronomical Data Analysis Software and Systems IV, 77 Chand V., Chattopadhyay T., Oganesyan G. et al. 2019, Astrophys. J., 874, 70 Chattopadhyay T., Vadawale S. V., Rao A. R., Sreekumar S., Bhattacharya D. 2014, Exp. Astron., 37, 555 Chattopadhyay T., Vadawale S. V., Aarthy E. et al. 2019, Astrophys. J., 884, 123 Chattopadhyay T., Gupta S., Sharma V. et al. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036021-09718-2 D’Avanzo P., Evans P. A., Kuin N. P. M. et al. 2017, GRB Coordinates Network, 22053, 1 Frederiks D., Golenetskii S., Aptekar R. et al. 2017a, GRB Coordinates Network, 20476, 1 Frederiks D., Golenetskii S., Aptekar R. et al. 2017b, GRB Coordinates Network, 21166, 1 Frederiks D., Golenetskii S., Aptekar R. et al. 2017c, GRB Coordinates Network, 22003, 1 Frederiks D., Golenetskii S., Aptekar R. et al. 2017d, GRB Coordinates Network, 22070, 1 Frederiks D., Golenetskii S., Aptekar R. et al. 2018a, GRB Coordinates Network, 22546, 1 Frederiks D., Golenetskii S., Aptekar R. et al. 2018b, GRB Coordinates Network, 23061, 1 Frederiks D., Golenetskii S., Aptekar R. et al. 2019a, GRB Coordinates Network, 24715, 1 Frederiks D., Golenetskii S., Aptekar R. et al. 2019b, GRB Coordinates Network, 23782, 1 Hunter J. D. 2007, Computing in Science & Engineering, 9, 90 Hurley K., Mitrofanov I. G., Golovin D. et al. 2017, GRB Coordinates Network, 21164, 1 Hurley K., Mitrofanov I. G., Golovin D. et al. 2018, GRB Coordinates Network, 22679, 1 Hurley K., Mitrofanov I. G., Golovin D. et al. 2019, GRB Coordinates Network, 23764, 1 Kasliwal M., Nakar E., Singer L. et al. 2017, Science, 358, 1559 Kozlova A., Golenetskii S., Aptekar R. et al. 2016, GRB Coordinates Network, 19842, 1 Kozlova A., Golenetskii S., Aptekar R. et al. 2017, GRB Coordinates Network, 21926, 1 Kozlova A., Golenetskii S., Aptekar R. et al. 2018, GRB Coordinates Network, 22680, 1 Landsman W. B. 1993, Astronomical Data Analysis Software and Systems II, 52 Marcinkowski R., Xiao H., Hajdas W. 2017, GRB Coordinates Network, 20387 Melandri A., D’Avanzo P., Page K. L. et al. 2017, GRB Coordinates Network, 21640, 1 Moss M. J., Barthelmy S. D., D’Elia V. et al. 2018, GRB Coordinates Network, 23105, 1 Palit S., Anumarlapudi A., Bhalerao V. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036-021-09759-7

93

Page 26 of 26

Poole C. M., Cornelius I., Trapp J. V., Langton C. M. 2012, Australas. Phys. Engg. Sci. Med., 35, 329 Rao A., Bhattacharya D., Bhalerao V., Vadawale S., Sreekumar, S. 2017, Curr. Sci., 113, https://doi.org/10. 18520/cs/v113/i04/595-598 Rao A. R., Chand V., Hingar M. K. et al. 2016, APJ, 833, 86 Robitaille T. P., Tollerud E. J., Greenfield P. et al. 2013, A&A, 558, A33 Sbarufatti B., Burrows D. N., Beardmore A. P. et al. 2017, GRB Coordinates Network, 20510, 1 Sharma V., Bhalerao V., Bhattacharya D., Rao A. R., Vadawale S. 2017, GRB Coordinates Network, 20389 Sharma Y., Marathe A., Bhalerao V. et al. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036-021-09714-6 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, in Proc. SPIE Vol. 9144, Space Telescopes and Instrumentation 2014: Ultraviolet to Gamma Ray, 91441S Singhal A., Srinivasan R., Bhalerao V. et al 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036-021-09743-1 Svinkin D., Golenetskii S., Aptekar R. et al. 2016a, GRB Coordinates Network, 19476, 1 Svinkin D., Golenetskii S., Aptekar R. et al. 2016b, GRB Coordinates Network, 19477, 1 Svinkin D., Golenetskii S., Aptekar R. et al. 2016c, GRB Coordinates Network, 19727, 1 Svinkin D., Golenetskii S., Aptekar R. et al. 2017, GRB Coordinates Network, 21679, 1 Svinkin D., Golenetskii S., Aptekar R. et al. 2018, GRB Coordinates Network, 23128, 1

J. Astrophys. Astr.

(2021) 42:93

Troja E., D’Ai A., D’Elia V. et al. 2018, GRB Coordinates Network, 22532, 1 Tsvetkova A., Golenetskii S., Aptekar R. et al. 2016a, GRB Coordinates Network, 19244, 1 Tsvetkova A., Golenetskii S., Aptekar R. et al. 2016b, GRB Coordinates Network, 19511, 1 Tsvetkova A., Golenetskii S., Aptekar R. et al. 2017, GRB Coordinates Network, 20806, 1 Tsvetkova A., Golenetskii S., Aptekar R. et al. 2018a, GRB Coordinates Network, 22513, 1 Tsvetkova A., Golenetskii S., Aptekar R. et al. 2018b, GRB Coordinates Network, 23254, 1 Tsvetkova A., Golenetskii S., Aptekar R. et al. 2019, GRB Coordinates Network, 24652, 1 Ukwatta T. N., Beardmore A. P., Evans P. A. et al. 2016, GRB Coordinates Network, 19502, 1 Vadawale S., Chattopadhyay T., Mithun N. et al. 2018, Nature Astron., 2, 50 Vadawale S. V., Chattopadhyay T., Rao A. R. et al. 2015, A&A, 578, A73 van der Walt S., Colbert S. C., Varoquaux G. 2011, Computing in Science & Engineering, 13(2), 22 Veres P., Meegan C., Mailyan B. 2018, GRB Coordinates Network, 23053, 1 von Kienlin A. 2019, GRB Coordinates Network, 24596, 1 von Kienlin A., Meegan C. A., Paciesas W. S. et al. 2020, APJ, 893, 46 Wanderman D., Piran T. 2010, MNRAS, 406, 1944 Wanderman D., Piran T. 2015, MNRAS, 448, 3026

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:73 https://doi.org/10.1007/s12036-021-09714-6

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

The search for fast transients with CZTI Y. SHARMA1,2 , A. MARATHE2,3 , V. BHALERAO2,* , V. SHENOY2 ,

G. WARATKAR2 , D. NADELLA3 , P. PAGE2 , P. HEBBAR2,4 , A. VIBHUTE5 , D. BHATTACHARYA5 , A. R. RAO6 and S. VADAWALE7 1

Division of Physics, Mathematics, and Astronomy, California Institute of Technology, Pasadena, CA 91125, USA. 2 Indian Institute of Technology Bombay, Powai, Mumbai 400 076, India. 3 National Institute of Technology Karnataka, Surathkal, Mangalore 575 025, India. 4 University of Alberta, Edmonton, AB T6G 2E1, Canada. 5 Inter-University Centre for Astronomy and Astrophysics, P.O. Bag 4, Ganeshkhind, Pune 411 007, India. 6 Tata Institute of Fundamental Research, Homi Bhabha Road, Mumbai 400 005, India. 7 Physical Research Laboratory, Ahmedabad 380 009, India. *Corresponding Author. E-mail: [email protected] MS received 3 November 2020; accepted 17 January 2021 Abstract. The Cadmium–Zinc–Telluride Imager on AstroSat has proven to be a very effective All-Sky monitor in the hard X-ray regime, detecting over three hundred GRBs and putting highly competitive upper limits on X-ray emissions from gravitational wave sources and fast radio bursts. We present the algorithms used for searching for such transient sources in CZTI data, and for calculating upper limits in case of nondetections. We introduce CIFT: the CZTI Interface for Fast Transients, a framework used to streamline these processes. We present details of 87 new GRBs detected by this framework that were previously not detected in CZTI. Keywords. Stars: gamma-ray burst: general—X-rays: bursts—methods: data analysis.

1. Introduction The Cadmium–Zinc–Telluride Imager (CZTI, Bhalerao et al. 2017b) is a high-energy coded aperture mask instrument on board AstroSat (Singh et al. 2014) CZTI comprises of four independent, identical quadrants giving a total physical area of 976 cm2 . Each quadrant consists of a 4  4 array of 5-mm thick Cadmium–Zinc–Telluride detectors, giving good sensitivity in the 20–200 keV energy range and an energy resolution of 11% at 60 keV. In nominal operations, all incident photons are saved in eventmode with 20 ls resolution. While primary coded field-of-view of CZTI is 4:6  4:6 , the collimators and support structure of CZTI become increasingly This article is part of the Special Issuse on ‘‘AstroSat: Five Years in Orbit’’.

transparent to radiation at energies above  100 keV, making it sensitive to sources all over the sky. The off-axis sensitivity depends on the effective area, which in turn is a strong function of energy and direction. Details of the effective area calculations are presented in Mate et al. (2021). As there are very few bright sources in this energy range, the net contribution of off-axis sources is small and simply manifests itself as a slightly elevated background. A special exception to this are bright, short-duration transient sources like gamma ray bursts (GRBs). GRBs with their high brightness and short durations (seconds to minutes) manifest themselves as an increase in the count rates in CZTI. Starting from the first GRB detection on the day the instrument was powered on (GRB 151006A; Bhalerao et al. 2015; Rao et al. 2016), CZTI has detected 325 GRBs in the five years since launch. On the other hand, the lack of a measurable change in count rates

73

Page 2 of 20

corresponding to a transient event can be mapped to an upper limit on the flux of the transient. With this technique, we have obtained stringent upper limits on X-ray emission from Fast Radio Bursts (Anumarlapudi et al. 2020), as well as from gravitational wave sources (Bhalerao et al. 2017a). In this paper, we describe the methods used for searching for such sources (called fast transients hereafter). In Section 2, we discuss the pre-processing of data for our searches. In Section 3, we discuss the search for ‘‘known’’ transients, where the time and possibly location are known from other sources. We also discuss methods for putting upper limits on the flux from such transients in case they are not detected in data. In Section 4, we discuss in detail the algorithms, software, and the interface developed for searching for transients in all of CZTI data. In Section 5, we discuss the performance of our software, and present the 87 transients detected in our searches. We conclude by discussing future improvements in Section 6.

2. Preparing the data The CZTI data reduction pipeline1 is designed for imaging and spectroscopy of sources in the primary field of view. There are two particular operations in the pipeline that are detrimental to the search and analysis of fast transients. First, the pipeline discards data from time intervals when the on-axis source being targeted by AstroSat is occulted behind earth — though CZTI might still detect fast transients that are located elsewhere in the sky. Second, sections of data where the count rates in detectors rise above a certain value are discarded as noisy: thus suppressing bright transients. For fast transient searches, we overcome these issues by changing a few pipeline parameters — thus ensuring that final data products are still compatible with any post-processing software. We follow the standard procedure to obtain Level-2 ‘‘bunch cleaned’’ data created by cztbunchclean. Next, when selecting good time intervals with cztgtigen, we change the config file mkfThresholds.txt to remove the earth occult condition (the ELV parameter), which would have discarded data when the on-axis target was occulted by the earth. The next stage is to reject noisy sections of data using cztpixclean. The default settings of cztpixclean discard intervals where a single pixel has more than 2 counts per second, or where a module has more than 35 counts per second. To 1

CZTI pipeline: http://astrosat-ssc.iucaa.in/?q=cztiData.

J. Astrophys. Astr.

(2021) 42:73

ensure that this step does not discard bright transients, we raise the detector count threshold to 1000 and the pixel count threshold to 100. Finally, we run cztevtclean to obtain cleaned event files. Since our processing is done independently for each quadrant, we use the _quad_clean.evt files. The next stage is to create light curves for each quadrant. Here we have to carefully correct for various sources of dead time in the instrument: for instance quadrant-level dead time (0.3 s dead time for collecting housekeeping data every 100 s), and module-wise dead time (arising from discarding particle– induced photon bunches). We use the pipeline module cztbindata to consider all these factors to correctly calculate the dead time for each time bin used. For certain searches, we also limit select the photon energy ranges in this step. The final step in data preparation is to remove the orbit-induced trends in the background. As AstroSat is in low earth orbit, the satellite sees a variable background count rate over different parts of the earth, rising near the South Atlantic Anomaly (SAA). We see that the background variations are relatively smooth, over timescales of hundreds of seconds. But, if a transient event were to evolve on comparable or longer timescales, we would not be able to distinguish it from background variations. Fortuitously, most transients of interest have timescales of tens of seconds or shorter. Hence, we can fit a smooth trend to the data and subtract it, effectively making the data ‘‘background-free’’ and greatly simplifying the task of transient detection. We have tested two methods for de-trending the data: in the first method, the trend is estimated by using a running median filter of 100 second width. In the second method, we estimate the background using a second order Savitzky-Golay (savgol) filter of 100 second width (for details see Anumarlapudi et al. 2020). Both trend estimates work well, and hence both are coded into our software. In preliminary testing, the savgol filter yielded better results for transient searches, hence it is set as the default filter.

3. Triggered searches In CZTI data analysis, searches for fast transients are broadly categorised into two types: ‘‘triggered’’ and ‘‘blind’’. Triggered searches are cases where the time of a transient, and possibly its position, are already known. For such cases, a qualitative search is carried out by pre-processing the data followed by visual

J. Astrophys. Astr.

(2021) 42:73

Page 3 of 20

examination. Blind searches, that are more quantitative, are discussed in Section 4.

3.1 Method Triggered searches start with pre-processing the data as discussed in Section 2, up to the creation of cleaned event files. We then create ‘‘spectrograms’’ or ‘‘timeenergy plots’’: two-dimensional histograms of the event data, and visually examine them for the transient (Fig. 1(a)). By default, the energy axis is binned in 10 keV bins from 20–200 keV. Searches are carried out by binning the time axis in 0.1 s, 1 s, and 10 s bins. We also calculate two further variants of this spectrogram to aid visual searches: we calculate the mean spectrum and subtract it from each time bin, thus highlighting any transient variations (Fig. 1(b)). In the third step, we take these mean-subtracted spectrograms and normalise the light curve in each energy bin by its standard deviation (Fig. 1(c)). This de-weights noisy energy bands, and gives a rough idea of the statistical significance of any transient. Light curves from a single quadrant occasionally show noise spikes which look similar to astrophysical transients. These events — often caused by charged particles or electronic noise — typically occur at low energies (.50 keV). Since the four quadrants of CZTI are electronically independent, the electronic noise events are always caused in just a single quadrant. Such noise candidates are readily rejected by requiring that any transient is considered ‘‘detected’’ only if it is

(a)

(b)

73

detected across multiple energy bins, and seen in more than one of the four independent quadrants of CZTI. Track-like events created by charged particles can sometimes be simultaneously seen in multiple quadrants. Such cases are always of short duration (\1 s), and can be discarded based on their track-like count distributions in the detector plane. Overall, four quadrant detections of transients are most unambiguous, but detections coincident in three or two quadrants are also considered acceptable if they pass the above cuts, are bright and broadband. CZTI also has caesium iodide scintillators as anticoincidence ‘‘Veto’’ detectors, to reject particle events. Veto detector spectra are sampled once per second, and downlinked along with CZT data. We generate similar spectrograms and light curves for Veto data and repeat the transient search. Since data are intrinsically binned at 1 s, the default searches are carried out only at 1 s and 10 s timescales. These searches are typically run by the Payload Operations Centre (POC) at IUCAA. Transients detected thus are reported in GCN circulars (see for instance, Gupta et al. 2020; Bhalerao et al. 2016) and announced on the CZTI GRB page at http://astrosat.iucaa.in/czti/ ?q=grb, along with the associated spectrograms.

3.2 Transient properties For every detected transient, we estimate its duration (T 90 ), peak rate (Rp ) above background (Rb ), and the total counts (Ctot ). We create a combined 20–200 keV

(c)

Figure 1. Spectrograms for quadrant C data for GRB 200306A, utilised in visual inspection of transient candidates (Section 3.1). Panel (a): The upper left frame shows raw data, binned in 1 s and 10 keV bins along the X and Y axes respectively. The upper right frame shows the spectrum, obtained by summing the spectrogram along the X axis. The lower left frame shows the light curve, obtained by summing the spectrogram along the Y axis. The lower right frame shows the distribution of count rates in the light curve. Panel (b): Mean-subtracted spectrogram, obtained by subtracting the average spectrum from each time bin. The four frames are analogous to Panel (a). Panel (c): Mean subtracted and sigma-normalised spectrogram. Note that the transient is brightest at the lowest energy bins (Panel (a)), but since those energies also have a higher sigma, the transient is statistically most significant around 60 keV (Panel (c)).

73

Page 4 of 20

J. Astrophys. Astr.

(b)

(a)

(c)

(2021) 42:73

(d)

Figure 2. Calculation of transient properties, illustrated with the light curve of GRB 200306A. Panel (a): The raw 20– 200 keV light curve summed across four quadrants, with the transient region marked in green. An initial background trend (orange) is fit to the background outside the transient region, and refined (purple) with sigma-clipping outlier rejection. Outliers are marked with red circles. The mean value of the refined trend is reported as the background count rate, Rb . Panel (b): De-trended light curve obtained by subtracting the background trend. The peak count rate (Rp ) and total counts (Ctot ) are measured from this de-trended light curve. Panel (c): A cumulative light curve calculated from Panel (b), normalised such that the median pre- and post-transient values are 0 and 1 respectively. Dashed lines indicate the points where data cross the 5% and 95% levels, which is used to calculate T 90 . Panel (d): Multiple light curves are generated from Panel (a) by assuming Poisson noise distribution, and the four parameters are measured for each of these. The four frames, clockwise from upper left, show distributions of Rb , Rp , T90 , and Ctot obtained from these light curves. These distributions are used to define 90% confidence error bars for the parameters actually measured in Panels (b) and (c). For GRB 200306A, we get þ51 þ4 þ449 1 1 Rb ¼ 495þ4 3 counts s , Rp ¼ 28919 counts s , T 90 ¼ 327 s, and Ctot ¼ 54441023 counts.

light curve from all quadrants that show a clear detection of the transient. ‘‘Pre-transient’’ and ‘‘post-transient’’ sections of the light curve are visually identified, and the background is estimated by fitting a quadratic to these. The best-fit quadratic is subtracted from the data to obtain a background-free light curve, and counts are summed to create a cumulative light curve. The posttransient part of this curve gives a measure of the total counts in the transient. The time taken for the cumulative curve to rise from 5% to 95% of the total counts is the T90 duration of the transient (Fig. 2). These details are included in the published GCN circulars.

We use a Monte-Carlo approach to estimate the uncertainties in the GRB parameters. We assume that observed photons follow a Poisson distribution, and for simplicity use the observed number of photons in each bin as the rate (k) parameter for the Poisson distribution in that bin. We create 5000 simulated light curves by drawing photons from such Poisson distributions for each bin, and measure the four parameters T90 , Rp , Rb and Ctot for each simulated light curve. We use 5–95% range in the histograms of these parameters (Fig. 2(d)) as the 90% credible intervals. For instance, the observed light curve of GRB 200306A

J. Astrophys. Astr.

(2021) 42:73

Page 5 of 20

73

yields T90 ¼ 32 s (Fig. 2(c)), while the central 90% credible region is from 25 s to 36 s (Fig. 2(d)). We report this as T90 ¼ 32þ4 7 s.

in the tw For all three bands, we process the window. In practice, we consider something a detection only if such spikes in counts are coincident across multiple quadrants, hence the actual FPP is even lower.

3.3 Count rate limits for non-detections

3.4 Flux calculations

In cases where no transient is seen, we can place upper limits on the maximum counts received from the transient that would be consistent with noise. Since the mean background level varies through the orbit, we cannot use a direct rate. Instead, we de-trend the data as discussed in Section 2. In addition, due to the noise spikes discussed in Section 3.1, the distribution of count rates deviates significantly from a simple Poisson or Normal distribution. In particular, there is a large tail of positive counts with respect to the mean rate which can mimic transient signals. To overcome this hurdle of an unmodeled count rate distribution, we estimate the upper limits (hereafter referred to as cutoff rates) using data from nearby orbits. The method is based on the assumption that the rate of astrophysical transients detectable by CZTI is low enough that nearby orbits are unlikely to have a large number of transients. We first decide the width of the window used for transient search, say tw ¼ 100 s, and an acceptable false positives probability (FPP, F ). We typically set F ¼ 0:1 for a single quadrant. Since we place limits using data from all four independent quadrants, the combined FPP is 104 . We now need to find a ‘‘cut-off rate’’ Rc such that the probability of this threshold being crossed by chance in tw is F . To calculate Rc , we select five orbits before and after the transient (excluding the orbit containing the transient) as ‘‘witness’’ orbits. We create light curves for these orbits using the same time bin as used in the original analysis, then de-trend them, and create histograms of the de-trended counts. Rc is defined as the point such that a fraction F of the data points have counts [ Rc . A typical orbit has 4000–5000 s of usable data, so that analysis of ten orbits with parameters F ¼ 0:1 and tw ¼ 100 s ensure that 40–50 data points are above Rc . This makes the method robust to the presence of another transient in the witness orbits. There are some caveats to be noted here. Occasionally, a quadrant can be extremely noisy in some orbit. If the candidate transient is in such an orbit, that quadrant is excluded from further analysis and there is a corresponding decrease in the FPP (for instance, Mate et al. 2017; Marathe et al. 2019). Our FPP estimates are derived from the probability of getting counts [ Rc in each of the four quadrants anywhere

Incident photons from off-axis transients are heavily re-processed (scattering, absorption, fluorescence, etc) by various satellite elements before they are incident on the detector. Hence, the mapping of incident spectra to measured spectra must be done by simulating these effects in software. We accomplish this by using a GEANT4-based mass model of the entire satellite (Mate et al. 2021). Since the effect of the satellite varies with direction, the simulations require knowledge of the source position in satellite coordinates. For transients where the position is known, Chattopadhyay et al. (2021) discuss a method of estimating the source spectrum and flux from CZTI data. While methods for calculating the source spectrum are still under development, we have found that source flux calculations based on the mass model are quite reliable if the source spectrum is known from other instruments. We leverage this by assuming a powerlaw or band model spectrum for sources, and calculating the flux corresponding to the number of counts in a quadrant. The total flux from the source is the sum of fluxes in all four quadrants. For certain transients, most notably gravitational wave events, the source location is not known precisely. Instead, discovery teams provide a sky-map with the source position probability distribution. For such sources, we evaluate the flux limit at each point on the sky map that is not occulted by the Earth at the instant of the transient. The overall flux limit is evaluated as a probability-weighted mean of these values (for instance, see Shenoy et al. 2020).

4. Blind searches for transients The triggered searches are complemented by a broad ‘‘blind’’ search over all of CZTI data to identify astrophysical transients. We have two pipelines for such searches — a pipeline based on Machine Learning (ML) (Abraham et al. 2019) and the CIFT.2 In this section, we discuss CIFT in detail. 2

CIFT is pronounced as sift.

73

Page 6 of 20

The broad outline for the CIFT searches is as follows: First, data are reduced and de-trended as discussed in Section 2. Various algorithms are used to identify outliers in light curves. These outliers are used to create ‘peak maps’ to identify candidate transients in data. Flagged candidates are displayed on an interface for human vetting. They undergo similar quality checks and inspection as discussed in Section 3, and final selected transients are saved in a database.

4.1 Preparing the data CZTI Level 2 bunch cleaned files are organised into ‘Obs-ID’s which have all the data taken during observations of any particular object requested by an observer. We undertake most of our searches Obs-ID wise, thus typically processing a few to a dozen orbits at a time. We see that noise events are more frequent in lower energies, while data are cleanest at higher energies. To leverage this factor, we divide CZTI data into three energy bands: 20–50 keV, for 50–100 keV, and 100–200 keV. For all three bands, we process the data following steps from Section 2, and create detrended light curves with 0.1 s, 1 s, and 10 s bins. We also use a 0.01 s binning when searching for counterparts to fast radio bursts. We use the entire energy range for the Veto detector, and create light curves at 1 s and 10 s binning. Thus, we generally create 36 light curves for CZTI data (3 time bins  3 energy bands  4 quadrants) and 8 light curves for Veto data (2 time bins  4 quadrants) per Obs-ID. We run a search algorithm on each light curve to identify outliers and create ‘peak maps’: boolean masks with value 1 for time bins containing the outliers, and 0 elsewhere. The twelve CZTI peak maps are added together, and any bin with a mask value of four or higher is flagged as a candidate transient. Similarly, the four Veto masks are combined and bins with mask value  3 are flagged as candidate transients. Next, we discuss the three outlier search algorithms currently implemented in CIFT. 4.2 Top-N The Top-N (TN) algorithm is based on a simple heuristic: a transient is expected to have among the highest count rates seen in a given light curve. We identify the brightest N bins in a light curve and flag them as outliers for the peak map.

J. Astrophys. Astr.

(2021) 42:73

While testing this algorithm, we obtained better results if the searches were carried out one orbit at a time (as opposed to Obs-ID wise searches for other algorithms). By varying values of N, we obtained the best results for N ¼ 3.

4.3 N-sigma The N-Sigma (NS) algorithm is a straightforward statistics-based method to select outliers in a time series. We identify outliers by using iterative sigma clipping as implemented in the Astropy sigma_clipped_stats module. Starting with a de-trended light curve, we calculate the median and standard deviation (r) values, and reject outliers that deviate more than 3r from the median. The process is repeated with the new light curve until convergence is attained, subject to a maximum cap of five iterations. The mean value l and the standard deviation r of the final iteration become the key parameters of algorithm. Using these values, outliers are defined as data points with counts [ l þ Nr, where our default value is N ¼ 5. The typical thresholds for flagging these outliers for various time bins, energy bands, and both detector types are given in Table 1. These values were calculated from data of entire five years of the search. We reiterate that the namesake N of this method is used only in identifying outliers for the peak map, while the iterative sigma estimation is always done at a three-sigma level.

4.4 Cutoff rates based on false positive probability The cutoff rate based search (CR) algorithm aims at attaining a given False Positive Probability (FPP) for candidate transients. Cutoff rates are determined following the procedure discussed in Section 3.3, with one important distinction. In Section 3.3, we assumed the presence of transient-free data of an order of magnitude larger duration than the timespan of interest. Since CIFT searches are meant to be conducted over all available data, this requirement clearly cannot be met. Instead, we set our FPP threshold based on the expected rates of transients, in particular, GRBs. The rate of detectable GRBs is a function of instrument sensitivity, energy range, and field-ofview. As a baseline, we note that on average Fermi GBM detects a GRB every 1.5 days (von Kienlin et al. 2020), while the BAT on the Neil Gehrels Swift

J. Astrophys. Astr.

(2021) 42:73

Page 7 of 20

Table 1. Combined cut-offs for cutoff rate and NSigma methods for each binning and band. These representative rates were calculated by using all the the five years of data used in this study. Note that rates are in units of counts/s, not counts per bin. Band-wise cutoff Method CZTI Cutoff rate

NSigma

Method Veto Cutoff rate NSigma

Binning (s)

0

1

2

0.1 1.0 10.0 0.1 1.0 10.0

1954 263 28 396 137 41

488 107 22 410 133 38

428 102 22 407 131 38

Binning (s)

Combined cutoffs

1.0 10.0 1.0 10.0

319 690 394 1150

Observatory averages one GRB every four days (Lien et al. 2016). Based on these we stipulate a rough upper bound of the rate of GRBs detectable by CZTI as 0.5 GRBs per day.3 We then stipulate that only 1% of our GRBs may be false positives (FPP = 0.01), corresponding to one false positive every 200 days. To arrive at an approximate solution for the FPP criterion, we consider the case of searching for a GRB with 1 s duration in light curves with 1 s binning. In this scenario, our false positive requirement of 1 per 200 days maps to one false positive in 1:728  107 bins. Since most basic acceptance criterion is coincident detection in two or more independent quadrants, each quadrant can have one false positive in pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 17280000 time bins, or 4156 s. This is a significant fraction of an orbit, and hence the robust estimation of Rc requires data from several orbits. Decreasing the time bin size increases the number of samples in the light curve, and owing to the random underlying process, makes outliers more likely. To correct for this, we change our cutoff rate requirements based on the bin size tbin : Rc is selected such that a fraction 0:01  ðtbin =4156sÞ of bins have a count rate [ Rc . We note that this is a highly simplified argument, which ignores the 12 light curves we make for every 3

We note that the subsequent arguments become stronger if the actual detected rate is lower as was expected. After completing the search, indeed we found a much lower GRB rate.

73

time bin and the [ 4 peak map condition. It also ignores the small effect of presence of transients in our ‘‘witness’’ data sets. However, it serves as a good approximate argument for selecting our Rc thresholds from data. The typical thresholds for flagging these outliers for various time bins, energy bands, and both detector types are given in Table 1. As in the NS method, these representative rates shown in the table were calculated with all five years included in this work. Some specialised searches use the entire 20–200 keV range as a single band. For such searches with 1 s binning, the cutoff rates for the 4 quadrants are 79, 68, 68, and 69 counts/sec respectively. For searches with 10 s binning, the rates drop to 10, 10, 10, and 12 counts/s respectively, corresponding to a total of 420 counts per 10 s bin.

4.5 The CIFT interface Once the peak maps have been created by any of the three algorithms discussed above, we apply our candidate selection criteria of requiring  4 matches out of 12 light curves for CZTI, and at least three matches out of four Veto light curves (Section 4.1). Candidate transients that meet this requirement are flagged as an ‘‘event’’, and entered into an SQL database. Certain basic properties like like number of quadrants and energy bands an event was detected in, their significance, rates above background, time since last SAA, time from next SAA, etc are also calculated and stored in the database. Events having the same trigger time (for instance, if they were detected by two different algorithms) are grouped, and their corresponding event-IDs are stored under a unique trigger-ID in a separate table. Furthermore, the trigger-events which are within 100 seconds of each other are grouped into a ‘‘superevent’’ and assigned a super-ID. These superevents are the final transient candidates, ready for human inspection. A separate program for plotting is run in parallel which takes input a list of Obs-IDs and fetches all the superevents in those Obs-IDs from the SQL database. For each superevent, it plots detailed time energy histograms, light curves and calculates T90 for each temporal binning. The CZTI Interface For Transients (CIFT) is a Flask4-based interface with SQL database as back4

https://pypi.org/project/Flask/.

73

Page 8 of 20

J. Astrophys. Astr.

(2021) 42:73

Figure 3. Screenshots of CIFT, showing functionality of various pages. Panel (a): The screenshot of the home page of CIFT, where the human scanner can input dates and the corresponding candidate tag which the scanner wants to see, refer Section 4.5. This main page also allows the user to navigate to other functionalities of the interface where one can add new tags, process the available unprocessed directories and access the diagnostics page, with few clicks. Panel (b): The SQL database displays all the candidates of the specified tag from all Obs-IDs contained within the date range specified on the CIFT main page. Each candidate has a dedicated row where the Superevent-ID, trigger time, T90 in both CZTI and Veto, number of sub-events are displayed along with relevant statistics like background rate, peak rate, number of quadrants where the candidate was detected for quick reference of the scanner. Each row also has a appropriate light curve thumbnail for both CZTI and Veto for visual inspection, allowing the scanner to discard the very obvious bogus candidates from this page itself (with the help of discard multiple option). This complete list of candidates is sorted in the ascending order of the Superevent-ID. Panel (c): Each candidate is linked to their inspection page which displays the break-down of all the computed characteristics shown on the Scanning page. The inspection page also contains links to five different lightcurves for different binnings of CZTI and Veto detectors. Based on the inspection of all these parameters and lightcurves, the scanner can classify the candidate and tag the candidate with the appropriate tag.

J. Astrophys. Astr.

(2021) 42:73

Page 9 of 20

73

Figure 3. Continued.

end, available to view the candidate transients (Fig. 3). The interface allows a human scanner to search all superevents by Obs-ID which are displayed in a table on a ‘scanning’ page (Fig. 3(b)). The scanning page has columns for Superevent-ID, trigger time, T90 in CZTI and Veto, number of sub-events, a column displaying relevant statistics like background rate, peak rate, number of quadrants the candidate was detected in, etc. and a check-box option to discard multiple superevents at once if bogus. Each superevent-ID is linked to an inspect page (Fig. 3(c)) which lists all characteristics of the superevent, and of each sub-events contained within it, along with several lightcurves of different binning sizes for CZTI and Veto. After inspection, a human scanner can tag the event with custom tags, including ‘‘known’’, ‘‘unknown’’, ‘‘ambiguous’’, ‘‘SAA Tentacle’’, etc. Superevents can be searched and filtered by tags from the main page (Fig. 3(a)). The CIFT interface also has other features like undertaking triggered searches and a front-end for initiating data processing.

5. Results We used our framework to search for GRBs in data from 06 October 2015 when CZTI was first powered on, till 10 October 2020 — spanning just over five years of data. ‘‘Slew’’ Obs-IDs are relatively short data sets acquired when AstroSat is slewing from one

source to another. These have been excluded from our search. We detected a total of 347 transients in CZTI data by using CIFT. Of these, 41 are GRBs or triggers previously reported by other missions but missed by POC triggered searches or ML pipeline (Section 5.2), while 46 are new discoveries (Section 5.3). In the same five-year span, triggered searches and the ML pipeline have detected 325 GRBs, of which our searches recovered 260. Two of these missed GRBs were in slew orbits. The reasons for missing  20% GRBs are discussed in Section 5.5.

5.1 Performance The processing code takes less than an hour to search for transient candidates in one month of data (approximately 130 GB). Creating diagnostic plots is a slower process which is spawned in parallel, and takes 3–4 hours to complete. Users remotely connect to the http-based interface for scanning the processed data. Visual examination of candidates from a month of data takes a few hours for an experienced user. Figure 8 shows the break-up of transient detections by the various algorithms. We see that most transients are detected by all three algorithms, followed by detections in both CS and TN. The TN method is solely responsible for the detection of 15% of Veto transients.

73

Page 10 of 20

J. Astrophys. Astr.

(2021) 42:73

Table 2. Comparison of the three search algorithms running on CZTI and Veto for different classifications of the candidates identified. The ‘Candidates’ column contains all potential transient candidates identified by our pipeline. The ‘Common with triggered or ML searches’ column contains all GRBs that were originally detected by triggered or ML searches on CZTI data. The ‘Known transients’ column contains all transients that had previously been reported by other instruments but had not been identified in CZTI or Veto data. The column ‘Discoveries’ comprises of all transients that have not been reported by any instrument before. Common events in various methods are shown in Fig. 8. Algorithm

Candidates

Common with triggered or ML searches

Known transients

New discoveries

CZTI Cutoff rate NSigma TopN

1290 2082 4199

206 164 210

16 7 19

29 19 30

Veto Cutoff rate NSigma TopN

10375 10625 13993

191 178 222

24 18 34

29 23 32

Total

22564

260

41

46

Table 2 summarises the performance of all algorithms. We see that there are a large number of false positives, particularly from the Veto detectors. This underscores the need for human vetting of the candidate superevents. On an average, CIFT flags about 339 candidates per month, adding up to 19628 candidates in 58 months of data. For the months of April and May 2020, we lowered the thresholds to search for even faint bursts associated with the outburst of the galactic magnetar/ FRB candidate SGR 1935?2154 (Mereghetti et al. 2020). We selected the top 5 peaks in the TN method, and required a coincidence of just 2 bands out of 12 in CR and NS methods. These reduced thresholds increased the number of candidates by a factor of 4.3, giving 2936 candidates in just 2 months. The most common type of false positives comprised of coincident detections in two Veto quadrants in just a single second, with no discernible signal in adjacent bins. These are most likely particle events, and are rejected. A closely associated class of Veto false positives are events that have a very sharp rise and an exponential decay: again a profile common for particle events. On the other hand, Veto light curves of GRBs that are also detected in CZT detectors show a wider variety. Hence we decided to keep the coincidence threshold for Veto as 3 out of 4 quadrants at the expense of missing possible real transients, and this was the number discussed at the start of Section 4. Other large number of bogus detections include false peaks near SAA due to bad de-trending or

inadequate SAA masking which can be ruled out during human vetting. In CZTI data, many false events are caused by a single pixel, generating noise events at all energies. Visual examination of the distribution of counts in the detector helps to quickly dismiss these as false positives. If the light curves are well-behaved with no real transients or noise spikes, then the TN algorithm often generates false positives by identifying ‘‘outliers’’ that are completely consistent with background. As human scanners gain more experience with the pathologies of false positives, we are working to improve automatic rejection of such candidates.

5.2 Known transients We detected 41 transients (referred as ‘Known’) that had previously been reported by other instruments but had not been identified in CZTI or Veto data (Fig. 4). These transients were matched to earlier reports in GCN Circulars,5 Fermi GBM Burst Catalog6 and the Fermi sub-threshold trigger lists.7; 8 Table 3 lists the key properties of these transients: a superevent ID, standard GRB name, trigger times (UTC), algorithms that detected the transient in CZTI or Veto data, 5

https://gcn.gsfc.nasa.gov/gcn3_archive.html. https://heasarc.gsfc.nasa.gov/W3Browse/fermi/fermigbrst.html. 7 https://gcn.gsfc.nasa.gov/fermi_gbm_subthresh_archive.html. 8 https://gammaray.nsstc.nasa.gov/gbm/science/sgrb_search.html. 6

J. Astrophys. Astr.

(2021) 42:73

Page 11 of 20

73

(a)

(b) Figure 4. The normalised lightcurves of GRBs detected by CIFT, that were reported by other instruments but had not been identified in CZTI or Veto data (Section 5.2). Each GRB lightcurve is normalised and labeled with the GRB name. Panel (a) shows normalised lightcurves for the GRBs detected in CZTI. The three sub-panels are with 0.1 s, 1 s and 10 s binning respectively, and each sub-panel is ordered by peak count rate above background, increasing from top to bottom. Panel (b) shows the normalised lightcurves of GRBs that were detected in Veto. These are plotted with a 1 s binning, and are also ordered by peak count rate above background, increasing from top to bottom.

temporal binning used in analysis, and the peak time (AstroSat time, measured as seconds since UT 2010-

01-01, 00:00:00). We then list the calculated parameters: the duration (T90 ), peak count rates above

73

Page 12 of 20

background, background count rates and total counts across all quadrants. We prefer using CZTI data to calculate these parameters. Even when our algorithms find a transient only in Veto detectors, we manually check if CZTI data can be used for calculation for uniformity. We use Veto data to calculate transient properties only if the transient is unseen in CZTI light curves. These cases are demarcated clearly in Table 3.

5.3 CIFT discoveries We discovered 46 new transients that have not been reported by any instrument before. As in Section 5.2, we show their light curves in Fig. 5 and list properties in Table 4. Six of these transients have been published already: GRB 180112B (Sharma et al. 2018), GRB 190628B (Marathe et al. 2019), GRB 191102A (Shenoy et al. 2019a), GRB 191105B (Shenoy et al. 2019b), GRB 191119A (Shenoy et al. 2019c), and GRB 200817B (Shenoy et al. 2020).

5.4 Properties of new transients The new transients detected by CIFT (Sections 5.2 and 5.3) span a wide range of properties. The shortest transient was GRB 200907A (T90 ¼ 0:13 s), while the longest was GRB 180809C with T90 of 290 sec. GRB 200510B had the highest count rate above background (6461.6 count/s), while GRB 200906B had the lowest (53.7 count/s). Figure 6 shows the distributions of T90 , peak count rate, and total counts for the four classes of transients: (a) Those reported in the past by CZTI POC, (b) transients reported by POC which were also found by CIFT, (c) CIFT-detected transients reported by other instruments, and (d) new CIFT discoveries. We observe that all four classes have similar distributions of T90 . A notable difference is seen in the total counts: transients with higher number of total CZT counts tend to be easily detected in regular triggered and ML searches. Also, GRBs with low peak count rates are more likely to be found in triggered searches undertaken by the POC but missed by CIFT. Note that although the three classes ‘‘POCGRBs’’, ‘‘Known GRBs’’, and ‘‘New Discovered’’ are mutually exclusive, the distributions overlap well at the faint end of the distribution.

J. Astrophys. Astr.

(2021) 42:73

We find that about 10% of all GRBs detected by CZTI are short GRBs, and the fraction remains the same for the 87 new bursts discovered with CIFT. The fraction of short GRBs is similar to the values for Swift-BAT (Lien et al. 2016), but smaller than the 26% measured in Fermi (von Kienlin et al. 2020). Here, we note an important caveat that we draw the line between short and long GRBs at the canonical value of T90 ¼ 2 s, but it is known that this can be different for different instruments and will have to be measured separately for CZTI. As an illustration, if we adapt the Fermi boundary of 6.1 s, we find that about one-third of all CZTI GRBs are short GRBs.

5.5 Transients missed by CIFT Sixty-five GRBs that were found in regular triggered ? ML searches were missed in the blind search with CIFT. Two of the missed GRBs were in AstroSat slew orbits which were skipped while processing, as mentioned in Section 5. We analysed the remaining cases to find the reasons why these were missed. The most common reason for the missed GRBs was that the transients were too faint in terms of their peak count rates. For instance, Fig. 7 shows the multi-quadrant, multi-band light curves for GRB 190605A. Visually, it is clear that the GRB is only weakly detected in all three search bands in CZTI data. In order to quantify this further, we calculated the count rates that would have been necessary to flag a data point as an outlier in the peak maps for this orbit. These rates for the CS method are shown with dashed lines, while the 5-r rates for NS are shown with dotted lines. It is clearly seen that the transient is well below these rates. Such transients are rather easily confirmed by a human scanner inspecting the spectrogram and finding similar patterns in multiple quadrants. For quantitative analysis with say the CS method, the search window for a triggered search is usually set to 100 s, much smaller than the 4156 s window used in blind searches. This results in a lower cutoff rate, and will make more such fainter transients detectable in the current CIFT framework. Similarly, a smaller search window enables lowering the NS threshold from 5-r to 4-r or 3-r thanks to the fewer data points present, thereby increasing the odds of detecting fainter transients.

J. Astrophys. Astr.

(2021) 42:73

Page 13 of 20

73

Table 3. The table contains the calculated parameters for all ‘known GRBs’, which are the GRBs that had previously been reported by other instruments but had not been identified in CZTI or Veto data.

SuperID

Time (UTC)

GRB name CZTI CZTI GRB160324A5

S200724370.0

GRB160512A3

04:46:08

S204095177.0

GRB160620A

05:06:15

S219728856.0

GRB161218A

3:47:34

S224556541.0

GRB170212A2

00:48:59

S229730480.0

GRB170412By

22:01:18

S230047201.0

GRB170416A4

13:59:59

S239852052.0

GRB170808Cy

01:34:10

S250719370.0

GRB171211By

20:16:08

S254815405.0

GRB180128B1

06:03:23

S269585348.0

GRB180718Cy

04:49:06

S271117239.0

GRB180804Ay

22:20:37

S272640643.0

GRB180822B2

S272942427.0

y

y

y

15:58:34

y

13:30:41

GRB180826A2

y

01:20:25

S279646944.0

GRB181111Ay

15:42:22

S280064235.0

GRB181116Ay

11:37:13

S280166126.0

GRB181117A3

15:55:24

S326222331.0

GRB200503B

17:18:50

S337184665.0

GRB200907A3

14:24:24

Detected: Analyzed: S185452630.0

Veto CZTI GRB151117Ay

10:37:08

S190192926.0

GRB160111A

07:22:04

S222932540.0

GRB170124B4

S241696600.0

GRB170829B2

S250224944.0

GRB171206A1

S254611360.0

y

Peak time (s)

Bin (s)

T90 (s)

Peak rate (cps)

Bkg rate (cps)

Total counts (counts)

4d

Detected: Analyzed: S196531116.0

y

Algorithm

y

05:42:18

y

09:56:38

y

2:55:42

GRB180125A3

y

21:22:38

S257333099.0

GRB180226A

09:24:57

S264915479.0

GRB180525Ay

03:37:57

C: CR, TN V: CR, NS, TN C: CR, TN V: CR C: CR, NS, TN V: CR, NS, TN C: CR, NS, TN V: NS, TN C: CR, NS, TN V: CR, TN C: CR, NS, TN V: None C: TN V: CR, NS, TN C: CR, NS, TN V: None C: CR, TN V: TN C: CR, TN V: NS, TN C: CR, TN V: None C: CR, NS, TN V: CR, NS, TN C: CR, TN V: None C: CR, TN V: CR, NS, TN C: TN V: None C: CR, TN V: CR, NS, TN C: CR, TN V: CR, NS, TN C: TN V: None C: CR, NS, TN V: CR, NS, TN

196531117.5

1

48þ38 17

120þ30 14

209:2þ0:9 1:0

1889þ570 589

200724382.5

1

26þ6 4

237þ43 26

336þ2 3

1854þ347 382

204095178.4

0.1

0:99þ0:62 0:17

1385þ218 222

447þ8 10

714þ101 103

219728856.1

0.1

6þ1 1

457þ190 18

423þ10 12

1134þ227 276

224556540.2

0.1

4þ2 1

353þ150 39

316þ8 6

536þ135 189

229730479.5

1

51þ7 17

318þ44 47

425þ3 3

2589þ667 689

230047201.2

0.1

5:9þ0:2 1:4

288þ141 26

333þ8 7

551þ115 135

239852052.4

0.1

4:5þ0:4 0:5

772þ196 86

422þ8 12

1106þ185 179

250719471.5

1

135þ8 8

278þ41 37

358þ2 2

4827þ697 641

254815407.2

0.1

5:7þ0:7 0:8

297þ195 17

466þ9 13

814þ190 200

269585341.5

1

10þ6 4

217þ38 40

363þ3 2

1024þ157 230

271117238.8

0.1

5:2þ0:6 0:8

695þ209 43

434þ7 8

1313þ152 142

272640638.0

10

126þ64 110

72þ11 10

352:4þ0:4 0:7

1589þ712 705

272942423.0

10

130þ11 14

119þ12 13

493:5þ0:8 1:1

6535þ565 551

279646935.5

1

14þ3 3

192þ41 42

480þ3 3

1138þ176 210

280064267.5

1

74þ9 4

316þ46 35

454þ2 2

5805þ590 626

280166128.5

1

12þ3 3

133þ37 23

328þ2 3

921þ208 174

326222323.5

1

62þ11 12

155þ43 25

155þ43 25

2372þ656 696

337184664.7

.01

0:13þ0:01 0:02

4718þ1176 1079

476þ16 47

260þ33 29

C: None V: CR, TN C: None V: TN C: None V: CR, TN C: None V: TN C: None V: TN C: None V: TN C: None V: NS, TN C: None V: TN

185452622.8

0.1

5þ2 1

267þ168 9

383þ5 8

493þ150 155

190192925.5

.01

903þ186 170

431þ8 10

126þ84 88

222932533.6

0.1

10:3þ0:5 1:2

264þ160 17

328þ6 6

897þ187 154

241696628.5

1

45þ4 3

182þ35 36

352þ3 2

2320þ297 430

250224944.7

0.1

3þ1 1

272þ170 13

477þ9 11

297þ156 155

254611351.5

1

24þ14 6

þ51:8 95:50:6

447þ3 3

1352þ385 394

257333099.3

0.1

0:78þ0:57 0:53

346þ159 50

340þ8 7

152þ61 57

264915479.1

.01

0:16þ0:02 0:05

1128þ798 120

487þ20 41

88þ27 25

73

Page 14 of 20

J. Astrophys. Astr.

(2021) 42:73

Table 3. Continued. Peak rate (cps)

Bkg rate (cps)

Total counts (counts)

18þ1 6

122þ47 6

391þ3 2

1164þ180 185

1

27þ5 5

242þ50 38

707þ3 4

2498þ457 474

273953548.1

0.1

5:8þ0:5 1:9

250þ208 35

517þ10 12

569þ183 177

288536007.5

1

10:4þ0:6 2:9

214þ40 24

341þ2 3

1030þ136 145

197170442.4

1

14þ14 4

385þ60 59

952þ5 4

1626þ459 581

207052810.8

1

16þ9 7

980þ80 86

1560þ5 6

3098þ632 666

224077218.4

1

218þ68 30

1511þ5 6

2193þ594 638

249867182.8

1

798þ76 86

1555þ5 7

853þ166 185

258277397.7

1

25þ5 8

310þ60 61

1041þ3 4

2484þ389 416

264108761.5

1

20þ4 2

441þ71 50

1555þ5 6

6091þ528 582

267468140.9

1

64þ2 4

518þ70 48

1163þ4 4

6986þ773 1088

278544285.6

1

10þ1 5

392þ66 70

1309þ5 5

1286þ278 307

280573745.5

1

44þ3 13

237þ60 45

1113þ4 5

2634þ566 623

289607814.3

1

199þ89 2

1563þ6 7

3597þ1454 1499

Peak time (s)

Algorithm

GRB name

Time (UTC)

S266353140.0

GRB180610Cy

18:58:58

266353130.5

1

S267371600.0

GRB180622B2

267371591.5

S273953547.0

GRB180906By

18:12:25

S288536007.0

GRB190222By

12:53:25

C: None V: CR, TN C: None V: TN C: None V: CR, TN C: None V: CR, NS, TN

Detected: Analyzed: S197170443.0

Veto Veto GRB160401By

01:34:01

C: None V: CR, NS, C: None V: CR, NS, C: None V: CR, NS, C: None V: TN C: None V: CR, TN C: None V: CR, NS, C: None V: CR, NS, C: None V: CR, NS, C: None V: CR, TN C: None V: CR, TN

SuperID

S207052803.0 S224077190.0

y

GRB160724Ay

13:53:18

10:40:03

GRB170206C

11:39:48

S249867183.0

GRB171201A

S258277398.0

GRB180309A1

07:43:16

S264108750.0

GRB180515Ay

19:32:28

S267468139.0 S278544286.0

23:33:01 y

GRB180623A

16:42:17

GRB181029Ay

21:24:44 y

S280573750.0

GRB181122A2

09:09:08

S289607820.0

GRB190306Cy

22:36:58

Bin (s)

TN TN

T90 (s)

TN

TN TN TN

The table is divided into three parts classified by what detector was used to detect (‘Detected’) and compute the parameters (‘Analysed’) given in the table. The column ‘SuperID’ gives the name of the superevent identified by the pipeline. The column ‘GRB Name’ contains the published name of the GRB, linked to the GRB report (more details in 5.2). Several of these entries in the ‘GRB Name’ column have a mark against their names, which gives the information on which quadrants were used for calculating the other parameters for that GRB. If there is no mark, then all four quadrants are used. Otherwise, marks ‘1’, ‘2’, ‘3’, ‘4’, and ‘5’ refer to the quadrant sets – ‘A, B, C’, ‘A, B, D’, ‘A, C, D’, ‘B, C, D’ and ‘C, D’ respectively. The column ‘Algorithm’ tells us what algorithms detected the GRB, where ‘TN’, ‘NS’, ‘CR’ stands for the three algorithms – TopN, N-sigma, and Cut-off rate respectively while ‘C’ and ‘V’ are the two detectors – CZTI and Veto. The time in AstroSat seconds where the GRB was brightest is given in the column ‘Peak time’. The bin size, that was used to generate the parameters – T90 , Peak count rate above background, Background rate and Total counts, is mentioned in the column ‘Bin’. GRBs marked with a dagger (y ) were obtained from the FERMIGBRST Catalogue (von Kienlin et al. 2020; Gruber et al. 2014; von Kienlin et al. 2014; Narayana Bhat et al. 2016).

6. Conclusions and future work CZTI has proven itself to be a sensitive transient detector, but our searches had largely been limited to triggered searches. The ML pipeline (Abraham et al. 2019) was the first major step towards detection of new transients with CZTI. The development of these algorithms, software, and the CIFT interface provide us with a powerful tool to extend our work further. Here, we have demonstrated the utility of this tool

with the discovery of 87 new transients that had been missed by previous searches, including 46 transients that had not been detected by any mission to date. This brings the total CZTI tally to 412 GRBs in the first five years of its operation since launch, or about  83 per year. For comparison, Swift BAT detects  92 GRBs per year from on-board triggers (Lien et al. 2016), while Fermi GBM detects  235 GRBs per year (von Kienlin et al. 2020).

J. Astrophys. Astr.

(2021) 42:73

Page 15 of 20

73

(a)

(b)

Figure 5. The normalised lightcurves of GRBs discovered by CIFT that have not been reported by any instrument before (Section 5.3). Each GRB lightcurve is normalised and labeled with the GRB name. Panel (a) shows normalised lightcurves for the GRBs detected in CZTI. The three sub-panels are with 0.1 s, 1 s and 10 s binning respectively, and each sub-panel is ordered by peak count rate above background, increasing from top to bottom. Panel (b) shows the normalised light curves of GRBs that were detected in Veto. These are plotted with a 1 s binning, and are also ordered by peak count rate above background, increasing from top to bottom.

The CIFT framework is constantly evolving. It has been designed to make it easy to incorporate new

features including search algorithms. We are working on metrics to quantify the statistical significance of a

73

Page 16 of 20

J. Astrophys. Astr.

(2021) 42:73

Table 4. The table contains the calculated parameters for all ‘Discovered GRBs’, which are all GRBs that have not been reported by any instrument before. Algorithm

GRB name

Time (UTC)

Detected: Analyzed: S184303433.0

CZTI CZTI GRB151104A

3:23:51

S184512194.0

GRB151106A3

13:23:12

S195263948.0

GRB160309A5

23:59:06

S197184964.0

GRB160401C1

05:36:02

S199212123.0

GRB160424B

16:42:01

S199449121.0

GRB160427A

10:31:59

S203963626.0

GRB160618A

16:33:44

S215358438.0

GRB161028A

13:47:16

S218385140.0

GRB161202B4

14:32:27

S226302000.0

GRB170304B

05:39:58

S232990228.0

GRB170520B

15:30:26

S239432831.0

GRB170803F

05:07:09

S242255598.0

GRB170904B

21:13:16

S247846852.0

GRB171108C3

14:20:50

S253483949.0

GRB180112B

20:12:27

S257929386.0

GRB180305B

07:03:04

S264583245.0

GRB180521B

07:20:43

S271507716.0

GRB180809C

10:48:34

S286741010.0

GRB190201A2

18:16:48

S288677154.0

GRB190224A

04:05:52

S298744222.0

GRB190620B

16:30:20

S299391822.0

GRB190628B3

04:23:40

S310393113.0

GRB191102A1

12:18:31

S310630593.0

GRB191105B

06:16:31

S311856067.0

GRB191119A

10:41:05

S317162956.0

GRB200119B

20:49:13

S324009902.0

GRB200408B

2:44:59

C: CR, NS, TN V: CR, NS, TN C: CR, NS, TN V: None C: CR, TN V: None C: CR, NS, TN V: CR, NS, TN C: CR, NS, TN V: None C: CR, TN V: CR, NS, TN C: CR, NS, TN V: CR, NS, TN C: CR, TN V: None C: CR, TN V: CR, NS, TN C: CR, NS, TN V: CR, NS, TN C: CR V: CR, NS, TN C: CR, NS, TN V: CR, NS, TN C: CR, NS, TN V: NS, TN C: CR, NS, TN V: None C: TN V: CR, TN C: CR, NS, TN V: CR, NS, TN C: CR, TN V: None C: CR, NS, TN V: CR, TN C: CR, TN V: CR, TN C: CR, NS, TN V: None C: CR, NS, TN V: None C: CR, TN V: None C: CR, TN V: None C: CR, NS, TN V: CR, NS, TN C: CR, NS, TN V: CR, NS, TN C: CR, TN V: CR, NS, TN C: CR, NS, TN V: None

SuperID

Peak time (s)

Bin (s)

T90 (s)

Peak rate (cps)

Bkg rate (cps)

Total counts (counts)

184303438.5

1

68þ2 4

911þ62 43

474þ3 4

28049þ744 827

184512190.5

1

39:4þ0:8 25:5

203þ39 25

319þ2 3

1596þ285 292

195263943.0

10

67þ27 16

62þ8 9

211:3þ0:4 0:6

2034þ521 618

197184965.2

0.1

3:1þ0:9 1:0

588þ189 20

386þ10 10

1090þ152 180

199212122.5

1

43:6þ0:7 2:2

691þ54 62

554þ4 3

4093þ342 511

199449124.5

1

34þ3 12

375þ46 50

476þ4 3

2669þ317 372

203963626.6

.01

0:27þ0:01 0:01

4918þ1328 366

444þ17 48

877þ61 61

215358429.5

1

25:7þ0:6 0:3

284þ44 36

434þ3 4

1883þ267 271

218385206.5

1

146þ1 4

131þ35 20

326þ2 2

3346þ767 848

226302047.5

1

41þ2 3

5091þ120 126

486þ2 2

23464þ495 511

232990228.1

0.1

2:3þ0:8 0:5

506þ176 61

490þ11 10

551þ101 162

239432833.4

0.1

6:0þ0:3 0:4

344þ172 68

452þ8 13

1494þ209 227

242255598.6

0.1

3:4þ0:4 0:5

966þ208 175

553þ10 12

1424þ199 160

247846851.7

0.1

0:88þ0:48 0:26

915þ183 143

374þ6 9

326þ69 63

253483948.3

0.1

9þ2 3

335þ178 3

409þ5 6

794þ210 159

257929428.5

1

32þ2 5

769þ56 59

436þ2 3

5170þ374 315

264583239.5

1

15þ4 9

167þ46 24

529þ3 3

1026þ287 289

271507720.0

10

290þ22 52

132þ13 14

535þ1 2

8438þ1314 1464

286741004.5

1

8:0þ0:4 2:9

136þ38 34

369þ2 4

558þ122 119

288677164.5

1

12:9þ0:4 0:4

843þ63 67

606þ4 5

5397þ254 261

298744221.9

0.1

4þ2 1

824þ238 55

549þ12 15

1504þ362 426

299391815.5

1.0

32þ15 24

196þ40 42

405þ3 3

1177þ410 513

310393104.5

0.1

4:2þ0:8 1:2

279þ189 25

392þ7 10

583þ126 124

310630601.8

0.1

13:2þ0:4 0:6

665þ202 77

739þ7 12

2200þ340 343

311856067.0

.01

0:15þ0:03 0:02

2702þ1130 230

449þ18 42

270þ39 34

317162955.5

1

195þ52 22

524þ4 4

1593þ1232 1522

324009901.5

1

217þ45 43

494þ4 4

590þ731 531

J. Astrophys. Astr.

(2021) 42:73

Page 17 of 20

73

Table 4. Continued. Algorithm

GRB name

Time (UTC)

S326787080.0

GRB200510B

6:11:17

S329736162.0

GRB200613B

09:22:40

S329806842.0

GRB200614C2

05:00:41

S334929280.0

GRB200812A

11:54:37

S335340170.0

GRB200817B

06:02:48

C: CR, NS, TN V: None C: CR, TN V: CR, NS, TN C: CR, TN V: CR, NS, TN C: CR, NS, TN V: CR, NS, TN C: CR, NS, TN V: CR, NS, TN

Detected: Analyzed: S213125894.0

Veto CZTI GRB161002A

17:38:12

S228861170.0

GRB170402C1

20:32:48

S242743550.0

GRB170910B

12:45:48

S255979768.0

GRB180210C3

17:29:26

S267689714.0

GRB180626D

06:15:12

S284139623.0

GRB190102B1

15:40:21

S285215230.0

GRB190115A2

02:27:08

S329776004.0

GRB200613C3

20:26:42

S329801810.0

GRB200614B4

03:36:47

S337054619.0

GRB200906B

02:16:56

S337615721.0

GRB200912A

14:08:38

Detected: Analyzed: S232918513.0

Veto Veto GRB170519B

19:35:11

S271825254.0

GRB180813A

03:00:52

S333985240.0

GRB200801D

13:40:38

SuperID

C: None V: CR, TN C: None V: CR, TN C: None V: TN C: None V: CR, NS, TN C: None V: CR C: None V: TN C: None V: CR, TN C: None V: NS C: None V: NS, TN C: None V: CR, NS C: None V: CR, NS, TN

C: None V: CR, TN C: None V: CR, NS, TN C: None V: CR, TN

Peak time (s)

Bin (s)

T90 (s)

Peak rate (cps)

Bkg rate (cps)

Total counts (counts)

326787121.8

.01

0:29þ0:01 0:01

8318þ1896 366

520þ20 50

1695þ84 75

329736159.5

1

11þ2 1

132þ58 3

618þ4 3

1190þ208 236

329806843.5

1

24þ7 10

104þ44 8

433þ3 3

1257þ288 306

334929324.5

1

10þ2 5

1677þ72 81

449þ5 3

4415þ194 224

335340230.5

1.0

23þ10 5

455þ48 43

490þ3 2

3847þ378 507

213125894.5

.01

0:87þ0:14 0:31

991þ792 56

413þ34 41

235þ116 105

228861062.0

10

119þ68 41

82þ11 9

360:9þ0:8 0:9

5072þ1150 1232

242743547.1

0.1

2:6þ0:2 1:8

400þ164 94

471þ9 10

259þ107 106

255979759.5

1

7þ3 1

43þ45 6

350þ3 3

110þ462 511

267689735.5

1

23þ6 3

133þ51 10

529þ3 3

1082þ314 344

284139622.4

0.1

3:1þ0:5 0:6

252þ154 12

345þ7 10

309þ88 86

285215220.5

1

19þ1 2

162þ40 24

384þ2 3

1225þ187 194

329776008.5

1

23þ4 10

97þ35 17

334þ3 3

925þ224 247

329801807.5

1

38:3þ0:5 24:2

112þ47 6

396þ2 3

1306þ379 294

337054499.5

1

35þ14 9

110þ45 6

462þ2 2

2033þ690 564

337615720.8

0.1

2:6þ0:0 1:3

415þ169 56

497þ9 10

396þ83 112

232918512.2

1

14þ6 4

377þ71 64

1583þ6 7

2251þ400 448

271825253.3

1

16þ9 7

506þ72 60

1464þ5 6

3144þ434 493

333985233.0

1

6þ1 3

283þ72 51

1687þ6 6

1198þ251 272

The table is also divided into three parts classified by what detector was used to detect (‘Detected’) and compute the parameters (‘Analysed’) given in the table. The column ‘SuperID’ gives the name of the superevent identified by the pipeline. Several of these entries in the ‘GRB Name’ column have a mark against their names that tells what quadrants were used for calculating all other parameters for that GRB. If there is no mark, then all four quadrants are used. Otherwise, marks ‘1’, ‘2’, ‘3’, ‘4’, and ‘5’ refer to the quadrant sets – ‘A, B, C’, ‘A, B, D’, ‘A, C, D’, ‘B, C, D’ and ‘C, D’ respectively. The column ‘Algorithm’ tells us what algorithms detected the GRB, where ‘TN’, ‘NS’, ‘CR’ stands for the three algorithms – TopN, N-sigma, and Cut-off rate respectively whereas ‘C’ and ‘V’ are the two detectors – CZTI and Veto. The time in AstroSat seconds where the GRB was brightest is given in the column ‘Peak time’. The bin size, that was used to generate the parameters – T 90 , Peak count rate above background, Background rate and Total counts, is mentioned in the column ‘Bin’.

73

Page 18 of 20

J. Astrophys. Astr.

(2021) 42:73

(a)

(b)

(c)

Figure 6. (a) Distribution of T90 values of all CZTI GRBs. (b) Distribution of peak count rates of transients in CZT detectors. Note that high peak rates are often obtained for short duration transients analysed with 0.1 s binning. (c) Distribution of total counts in CZT detectors. Comparing the distributions of the duration (T90 ), peak count rate, and total counts in three search classes. Blue lines (‘‘POC GRBs’’) denote transients detected in regular triggered searches and ML pipeline searches. Orange lines (‘‘Common GRBs’’) denote the transients CIFT detected among the ‘‘POC GRBs’’. Green lines (‘‘known GRBs’’) denote transients that have been reported by other instruments (Section 5.2) but missed by POC searches or ML pipeline, while red lines (‘‘Discovered GRBs’’) denote the new transients discovered with CIFT.

Figure 7. Diagnostic light curves for GRB 190605A. Top panel: 20–50 keV light curves for all four CZTI quadrants. The shaded green region denotes the GRB. Dashed and dotted lines denote the outlier threshold for CS and NS methods respectively, for each of the four quadrants. Middle panel: Same as top panel, but for 50–100 keV. Bottom panel: Same as top panel, but for 100–200 keV. The transient light curve looks similar in all quadrants, but it is too faint to qualify as an outlier in any of the methods.

transient, so that we can lower the FPP. We have developed and tested a new search based on Bayesian Blocks (BB; Scargle et al. 2012). We use the astropy.stats.bayesian_blocks module to obtain block representations of de-trended light curves, and search for blocks that are 3-r outliers. These outliers then form the peak maps discussed in Section 4.1, so the BB search can easily be integrated into CIFT as a fourth algorithm. Preliminary testing has shown promising results with significantly lower false positive rates as compared to other algorithms. We will now run the BB search on the full data set.

J. Astrophys. Astr.

(2021) 42:73

(a)

Page 19 of 20

73

(b)

Figure 8. Statistics of transients detected in CZTI (Panel (a)) and Veto (Panel (b)) detectors. The three bottom rows show the number of transients detected by each method: for instance, the N-sigma method detected 115 transients in CZTI. The bar charts at the top show overlaps between various combinations of methods. We see that of the 263 transients detected by CZTI detectors, 190 were detected by all three methods, while another 58 were detected by both the ‘‘cutoff rates’’ and ‘‘Top-N’’ methods. Among the 297 transients found in the Veto detector, 204 transients were detected by all three methods, 41 were detected only by the ‘‘Top-N’’ method, and 35 were detected by both the ‘‘cutoff rates’’ and ‘‘Top-N’’ methods.

Searches for fast transients also stand to benefit from other developments in CZTI data processing. New methods for rejecting noise from raw data (Ratheesh et al. 2021) are improving the quality of light curves. These promise to lower the cut-off rates for CS by a factor of a few and will give a proportional boost to the count rate sensitivity of CZTI. Another notable change to be introduced is the nonremoval of Veto-tagged events. The default CZTI pipeline attributes coincident events between CZT and Veto detectors to charged particles, and discards them. In case of bright GRBs, large numbers of photons are incident both on CZT and Veto detectors, greatly increasing the chance coincidence rates. Since these are real GRB photons which should not be discarded, future CIFT-based searches will disable Veto-event filtering. We have also added functionality to undertake specialised searches for X-ray counterparts to Fast Radio Bursts (FRBs) and Gravitational Wave (GW) sources. For instance, the magnetar source SGR 1935?2154 became active in early 2020, creating a series of bursts including one coincident with Fast Radio Burst (Li et al. 2020). We used the CIFT interface to efficiently search CZTI data for any bursts from this source. A first blind search was conducted with the default thresholds and we found three bursts, coincident with times reported by other

instruments. We then lowered the search thresholds and found an additional four bursts, corresponding to those reported by other missions. We are in the process of analysing properties of these CZTI-detected bursts, and the results will be reported separately (Raman et al., in prep.). We have also incorporated the ability to process GW localisation maps to calculate direction-dependent sensitivity. These features will streamline and boost the effort to search for X-ray counterparts to GW sources from the third observing run of advanced gravitational wave detectors (Abbott et al. 2020). Acknowledgements CZT-Imager is built by a consortium of Institutes across India. The Tata Institute of Fundamental Research, Mumbai, led the effort with instrument design and development. Vikram Sarabhai Space Centre, Thiruvananthapuram provided the electronic design, assembly and testing. ISRO Satellite Centre (ISAC), Bengaluru provided the mechanical design, quality consultation and project management. The Inter University Centre for Astronomy and Astrophysics (IUCAA), Pune did the Coded Mask design, instrument calibration, and Payload Operation Centre. Space Application Centre (SAC) at Ahmedabad

73

Page 20 of 20

provided the analysis software. Physical Research Laboratory (PRL) Ahmedabad, provided the polarisation detection algorithm and ground calibration. A vast number of industries participated in the fabrication and the University sector pitched in by participating in the test and evaluation of the payload. The Indian Space Research Organisation funded, managed and facilitated the project. This work utilised various software including Python, AstroPy (Robitaille et al. 2013), NumPy (van der Walt et al. 2011), Matplotlib (Hunter 2007), https://github.com/jnothman/upsetplot/ UpSetPlot (Lex et al. 1983), and ngrok.

References Abbott R., Abbott T. D., Abraham S. et al. 2020, GWTC-2: Compact Binary Coalescences Observed by LIGO and Virgo During the First Half of the Third Observing Run, arXiv:2010.14527 Abraham S., Mukund N., Vibhute A. et al. 2019, ArXiv e-prints, 1906.09670, arXiv:1906.09670 Anumarlapudi A., Bhalerao V., Tendulkar S. P., Balasubramanian A. 2020, Astrophys. J., 888, 40 Bhalerao V., Bhattacharya D., Rao A. R., Vadawale S. 2015, GRB Coordinates Network, 18422, 1 Bhalerao V., Kumar V., Bhattacharya D., Rao A. R., Vadawale S. 2016, GRB Coordinates Network, 19519, 1 Bhalerao V., Kasliwal M., Bhattacharya D. et al. 2017a, Astrophys. J., 845, arXiv:1706.00024 Bhalerao V., Bhattacharya D., Vibhute A. et al. 2017b, J. Astrophys. Astr., 38, 31 Chattopadhyay T., Gupta S., Sharma V. et al. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036021-09718-2 Gruber D., Goldstein A., Weller von Ahlefeld V. et al. 2014, 211, 12 Gupta S., Sharma V., Bhattacharya D. et al. 2020, GRB Coordinates Network, 28451, 1 Hunter J. D. 2007, Computing in Science & Engineering, 9, 90 Lex A., Gehlenborg N., Strobelt H., Vuillemot R., Pfister H. 2014, IEEE Trans. on Visualization and Computer Graphics, 20, 1983

J. Astrophys. Astr.

(2021) 42:73

Li C. K., Lin L., Xiong S. L. et al. 2020, arXiv e-prints, arXiv:2005.11071 Lien A., Sakamoto T., Barthelmy, S. D. et al. 2016, Astrophys. J., 829, 7 Marathe A., Sharma Y., Bhalerao V. et al. 2019, GRB Coordinates Network, 24972, 1 Mate S., Bhalerao V., Bhattacharya D. et al. 2017, GRB Coordinates Network, 20796, 1 Mate S., Chattopadhyay T., Bhalerao V. et al. 2021, submitted to J. Astrophys. Astr., 42, https://doi.org/10. 1007/s12036-021-09763-X Mereghetti S., Savchenko V., Ferrigno C. et al. 2020, Astrophys. J. Lett., 898, L29 Narayana Bhat P., Meegan C. A., von Kienlin A. et al. 2016, 223, 28 Rao A. R., Chand V., Hingar M. K. et al. 2016, Astrophys. J., 833, 86 Ratheesh A., Rao A., Mithun N. et al. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036-021-09716-4 Robitaille T. P., Tollerud E. J., Greenfield P. et al. 2013, A&A, 558, A33 Scargle J. D., Norris J. P., Jackson B., Chiang J. 2012, Astrophys. J., 764, 167 Sharma Y., Bhalerao V., Khanam T. et al. 2018, GRB Coordinates Network, 23511, 1 Shenoy V., Aarthy E., Bhalerao V. et al. 2020, GRB Coordinates Network, 27315, 1 Shenoy V., Sharma Y., Bhalerao V. et al. 2019a, GRB Coordinates Network, 26378, 1 Shenoy V., Sharma Y., Bhalerao V. et al. 2019b, GRB Coordinates Network, 26376, 1 Shenoy V., Sharma Y., Bhalerao V. et al. 2019c, GRB Coordinates Network, 26268, 1 Shenoy V., Bhalerao, V., Gupta, S. et al. 2020, GRB Coordinates Network, 28354, 1 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, in Takahashi T., den Herder J.-W. A., Bautz M., eds, Space Telescopes and Instrumentation 2014: Ultraviolet to Gamma Ray, vol. 9144, 91441S van der Walt S., Colbert S. C., Varoquaux G. 2011, Computing in Science & Engineering, 13, 22 von Kienlin A., Meegan C. A., Paciesas W. S. et al. 2014, 211, 13 von Kienlin A., Meegan C. A., Paciesas W. S. et al. 2020, Astrophys. J., 893, 46

J. Astrophys. Astr. (2021)42:65 https://doi.org/10.1007/s12036-021-09702-w

Ó Indian Academy of Sciences Sadhana(0123456789().,-volV)F T3](0123456789().,-volV)

SCIENCE RESULTS

Coronae of an active fast rotator FR Cnc JEEWAN C. PANDEY1,*, GURPREET SINGH1, SUBHAJEET KARMAKAR1,

ARTI JOSHI1,2, I. S. SAVANOV3, S. A. NAROENKOV3 and M. A. NALIVKIN3 1

Aryabhatta Research Institute of Observational Sciences (ARIES), Manora Peak, Nainital 263 001, India. School of Physics and Technology, Wuhan University, Wuhan 430072, China. 3 Institute of Astronomy, Russian Academy of Sciences (INASAN), Pyatnitskaya ul. 48, Moscow 119017, Russia. *Corresponding author. E-mail: [email protected] 2

MS received 29 October 2020; accepted 29 December 2020 Abstract. We present the first detailed X-ray study and simultaneous optical observations of the active fast rotating star FR Cnc. The X-ray spectra are found to be explained by two-temperature plasma model with temperature of cool and hot components of 0.34 and 1.1 keV, respectively. The X-ray light curve in the 0.5– 2.0 keV energy band is found to be rotationally modulated with the degree of rotational modulation of  17%. We have also found that the X-ray light curve is anti-correlated with the optical light and colour curves in the sense that maximum X-ray light corresponds to minimum optical light and cooler region on the surface of FR Cnc. The X-ray luminosity of FR Cnc is found to be almost consistent in the last 30 years with an average value of 4:85  1029 erg s1 in the 0.5–2.0 keV energy band. Keywords. Star—late-type—active—X-ray—individual (FR Cnc).

1. Introduction Stars with an outer convective envelope usually show magnetic field induced activities like dark spots on the surface, flares, activity cycles, and hot outer atmosphere (i.e. chromospheric and coronal emission). These activities produce a range of emission from Xray to radio domain (e.g. see the review articles by Strassmeier 2009; Gu¨del 2004; Favata & Micela 2003). The activity phenomena are found to be more in stars with faster rotation periods. However, for the rapidly rotating active stars, the magnetic activities reach the saturation level where the X-ray emission becomes independent of the rotation period (e.g. Pizzolato et al. 2003; Wright et al. 2011). The ratio of X-ray to bolometric luminosity (LX =Lbol ) at the saturation level is generally found as  10-3 with saturation period between  1 to  4 days for different type of active stars (see Pizzolato et al. 2003). This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

Various explanations have been proposed for the saturation in active stars, e.g. internal dynamo itself saturates and produces no more magnetic flux with increasing rotation (e.g. Vilhu & Walter 1987), maximum possible coverage of active regions on the stellar surface (Vilhu 1984), and a centrifugal stripping of the corona (Jardine & Unruh 1999). Thus rapidly rotating late-type stars are important as they display an extremely enhanced level of stellar activity at saturation when compared with that of the Sun and other slow rotating stars. FR Cnc is a single, young, and fast rotating active star, which shows its magnetic activity at saturation level. It is a K7V-type star with the rotational period of 0.826518 d and located at a distance of 35:56  0:08 pc (Lindegren et al. 2018). It was identified as a probable active star when it was found to be an optical counterpart of X-ray source 1ES 0829?15.9 (or 1RXS J083230.9?154940) in Einstein Slew Survey and later in ROSAT All-Sky Survey (Elvis et al. 1992; Voges et al. 1999). The X-ray flux during the ROSAT AllSky Survey was found to be weaker than that found in

65

Page 2 of 7

the Einstein Slew Survey, showing its variable active nature. Using long-term optical photometry and spectroscopy, Pandey et al. (2002, 2005), for the first time, derived its rotational period to be 0.8267 days. They also showed that it consists of two long-lasting groups of spots and a presence of Hb, Ha, and CaII H and K emission lines in the optical spectra. A few flaring events were also caught in the past in the optical band. Golovin et al. (2012, 2007) detected a strong flare in B, V, R, and I filters, where peak flare flux in B band was found to be  37 times more than the quiescent flux. Later in 2010, Kozhevnikova et al. (2018) reported one flare with flare duration of 32.5 minutes. Savanov et al. (2019) also detected a flare in 2019 with a flare energy of the order of 1033 erg. Its short rotational period along with the highly variable chromospheric features and optical light curves imply that this star should manifest strong magnetic activity including flaring activity in X-rays. Apart from being well studied in the optical band (photometry and spectroscopy), its coronal activities and properties are not yet probed. The logðLX =Lbol ) of –3 of FR Cnc indicates its high level of X-ray activity at the saturation level. Further, it would be interesting to see its long term X-ray activity as previous X-ray observations from Einstein Observatory and ROentgen SATellite (ROSAT) are approximately 30–40 years back. Moreover, the weaker X-ray flux during RASS observations than that of the Einstein Observatory indicates a variable X-ray emission. In addition to that, the simultaneous X-ray and optical observations will help us to understand better any correlated photospheric and coronal emission. With this aim, we have carried out X-ray observations using the AstroSat and coordinated optical observations using Zvenigorod Observatory (Savanov et al. 2018). Results from ground-based optical observations have been published in Savanov et al. (2019). In the forthcoming sections, we present observations and data reduction in Section 2, spectral and timing analyses in Section 3, discussion on the results obtained in Section 4, and conclusions in Section 5.

J. Astrophys. Astr. (2021)42:65

2014; Agrawal 2017) on March 1–3, 2019. The observations were started at 18:17:08 UT on March 1, 2019, and ended at 06:31:17 UT on March 3, 2019. The SXT is equipped with a CCD detector in its focal plane, which covers the energy band of 0.3–8.0 keV with an effective area of  120 cm2 and energy resolution of  150 eV at 1.5 keV (Singh et al. 2017). We have used Level 2 data which provides filtered events for each orbit of the AstroSat. The Level 2 data are collected from ISRO science data archive/AstroSat archive.1 The Level 2 event files were generated by filtering any contamination from the charged particles due to satellite’s excursions through the South Atlantic Anomaly region, event grading of 0–12, bias subtraction, and bad pixel flagging etc. Individual event files from each orbit were merged to avoid any time-overlapping events from the consecutive orbits using the SXTMERGERTOOL provided by AstroSat science support cell.2 After all the filtering the total useful exposure time was 40781 s. The X-ray spectra and light curves from the clean merged Level 2 event file were extracted using the xselect (V2.4j) package of HEASOFT.3 In order to extract the light curve and spectra, we have chosen the circular region of the radius 130 centring the source for the source. More than 90% of source counts are found to lie within this radius. However, for the background several small source-free circular regions with a radius of 2:50 each around the source were taken at the offset of [ 16:50 from the centre. The response file ‘‘sxt_pc_mat_g0to12.rmf’’ was used for the spectral analysis. Simultaneous optical observations of FR Cnc were planned from ARIES, Nainital and Zvenigorod Observatory of INASAN, Moscow Russia (Savanov et al. 2018). Unfortunately, we could not observe from ARIES due to bad weather, however, we could observe in B and V bands from Zvenigorod Observatory on March 2, 3, 6, 9, 11, and 13, 2019 of which observations carried out on March 2–3, 2019 with a total of 17.2 ks exposure were almost simultaneous with AstroSat observations. Detailed observations, data reduction, and analysis are given in Savanov et al. (2019).

2. Observations and data reduction We have observed (Observation ID. 9000002748) FR Cnc using Soft X-ray Telescope (SXT; Singh et al. 2016, 2017) onboard AstroSat (Singh et al.

1

https://astrobrowse.issdc.gov.in/astro_archive/archive/Home.jsp. http://astrosat-ssc.iucaa.in/?q=sxtData. 3 https://heasarc.gsfc.nasa.gov/docs/software/heasoft/. 2

J. Astrophys. Astr. (2021)42:65

Page 3 of 7

65

3. Analysis and results

3.2 X-ray light curve

3.1 X-ray spectra

X-ray light curve was extracted in the energy band 0.5–2.0 keV. The background-subtracted X-ray light curve in the energy band 0.5–2.0 keV is shown in the Fig. 2. Binning of X-ray light curve was 711 s. The light curve appears to be variable during the observations. In order to check the variability of X-ray light curve, we have applied v2 -test to the time series data, P where v2 is defined as v2 ¼ ½fCðtÞ  CðtÞg2 =rðtÞ2 . Here C(t) is count rate at time t, CðtÞ is average count rate, and rðtÞ is error on C(t). The derived value of v2 of 168 for 79 degrees of freedom is higher than the critical value of the v2 of 111 for 99% of the confidence limit. This indicates that the X-ray light curve is variable with confidence limit of 99%. The X-ray light curve was folded with a period of 0.826518 days and at JD = 2452635.72669 as given Golovin et al. (2012). The phase bin of the folded light curve was 0.05. Top panel of Fig. 4 shows the folded X-ray light curve in the 0.5–2.0 keV energy

We have extracted the X-ray spectra for the energy range of 0.5–7.0 keV. Beyond 2.0 keV the source spectra was found to be merged with the background. Therefore, for further analysis, we have taken the Xray spectra from 0.5 to 2.0 keV. The X-ray spectra of FR Cnc is shown in Fig. 1 along with its best fit model. The X-ray spectral fitting was performed with xspec (version: 12.11.0; Arnaud 1996; Dorman & Arnaud 2001) using the v2 statistics. Firstly, the X-ray spectra were fitted with single temperature (1T) plasma model APEC (Smith et al. 2001) with solar abundance, which yielded an unacceptably high value of minimum reduced v2 (v2m ) of 3.3. The abundances were then freed to depart from the solar values and the fit was improved significantly with v2m ¼ 1:3. Further, we have fit the spectra with two temperature plasma model and found that the fit improved with v2m ¼ 1:2. The best-fit parameters are given in Table 1. For all models the solar photospheric abundances were according to Anders and Grevesse (1989). The best-fit model yielded the plasma temperatures of 0.34 keV and 1.1 keV with abundances of 0:07Z , where Z is solar photospheric abundances. In the fitting procedure the value of hydrogen column density (NH ) was fixed to a low value of 3  1020 cm2 , which is similar to the value of NH towards the direction of FR Cnc (see Dickey & Lockman 1990). The flux in the energy band of 0.5–2.0 keV was derived using the CFLUX model. The unabsorbed flux in the energy band of 0.5– 2.0 keV was found to be 4:5  1029 erg s1 .

Table 1. Best-fit parameters obtained from X-ray spectral fitting. All of the errors are with the 90% confidence interval for a single parameter. Model! Parameters# kT1 (keV) kT2 (keV) EM1 ð1052 cm3 ) EM2 ð1052 cm3 ) ZðZ Þ FX ð1012 erg s1 ) LX ð1030 erg s1 ) v2m (dof)

APEC (1T)

APEC (2T)

0:84þ0:03 0:03 ... 10:7þ2 2 ... 0:029þ0:007 0:007 2:8  0:1 4:2  0:2 1.3 (116)

0:34þ0:06 0:04 1:1þ0:1 0:1 9:2þ1 1 4:0þ1 1 0:07þ0:02 0:02 2:8  0:1 4:2  0:2 1.21 (114)

0.1 0.05

0.02 0.01

2 0 −2 0.5

1.5

1

2

Energy (keV)

Figure 1. X-ray spectra along with the best-fit 2-temperature plasma model APEC.

Figure 2. Background subtracted X-ray light curve of FR Cnc in 0.5–2.0 keV energy band.

65

Page 4 of 7

band. A clear rotational modulation is present in the X-ray light curve, where peak-to-peak amplitude was found to be  0.04 counts s1 . To get the exact degree of modulation, we have fitted the folded X-ray light curve with a sinusoidal function, which yielded an amplitude of 0:011  0:001 counts s1 (see the top panel of Fig. 4). This corresponds to 17  2% modulation of the mean value of X-ray flux.

3.3 Optical light curves The optical light curves in B and V filters are shown in Fig. 3. The flare observed on March 11, 2019 (HJD = 2458554.25891) is also shown in the insets. The analysis of the optical V band light curve is published in Savanov et al. (2019) where we show the surface coverage of dark spots during the observations was  12% of the total surface area of FR Cnc. We also noticed that spots are concentrated on two active longitudes of FR Cnc. The flare was observed near phase 0.8, which is the maximum of the V-band light curve. The flare duration was found to be  50

Figure 3. B and V bands light curves of FR Cnc. Insets in each panel show the flare light curves in the respective band as observed on March 11, 2019.

J. Astrophys. Astr. (2021)42:65

minutes in B-band whereas the total flare energy in B and V were found to be 2:2  1033 and 1:4  1033 erg, respectively. Light curve and colour curve are folded with an ephemeris as used for X-ray light curve folding. Middle and bottom panels of Fig. 4 shows the folded light curve in V -band and (B–V) colour curves. We have also derived the Spearman rank correlation coefficient between X-ray and V-band fluxes, X-ray and (B–V) colour, and V and (B–V). The resultant values of the correlation coefficient with the probability of no correlation are given in Table 2. The V-band light curve was found to be strongly correlated with (B–V) colour curve in the sense that during maximum V-band magnitude the (B–V) colour is also found to be maximum. However, the X-ray flux appeared to be anti-correlated with V-band magnitude

Figure 4. Folded X-ray light curve of FR Cnc in X-ray and optical bands. The colour curve is shown in the bottom panel. The error in single measurement of V is 0.008 mag. The continuous blue curve in the top panel is the best-fit sinusoidal curve to the folded X-ray light curves.

J. Astrophys. Astr. (2021)42:65

Page 5 of 7

Table 2. Spearman rank correlation coefficients between X-ray and optical. Parameters X-ray – (B–V) X-ray – V V – (B–V) 

Correlation coefficient

Probability

–0.61 –0.54 0.77

0.0067 0.0020 0.000016

Probability of no correlation.

and (B–V) colour. X-ray flux was found to be minimum near the maximum V-band flux and (B–V) colour.

4. Discussion For the first time, we present the detailed X-ray analysis and simultaneous optical observations of the active star FR Cnc. The X-ray light curve is found to be rotationally modulated. The degree of X-ray modulation was found to be 17% in FR Cnc. Rotational modulation in X-ray was also found in the several other active stars like V711 Tau (Agrawal & Vaidya 1988; Audard et al. 2001), AB Dor (Hussain et al. 2005; Kuerster et al. 1997) and V1147 Tau (Patel et al. 2013), where the degree of rotational modulation was found up to 30%. The maximum of the V-band light curve is found near the minimum of the X-ray light curve. Also, we found that both V band flux and colour are anti-correlated with the X-ray flux. This indicates the high level of X-ray activity has resulted from the high magnetic regions at the surface of FR Cnc. Also in the case of Sun, significant variability of the average X-ray flux is found to be due to the rotational modulation of the active region (Orlando et al. 2004). While comparing the spectral parameters of FR Cnc with other similar investigation, we found that the coronal activity of FR Cnc lies among the other active dwarf stars. Using the ROSAT X-ray observations Dempsey et al. (1997) found that coronae of active dwarfs show two temperature plasma with the lowtemperature component in the range of 0.13–0.30 keV and high-temperature component in between 1.07 to 2.81 keV. Pandey & Singh (2008) also found the quiescent coronae of active dwarfs consist of two temperature of 0.2–0.5 and 0.6–1.0 keV using observations from X-ray Multi-Mirror Mission (XMMNewton). In the case of FR Cnc, we found similar values of low and high temperatures. The average values of volume emission measures EM1 and EM2

65

for FR Cnc were found to be 9:2  1052 and 4:0  1052 , respectively. These values are also very close to those for similar stars (Dempsey et al. 1997; Pandey & Singh 2008). The X-ray luminosity of FR Cnc is found to be similar to the average value of 4:0  1029 erg s1 for 101 active dwarf stars (Pandey et al. 2005). Also, FR Cnc has a similar X-ray activity level to that of single rapidly rotating active stars (e.g. LO Peg, BO Mic, AB Dor etc.; Karmakar et al. 2016; Garcı´a-Alvarez et al. 2008) whereas its X-ray luminosity is less than the other fast rotating (period \1 d) active dwarf binaries (see Catalogue of Chromospherically Active Binaries; Eker et al. 2008). If we compare our results with evolved active stars, we found that FR Cnc is less active with respect to them. Pandey and Pant (2012) showed that quiescent coronae of evolved active stars are explained by 3 temperature plasma with the median values of 0.38, 0.99 and 2.92 keV. Also, the median value of X-ray luminosity for evolved active stars is an order higher than that of FR Cnc and other active dwarfs. The abundance derived for FR Cnc is little lower than to that found in other active stars (Imanishi et al. 2001; Pandey & Singh 2008, 2012). We have also investigated the long-term behaviour of FR Cnc in the X-ray band. For this, we have taken the count rates of the past X-ray observation from Einstein slew survey (Elvis et al. 1992), ROSAT AllSky Survey (Voges et al. 1999), and XMM-Newton Slew Survey (Saxton et al. 2008). These count rates are converted into flux using WebPIMMS.4 The flux is estimated in the 0.5–2.0 keV energy band by assuming a thermal plasma of 1.0849 keV and an abundance of 0:2Z . In Table 3, we present the count rates from these observations and corresponding X-ray luminosity in 0.5–2.0 keV energy band along with the flux derived from AstroSat observations. Figure 5 shows the variation of X-ray luminosity with time. The X-ray luminosity during the AstroSat, ROSAT and XMMNewton were found to be consistent within a 1.5r level indicating the almost constant level of the X-ray activity in the last 30 years. However, the X-ray luminosity of FR Cnc,  10 years earlier than the ROSAT observations was 2–4 times more than the other four observations. This indicates that coronae of FR Cnc is highly active during the observations from the Einstein Observatory and became less active during later observations from the year 1990. Due to 4

https://heasarc.gsfc.nasa.gov/cgi-bin/Tools/w3pimms/w3pimms. pl.

65

Page 6 of 7

J. Astrophys. Astr. (2021)42:65

Table 3. X-ray flux and luminosity of FR Cnc during different observations in the energy band 0.5–2.0. Observatory

Date of Obs. (dd/mm/yyyy)

Einstein ROSAT XMM-Newton ... AstroSat

1978–1981 09–29/10/1990 30/10/2006 05/11/2008 01–03/03/2019

(10

12

Flux erg s1 cm2 )

10.9 ± 4.5 2.5 ± 0.6 4.4 ± 0.9 2.9 ± 0.6 3.0 ± 0.1

LX (10 erg s1 ) 29

16.4 ± 6.7 3.8 ± 0.9 6.7 ± 1.4 4.4 ± 0.9 4.5 ± 0.2

Count rate (counts s1 ) 0.56 0.29 4.00 2.65 0.112

± ± ± ± ±

0.23 0.07 0.86 0.59 0.001

References Elvis et al. (1992) Voges et al. (1999) Saxton et al. (2008) Saxton et al. (2008) Present study

Count rates from different observations are in the respective energy band of the instrument of corresponding observatories.

variation was found to be anti-correlated with the (B– V) colour and V-band flux indicating the higher X-ray flux corresponds to the more active regions on the surface of FR Cnc. With the limited and sparse data points, the X-ray activity of FR Cnc since last 30 years appears to be almost constant. The X-ray spectra are explained by two temperatures plasma model with cooler and hotter temperatures of 0.34 and 1.1 keV, respectively. The X-ray luminosity during the AstroSat observations is found to be 4:5  1029 erg s1 in the 0.5–2.0 keV energy band. Figure 5. Long-term X-ray variation of FR Cnc in 0.5–2.0 keV energy band.

limited data points, it is not possible to comment on any X-ray activity cycle in FR Cnc. There are only a few examples which show the convincing X-ray cycles such as a Cen B (Ayres 2009), HD 81809 (Favata et al. 2008), and 64 Cyg A (Hempelmann et al. 2006). All these stars have relatively low magnetic activity levels. However, the past observational results have shown a lack (or a little-evidence) of presence of long-term X-ray activity cycle in highly active stars (e.g. Stern 1998). It has been suggested that the general lack of the observed cyclic activity in most active rapidly rotating stars could be due to the turbulent or distributive dynamo (Drake et al. 1996; Kashyap & Drake 1999). We encourage continuous X-ray monitoring of FR Cnc to know its long term active nature.

Acknowledgements This research is based on the results obtained from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). This work has been performed utilizing the calibration data-bases and auxiliary analysis tools developed, maintained and distributed by AstroSat-SXT team with members from various institutions in India and abroad. This research has been done under the Indo-Russian DST-RFBR Project reference INT/RUS/RFBR/P-271 (for India) and Grant RFBR Ind a 14-02-92694, 17-52-45048 and Grant No. 075-15-2020-780 (for Russia). We acknowledge the referee for reading our paper and his/her comments.

References 5. Conclusions The first pointed X-ray observations of the active fast rotator FR Cnc show a variable coronal emission. It is found to be rotationally modulated with 17% degree of modulation around the mean flux. The X-ray

Agrawal P. C. 2017, J. Astrophys. Astr., 38, 27 Agrawal P. C., Vaidya J. 1988, MNRAS, 235, 239 Anders E., Grevesse N. 1989, Geochim. Cosmochim. Acta, 53, 197 Arnaud K. A. 1996, in Jacoby G. H., Barnes J., eds, Astronomical Society of the Pacific Conference Series,

J. Astrophys. Astr. (2021)42:65 Vol. 101, Astronomical Data Analysis Software and Systems V, 17 Audard M., Gu¨del M., Mewe R. 2001, A&A, 365, L318 Ayres T. R. 2009, ApJ, 696, 1931 Dempsey R. C., Linsky J. L., Fleming T. A., Schmitt J. H. M. M. 1997, ApJ, 478, 358 Dickey J. M., Lockman F. J. 1990, ARA&A, 28, 215 Dorman B., Arnaud K. A. 2001, in Harnden J., Primini F. R. F. A., Payne H. E., eds, Astronomical Society of the Pacific Conference Series, Vol. 238, Astronomical Data Analysis Software and Systems X, 415 Drake J. J., Stern R. A., Stringfellow G. et al. 1996, ApJ, 469, 828 Eker Z., Ak N. F., Bilir S. et al. 2008, MNRAS, 389, 1722 Elvis M., Plummer D., Schachter J., Fabbiano G. 1992, ApJS, 80, 257 Favata F., Micela G. 2003, Space Sci. Rev., 108, 577 Favata F., Micela G., Orlando S. et al. 2008, A&A, 490, 1121 Garcı´a-Alvarez D., Drake J. J., Kashyap V. L., Lin L., Ball B. 2008, ApJ, 679, 1509 Golovin A., Pavlenko E., Kuznyetsova Y., Krushevska V. 2007, Information Bulletin on Variable Stars, 5748, 1 Golovin A., Ga´lvez-Ortiz M. C., Herna´n-Obispo M. et al. 2012, MNRAS, 421, 132 Gu¨del M. 2004, A&A Rev., 12, 71 Hempelmann A., Robrade J., Schmitt J. H. M. M. et al. 2006, A&A, 460, 261 Hussain G. A. J., Brickhouse N. S., Dupree A. K. et al. 2005, ApJ, 621, 999 Imanishi K., Koyama K., Tsuboi Y. 2001, ApJ, 557, 747 Jardine M., Unruh Y. C. 1999, A&A, 346, 883 Karmakar S., Pandey J. C., Savanov I. S. et al. 2016, MNRAS, 459, 3112 Kashyap V., Drake J. J. 1999, ApJ, 524, 988 Kozhevnikova A. V., Kozhevnikov V. P., Alekseev I. Y. 2018, Astrophysics, 61, 30 Kuerster M., Schmitt J. H. M. M., Cutispoto G., Dennerl K. 1997, A&A, 320, 831 Lindegren L., Herna´ndez J., Bombrun A. et al. 2018, A&A, 616, A2 Orlando S., Peres G., Reale F. 2004, A&A, 424, 677 Pandey J. C., Pant V. 2012, in Prugniel P., Singh H. P., eds, Astronomical Society of India Conference Series, Vol. 6, Astronomical Society of India Conference Series, 239

Page 7 of 7

65

Pandey J. C., Singh K. P. 2008, MNRAS, 387, 1627 Pandey J. C., Singh K. P. 2012, MNRAS, 419, 1219 Pandey J. C., Singh K. P., Drake S. A., Sagar R. 2005, AJ, 130, 1231 Pandey J. C., Singh K. P., Sagar R., Drake S. A. 2002, Information Bulletin on Variable Stars, 5351, 1 Patel M. K., Pandey J. C., Savanov I. S., Prasad V., Srivastava D. C. 2013, MNRAS, 430, 2154 Pizzolato N., Maggio A., Micela G., Sciortino S., Ventura P. 2003, A&A, 397, 147 Savanov I. S., Naroenkov S. A., Nalivkin M. A., Puzin V. B., Dmitrienko E. S. 2018, Astrophys. Bull., 73, 344 Savanov I. S., Naroenkov S. A., Nalivkin M. A., Pandey J. C., Karmakar S. 2019, Astron. Lett., 45, 602 Saxton R. D., Read A. M., Esquej P. et al. 2008, A&A, 480, 611 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, in Takahashi T., den Herder J.-W. A., Bautz M., eds, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 9144, Space Telescopes and Instrumentation 2014: Ultraviolet to Gamma Ray, 91441S Singh K. P., Stewart G. C., Chandra S. et al. 2016, in den Herder J.-W. A., Takahashi T., Bautz M., eds, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 9905, Space Telescopes and Instrumentation 2016: Ultraviolet to Gamma Ray, 99051E Singh K. P., Stewart G. C., Westergaard N. J. et al. 2017, J. Astrophys. Astr., 38, 29 Smith R. K., Brickhouse N. S., Liedahl D. A., Raymond J. C. 2001, ApJ, 556, L91 Stern R. A. 1998, in Donahue R. A., Bookbinder J. A., eds, Astronomical Society of the Pacific Conference Series, Vol. 154, Cool Stars, Stellar Systems, and the Sun, 223 Strassmeier K. G. 2009, A&A Rev., 17, 251 Vilhu O. 1984, A&A, 133, 117 Vilhu O., Walter F. M. 1987, ApJ, 321, 958 Voges W., Aschenbach B., Boller T. et al. 1999, A&A, 349, 389 Wright N. J., Drake J. J., Mamajek E. E., Henry G. W. 2011, ApJ, 743, 48

J. Astrophys. Astr. (2021) 42:58 https://doi.org/10.1007/s12036-021-09710-w

Ó Indian Academy of Sciences Sadhana(0123456789().,-volV)F T3](0123456789().,-volV)

SCIENCE RESULTS

A new measurement of the spin and orbital parameters of the high mass X-ray binary Centaurus X-3 using AstroSat PARISEE SHIRKE* , SUMAN BALA, JAYASHREE ROY and DIPANKAR BHATTACHARYA Inter-University Centre for Astronomy and Astrophysics (IUCAA), Post Bag 4, Ganeshkhind, Pune 411 007, India. *Corresponding author. E-mail: [email protected] MS received 30 October 2020; accepted 6 January 2021 Abstract. We present the timing results of out-of-eclipse observations of Centaurus X-3 spanning half a binary orbit, performed on 12–13 December, 2016 with the Large Area X-ray Proportional Counter (LAXPC) on-board AstroSat. The pulse profile was confirmed to exhibit a prominent pulse peak with a secondary inter-pulse. The systemic spin period of the pulsar was found to be 4:80188  0:000085 s in agreement with its spin up trend. The spin up timescale seems to have increased to 7709  58 yr that points to negative torque effects in the inner accretion disk. We also report the derived values of projected semimajor axis and orbital velocity of the neutron star. Keywords. X-rays: binaries—stars: neutron—pulsars: individual: Centaurus X-3—X-rays: stars—methods: data analysis—AstroSat: LAXPC.

1. Introduction The X-ray source Centaurus X-3 was detected by Chodil et al. (1967) in a rocket experiment. From the X-ray and optical studies, it has been established that it is an eclipsing High Mass X-ray Binary (HMXB) comprising of an X-ray pulsar with a spin period of  4.8 s and an optical companion star with an orbital period of  2.1 days (Giacconi et al. 1971; Schreier et al. 1972b). The optical companion is an O6.5 II-III supergiant known as Krzeminski’s star (Verbunt & van den Heuvel 1995; Schreier et al. 1972a; Krzeminski 1974). The X-ray pulsar has a mass of 1:49  0:08M  (Rawls et al. 2011) and the optical counterpart has a mass of 20:5  0:7M  (Hutchings et al. 1979; Ash et al. 1999) and radius of 11:4  0:7R (Falanga et al. 2015). The eclipse lasts for  20% of the orbit (Nagase 1989). The latest estimate of the distance to This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

Centaurus X-3 is 5:7  1:5 kpc (Thompson & Rothschild 2009). The high, persistent luminosity of the X-ray pulsar L210 keV  51037 ergs/s is sustained by mass transfer arising from a combination of predominantly disk accretion and an excited stellar wind (Petterson 1978; Bildsten et al. 1997). The stellar wind seems to be driven thermally from the X-ray heated face of the atmosphere of the Krzeminski’s star (Day & Stevens 1993). This is supported by the observation of ellipsoidal variations in the optical light curve produced by the tidally deformed supergiant (Tjemkes et al. 1986) indicative of a star filling its Roche lobe. Centaurus X3 is one of the very few X-ray binaries that exhibit a secular decay of the orbital period, which could be attributed to tidal dissipation (Burderi et al. 2000). The mass transfer causes a secular spin up trend with wavy fluctuations over several years. The resultant tidal interaction between a distorted supergiant and its companion neutron star results in an orbital decay characterised by jP_b =Pb j  1:8106 yr1 (Kelley et al. 1983; Nagase et al. 1992).

58

Page 2 of 8

The out-of-eclipse phase-averaged spectrum in 1-40 keV band is usually modelled by a power law with a high energy cut-off (Suchy et al. 2008) and a Gaussian iron line along with a soft X-ray excess below 5 keV (White & Swank 1982). The soft excess seems to arise due to the scattering of X-rays by ambient wind around an obscuring gas stream (Nagase et al. 1992). Burderi et al. (2000) interpreted it as blackbody radiation with kT  0:1 keV. A cyclotron resonance scattering feature (CRSF) around 30 keV was first confirmed by Santangelo et al. (1988), who estimated the magnetic field to be  ð2:43Þ  1012 Gauss. Pulse phase-resolved spectroscopy reveals asymmetric variation of the magnetic cyclotron line energy which could be due to an offset of the dipole with respect to the neutron star center (Burderi et al. 2000). Burderi et al. (2000) undertook broad-band (0.1–100) spectral and timing analysis of out-of-eclipse Centaurus X-3 using observations by BeppoSAX in the year 1997. Suchy et al. (2008) presented a detailed analysis of PCA-RXTE data from observations over two consecutive binary orbits. This was followed by an in-depth pulse arrival time analysis by Raichur and Paul (2010) from the same observation. Naik et al. (2011) carried out the spectral analysis of Centaurus X-3 using Suzaku observations over one orbital period. The AstroSat mission launched in 2015 promises powerful timing and spectral capabilities for studies of compact objects (Bhattacharya 2017). Out of this class of targets, observations of neutron stars in X-ray binary systems can provide accurate measurements of orbital parameters, thus aiding the characterisation of the orbital evolution of stars in binary systems (Paul 2017). Centaurus X-3 is an example of such a system. In this paper, we present the timing parameters of Centaurus X-3 derived from an AstroSat/LAXPC observation of 12–13 December, 2016. In Section 2, we provide details about the data and its reduction. Section 3 describes the timing analysis performed for estimation of the spin and orbital parameters. Section 4 summarises the results and discusses their physical significance. Finally, we present the conclusions in Section 5.

2. Observations and data reduction We make use of the AstroSat/LAXPC observations from 12 December, 2016 12:31:56 (hh:mm:ss) UTC to 13 December, 2016 10:05:27 (hh:mm:ss) UTC with a

J. Astrophys. Astr.

(2021) 42:58

useful exposure time of 53 ks avoiding eclipses. This data, with Obs ID 9000000880, was retrieved from the AstroSat open public data archive.1 AstroSat was launched on September 28, 2015 by the Indian Space Research Organisation (ISRO) and is the first dedicated Indian astronomy mission aimed at simultaneous multi-wavelength study of astronomical sources in X-ray, optical and UV spectral bands (Agrawal et al. 2006). It is aimed at examining their wavelength-dependent intensity variations and the underlying physical processes (Singh et al. 2014). Some of its prime scientific objectives are to understand high energy processes in X-ray binary systems and estimate magnetic fields of neutron stars. This work makes use of Level 1 data of the Large Area X-ray Proportional Counter (LAXPC), which is a major payload on-board AstroSat with one of its primary objectives being the conduct of timing studies of X-ray binaries like Centaurus X-3 (Agrawal et al. 2017). LAXPC consists of three identical, co-aligned, independent proportional counters that register X-ray photons in the wide energy range of 3–80 keV. It has a total effective area of  6000 cm2 at 10 keV, a timing precision of 10 ls and a sensitivity of 1 milliCrab in 1000 s (Antia et al. 2017; Yadav et al. 2016). Due to a larger area (four to five times more effective area above 30 keV compared to RXTE-PCA), the sensitivity of LAXPC is unmatched in medium energy X-rays. The Level 1 data was converted into Level 2 data with LaxpcSoft: Format (A) software2 (Ver. 2018, May 21) using laxpc_make_event. We made use of the event mode data (modeEA) from the normal (default) mode of operation which is suitable for bright X-ray sources. The Level 2 data contains the arrival time, energy and identity of the detecting element (detector number and anode layer) for each X-ray photon detection event. A file specifying the time segments called Good Time Intervals (GTI), containing reliable data by filtering out target occultation by the Earth and passage through the South Atlantic Anomaly (SAA) regions, was produced using laxpc_make_stdgti. While the standard LAXPC software generates the GTI table based on the set criterion, some stray bad points can still remain within the GTI. These anomalies were identified by fine inspection of the light curve and manually rejected by suitably re-defining the GTI. The

1

http://astrosat-ssc.iucaa.in:8080/ObservationFinder/. http://astrosat-ssc.iucaa.in/?q=laxpcData.

2

J. Astrophys. Astr.

(2021) 42:58

long observation was thus divided into 17 such time segments. Barycentric correction was applied to refer the photon arrival times to the barycenter of the solar system in order to correct for the relative motion of the satellite and the Earth with respect to the target source. The AstroSat Orbit File Generator Utility3 was used to generate the orbit file for the observation duration. The resulting .orb file was provided to AstroSat’s as1bary4 tool for barycentric correction after applying the Barycentric Correction Code (Ver. 2017, Jan. 27).

3. Data analysis 3.1 Timing analysis The LaxpcSoft Format (A) Suite2 was used to analyse the LAXPC data. Light curves were generated using laxpc_make_lightcurve command in LaxpcSoft. The 10s binned complete light curve of the AstroSat/LAXPC observation of Centaurus X-3 on MJD 57734.57-57735.41 (12–13 December, 2016) combining observations of all the three LAXPC detectors, LAXPC10, LAXPC20 and LAXPC30 is depicted in Fig. 1. The corresponding estimated background rates are over-plotted in red. The systematic LAXPC error for the X-ray background is 2% as per the laxpc_make_backlightcurve task for generating background light curves. Timing analysis was carried out using NASA/GSFC’s HEASoft5 software package (Ver. 6.26.1 released on 2019, May 21) that comprises of FTOOLS,6 a general-purpose tool to manipulate FITS files and XRONOS7 timing analysis software package (Stella and Angelini 1992). The eclipses of Centaurus X-3 are expected to last for  10 h, as it has an orbital period of 2.1 days (Giacconi et al. 1971) and is known to show eclipses for  20% of the orbit (Nagase 1989), during which its X-ray flux drops by an order of magnitude (Giacconi et al. 1971; Schreier et al. 1972b). The absence of such an eclipsing event during the observation time span was confirmed by using a light curve with hourly resolution (time bin size = 3600s).

3

http://astrosat-ssc.iucaa.in:8080/orbitgen/. http://astrosat-ssc.iucaa.in/?q=data_and_analysis. 5 https://heasarc.gsfc.nasa.gov/lheasoft/download.html. 6 https://heasarc.gsfc.nasa.gov/ftools/. 7 https://heasarc.gsfc.nasa.gov/docs/xanadu/xronos/xronos.html. 4

Page 3 of 8

58

The detection of periodicity in data was accomplished by performing a Fourier transform of the light curve. This was done using the powspec command in HEASoft that produces the power density spectrum (PDS). The 0.01s binned light curve was divided into stretches of 32768 bins per interval and results from all intervals (except the last one in which the data is insufficient to fill all the time bins) were averaged in a single frame. The resultant PDS was observed to have a sharp peak at 0.2085 Hz. Higher harmonics were also detected upto fourth order. Using the reciprocal of this value as an initial approximate estimate of spin period, we searched for pulsations in the light curve with higher precision using efsearch in HEASoft, that determines the precise value of the local spin period Ps for each pointed observation by the standard v2 -maximisation technique. The efsearch task runs recursive finesearches for periodicities in a time series by folding the data over a range of periods around an estimated period at high resolution. The 1s binned light curve was folded with 65536 pulsations around the approximate value of 4.796s obtained from the power density spectrum with a 0.00001s time resolution and 64 phase bins/period. The resultant output is a distribution between the v2 value and the spin period of the pulsar that contains a Gaussian peak located at the best fit period which can be measured by fitting a Gaussian model to this pulsation peak. For example, for the 4th time segment, the value of spin period was found to be 4.79534 s. The width of the fitted Gaussian model to the peak is a measure of the error associated with the value of Ps . Using this exact value of spin period, the folded and stacked pulse profile was generated with the 1s binned light curve using the FTOOL efold by averaging over all consecutive pulses available in each time segment. The profile clearly consists of a prominent, highly-peaked feature followed by a minor shoulder or bump (Fig. 2). We refer to these as the primary peak and the secondary inter-pulse, respectively. Further, we resolve this pulse profile into different energy ranges of 3.0–6.0 keV, 6.0–9.0 keV, 9.0–15.0 keV and 15.0–40.0 keV as shown in Fig. 3 at a phase resolution of 64 phase bins per period. As the incident flux varies over the observation due to the motion of the pulsar in its binary orbit, the profiles are plotted with normalised intensity in order to discount the effect of such a variation to facilitate comparison. The secondary inter-pulse can be seen clearly in panels (a) and (b) and gradually subsides in panels (c) and (d),

Page 4 of 8

J. Astrophys. Astr.

5000

58

(2021) 42:58

2000

3000

3.2 Estimation of spin and orbital parameters

1000

Counts/s

4000

maximum and minimum intensity of the profile, respectively (Roy et al. 2020; Yang et al. 2018).

2×104

0

4×104

6×104

8×104

Time (s)

Figure 1. The complete 10s binned AstroSat/LAXPC light curve of Centaurus X-3 in 3.0–80.0 keV energy range using all the three LAXPC10, LAXPC20 and LAXPC30 detectors is presented in black colour. The gaps present in the light curve arise from the passage of AstroSat through the South Atlantic Anomaly (SAA) regions. The corresponding estimated background count rates are over-plotted in the same panel in red colour. The LAXPC error on these is 2% systematic error.

For an accreting X-ray pulsar, the observed spin period is modulated by the Doppler shift arising out of its orbital motion which suffers smearing when integrated over a long time interval. To obtain the accurate and refined results of the values of systemic spin and orbital parameters, we corrected for this Doppler modulation using phase calibration that assigns a value of spin phase to each detected X-ray photon. The photons were then categorised into 10 spin phase bins for further analysis. The Doppler formula for change in frequency m is given as Dm vr ¼ : m c

As the orbit of this binary is nearly circular, the radial component of the velocity along the line-of-sight is vr ¼ vb sin xb ðt  T0 Þ;

ð2Þ

where vb is the orbital velocity, T0 is a reference epoch corresponding to a vanishing vr and xb is the orbital angular velocity. Using Equations (1) and (2), one can write ( #) " vb 2pðt  T0 Þ ; ð3Þ mðtÞ ¼ m0 1 þ sin Pb c

1.0

Normalised Intensity (counts/s)

ð1Þ

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.0

0.2

0.4

0.6

0.8

1.0

Spin Phase

Figure 2. The pulse profiles, shown in different colours, obtained individually for all time segments (to the right) are overlapped on top of each other (to the left) for phase locking. The pulse profiles are averaged over the full LAXPC energy bandwidth of 3.0 to 80.0 keV.

where m0 is the systemic spin frequency of the pulsar observed at the reference epoch T0 and Pb is the orbital period of the binary system. Such a sinusoidal fit to the spin frequencies shown in Fig. 4 was obtained using the Levenberg–Marquardt algorithm, with m0 , vb , T0 and Pb as free parameters. We integrate Equation (3) to derive the value of phase with time given as " ( !#) vb Pb 2pðt  T0 Þ /ðtÞ ¼ m0 ðt  T0 Þ  1  cos : Pb 2pc ð4Þ

leaving just the primary peak to be observed at higher energies. The pulsed fraction of the pulse profiles was computed using the expression ðfmax  fmin Þ= ðfmax þ fmin Þ, where fmax and fmin are the values of

Applying Equation (4) to the time stamp of each detection, we assigned a phase value to each event and constructed phase histograms individually for each time segment, thus yielding the corresponding pulse profiles. It can be seen in Fig. 2 that the profiles do not

J. Astrophys. Astr.

(2021) 42:58

Page 5 of 8

and thus, provides a unique time marker for each of the pulse peaks. A systematic sinusoidal trend is evident, which we fit with a model with a periodicity of Pb =2 which is characteristic of a small deviation of the shape of the binary orbit from a circular approximation (Fabbiano & Schreier 1977). Despite the upper constraint on the orbital eccentricity of Centaurus X-3 being placed at a very small 0.0001 by Raichur and Paul (2010), we corrected for this effect nevertheless in order to derive precise systemic parameter measurements.

Normalised Intensity (counts/s)

1.5

(a) 3-6 keV

1.0 1.5

(b) 6-9 keV

1.0 1.5

(c) 9-15 keV

1.0 1.5

(d) 15-40 keV 1.0

0

0.5

1

58

1.5

2

Spin Phase

Figure 3. Energy resolved pulse profile of Centaurus X-3 in different X-ray energy bands (a) 3.0–6.0 keV, (b) 6.0– 9.0 keV, (c) 9.0–15.0 keV and (d) 15.0–40.0 keV obtained by folding and stacking the combined 1s time binned LAXPC10, LAXPC20, LAXPC30 light curve with a pulse period of 4.79534 s. The pulse profile in the figure was created using efold ftool with 64 phase bins per period and a reference time epoch of T0 ¼ 57734:1 MJD. The task normalises the output profiles for comparison by dividing by the average count rate.

Figure 4. The observed spin frequencies have been plotted for all time segments over the full observation time span of MJD 57734.57-57735.41. The errors in the measured spin frequencies are smaller than the marker size. The sinusoidal Doppler variation expected to arise from the motion of the pulsar in its binary orbit is overplotted in a solid black line.

align exactly. The differences in phases of the pulse profiles were obtained by locating the peak of the pulse profile for each time segment and is shown as a function of time in Fig. 5. The validity of this method is attributed to the stability of the shape of the pulse as verified in Fig. 2 which prevents ambiguities in measurements of phase due to pulse shape variations

4. Results and discussions The timing characteristics of the high mass X-ray binary Centaurus X-3 have been presented from AstroSat/LAXPC observations on MJD 57735. The count rate measured during our observation varied between 1187 counts/s and 1356 counts/s. The minimum value is still much larger than the expected drop of an order of magnitude during an eclipse (Giacconi et al. 1971; Schreier et al. 1972b). Thus, it was confirmed that an eclipsing event did not occur during the time span of our observation. A prominent pulse peak with a secondary interpulse was seen in the pulse profile in Fig. 2 which is in agreement with that reported in previous literature (Suchy et al. 2008; Burderi et al. 2000). Suchy et al. (2008) interpret this to be due to a larger fan out of higher energy emission in comparison to the more sharply beamed lower energy X-rays. The pulse profile was observed to have a significant dependence on photon energy as seen in Fig. 3. The pulsed fractions (%) are 51:8  0:34, 53:6  0:37, 58:2  0:35, 48:4  0:43 in the four energy bands in Fig. 3, respectively. A double-peaked pulse profile is seen in the 1–10 keV band which matches the overall pulse profile observed over the full energy band of 3.0–80.0 keV shown in Fig. 2. This double-peaked nature gradually evolves into a single-peaked behaviour above 10 keV, consistent with previous reports (Nagase et al. 1992). Kraus et al. (1996) have interpreted the shape and the energy-dependence of the pulse profile to point to a distorted magnetic dipole, as also confirmed by Suchy et al. (2008). The measured values of spin and orbital parameters are summarised in Table 1. A sinusoidal variation in the measured values of local spin period was observed which is consistent with our expectation from the orbital Doppler effect. The second

58

Page 6 of 8

J. Astrophys. Astr.

Figure 5. The figure shows the residuals in phase for all time segments after preliminary calibration carried out by correcting for the Doppler variation in measured spin frequencies. The residuals show a sinusoidal trend with a periodicity of Pb =2 typical of a small deviation from the circular orbit approximation. The root-mean-square (rms) variation is 0.033. Table 1. Measured values of the spin and orbital parameters of Centaurus X-3 using AstroSat 2016 observations. Parameter

Value

Pulsar spin Ps T0 a; b hP_s ij2016 1997 _ jPs =Ps j

4:80188  0:000085 s 57734:10  0:01 MJD ð1:978  0:015Þ1011 s/s 7709  58 yr

Binary orbit Pb a; c hP_b ij2016 1971 P_b =Pb v sin i a sin i

2:033  0:029 days ð1:19  0:63Þ103 days/yr ð5:78  2:9Þ104 yr1 410:2  6:2 km/s 38:23  0:58 lt-s

a

These represent average values estimated from the difference of the measured periods at the indicated epochs and dividing them by the time interval. b From Burderi et al. (2000) to this work. c From Schreier et al. (1972b) to this work.

sinusoidal nature of the phase residuals seen in Fig. 5 was found to have a periodicity of 0:964  0:011 days which agrees with  Pb =2 expected due to a small deviation from the circular orbit approximation (Fig. 6). Highly accurate phase calibration was achieved by correcting for both of the above effects as verified by a decrease in the rms of the phase residuals from 0.033 to 0.013. The small randomness remaining in the residuals arises from the slight variation in pulse profile during the observation span (Raichur & Paul 2010).

(2021) 42:58

Figure 6. The figure shows the residuals in phase for all time segments after fine calibration that corrects for the small deviation from circular orbit approximation to ensure accuracy of results. The residuals are randomly distributed and do not exhibit any systematic trend, which is evident from the slight variation in pulse profile during the observation span. The root-mean-square (rms) variation is 0.013.

The systemic spin period of the pulsar in Centaurus X-3 was found to be 4:80188  0:000085 s at the epoch of our observation. It has decreased from the last reported value of 4:81423  0:00001 s (Burderi et al. 2000) indicating that the neutron star is continuing to spin up, which is consistent with its known behaviour. Assuming a constant spin up rate, an estimate of the average spin up timescale over a  20 year stretch from MJD 50507.14 (Burderi et al. 2000) is found to be jPs =P_s j ¼ 7709  58 yr, more than twice of 3571.428 yr reported earlier by Schreier et al. (1972b). This retardation in the spin up rate may indicate variation in the transfer of angular momentum to the neutron star induced by a corresponding variation in the mass accretion rate (Tsunemi et al. 1996). A constant spin up rate is indeed an approximation— variations including glitches have been observed to be present (Bildsten et al. 1997). The orbital period was found to be 2:033  0:029 days, giving an average value of the orbital decay rate P_b =Pb as ð5:78  2:9Þ104 yr1 , with the largest possible time baseline of 45 years available in the literature from MJD 41131.58 (Schreier et al. 1972b) till this work. Although the significance of our measurement is somewhat limited, this value appears to be substantially higher than that of ð1:799  0:002Þ106 yr1 reported by Raichur and Paul (2010). This could indicate short time scale variations in the orbital period as noticed by Kelley et al. (1983). The orbital velocity projected along our line-ofsight was found to be 410:2  6:2 km/s which is similar to v sin i ¼ 415:1  0:4 km/s reported by Schreier et al. (1972b). The projected orbital radius

J. Astrophys. Astr.

(2021) 42:58

a sin i earlier measured to be 39:6612  0:0009 lt-s by Raichur and Paul (2010) was found to have decreased to 38:23  0:58 lt-s in accordance with the orbital decay of the Centaurus X-3 binary system.

Page 7 of 8

58

suggestions from an anonymous referee improved the presentation of this paper.

have

References 5. Conclusions We have carried out timing analysis of an out-ofeclipse X-ray observation of Centaurus X-3 using AstroSat/LAXPC. We find that the broad-band 3–80 keV pulse profile of Centaurus X-3 has a prominent pulse peak with a secondary inter-pulse which is consistent with previous observations (Suchy et al. 2008). The systemic spin period, corrected for the orbital motion of the X-ray pulsar around its optical counterpart and the small value of orbital eccentricity, was found to have decreased to 4:80188  0:000085 s from 4:81423  0:00001 s (Burderi et al. 2000) which is in agreement with the spin up trend observed in Centaurus X-3. The projected semi-major axis and orbital velocity were found to be 38:23  0:58 lt-s and 410:2  6:2 km/s, respectively. We observe an increase in the spin up timescale by over a factor of 2 to 7709  58 yr. This deceleration in the spin up rate of the pulsar could result from a re-proportioning of positive and negative torque action in the innermost regions of the accretion disk.

Acknowledgements The research is based to a significant extent on the results obtained from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). This work has used data from the Large Area X-ray Proportional Counter (LAXPC) detectors developed at TIFR, Mumbai and we thank the LAXPC Payload Operation Centre for verification and release of the data on the ISSDC data archive and providing the requisite software for analysis. We thank the AstroSat Science Support Cell (ASSC) hosted at IUCAA for technical assistance. This research has made use of the High Energy Astrophysics Software obtained through the High Energy Astrophysics Science Archive Research Center (HEASARC) Online Service, provided by the NASA/Goddard Space Flight Center (GSFC), in support of NASA’s High Energy Astrophysics Programs. This research has also made use of the NumPy and SciPy packages in Python. Valuable

Agrawal P. C. 2006, Adv. Space Res., 38, 2989 Agrawal P. C. et al. 2017, J. Astrophys. Astr., 38, 30 Antia H. M. et al. 2017, ApJS, 231, 10 Ash T. D. C., Reynolds A. P., Roche P. et al. 1999, MNRAS, 307, 357 Bhattacharya D. 2017, J. Astrophys. Astr., 38, 51 Bildsten L., Chakrabarty D., Chiu J. et al. 1997, ApJS, 113, 367 Burderi L., Di Salvo T., Robba N. R. et al. 2000, ApJ, 530, 429 Chodil G., Mark H., Rodrigues R. et al. 1967, Ph. Rev. Lett., 19, 681 Day C. S. R., Stevens I. R. 1993, ApJ, 403, 322 Fabbiano G., Schreier E. J. 1977, ApJ, 214, 235 Falanga M., Bozzo E., Lutovinov A. et al. 2015, A&A, 577, A130 Giacconi R., Gursky H., Kellogg E. et al. 1971, ApJL, 167, L67 Hutchings J. B., Cowley A. P., Crampton D. et al. 1979, ApJ, 229, 1079 Kelley R. L., Rappaport S., Clark G. W. et al. 1983, ApJ, 268, 790 Kraus U., Blum S., Schulte J. et al. 1996, ApJ, 467, 794 Krzeminski W. 1974, ApJL, 192, L135 Nagase F. 1989, PASJ, 41, 1 Nagase F., Corbet R. H. D., Day C. S. R. et al. 1992, ApJ, 396, 147 Naik S., Paul B., Ali Z. 2011, ApJ, 737, 79 Paul B. 2017, J. Astrophys. Astr., 38, 39 Petterson J. A. 1978, ApJ, 224, 625 Raichur H., Paul B. 2010, MNRAS, 401, 1532 Rawls M. L., Orosz J. A., McClintock J. E. et al. 2011, ApJ, 730, 25 Roy J., Agrawal P. C., Singari B. et al. 2020, RAA, 20, 155 Santangelo M., Benedetti W., Donati S. et al. 1988, Astronomia UAI, 6, 36 Schreier E., Levinson R., Gursky H. et al. 1972a, ApJL, 172, L79 Schreier E., Tananbaum H., Kellogg E. et al. 1972b, in Bulletin of the AAS, Vol. 4, 261 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, AstroSat Mission, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 9144, 91441S Stella L., Angelini L. 1992, in Data Analysis in Astronomy, 59 Suchy S., Pottschmidt K., Wilms J. et al. 2008, ApJ, 675, 1487 Thompson T. W. J., Rothschild R. E. 2009, ApJ, 691, 1744

58

Page 8 of 8

Tjemkes S. A., Zuiderwijk E. J., van Paradijs J. 1986, A&A, 154, 77 Tsunemi H., Kitamoto S., Tamura K. 1996, ApJ, 456, 316 Verbunt F., van den Heuvel E. P. J. 1995, in X-ray Binaries, 457 White N. E., Swank J. H. 1982, ApJL, 253, L61

J. Astrophys. Astr.

(2021) 42:58

Yadav J. S., Agrawal P. C., Antia H. M. et al. 2016, in den Herder J. W. A., Takahashi T., Bautz M., eds, SPIE Conf. Ser., Vol. 9905, Space Telescopes and Instrumentation 2016: Ultraviolet to Gamma Ray, 99051D Yang J., Zezas A., Coe M. J. et al. 2018, MNRAS, 479, L1

J. Astrophys. Astr. (2021) 42:83 https://doi.org/10.1007/s12036-021-09756-w

Ó Indian Academy of Sciences Sadhana(0123456789().,-volV)F T3](0123456789().,-volV)

SCIENCE RESULTS

Observations of AR Sco with Chandra and AstroSat soft X-ray telescope K. P. SINGH1,* , V. GIRISH2, J. TIWARI1, P. E. BARRETT3, D. A. H. BUCKLEY4,

S. B. POTTER4,8, E. SCHLEGEL5, V. RANA6 and G. STEWART7 1

Indian Institute of Science Education and Research Mohali, SAS Nagar, Sector 81, Mohali 140 306, India. Indian Space Research Organisation HQ, New BEL Road, Bengaluru 560 094, India. 3 The George Washington University, 725 21st St. NW, Washington, DC 20052, USA. 4 South African Astronomical Observatory, PO Box 9, Observatory Road, Observatory, Cape Town 7935, South Africa. 5 The University of Texas at San Antonio, One UTSA Circle, San Antonio, TX 78249, USA. 6 Raman Research Institute, C. V. Raman Avenue, Sadashivanagar, Bengaluru 560 080, India. 7 Department of Physics and Astronomy, The University of Leicester, University Road, Leicester LE1 7RH, UK. 8 Department of Physics, University of Johannesburg, PO Box 524, Auckland Park 2006, South Africa. *Corresponding Author. E-mail: [email protected] 2

MS received 5 November 2020; accepted 16 April 2021 Abstract. We present our AstroSat soft X-ray observations of a compact binary system, AR Sco, and analysis of its X-ray observations with Chandra that were taken only about a week before the AstroSat observations. An analysis of the soft X-ray (0.3–2.0 keV) data limits the modulation of the spin, orbital, or beat periods to less than 0.03 counts s1 or \10% of the average count rate. The X-ray flux obtained from both observatories is found to be almost identical (within a few percent) in flux, and about 30% lower than reported from the nine months older observations with XMM-Newton. A two-temperature thermal plasma model with the same spectral parameters fit Chandra and AstroSat data very well, and requires very little absorption in the line of sight to the source. The low-temperature component has the same temperature (  1 keV) as reported earlier, but the high-temperature component has a lower temperature of 5.0þ0:8 0:7 keV as compared to 8.0 keV measured earlier, however, the difference is not statistically significant. Keywords. Stars: individual: AR Sco—X-rays: binaries: novae, cataclysmic variables—white dwarfs.

1. Introduction A uniquely variable star, AR Scorpii (AR Sco) was initially mis-classified as a pulsating d-Scuti type (Satyvaldiev 1971). The star is at a distance of 117±1 pc based on parallax measurements given in Gaia Early Data Release 3 by Gaia Collaboration, Brown et al. (2020), with statistical error as given at the Gaia This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’. 1

https://gea.esac.esa.int/archive/.

website.1 It gained prominence when it was identified as a rare radio pulsing white dwarf (WD) binary system by Marsh et al. (2016). A very short spin period of Ps ¼ 117:1 s, and a synodic (spin-orbital beat) period, Pb ¼ 118:2 s and its harmonics, with very high amplitudes of Pb in the ultraviolet, optical and infra-red were found by Marsh et al. (2016). These periods were also seen at radio frequencies, and a slowing of the synodic period ðP_b ¼ 3:92  1013 s s1 ) was also reported by Marsh et al. (2016), thus implying a time-scale of 107 years for synchronization.

83

Page 2 of 10

However, in more recent photometric studies by Stiller et al. (2018) and Gaibor et al. (2020), the latter from data taken over a 5 year baseline, a larger spin down rate of P_b ¼ 6:82  1013 s s1 is derived, implying a commensurate decrease in the synchonization time. Marsh et al. (2016) concluded that AR Sco is a close binary system consisting of a WD and a cool low-mass M-star, with an orbital period of 3.56 h. They further pointed out that AR Sco is primarily spin-down powered, and its broadband spectral energy distribution is well explained by synchrotron radiation from relativistic electrons which likely originate from near the WD or are generated in situ at the M star surface through a direct interaction with the WD’s magnetosphere. Marsh et al. (2016) also reported on a short X-ray observation of AR Sco using the Swift X-ray Telescope (XRT) that showed the X-ray luminosity of AR Sco is very low ( 5  1030 erg s1 Þ and therefore, accretion is not sufficient to power the system. In some respects AR Scorpii resembles the cataclysmic variable AE Aquarii (Ikhsanov 1998; Meintjes and Venter 2005; Oruru and Meintjes 2012), which has a 33 s white dwarf spin period and a P_b ¼ 5:66  1014 s s1 and is in a propeller mass ejection phase, i.e., a system with very little or no mass accretion. However, as pointed out by Marsh et al. (2016) and Buckley et al. (2017), no flickering or broad emission lines in the optical are seen in AR Sco, in contrast to AE Aqr. This implies that the system is detached, with little or no mass loss from the secondary star, and therefore by definition is not a cataclysmic variable. AR Sco also exhibits strong (up to  40%) linear polarization in the optical, modulated strongly at Ps and Pb (Buckley et al. 2017; Potter and Buckley 2018a), which Buckley et al. (2017) used to derive an estimate of the white dwarf’s magnetic field of B  108 Gauss at its surface. Such a large field was also shown to be consistent with the system being powered by dipole radiation of the slowing down white dwarf, caused by the magnetic torque between the white dwarf and secondary star. A recent model proposed by Lyutikov et al. (2020) attempts to explain the properties of AR Sco with a lower white dwarf magnetic field, closer to what is seen in the intermediate polar sub-class of magnetic CVs, which would more easily explain the initial spin-up of the white dwarf. However, the current large spin-down rate, coupled to the very different polarimetric properties of AR Sco compared to intermediate polars, may be argue against a lower field strength. Optical photo-

J. Astrophys. Astr.

(2021) 42:83

polarimetry and photometry by Potter and Buckley (2018a, b) has identified several harmonic components of the spin and beat periods, and they have suggested a model in which the optical polarized emission comes from two diametrically opposed synchrotron emission regions in the magnetosphere of the WD, consistent with the model proposed by Takata et al. (2018). Finally, work by du Plessis et al. (2019), where the RVM (rotating vector model) for pulsars was successfully applied to the AR Sco polarimetric observations, is added evidence for the presence of a rotating magnetic dipole in AR Sco. AR Sco was observed in the UV and X-rays with XMMNewton by Takata et al. (2018), who found orbital modulation, but with no evidence of eclipses or absorption features. The XMM-Newton data also showed strong Xray pulsations at the synodic beat frequency mb ð¼ ms  mo Þ of 8.461100 mHz (Pb ¼ 118:2 s), where ms is the spin frequency at 8.5390 mHz (117.11 s) and mo is the orbital frequency at 0.07792 mHz (3.56 h). These modulations are seen mostly in the soft energy band of 0.15–2.0 keV (Takata et al. 2018). They also reported a very weak signal corresponding to the side-band frequency of ms þ mo . The pulse profiles were similar in UV and soft X-rays thus suggestive of similar origin, though the pulse modulation in UV is about 3 times higher than in X-rays. The phase averaged X-ray spectrum reported by Takata et al. (2018) was best fit by two-temperature (1.1 keV and 8.0 keV) plasma emission models and even better fit by a three-temperature (0.6, 1.7 and 8.0 keV) plasma model with solar abundance, with total flux of 3:2  1012 ergs cm2 s1 in the energy band of 0.15–12 keV (Takata et al. 2018). The low absorption column density, NH of 4  1020 cm2 and X-ray luminosity of 4  1030 ergs s1 are very different from that seen in intermediate polars. The X-ray emission likely arises from a non-thermally heated plasma and the system appears to be a WD analogue of a neutron star radio pulsar, although no direct evidence is found for the presence of a non-thermal component in the X-ray spectrum. Data analyzed in the energy range from 100 MeV to 500 GeV from Fermi Large Area Telescope (LAT) for the period August 4, 2008 to March 31, 2019, did not lead to any statistically significant detection either (Singh et al. 2020). The exact emission mechanism operating in AR Sco, therefore, continues to be a mystery. Here, we present X-ray observations of AR Sco with the Chandra X-ray Observatory and the AstroSat Soft X-ray telescope carried out just a few days apart from each other.

J. Astrophys. Astr.

(2021) 42:83

Page 3 of 10

83

2. Observations

3. Analysis and results

2.1 Chandra X-ray observatory

An X-ray image of AR Sco obtained with the Chandra ACIS is shown in Fig. 1. The source lies towards the edge of the field-of-view. A circular region of 7 arcsec radius was used for extraction of counts from the source. A box region of size 13600  9300 in the vicinity of the source was selected as the background region. An X-ray image of AR Sco taken with the SXT is shown in Fig. 2. Source counts were extracted from a circular region of 10 arcmin radius centered on AR Sco. Both a light curve and a spectrum were extracted from this region for further analysis.

AR Sco was observed with the Advanced CCD Imaging Spectrometer (ACIS-I) onboard the Chandra X-ray Observatory (CXO) from 2017 June 23, 23 h : 36 m : 13 s to June 24, 07 h : 25 m : 59 s. The data from this observation (Observation ID: 19711) were downloaded from archives maintained by NASA and analysed with Chandra Interactive Analysis of Observations (CIAO) software version 4.12. Each dataset was reprocessed with the most recentlyavailable calibration (CALDB version 4.9.1) applied to it using the chandra repro script. The reprocessing created a new Level 2 event file for each dataset which was used in all further analyses. The Level 2 event files were filtered for Solar Particle flares by analysing their light curves (LCs), and a useful exposure time of 24790 s was obtained.

2.2 AstroSat soft X-ray telescope AstroSat (Singh et al. 2014) observed AR Sco with the Soft X-ray Telescope (SXT) (Singh et al. 2016, 2017) as the prime instrument in the photon counting (PC) mode. SXT observed the source throughout an orbit of the satellite taking care that the Sun avoidance angle is  45 and RAM angle (the angle between the payload axis to the velocity vector direction of the spacecraft) [ 12 to ensure the safety of the mirrors and the detector. The observation (Observation ID 9000001350) was started on 2017 June 30 at 06 h:36 m:51 s UT and ended on 2017 July 2 at 10 h:13 m:53 s UT. Level 1 Data from individual orbits were received at the SXT POC (Payload Operation Centre) from the ISSDC (Indian Space Science Data Center). These data were processed with the sxtpipeline, available in the SXT software (AS1SXTLevel2, version 1.4b), thus calibrating the source events and extracting Level-2 cleaned event files for the individual orbits. The cleaned event files of all the orbits were then merged into a single cleaned event file using a Julia-based merger tool developed by G. C. Dewangan to avoid the time-overlapping events from the consecutive orbits. The XSELECT (V2.4d) package was used to extract the source spectra and light curves from the processed Level-2 cleaned event files and a useful exposure time of 53610 s was obtained.

3.1 X-ray light curves The X-ray light curve in the energy range of 0.3–7.0 keV obtained from Chandra observations is shown in Fig. 3. The task ‘dmextract’ was used to generate the background subtracted source light curve shown. A binsize of 3.364 s was used for extraction. The background region was supplied via the ‘dmextract’ task parameter ‘bkg’. The SXT light curve of AR Sco was extracted in the useful energy band of 0.3–7.1 keV and is shown in Fig. 4. The light curve was extracted with the highest time resolution of 2.3775s. We also extracted the light curves in the soft energy band of 0.3–2.0 keV from both the Chandra and SXT data, where most of the modulation was seen by Takata et al. (2018), and studied their power spectra. Barycentric corrections are applied to both of the light curves shown in Figures 3 and 4. The light curves are analyzed using the algorithm of Bretthorst (1988). This algorithm is a formal derivation of the discrete Fourier transform (DFT) using Bayesian statistics and is a generalization of the Lomb–Scargle periodogram (see Appendix A). The signal-to-noise (S/N) ratio of the resulting power spectra for the 0.3– 2.0 keV energy band of Chandra and AstroSat are shown in Figures 5 and 6, respectively. Neither figure shows a signal (or peak) having a S/N ratio of [5, or equivalently a significance of[99%, indicating that no signal is detected. Figure 7 shows S/N obtained for the combined Chandra and AstroSat light curves that include an injected sinusoidal Poisson signal with an amplitude of 0.03 counts s1 at a frequency of 5 mHz. The S/N ratio of the dominant signal is about 3 or a likelihood of [90%. Note that the S/N ratio decreases quickly from about 3 to 1 between 0.03 to 0.025

83

Page 4 of 10

J. Astrophys. Astr.

(2021) 42:83

Figure 1. Chandra X-ray image of AR Sco in the energy band of 0.3–7.0 keV showing the source region (circle) and the background region (rectangle) for the extraction of photons used in the analysis.

Figure 2. AstroSat SXT image of AR Sco in the energy band of 0.3–7.1 keV showing the circular extraction region of 10 arcmin radius for source counts. Smoothing by a Gaussian with kernel of 3 pixel radius (1 pixel = 4 arcsec) has been applied.

J. Astrophys. Astr.

(2021) 42:83

Figure 3. X-ray light curve of AR Sco obtained with the Chandra in the energy range of 0.3–7.0 keV without background subtraction.

Page 5 of 10

83

Figure 5. Signal-to-noise ratio at each frequency in mHz for the Chandra 0.3–2.0 keV data, obtained after applying the DFT. The S/N is\1 for all frequencies, so the data show no obvious signal at the spin, orbital, or beat periods. The two vertical lines mark the spin and beat frequencies.

the lower energy cutoff of the XMM-Newton X-ray band compared to that of AstroSat (0.15 keV vs. 0.3 keV), and a small change in the X-ray pulse fraction between the 2016 and 2017 observations, is the reason for the non-detection of the beat period by AstroSat and Chandra.

3.2 X-ray spectra

Figure 4. X-ray light curve of AR Sco obtained with the AstroSat SXT in the energy range of 0.3–7.1 keV without any background subtraction.

counts s1 . Based on this simulation, any modulation of the spin, orbital, and beat periods in the combined Chandra and AstroSat soft X-ray data are limited to less than about 0.03 counts s1 , or a modulation amplitude of less than about 10%, assuming an average count rate of 0.3 counts s1 . Takata et al. (2018) report an 14% pulse fraction of the beat period in the XMM-Newton 0.15–0.20 keV X-ray data and essentially no modulation in the 2–12 keV data from XMMNewton, thus indicating that the pulse fraction decreases rapidly with increasing energy. We conclude that this spectral characteristic combined with

X-ray spectra were extracted from both the Chandra and SXT observations. The filtered events file from Chandra observation was used for extraction of spectra and responses (Ancillary Response Function (ARF) and spectral Response Matrix File (RMF)) with the CIAO task ‘specextract’. The background region was supplied via the task parameter ‘bkgfile’. The parameter ‘correctpsf=yes’ was used to apply pointsource aperture correction to the source ARF file. For the SXT data, a background spectral file SkyBkg comb EL3p5 Cl Rd16p0 v01:pha, derived from a composite of several deep blank sky observations, distributed by the instrument team is used for spectral analysis for all the spectra analysed here. We used sxt arf excl00 v04 20190608:arf as the ARF, and sxt pc mat g0to12:rmf as the RMF for the SXT in this work. All these files are available at the SXT POC website https://www.tifr.res.in/*astrosat_sxt/ index.html. The source spectra from both the observations were grouped using the grppha tool to ensure a minimum of

83

Page 6 of 10

Figure 6. Similar to Fig. 5 for SXT 0.3–2.0 keV light curve. The S/N is \1.5 for all frequencies showing no obvious signal at the spin, orbital, or beat periods. The two vertical lines mark the spin and beat frequencies.

Figure 7. Same as above but with the combined Chandra and AstroSat data injected with a 0.03 counts s1 Poisson sinusoid. This simulation implies that any modulation of the spin, orbital, and beat periods is less than about 10% in soft X-rays.

25 counts per energy bin, prior to further analysis here and below. The average spectra obtained after background subtraction are shown in Fig. 8. We obtained a net count rate of 0:17  0:0026 s1 for the Chandra data and 0:034  0:0012 s1 in the SXT spectrum in the energy range 0.7-7.0 keV, used for a joint spectral analysis. We confined ourselves to the low energy limit of 0.7 keV instead of 0.3 keV as the background subtraction for this very weak source in the SXT indicated almost negligible flux below 0.7 keV while the flux in the Chandra data at this low energy was

J. Astrophys. Astr.

(2021) 42:83

slightly higher. This small discrepancy below 0.7 keV, though not statistically significant, would have compromised the value of the column density obtained. A line at 6.7 keV is observed clearly (Fig. 8) in the X-ray spectrum obtained from the Chandra observations. The line which is very close to the line expected from ionised Fe XXV indicates thermal emission in the source. The two spectra were fitted jointly with a simple plasma emission models apec using the xspec program (version 12.9.1; Arnaud 1996 distributed with the heasoft package (version 6.20)). The atomic data base AtomDB version 3.0.7 (http://www.atomdb.org), was used. We used a common absorber model Tbabs as a multiplicative model with the model parameter N H , i.e., the equivalent Galactic neutral hydrogen column density. The plasma temperature (kT) and N H parameter were kept free, at least initially. We used the abundance table ‘aspl’ given by Asplund et al. 2009 for our analysis, and the abundances of all the elements were tied together and could be varied together with respect to the solar values as one parameter. The normalisation for the plasma component was kept free for the two spectra, and v2 minimisation technique was used to find the best fit parameters. The column density, N H , was fixed at the low value of 0:1  1021 cm2 after it was found to be tending toward zero. This simple model, when fitted to the spectra in the energy range of 0.7–7.0 keV, gave a good fit with a plasma temperature  4 keV and solar elemental abundances. The best-fit values for this and other models used for fitting are given in Table 1. The normalisation of the apec model for the two sets of data differ only by  7%, most probably due to systematic differences in the calibration of the two instruments used. The total flux obtained for the energy band of 0.5–10 keV is found to be ð2:5  0:8Þ  1012 ergs cm2 s1 . Assuming non-solar abundances for all the elements improved the fit only slightly. We have explored the possibility of an additional thermal component: a black body or a plasma component at different temperature. Addition of either of these gave a significant improvement in the v2 (and based on the F-statistic values) as can be seen in Table 1, with the maximum improvement coming from the addition of another thermal component. The additional blackbody component was best fit for a temperature of 0.13 keV, but it required a significant absorption column density of 1:8  6:7  1021 cm2 for solar abundances and 2:3  8:2  1021 cm2 (90% confidence range) for

J. Astrophys. Astr.

(2021) 42:83

Page 7 of 10

83

Figure 8. X-ray Spectra of AR Sco (top panel) along with the best-fit 2-temperature plasma models (apec) shown as histograms. Contribution of each component of the plasma model is shown separately as dotted and dash curves, while the solid line histogram shows the combined contribution from the two components. The rms-normalised residuals per channel are shown in the bottom panel. Table 1. Spectral parameters obtained from a joint fit to the Chandra and SXT spectra (0.7–7.0 keV). Spectral model NH tbabs*apec tbabs*apec tbabs*(bbody?apec) tbabs*(bbody?apec) tbabs*(apec?apec) tbabs*(apec?apec) tbabs*(mekal?mekal)

a

0.1 0.1 4.7þ2:0 2:9 5.4þ2:8 3:1 0.2þ2:0 0:2 0.2þ1:9 0:2 0.3þ2:6 0:3

Parameters b

kTapec keV

Z

4.3þ0:4 0:3 4.1þ0:4 0:3 4.1þ1:1 0:6 4.1þ0:8 0:6 5.0þ0:8 0:7 5.0þ0:8 0:7 4.65þ0:7 0:7

1.0 0.6þ0:2 0:3 1 0.7þ0:3 0:3 1 0.6þ0:3 0:3 0.7þ0:2 0:3

c

A1 (C, S)

kTbb=apec keV

1.55 ,1.66 1.68 ,1.80 1.71,1.70 1.87,1.85 1.5,1.4 1.5,1.4 1.44,1.37

0.13þ0:03 0:01 0.13þ0:02 0:01 0.94þ0:15 0:18 0.95þ0:20 0:19 0.77þ0:18 0:16

A2c (C, S)

0.05, 0.07, 0.06, 0.07, 0.06,

0.06 0.08 0.13 0.15 0.14

v2m /dof

FluxAd (C, S)

FluxBd (C, S)

1.39/274 1.30/273 1.21/270 1.20/269 1.20/270 1.20/269 1.23/269

0.97,1.04 0.97,1.04 0.95,1.08 0.93,1.06 0.97,1.04 0.97,1.04 0.99,1.11

1.43,1.53 1.37,1.47 1.49,1.46 1.42,1.39 1.5,1.4 1.5,1.4 1.44,1.38

a

N H is in units of 1021 cm2 . b Abundance, Z, is relative to solar values for all the elements. c

A1 and A2 are the normalizations for the additive models (as listed) in units of 103 photons cm2 s1 ; The letters C and S for A1, A2 and for the flux values refer to Chandra and SXT respectively.

d

Fluxes are in units of 1012 ergs cm2 s1 and are quoted for two energy bands: A for 0.5–2.0 keV and B for 2.0–10 keV; All errors quoted are with 90% confidence.

non-solar abundances. The best fit to the two-temperature plasma models requires a low temperature component of 0.95þ0:20 0:19 keV and a high temperature

þ0:8 keV, elemental abundance of component of 5.00:7 0:6  0:3 solar, and a negligible N H in the range of ð0:0  2:1Þ  1021 cm2 . We also tried two-

83

Page 8 of 10

component mekal plasma models to compare with the results of Takata et al. (2018), and found very similar results, except for the low-temperature component which gave the best fit for a temperature of 0.77þ0:18 0:16 , slightly lower but consistent within the errors with that obtained from XMM-Newton data by Takata et al. (2018).

J. Astrophys. Astr.

(2021) 42:83

Poissonian because of the low count rates, which may increase the noise in the data caused by quantization error. It is possible that longer exposures by either of these two telescopes may increase the sensitivity enough to detect a modulation of the soft X-ray flux.

Acknowledgements

4. Discussion and conclusion Our X-ray observations of AR Sco were carried out in 2017, June–July within a few days of each other with Chandra and AstroSat SXT, and about nine months after the XMM-Newton observations (2016 September 19). We find that although on average AR Sco appears about 30% fainter this is not statistically significant given our errors. We find that the spectral data from the two instruments can be fit with the same models with remarkable agreement of normalization to within 7% of each other. The spectra are clearly thermal as significant emission due to the presence of an ionised iron (Fe XXV) line is seen clearly in the Chandra spectrum. The Fe line was also reported earlier by Takata et al. (2018) in the XMM-Newton spectrum. The best-fit spectral models favour two temperature plasma apec models with the best-fit high temperature (  5 keV) component differing from  8 keV obtained from XMM-Newton observations. It should be noted, however, that the SXT has a negligible effective area above the 7.1 keV limit used here, and Chandra has a very low area above 8.0 keV, and in this particular case the Chandra data above 7.0 keV were flagged as unusable. This could be the reason for the difference in the estimated high energy component. The low-temperature component (  1 keV) and the low column density (negligible) are in agreement, however. The time series analysis of the Chandra and AstroSat soft X-ray data show no periodicities at the WD spin period, the orbital period, or the beat period of the two. We believe that the lack of such periodicities is due to the lower sensitivity of the Chandra and AstroSat X-ray telescopes compared to that of XMM-Newton. The XMM light curve contains about 20000 counts, while those of Chandra and AstroSat SXT contain about a factor of ten less or about 2000 counts. Therefore, the relative noise in the Chandra and AstroSat periodograms are about a factor of three larger. We also note that the counting statistics are

We thank the Indian Space Research Organisation for scheduling the observations and the Indian Space Science Data Centre (ISSDC) for making the data available. This work has been performed utilizing the calibration data-bases and auxillary analysis tools developed, maintained and distributed by AstroSatSXT team with members from various institutions in India and abroad and the SXT Payload Operation Center (POC) at the TIFR, Mumbai for the pipeline reduction. The work has also made use of software, and/or web tools obtained from NASA’s High Energy Astrophysics Science Archive Research Center (HEASARC), a service of the Goddard Space Flight Center and the Smithsonian Astrophysical Observatory.

Appendix A Arthur Schuster (1905) introduced the periodogram near the beginning of the 20th century and this algorithmic technique has been discussed by many authors since that time; including Lomb (1976) and Scargle (1982) for which the Lomb–Scargle algorithm is known by the astronomical community. The periodogram CðxÞ is defined as the squared magnitude of the discrete Fourier transform of the data D ¼ d1 ; d2 ; . . .; dN :  2  N 1 1  X  2 2 CðxÞ ¼ ½RðxÞ þ IðxÞ ¼  dj eiwtj  ; ðA1Þ  N N  j¼1 where RðxÞ ¼

N X

di cosðxti Þ

ðA2Þ

di sinðxti Þ:

ðA3Þ

i¼1

and IðxÞ ¼

N X i¼1

A generalization of the periodogram for any time series function was derived using Bayesian principles

J. Astrophys. Astr.

(2021) 42:83

Page 9 of 10

by Bretthorst in his Ph.D. thesis (1988). The monograph is available for free download at https://bayes. wustl.edu/glb/book.pdf. The following is a brief presentation of Chapters 3 and 4, ‘The General Model Equation Plus Noise’. In the most general model, the hypothesis H is H f ðtÞ ¼

m X

Bj Gj ðt; fxgÞ;

ðA4Þ

where f(t) is some analytic expression of the time series, Gj ðt; fxgÞ is a particular model function, Bj is the amplitude of the model, and fxg is a set of parameters (in our case, frequencies). The Bayesian likelihood function for data D is ) ( NQ N ðA5Þ LðfBg; fxg; rÞ / r exp  2 ; 2r where m X N m X m 2X 1X Bj di Gj ðti Þ þ gjk Bj Bk ; N j¼1 i¼1 N j¼1 k¼1

ðA6Þ gjk ¼

N X

Gj ðti ÞGk ðti Þ;

ðA7Þ

i¼1

and r is the standard deviation of the noise. Q is an m  m matrix that can be orthogonalized, resulting in a set of eigenvectors of the likelihood equation. Therefore f(t) can be written in terms of a set of othonormal functions: f ðtÞ ¼

m X

Ak Hk ðtÞ;

ðA8Þ

k¼1

where Ak ¼

m pffiffiffiffiffi X kk Bj ekj ;

ðA9Þ

j¼1

Bk ¼

Lðgg; fxg; rÞ / r

þ

exp

" m N 2X  2 d2  Aj hj 2r N j¼1 #)

m 1X A2 N j¼1 j

; ðA12Þ

j¼1

Q d2 

( N

83

m X Aj ejk pffiffiffiffi kj j¼1

ðA10Þ

and m 1 X ejk Gk ðtÞ: Hj ðtÞ ¼ pffiffiffiffi kj k¼1

where hj

N X

di Hj ðti Þ; ð1  j  mÞ:

hj is the projection of the data onto the orthonormal model function Hj . Because we have no knowledge of the amplitudes Aj , they need to be marginalized, or integrated out, by performing j integrations, which gives the likelihood ) ( Nd 2  mh2 Nþm ; ðA14Þ exp  Lðfxg; rÞ / r 2r2 where d2

N 1X

d2 N i¼1 i

A change of functions (A6) and variables (A8) gives the joint likelihood of the new parameters

ðA15Þ

and h2

m 1X h2 : m j¼1 j

ðA16Þ

If r is unknown, then it’s considered a nuisance parameter and can be eliminated by integrating over all values using the Jeffreys prior 1=r. This gives an expression for the probability of the form of the ‘‘Student t-distribution’’: #mN " 2 mh2 : ðA17Þ PðfxgjD; IÞ / 1  Nd2 Although fAg and r are described as nuisance parameters when deriving the likelihood of the frequency parameters, they are actually of some interest. Bretthorst shows in Chapter 4 that the first posterior moments, i.e., the expected amplitudes are hAj i ¼ hj ;

ðA11Þ

ðA13Þ

i¼1

ðA18Þ

the second posterior moments are hAj Ak i ¼ hj jk þ r2 djk ; the noise variance is

ðA19Þ

83

Page 10 of 10

" # N m X X 1 2 2 hr i ¼ d  hj ; N  m  2 i¼1 i j¼1 2

and the signal-to-noise ratio is ( " #)12 Signal m h2 : ¼ 1þ 2 Noise N r

J. Astrophys. Astr.

ðA20Þ

ðA21Þ

References Arnaud K. A. 1996, in Jacoby G., Barnes J., eds, Astronomical Data Analysis Software and Systems V, ASP Conference Series, Vol. 101, p. 17 Asplund M., Grevesse N., Sauval A. J., Scott P. 2009, ARAA, 47, 481 Bretthorst G. L. 1988, in Berger J., Fienberg S., Gani J., Krickeberg K., Singer B., eds, Bayesian Spectrum Analysis and Parameter Estimation, Lecture Notes in Statistics, Springer-Verlag, Vol. 48 Buckley D. A. H., Meintjes P. J., Potter S. B., Marsh T. R., Ga¨nsicke B. T. 2017, Nature Astron., 1, 0029 du Plessis L, Wadiasingh Z., Venter C., Harding A. K. 2019, ApJ, 887, 44 Brown A. G. A., Vallenari A., Prusti T., de Bruijne J. H. J. et al. 2020, A&A (in press); https://doi.org/10.1051/ 0004-6361/202039657 Gaibor Y., Garnavich P. M., Littlefield C., Potter S. B., Buckley D. A. H. 2020, MNRAS, 496, 4849

(2021) 42:83

Ikhsanov N. R. 1998, Astron. Astrophys., 338, 521 Lomb N. R. 1976, Adv. Space Sci., 39, 447 Lyutikov M., Barkov M., Route M., Balsara D., Garnavich P., Littlefield C. 2020, arXiv:2004.11474 Marsh T. R., Ga¨nsicke B. T., Hu¨mmerich S. et al. 2016, Nature, 537, 374 Meintjes P. J., Venter L. A. 2005, MNRAS, 360, 573 Oruru B., Meintjes P. J. 2012, MNRAS, 421, 1557 Potter S. B., Buckley D. A. H. 2018, MNRAS, 478, L78 Potter S. B., Buckley D. A. H. 2018, MNRAS, 481, 2384 Satyvaldiev V. 1971, Astron. Tsirk. 633, 7 Scargle J. D. 1982, ApJ, 263, 835 Schuster A. 1905, Proc. Roy. Soc. London, 77, 136 Singh K. K., Meintjes P. J., Kaplan Q., Ramamonjisoa F. A., Sahayanathan S. 2020, astro-ph.HE, 2006. 12950v1 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, Proc. SPIE, Space Telescopes and Instrumentation 2014: Ultraviolet to Gamma Ray. 9144, 91441S, https://doi. org/10.1117/12.2062667 Singh K. P., Stewart G. C., Chandra S. et al. 2016, Proc. SPIE, in Space Telescopes and Instrumentation 2016: Ultraviolet to Gamma Ray 9905, p. 99051E, https://doi. org/10.1117/12.2235309 Singh K. P., Stewart G. C., Westergaard N. J. et al. 2017, J. Astrophys. Astr., 38, 29 Stiller R. A., Littlefield C., Garnavich P. et al. 2018, AJ, 156, 150 Takata J., Hu C.-P., Lin L. C. C., Tam P. H. T., Pal P. S., Hui C. Y., Kong A. K. H., Cheng K. S. 2018, ApJ, 853, 106

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:53 https://doi.org/10.1007/s12036-021-09725-3

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

Study of dynamical status of the globular cluster NGC 1851 using ultraviolet imaging telescope GAURAV SINGH1,2,* , R. K. S. YADAV1, SNEHALATA SAHU3

and ANNAPURNI SUBRAMANIAM3 1

Aryabhatta Research Institute of Observational Sciences (ARIES), Manora Peak, Nainital 263 001, India. Department of Physics and Astrophysics, University of Delhi, Delhi 110 007, India. 3 Indian Institute of Astrophysics, Koramangala II Block, Bangalore 560 034, India. *Corresponding Author. E-mail: [email protected] 2

MS received 6 November 2020; accepted 3 February 2021 Abstract. We present the study of dynamical status of the globular cluster NGC 1851. A combination of multi-wavelength space and ground-based data sets are used for the present analysis. In order to select the genuine cluster members, we used the astro-photometric data available from HST and GAIA-DR2 catalogs. The BSS radial distribution of the cluster is plotted from the center of the cluster to the outskirts. The radial distribution of BSS shows a central peak, followed by a dip at the intermediate radii (rmin  9000 ) and a rising trend in the outskirts. We also estimated Aþ rh parameter as 0.391 ± 0.006 to validate the findings of the radial distribution study. On the basis of the minima in the BSS radial distribution and the value of Aþ rh parameter, we conclude that NGC 1851 belongs to Family II classification and is an intermediate dynamical state cluster. Keywords. Galaxy: globular—clusters: individual: NGC 1851—stars: blue stragglers—stars: Hertzsprung– Russell and colour-magnitude diagrams.

1. Introduction Globular clusters (GCs) are compact, centrally concentrated and gravitationally bound systems of stars. The high density in the central region of globular clusters lead to frequent gravitational interactions. The gravitational interactions among stars result in various dynamical processes i.e., two-body relaxation, core collapse, mass segregation, stellar collisions, stellar mergers etc (Meylan & Heggie 1997). These dynamical processes give rise to several exotic populations, i.e., millisecond pulsars, cataclysmic variables and blue straggler stars (BSSs) (Ferraro et al. 2001; Bailyn 1995). BSSs are bright and are more populous among these exotic population, therefore serve as a crucial probe to understand the internal dynamics of the GCs. This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

In 1953, Sandage discovered BSSs populating an unusual location in the optical color-magnitude diagram (CMD) of the GC, M3 (Sandage 1953). He found the location of these stars to be bluer with respect to the main sequence stars and they appear as an extension of the main sequence turnoff (MS-TO). Based on isochrone fitting technique, Shara et al. (1997) found BSSs to be more massive (M  1:2M ) than the average mass of stars in the GCs (M  0:3M ). Using Spectral Energy Distribution (SED) fitting technique, Raso et al. (2019) found similar estimates of BSS masses, with a few BSSs having masses larger that 2 times of MS-TO. Being more massive than the normal stars, they are subject to dynamical friction, which segregates the BSSs towards the centre of the cluster (Ferraro et al. 2006). At the very centrally dense regions of GCs, HST astro-photometric catalog provides useful information to select cluster members. Also, after the release of

53

Page 2 of 9

GAIA DR2 it has now become feasible to select genuine BSS population from the outer region of the cluster. Once, the selection of genuine BSS population are completed, their radial distribution can be studied to infer the dynamical status of the GCs. On the basis of the shape of the observed radial distribution, Ferraro (2012) classified GCs into three main categories: (1) Family I: The radial distribution of BSSs show a flat distribution, and are classified as dynamically young systems. The examples of GCs classified in Family I are e.g., Palomar 14 (Beccari et al. 2011), NGC 2419 (Dalessandro et al. 2008b), and x Centauri (Ferraro et al. 2006). (2) Family II: The radial distribution of BSSs show a bimodal distribution with a central peak, followed by a minima at some intermediate radii (rmin ), and an external rising trend. These clusters are classified as dynamically intermediate clusters. The example of the clusters showing bimodal distribution are: NGC 6388 studied by Dalessandro et al. (2008a), M53 by Beccari et al. (2008), M5 by Lanzoni et al. (2007a), and 47 Tuc by Ferraro et al. (2004), M55 by Lanzoni et al. (2007c), NGC 6752 by Sabbi et al. (2004) and NGC 5824 by Sanna et al. (2014). (3) Family III: The radial distribution of BSSs show a monotonic behaviour, with a central peak followed by a decreasing trend and no signs of an external rise. These clusters are classified as dynamically old systems. The examples of GCs showing unimodal behavior are e.g., M79 studied by Lanzoni (2007b), M80 and M30 by Ferraro (2012), and M75 by Contreras et al. (2012). In the recent years, a new parameter (Aþ rh ) has been proposed to measure the dynamical segregation of BSSs by Alessandrini et al. (2016), which is given as the area between the cumulative distribution curves of the reference population (/REF ðxÞ) and BSSs (/BSS ðxÞ): Z x þ /BSS ðx0 Þ  /REF ðx0 Þdx0 ð1Þ Arh ðxÞ ¼ xmin

In the above equation, (x ¼ logðr=rh )) is defined as the logarithmic distance from the center of the cluster and scaled over the half-mass radius rh of the cluster. Lanzoni et al. (2016) found a direct correlation between rmin and Aþ rh parameter, suggesting that both the parameters are governed by a basic mechanism i.e., dynamical friction.

J. Astrophys. Astr. (2021)42:53

UVIT/AstroSat observations have also been beneficial in the BSS radial distribution studies. Sahu et al. (2019) studied the specific frequency of BSS in the cluster NGC 288 and found a bimodal radial distribution using UVIT/AstroSat data. In this paper, we present the study of dynamical status of the cluster, NGC 1851. This is a high density globular cluster with the core (rc ) and tidal (rt ) radii around 500 :4 and 48100 :6, respectively (Ferraro et al. 2018). NGC 1851 (aJ2000 ¼ 5h 14m 6s :76, dJ2000 ¼ 40 20 4700 :6; l ¼ 244 :51, b ¼ 35 .03, Harris 1996 (2010 edition)1) is located at a distance of 12.1 kpc from the Sun. NGC 1851 is an intermediate metallicity cluster ([Fe/H] = -1.18). In Table 1, we list all the known properties of the cluster NGC 1851. Subramaniam et al. (2017) and Singh et al. (2020) presented the UV and optical CMDs of the cluster using UVIT/AstroSat data. The paper is organized in the following manner: data used for the present analysis is presented in Section 2, the selection of BSS and reference population followed by the results obtained from radial distribution in Section 3 and we discuss the results, followed by summary and conclusions in Sections 4 and 5.

2. Data sets To study the dynamical status and to create the radial distribution of BSS in the cluster NGC 1851, we use space and ground-based data sets in NUV and optical wavelengths. We aim to study the BSSs located in the entire cluster extension, i.e., from the cluster centre to the tidal radii (rt ). We use the astro-photometric catalog from Nardiello et al. (2018) in the central dense region (r  9000 ) of the cluster, observed as a part of the HST UV Legacy Survey of Galactic Globular Cluster (HUGS) program (Piotto et al. 2015). The catalog contains the photometric data sets in F275W, F336W, and F438W pass-bands, which were observed through WFC3/ UVIS channel. The F606W and F814W pass-bands were observed through ACS/WFC channel. The catalog also contains the membership information of all the stars that are common to WFC3/UVIS and ACS/ WFC field-of-view (FoV). To select the BSSs from the outer region (9000  r  48000 ), we use Near-UV (NUV) data in N279N filter, observed through Ultra Violet Imaging Telescope (UVIT) on board AstroSat satellite during 19th–21st 1

http://physwww.physics.mcmaster.ca/*harris/mwGC.dat.

J. Astrophys. Astr. (2021)42:53

Page 3 of 9

Table 1. Parameters of the cluster NGC 1851 used in this paper. Parameter RA (J2000) DEC (J2000) Distance Metallicity, [Fe/H] Distance modulus Age Core radius (rc ) Tidal radius (rt )

Value

References

5h 14m 6s :76 40 20 4700 :6 12.1 kpc -1.18 dex 15.47 mag 10 Gyr 500 :4 48100 :6

Harris (1996) Harris (1996) Harris (1996) Kunder et al. (2013) Cassisi et al. (2008) Cassisi et al. (2008) Ferraro et al. (2018) Ferraro et al. (2018)

March 2016, as a part of Performance Verification (PV) phase. UVIT/AstroSat provides a simultaneous observations in a wide range of electromagnetic spectrum from Far-UV (130–180 nm) to NUV (200–300 nm) wavelengths. It provides a circular field-of-view of about  280 and an angular resolution better than 1:00 8 in both the channels. Subramaniam et al. (2016) and Tandon et al. (2017) presented the details of instrument and calibration of the UVIT data. For this cluster, the data acquisition and reduction procedures are described in detail in Subramaniam et al. (2017). For the optical wavelength in the outer region, we used the UBVRI photometric catalog provided by Stetson et al. (2019). The Gaia DR2 catalog provides the information of photometry and astrometry of all the stars down to G  21 mag (Collaboration 2016, 2018). We, therefore, obtained the membership information using the method given in Sanders (1971) based on the maximum likelihood principle for all the stars located outside the HST region and using Gaia DR2 catalog. Figure 1 shows the histogram plot (frequency vs. membership probability) for all the stars lying in the outer region. Since, in the histogram star counts start rising after 50% membership probability. We have used membership criteria of P  50% for the selection of cluster members in the further analysis. So, by combining the HST membership catalog with the Gaia DR2 membership data, we now have membership information of most of the stars used in the present analysis.

3. Analysis and results 3.1 BSS and reference population selection The first step towards studying the BS radial distribution is to carefully select the genuine BSS and a

53

reference population. For this purpose, we will use NUV pass-bands for the primary selection of the BSSs. To select the reference population and complete BSS population, we will use the optical pass-bands for the selection, as described in (Singh & Yadav 2019).

3.2 BSS population selection For selecting the BSSs in the inner region, we use HST CMD (F275W, (F275W-F606W)) and for selecting BSSs from the outer region, we use UVIT CMD (N279N, (N275N-V)). The NUV-optical CMDs are fitted with the hybrid models i.e., the isochrone is fitted with the the Flexible Stellar Population Synthesis (FSPS) models of Conroy et al. (2009) and HB sequence is fitted with the HB model generated using the updated Bag of Stellar Tracks and Isochrones (BaSTI-IAC,2 Hidalgo et al. 2018). Both the FSPS and BaSTI-IAC model are generated for a metallicity of [Fe/H] = 1:2 dex (Kunder et al. 2013), a distance modulus of 15.47 mag and an assumed age of 10 Gyr (Cassisi et al. 2008). The theoretical isochrones are fitted well with the observed data points. Raso et al. (2017) have used UV CMD (F275W, (F275W-F336W)) to select the BSSs from the central region of the cluster. In the UV CMD, BSS sample can be selected both efficiently and reliably, since it can be easily separated from the optical blends just above the MS-TO. Therefore, to select the BSSs from the entire region, we use NUV-optical CMD (HST and UVIT CMDs) as our primary selection criteria where the contamination from the optical blends can be minimized. In the NUV-optical CMD, the BSSs can be separated from the optical blends which extends a vertical sequence just above the main sequence while in optical CMD it is difficult to separate these two sequences. The selection box criteria is shown in the Fig. 2. Since, in NUV-optical CMDs the BSSs can be easily distinguished and are separated from the optical blends and visible plumes. The evolutionary track of BSS in the theoretical isochones also provides useful information for defining the selection criteria for the selection of BSSs. In Fig. 2, the BSSs defines a vertical sequence, while the cooler giants like SGB, RGB are suppressed. The selected BSSs are shown with filled circles. To minimize the contamination from the MS-TO and the sub-giant branch, we adopted a limiting magnitude of F275W = 19.85 mag and N279N = 19.30 mag in the HST and UVIT CMDs, respectively. 2

http://basti-iac.oa-abruzzo.inaf.it/hbmodels.html.

53

Page 4 of 9

Figure 1. The frequency vs. membership probability for all the stars lying in the outer region is shown. In this histogram star counts start rising after 50% membership probability. Therefore, we can use membership criteria of P  50% for the selection of cluster members in the further analysis.

J. Astrophys. Astr. (2021)42:53

The upper limit extends up to the brightest BSSs found in both the samples. We have identified 154 BSSs in the inner region and 10 BSSs in the outer region using NUV-optical CMDs. Once the BSSs are selected from the NUV-optical CMDs, they are used to define the selection box criteria in the optical CMD (V, (V-I)), as shown in Fig. 3. By combining the Stetson’s catalog with the Gaia DR2 data and HST photometry, it is possible to cover both the inner and outer regions (Stetson et al. 2019). Therefore, we have used Stetson’s catalog in the outer region, and HST in the inner region. In order to select the BSS and reference population from a common optical CMD, the F606W and F814W magnitudes are transformed to calibrated V and I magnitudes available in the Stetson’s catalog, using the relationship derived by Sirianni et al. (2005). We found 3 additional BSSs from the optical CMD in the inner region and 5 additional BSSs in the outer region. The reason for getting these additional BSSs is attributed to the fact that there are 3 BSSs that lies in the ACS FoV but are not present in the WFC3 FoV. However, the 4 additional BSSs in the outer region are due to the incompleteness of the UVIT N279N and one is the BSS?EHB photometric binary companion identified by Singh et al. (2020), that is located in the EHB sequence in the UVIT CMD. Therefore, in total we have found 172 BSSs throughout the entire cluster region, 157 in the inner and 15 in the outer region. All the BSSs are genuine cluster members with a membership probability, P  80%.

3.3 Reference population selection

Figure 2. The selection of BSSs from the NUV-optical CMDs, with (F275W, (F275W-F606W)) in the left panel and (N279N, (N275N-V)) in the right panel, respectively. The selected BSSs are shown with filled circles and they extends a vertical sequence in the NUV-optical CMDs. The NUV-optical CMDs are fitted with the hybrid models i.e., the isochrone is fitted with the FSPS models and the HB sequence is fitted with the HB model generated through BaSTI-IAC models. The selection box criteria is shown in both the panels.

Reference populations are important to understand the segregation of BSSs and we used post-MS stars, since they define a natural trend in radial distribution against which the radial distribution of the BSSs can be compared. The number of stars in a Post-MS stage is directly proportional to its evolutionary time scales (Lanzoni et al. 2007c; Renzini et al. 1988). These post-MS stars are, therefore, important for qualitative study of BSS specific frequency and radial distribution studies. The specific frequency of these branches are constant throughout the entire cluster region and are equal to the evolutionary time scales of the HB and GB (SGB?RGB) phase respectively. Therefore, these branches are considered as the reference population.

J. Astrophys. Astr. (2021)42:53

Page 5 of 9

53

3.4 BSS radial distribution

Figure 3. The selection of the BSS and the reference population using optical CMD (V, (V-I)). The selection box criteria of BSS and reference population are shown for both the inner and outer regions in the left panel and the right panel, respectively. The selected BSSs are marked by filled circles.

For the reference population selection, we consider the same limiting magnitudes subtended by the BSS population in the optical CMD. Therefore, our selection of reference population are not affected by the completeness of the sample. To select the genuine reference population, we have adopted the membership criteria of P  50%, since at higher membership cutoffs, we found a significant change in the number of reference population. In Fig. 3, the selection criteria for reference population is shown. We used the same selection criteria followed by Singh and Yadav (2019) to select the reference population. Hence, we found 394 and 127 HB stars in the inner and outer sample respectively. Also, 2690 and 634 GB stars from the inner and outer regions, respectively. In total, we found, 521 HB and 3324 GB stars from the entire cluster region. Figure 4 shows the spatial distribution of BSSs selected from the entire cluster region. The BSSs are more concentrated towards the center and a large fraction of them are located in the inner region, much within the half-mass radius (rh ). However, significant number of BSSs are present in the outer region of the cluster as well. We also plotted the spatial distribution of GB population which is nearly symmetric in comparison to the spatial distribution of BSSs.

In this section, we present the radial distribution of BSS with respect to the reference population. To obtain the radial distribution plot of BSS with respect to the reference population, we divided the cluster area into six concentric circles. In the Table 2, we listed the number of genuine BSS and reference population corresponding to each of the radial bin (NBSS , NGB and NHB ). We also obtain the specific BSS BSS ¼ NBSS /NGB , FHB ¼ NBSS /NHB and frequency (FGB HB FGB ¼ NHB /NGB ) of BSS and plotted them in Fig. 5. The errors are plotted with 1 sigma error bar. The specific frequency of BSS show a bimodal distribution with a central peak followed by a minima and an external rising trend in the outer region, while the reference population show a flat radial distribution. In order to check the significance of the rise in the external region, we performed Z-test3 and found the significance level of rise in the outer region to be  77%. We also obtained the doubly normalized ratio for BSS with respect to the reference population. These population (‘‘Pop’’) could be BSS, HB or RGB. It is defined as the number of ‘‘Pop’’ observed in a region to the total number of ‘‘Pop’’ divided by the fraction of light sampled in the same region with respect to the total measured luminosity (Ferraro et al. 1993). It is written as RPop ¼

tot NPop =NPop

Lsamp =Ltot samp

:

ð2Þ

We estimated the sampled to the total luminosity for each radial bin by integrating the isotropic single-mass King profile using parameters taken from Harris (1996) catalog. We assumed the Poisson error in the values of luminosities and numbers. Using the formula described in Sabbi et al. (2004), we considered propagation of errors to estimate the errors in the double normalized ratios. In Table 2, we listed the luminosity ratios computed in the corresponding annulus. In Fig. 6, we plot the radial distribution of BSS, HB and GB, using double normalized ratios with respect to the radial distance scaled over rc . In the upper panel, double normalized ratio of BSS with respect to HB is plotted and the lower panel shows the variation of double normalized ratio of BSS with respect to GB. The value of RBSS shows a bimodal distribution with a peak in the center, a minima at 3

http://www.stat.yale.edu/Courses/1997-98/101/sigtest.htm.

53

Page 6 of 9

J. Astrophys. Astr. (2021)42:53

Figure 4. The plot show the spatial distribution of BSSs in the entire cluster region and are marked by filled blue circles. The GB population is shown as black dots. Table 2. The log of the Number counts for BSS and reference population. Radial bin (arcsec)

NBSS

NHB

NGB

Lsamp =Ltot samp

0–15 15–30 30–60 60–120 120–240 240–480

118 18 15 7 11 3

125 96 103 113 60 22

839 619 791 591 342 141

0.34 0.20 0.19 0.15 0.10 0.06

r  17rc and an outward rising trend. The double normalized ratios of the reference population (RHB and RGB ), however show a flattened behaviour (  1), as expected from the stellar evolution theories (Renzini et al. 1988). To recheck the dynamical segregation of the BSSs, we obtained the Aþ rh parameter defined by Lanzoni et al. (2016) as the area enclosed between the cumulative radial distribution of BSS and the reference populations scaled over rh . In Fig. 7, the cumulative radial distributions of BSS with respect to HB and GB are plotted in the left and right panels, respectively. The value of Aþ rh are estimated as 0.394 and 0.386 with respect to GB and HB respectively. The corresponding values of mean and standard deviation of Aþ rh are therefore 0.391 and 0.006, respectively.

Figure 5. The specific frequency of BSS with respect to HB and GB are plotted in the upper panel and middle panel, respectively, while the specific frequency of HB and GB is plotted in the lower panel. The specific frequency distribution of BSS show a bimodal distribution, whereas for the reference population, it shows a constant value throughout the entire cluster region.

In the upper panel of Fig. 8, we plot the empirical dynamical clock relation defined by Ferraro (2012), which correlates the position of minima of the BSS radial distribution (rmin =rc ) with the core relaxation time (trc /tH ). In the lower panel of Figure 8, we plot the correlation of Aþ rh parameter and trc /tH . The cluster NGC 1851 is shown with filled circle. The value of error in rc is taken from Miocchi et al. (2013) for estimating the error in the rmin =rc . The values of rmin , trc , and rc of other clusters are adopted from Ferraro (2012).

4. Discussions The specific frequency and double normalized ratio of BSS show bimodal distribution, with a peak in the central region, followed by a clear dip at the intermediate radii (rmin  9000 ) and a rising trend in the external region. The BSSs are more massive than the reference population and therefore are subjected to the dynamical friction which plays a crucial role in segregating the BSSs towards the cluster center. However, the BSSs that are located in the outskirts, show a flat distribution and are not yet affected by the action

J. Astrophys. Astr. (2021)42:53

Figure 6. The double normalized ratios of BSS are plotted with respect to HB and GB in the upper panel and lower panels, respectively. The BSS radial distribution show a peak in the center, a minima at the intermediate radii (rmin  9000 ), and the external rising trend.

Figure 7. The cumulative distribution plot of BSS, HB and GB. The Aþ rh parameter is obtained as area between the curves of BSS with respect to HB and GB and are found to be 0.386 and 0.394, respectively.

of dynamical friction. The radial distribution of BSS observed in NGC 1851 are in agreement with the previous studies (i.e., Ferraro 2012) on the cluster showing bimodal radial distribution.

Page 7 of 9

53

Figure 8. The plot showing the location of the cluster NGC 1851 in the empirical dynamical clock relation defined by Ferraro (2012) in the upper panel. The filled triangles are dynamically old clusters while open circles are the dynamically intermediate age clusters. The dynamically young clusters are plotted as lower-limit arrows at rmin  0:1. The cluster location in the empirical relation defined by Lanzoni et al. (2016) is shown in lower panel.

Ferraro et al. (2018) found the value of Aþ rh parameter of the cluster NGC 1851 to be 0.48 ± 0.04. Our estimation of Aþ rh parameter is slightly less than the value estimated by Ferraro et al. (2018), and it could be due to the strict condition of membership probability, P  50% for the selection of BSS and reference population. Although, Ferraro et al. (2018) included proper motion selection criteria using the Vector-Point diagram (VPD) from the HUGS data set. The difference in the Aþ rh parameter could be largely due to the selection of reference population. Ferraro et al. (2018) used 5r selection box criteria for the selection of MS-TO as the reference population in the UV-CMD, while in the present analysis, we have done the selection of reference population (HB and GB) from the optical CMD, due to incompleteness of the MS-TO sample in the UVIT CMD. The Aþ rh parameter of the NGC 1851, studied by Ferraro et al. (2018) does provides the information about the level of segregation of BSSs in the cluster sample, but in the central dense region of the cluster (up to rh ), therefore, BSS radial distribution is need to understand the level of segregation of BSSs located in the outskirts (after rh ) of the cluster. Also, it provides a very useful information about the cluster dynamical

53

Page 8 of 9

state. Therefore, we have estimated both the rmin and Aþ rh parameters to obtain the dynamical state of the cluster, NGC 1851. The position of the rmin is defined by Ferraro (2012) as an indicator of the level of segregation of BSSs or dynamical status of the cluster. There is a well established empirical dynamical clock relation between rmin and core or half-mass relaxation times (trc or trh ) defined by Ferraro (2012). Also, Lanzoni et al. (2016) uses Aþ rh parameter and show that there exists a direct correlation between rmin and Aþ rh . Therefore, based on the position of the minima in the BSS radial distribution and Aþ rh parameter, we suggest that NGC 1851 belongs to Family II classification and is an intermediate dynamical state cluster. The bimodal radial distribution suggest that, BSSs located in the outskirts of the cluster are still not affected by the role of dynamical friction. The spatial distribution of the BSSs also suggest that most of the BSSs are located in the central region, while some BSSs are located in the outskirts, which validates our findings. In the outskirts of the GC, NGC 1851, Singh et al. (2020) found a candidate BSS?EHB binary system, suggesting that some BSSs are located in the outskirts of NGC 1851, where density is low and binary BSSs can form through transfer mass from the companion star.

J. Astrophys. Astr. (2021)42:53

Acknowledgements We are thankful to the reviewer for the thoughtful comments and suggestions that improved the quality of the manuscript. This publication uses the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). This publication uses UVIT data processed by the payload operations centre at IIA. The UVIT is built in collaboration between IIA, IUCAA, TIFR, ISRO and CSA. This work has made use of data from the European Space Agency (ESA) mission Gaia (https://www.cosmos.esa.int/ gaia), processed by the Gaia Data Processing and Analysis Consortium (DPAC, https://www.cosmos. esa.int/web/gaia/dpac/consortium). Funding for the DPAC has been provided by national institutions, in particular, the institutions participating in the Gaia Multilateral Agreement. This research has made use of data, software and/or web tools obtained from the High Energy Astrophysics Science Archive Research Center (HEASARC), a service of the Astrophysics Science Division at NASA/GSFC and of the Smithsonian Astrophysical Observatory’s High Energy Astrophysics Division.

References 5. Summary and conclusions (1) We present the dynamical status of the cluster NGC 1851, using data from HST, UVIT/AstoSat, GAIA DR2 and ground-based catalogs. All the BSS and reference population used in the study are bonafide members of the cluster. (2) We found in total 172 number of BSSs from the entire cluster sample using NUV-optical and optical CMDs for the selection. We also identified 521 and 3324 number of HB and GB stars as reference populations. (3) The observed radial distribution of BSS shows a peak in the central region, a dip at the intermediate radii (rmin  9000 ) and a rising trend in the external region. We also estimated Aþ rh parameter to be 0.391 ± 0.006. The values of both the rmin and Aþ rh parameter therefore indicate that NGC 1851 is an intermediate dynamical state cluster and it belongs to Family II classification. This indicates that the BSSs located in the outskirts of the cluster are still not affected by the action of dynamical friction.

Alessandrini E., Lanzoni B., Ferraro F. R., Miocchi P., Vesperini E. 2016, ApJ, 833, 252 Bailyn C. D. 1995, ARA&A, 33, 133 Baldwin A. T., Watkins L. L., van der Marel R. P., Bianchini P., Bellini A., Anderson J. 2016, ApJ, 827, 12 Beccari G., Lanzoni B., Ferraro F. R., Pulone L., Bellazzini M., Fusi Pecci F., Rood R. T., Giallongo E. 2008, ApJ, 679, 712 Beccari G., Sollima A., Ferraro F. R., Lanzoni B., Bellazzini M., De Marchi G., Valls-Gabaud D., Rood R. T. 2011, ApJ, 737, L3 Cassisi S., Salaris M., Pietrinferni A. et al. 2008, ApJL, 672, L115 Conroy C., Gunn J. E., White M. 2009, ApJ, 699, 486 Contreras Ramos R., Ferraro F. R., Dalessandro E., Lanzoni B., Rood R. T. 2012, ApJ, 748, 91 Dalessandro E., Lanzoni B., Ferraro F. R., Rood R. T., Milone A., Piotto G., Valenti E. 2008, ApJ, 677, 1069 Dalessandro E., Lanzoni B., Ferraro F. R., Vespe F., Bellazzini M., Rood R. T. 2008, ApJ, 681, 311 Ferraro F. R., Pecci F. F., Cacciari C., Corsi C., Buonanno R., Fahlman G. G., Richer H. B. 1993, AJ, 106, 2324 Ferraro F. R., D’Amico N., Possenti A., Mignani R. P., Paltrinieri B. 2001, ApJ, 561, 337

J. Astrophys. Astr. (2021)42:53 Ferraro F. R., Beccari G., Rood R. T., Bellazzini M., Sills A., Sabbi E. 2004, ApJ, 603, 12 Ferraro F. R., Lanzoni B., Raso S. et al. 2018, ApJ, 860,36 Ferraro F. R., Sollima A., Rood R. T., Origlia L., Pancino E., Bellazzini M. 2006, ApJ, 638, 433 Ferraro F. R. et al. 2012, Nature, 492, 393 Gaia Collaboration et al. 2016, A&A, 595, A1 Gaia Collaboration et al. 2018, A&A, 616, A1 Harris W. E. 1996, AJ, 112, 1487 Hidalgo S. L., Pietrinferni A., Cassisi S. et al. 2018, ApJ, 856, 125 Kunder A., Salaris M., Cassisi S. et al. 2013, AJ, 145, 25 Lanzoni B., Dalessandro E., Ferraro F. R., Mancini C., Beccari G., Rood R. T., Mapelli M., Sigurdsson S. 2007, ApJ, 663, 267 Lanzoni B. et al. 2007, ApJ, 663, 1040 Lanzoni B., Dalessandro E., Perina S., Ferraro F. R., Rood R. T., Sollima A. 2007, ApJ, 670, 1065 Lanzoni B., Ferraro F. R., Alessandrini E., Dalessandro E., Vesperini E., Raso S. 2016, ApJ, 833, L29 Meylan G., Heggie D. C. 1997, A&AR, 8, 1 Miocchi P., Lanzoni B., Ferraro F. R. et al. 2013, ApJ, 774, 151 Nardiello D., Libralato M., Piotto G. et al. 2018, MNRAS, 481, 3382 Paresce F., Meylan G., Shara M., Baxter D., Greenfield P. 1991, Nature, 352, 297 Piotto G. et al. 2015, AJ, 149, 91

Page 9 of 9

53

Raso S., Ferraro F. R., Dalessandro E. et al. 2017, ApJ, 839, 64 Raso S., Pallanca C., Ferraro F. R. et al. 2019, ApJ, 879,56 Renzini A., Fusi Pecci F. 1988, ARA&A, 26, 199 Sabbi E., Ferraro F. R., Sills A., Rood R. T. 2004, ApJ, 617, 1296 Sahu S., Subramaniam A., Cote P., Rao N. K., Stetson P. B. 2019, MNRAS, 482, 1080 Sandage A. R. 1953, AJ, 58, 61 Sanders W. L. 1971, A&A, 14, 226 Sanna N., Dalessandro E., Ferraro F. R., Lanzoni B., Miocchi P., O’Connell R. W. 2014, ApJ, 780, 90 Shara M. M., Saffer R. A., Livio M. 1997, ApJ, 489, L59 Singh G., Sahu S., Subramaniam A., Yadav R. K. S. 2020, ApJ, 905, 44 Singh G., Yadav R. K. S. 2019, MNRAS, 482, 4874 Sirianni M., Jee M. J., Benitez N. et al. 2005, PASP, 117, 1049 Stetson P. B., Pancino E., Zocchi A., Sanna N., Monelli M. 2019, MNRAS, 485, 3042 Subramaniam A., Tandon S. N., Hutchings J. et al. 2016, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, vol. 9905, In-orbit Performance of UVIT on ASTROSAT, 99051F Subramaniam A., Sahu S., Postma J. E. et al. 2017, AJ, 154, 233 Tandon S. N., Subramaniam A., Girish V. et al. 2017, AJ, 154, 128

 Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:82 https://doi.org/10.1007/s12036-021-09718-2

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

Sub-MeV spectroscopy with AstroSat-CZT imager for gamma ray bursts TANMOY CHATTOPADHYAY1,* , SOUMYA GUPTA2, VIDUSHI SHARMA2,3,

SHABNAM IYYANI2, AJAY RATHEESH4,5,6, N. P. S. MITHUN7, E. AARTHY7, SOURAV PALIT8, ABHAY KUMAR7, SANTOSH V. VADAWALE7, A. R. RAO2,9, VARUN BHALERAO8 and DIPANKAR BHATTACHARYA2 1

Kavli Institute of Astrophysics and Cosmology, 452 Lomita Mall, Stanford, CA 94305, USA. The Inter-University Centre for Astronomy and Astrophysics, Pune 411 007, India. 3 Department of Physics, KTH Royal Institute of Technology, AlbaNova, 10691 Stockholm, Sweden. 4 Dipartimento di Fisica, Universita di Roma Tor Vergata, Via della Ricerca Scientifica 1, 00133 Roma, Italy. 5 INAF Istituto di Astrofisica e Planetologia Spaziali, Via del Fosso del Cavaliere, 00133 Roma, RM, Italy. 6 Dipartimento di Fisica, Universita La Sapienza, 00185P. le A. Moro 2, Roma, Italy. 7 Physical Research Laboratory, Ahmedabad 380 009, India. 8 Indian Institute of Technology Bombay, Mumbai 400 076, India. 9 Tata Institute of Fundamental Research, Mumbai 400 005, India. Corresponding author. E-mail: [email protected]; [email protected] * 2

MS received 7 November 2020; accepted 27 January 2021 Abstract. Cadmium–Zinc–Telluride Imager (CZTI) onboard AstroSat has been a prolific Gamma-Ray Burst (GRB) monitor. While the 2-pixel Compton scattered events (100–300 keV) are used to extract sensitive spectroscopic information, the inclusion of the low-gain pixels (  20% of the detector plane) after careful calibration extends the energy range of Compton energy spectra to 600 keV. The new feature also allows single-pixel spectroscopy of the GRBs to the sub-MeV range which is otherwise limited to 150 keV. We also introduced a new noise rejection algorithm in the analysis (‘Compton noise’). These new additions not only enhances the spectroscopic sensitivity of CZTI, but the sub-MeV spectroscopy will also allow proper characterization of the GRBs not detected by Fermi. This article describes the methodology of single, Compton event and veto spectroscopy in 100–900 keV combined for the GRBs detected in the first year of operation. CZTI in last five years has detected  20 bright GRBs. The new methodologies, when applied on the spectral analysis for this large sample of GRBs, has the potential to improve the results significantly and help in better understanding the prompt emission mechanism. Keywords. AstroSat—CZT imager—sub-MeV spectroscopy—gamma ray burst.

1. Introduction Cadmium–Zinc–Telluride Imager (hereafter CZTI) on board AstroSat (Singh et al. 2014; Paul 2013), India’s first dedicated astronomical satellite, has been demonstrated as a prolific Gamma Ray Burst (GRB) monitor, since the launch of AstroSat (Rao et al. 2016; Chattopadhyay et al. 2019). CZTI is one of the two hard XThis article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

ray detectors sensitive in 20–150 keV. The instrument employs an array of CZT detectors, each 40 mm  40 mm  5 mm in size, totalling to a collecting of 924 cm2 . Each detector is further segmented spatially to 256 pixels with a pitch of  2.5 mm. Use of collimator made of 0.07 mm tantalum and 1 mm aluminum sheets restricts the field of view of the instrument to  4 . Details of the payload design and function are given in Bhalerao et al. (2017) and Rao et al. (2016). At energies beyond 100 keV, the increasing transparency of the collimators and the supporting structure enables CZTI to work as an all-

82

Page 2 of 20

sky monitor. Because of this all-sky sensitivity, CZTI instrument since the launch of AstroSat has been working as an efficient GRB monitor with around  83 GRB detections per year.1 In the last one year, we have explored a number of new techniques in the spectral analysis for bright ONaxis sources like Crab and Cygnus X-1 (Chattopadhyay et al. 2021, under preparation). We also identified a number of possible improvements in the AstroSat mass model for better spectroscopic and polarimetry analysis for these sources. Implementation of these new techniques (listed below) will yield a significant improvement in the overall spectro-polarimetric sensitivity of GRBs detected by the CZTI. • After the launch of AstroSat,  20% of the CZTI pixels were found to have electronic gains significantly lower than the ground calibrated gain values. Majority of these pixels now possesses gain around 2–4 times lower than expected. However, the gain for these pixels is now stable since launch i.e. the gain change was a one-time phenomenon whose origin remains unknown. As a result of the lower gain, these pixels have a higher energy threshold of  60 keV for X-ray photon detection but are also sensitive to photons of much higher energies up to  800 keV. We refer to these pixels as low-gain pixels, which were originally excluded from any scientific analysis. However, after a careful and detailed analysis of the events from these pixels, here we attempt to include these pixels to increase the spectroscopic energy ranges for GRBs. • From the Detector Plane Histogram (DPH) images of the valid Compton scattered events, we further identify the noisy pixels giving rise to 2-pixel events. Filtering out this ‘Compton noise’ is otherwise not removed from the standard noise rejection algorithm. The new techniques allow us to explore the capability of CZTI as a sub-MeV GRB spectrometer. In the standard CZTI analysis pipeline, the prompt emission spectroscopy of the bursts are limited only in 100–200 keV whereas even with the 2-pixel Compton scattering events, the spectroscopy can only be extended up to  350 keV. With the utilization of the low-gain pixels, the spectroscopy of the GRBs are now extended all the way up to  1 MeV. There are three different ways the CZTI instrument provides spectroscopic information for the GRBs: (1) 1-pixel or 1

http://astrosat.iucaa.in/czti/?q=grb.

J. Astrophys. Astr. (2021)42:82

single pixel events from CZT detectors in 100–900 keV, (2) 2-pixel or Compton scattering events from CZT detectors in 100–700 keV which are used to extract polarization information and (3) four CsI-Veto detectors below the CZTI sensitive in 100–500 keV. We use the AstroSat mass model to generate the effective area as a function of energy and response matrix for each of these spectroscopic techniques and perform broadband spectral analysis along with Fermi and Swift-BAT data. Proper spectral fits and constraining the spectral parameters critically depend on the correct estimation of response matrix elements which are different for different GRB direction with respect to the satellite orientation. Although the mass model has been validated and tested in detail using imaging method (Mate et al. 2021), spectroscopic analysis of the eleven GRBs which cover the full sky with respect to the AstroSat satellite indirectly tests the mass model further. This also helps in identifying the shortcomings in some parts of the mass model and quantifying those from the spectral fits. CZTI subMeV spectroscopy is particularly valuable for those GRBs which are detected by AstroSat and Niel Gehrels Swift BAT but not by Fermi, as it allows us to constrain the spectral parameters including the peak energy in the energy range (15–900 keV), which otherwise generally is not possible because of the narrow energy range of BAT. In this article, we explore CZTI as a sub-MeV spectroscopy and report the spectroscopic measurements for the eleven bright GRBs detected in the first year of CZTI operation with the implementation of these new developments for the entire burst time interval which is obtained using the Bayesian block technique on the GRB single event data. The new techniques and the burst selection methods are described in Section 2 In Section 3, we describe the spectroscopy methods followed by broadband spectral analysis in Section 4. While this article primarily outlines the methodologies of subMeV spectroscopy for GRBs, we plan to apply the new techniques to a sample of  20 bright GRBs detected in last five years of operation of AstroSat.

2. New techniques in the spectrum analysis In this section, we describe the new techniques implemented in the spectral analysis compared to that discussed in Chattopadhyay et al. (2019). In the previous polarimetry reports on AstroSat GRB by Chattopadhyay et al. (2019); Chand et al. (2018, 2019); Sharma et al. (2019), we utilized only 75–80% of the

J. Astrophys. Astr. (2021)42:82

Page 3 of 20

82

CZTI collecting area consisting only the ‘spectroscopically good’ pixels. A fraction of CZTI pixels are found to have lower gains (gain value 3–4 times lower than that of the normal or good pixels) and therefore are sensitive at higher energies. We performed a detailed characterization of these pixels and utilize them in the spectrum analysis extending the overall spectroscopic energy range to  1 MeV. We discuss these new developments below along with our new strategy of selecting the burst interval for spectroscopic analysis of the GRBs.

2.1 Characterization of low-gain pixels From the detector plane histogram (DPH) of the onboard data, it was seen that the count rate in some spatially clustered pixels were significantly lower compared to the mean count rate (see Fig. 1). Even though most of the pixels are found in clusters, there are instances of isolated pixels as well. These pixels also did not show the alpha tagged line at 60 keV from the on board calibration source 241 Am, indicating that the gain has shifted at least by a factor of two or three (hereafter we refer to these pixels as low-gain pixels). The reason for the shift is unknown, however since no shift was seen in the laboratory measurements during calibration and appeared right after the launch, mechanical stress during the launch is thought to be one of the possibilities. From the light curve analysis from the low-gain pixels with different time bins, we found that the count rates detected by these pixels are of Poissonian nature and therefore the detected events are not spurious pixel noise and could be real X-ray events. Because these pixels consist almost 20% of the CZTI active area, we explored the possibility of characterizing the pixels in detail. In absence of any mono-energetic line at higher energies to calculate the correct gain for the low-gain pixels, we compare the overlapping region of the continuum spectrum of these pixels and the spectroscopically good pixels. For that purpose we first fitted the good pixel spectra for each of the 64 detector modules with an empirical model in 45–180 keV range using three Gaussian (tantalum ka line at 54 keV, a bump structure around 65 keV source of which is unknown and a tellurium activation line at around 88 keV) and a broken power law with a break energy around 140 keV as shown in Fig. 2. The break energy denotes the onset of falling detection efficiency for a 5-mm

Figure 1. The detector plane histogram (DPH) of all the CZT detector quadrants for the obsID: 9000000618 (data from 2016 August). The lower count rates detected in a fraction of the pixels are seen as patches in the DPH which is because of the relatively higher gain values of the pixels. The color bars indicate the count rate.

Figure 2. Continuum spectra from the spectroscopically good pixels (in blue) for one of the detector modules (data taken from July 2016). The spectrum is fitted with an empirical model (red) consisting of three Gaussian: (1) Tantalum line at 54 keV, (2) a bump structure near the tantalum line and (3) an arbitrary line around 90 keV which is most likely a proton induced background feature (Odaka et al. 2018) and a broken power law (break energy around 140 keV which denotes the onset of falling detection efficiency for the 5 mm thick CZT detectors). This template has been used to compare the spectra of the low-gain pixels to estimate their gains (see text for more details).

thick CZT detector and therefore can also be used to calibrate the low-gain pixels along with the continuum comparison. The strong line around 88 keV seen in the spectra is supposed to originate from high

82

Page 4 of 20

J. Astrophys. Astr. (2021)42:82

Figure 3. Spectra of three low-gain pixels of type I before and after applying gain correction given in left and right panels respectively. Top: Pixel number 161 from module 4, middle: pixel number 248 from module 5, and bottom: pixel number 45 from module 13. The red lines are the empirical models used to compare the overlapping region of 45–180 keV of the low gain pixels. After comparison, the fitted gain shift factors (a multiplication factor to the ground calibrated gain) are found to be between 0.8 and 1.5 for the type I low gain pixels.

energy particle induced tellurium activation (127m Te with half life of 9:17  106 seconds (Odaka et al. 2018). We also see a hint of a line feature around 145 keV which could also be from activated tellurium (125m Te with half life of 4:96  106 seconds). Because of the large half life of the isotopes, we see the lines even far from the SAA region where the

activation is supposed to take place (Odaka et al. 2018). Since the number of good pixels vary in each module, the count rate was normalised by the total number of good pixels in that module. In order to have sufficient statistics in the spectra of both good and low-gain pixels, we took a long one month data (  1 million seconds of exposure). The South

J. Astrophys. Astr. (2021)42:82

Page 5 of 20

82

Figure 4. Same as Fig. 3 but for three type II low-gain pixels (top: pixel number 223 from module 13, middle: pixel number 0 from module 5, and bottom: pixel number 1 from module 5). For type II low-gain pixels, we found the fitted gain shift factors between 1.5 and 4.

Atlantic Anamoly (SAA) regions are normally excluded in the raw data itself based on the count rates from an on board charge particle monitor. In addition to that, a time interval of 500 seconds was ignored before and after the already excluded SAA region in order to filter out the high particle background regions. The fitted module wise good pixel models were then used to compare the spectra of each of the low-gain pixels in 100–180 keV range and reduced v2 values were calculated by varying a

multiplication factor to the ground calibrated gain of the low-gain pixels in the range of 0.8–5.0 at an interval of 0.01. We call this multiplication factor to the gain as ‘gain shift factor’. Based on the fitting results, we classified the low-gain pixels into three subcategories: (1) Low-gain pixels type I: These pixels were found to have gain shift factor between 0.8 and 1.5 and are seen to have spectral features like the tantalum

82

Page 6 of 20

J. Astrophys. Astr. (2021)42:82

gains very close their ground calibration values, we plan to calibrate them with the on board calibration source for further validation, details of which will be presented elsewhere (Mithun et al., in prep.). (2) Low-gain pixels type II: These pixels were found to have relatively higher gain shift values between 1.5 and 5.0. Comparison of the continuum spectra in 100–180 keV before and after gain correction is shown in Fig. 4. (3) Low-gain pixels type III: For a fraction of pixels we could not get satisfactory fit in the common 100–180 keV range even for the maximum gain shift values. These pixels are ignored from any Figure 5. The pulse profile of Crab pulsar in low-gain further analysis. pixels (blue) of all the CZT quadrants after gain correction. For comparison, the pulse profile in the spectroscopically good pixels are plotted against it (red).

Figure 6. Light curve of GRB 160821A in the low-gain pixels with corrected gains. Different colors represent the four different CZTI quadrants as indicated inside the plot. The time axis is plotted from AstroSat time 209507728 seconds (marked as zero). Each CZTI quadrant is shadowed by different degree for each GRB according to its location with respect to the spacecraft giving rise to unequal flux levels in different CZTI quadrants.

line, the 90 keV background and spectral break at around 140 keV as also seen in the good pixel spectra. Left column of Fig. 3 shows the comparison of the spectra (shown in blue) with the good pixel model (shown in red) for three pixels whereas the right panel shows the same comparison after correcting for the gains of the pixels. These pixels were previously identified as lowgain pixels in CALDB. Since we find them to have

We carried out the analysis for each of the five years of CZTI data (normally June/July of each year depending on the data availability) to check for repeatability or any possible time evolution in the obtained gain values for the type I and type II lowgain pixels. We use the gain list from the year of detection of a given GRB (note for this paper we use gain list of the year 2016 as all the GRBs analyzed here are detected in 2016). In future, we plan to characterize the low-gain pixels (particularly the type II pixels) using various particle-induced radioactivation background lines (Odaka et al. 2018) for further verification. It might be possible to further verify the gain values by looking at the Crab pulse profile from these pixels and calculate the ratio of pulsed fractions in two pulses as they are known to be energy dependent. We also plan to validate the gain values of the type I pixels by investigating the alpha tagged spectra from these pixels. In order to boost the confidence in the use of lowgain pixels, we attempted to reconstruct the Crab pulse profile using these pixels after gain correction. Figure 5 shows the pulse profile of Crab pulsar in lowgain pixels from all the four CZTI quadrants during a  78 ks observation on 14th January 2017. This further verifies that the events from these pixels are genuine X-ray events and not random noise. We used the crab ephemeris at MJD 57769.0 from Lyne et al. (1999). The events are folded from AstroSat time of 222220803.426 seconds. The background is subtracted by the counts in the off pulse region and the pulse profile is normalised by the maximum peak counts for the purpose of visualization. We could also detect the GRBs in these pixels as shown in Fig. 6 for GRB 160821A.Since the number of low-gain pixels vary in different quadrants, the

J. Astrophys. Astr. (2021)42:82

Figure 7. Detector plane histogram of the neighbouring 2-pixel Compton events for the 3rd CZTI quadrant. The plotted data belongs to obsID: 9000000618 (data from 2016 August). The color bar indicates count rate. The brighter spots in the image correspond to the Compton noisy events arising from noisy neighboring pixels. These events are removed from further analysis.

count rate is normalized by the total number of lowgain pixels in a quadrant. Detection of astrophysical sources in low-gain pixels therefore presents a strong case in using them for future spectral analysis.

2.2 Compton noise CZTI has already been demonstrated as a sensitive ON-axis and GRB polarimeter in 100-350 keV in Vadawale et al. (2018) and Chattopadhyay et al. (2019) respectively, where the Compton scattered events are used to generate the azimuthal angle histogram. The same Compton events can be used in spectroscopy of the GRBs. These events are selected through strict Compton kinematics criteria: • identify the adjacent 2-pixel events from 20 ls coincidence window both from the spectroscopically good pixels and low-gain pixels, • impose criteria of ratio of the energies deposited in two pixels between 1 and 6 in order to filter out the noisy chance events. This is motivated by the fact that in a true Compton scattering event, the electron recoil energy deposited in one of the pixels, is much lower than the scattered photon energy deposited in the other pixel.

Page 7 of 20

82

In spite of the strict selection criteria, there is still a significant amount of overlapping noise events. Neighbouring pixels can flicker at time scales lower than the coincidence time window of 20 ls causing some of these events to permeate into the Compton event selection and thus causing instrumental artifacts in the modulation curve. A DPH showing outliers in 2-pixel events is shown in Fig. 7. These events can be identified as outliers from the DPH of neighbouring 2-pixel events and can be removed from further analysis. Threshold for an outlier is kept at four sigma and three sigma from the mean for normal and low-gain pixels respectively. Due to the difference in count rate between the side and corner pixel double events, we identify the noisy pixels in the side and corner pixels separately. When a pixel is identified as noisy, no events from that pixel is considered for further analysis. Further details on the Compton noise analysis can be found in Ratheesh et al., 2020, this issue.

2.3 Selection of burst interval

In this work, the spectrum analyses are conducted on the time integrated emission of the bursts. The time interval corresponding to the integrated emission is chosen by employing the Bayesian block algorithm (Scargle 1998; Scargle et al. 2013; Burgess 2014) of time binning on the single pixel event data of the bursts. The block with the minimum probability density value corresponding to the background region is taken as the guide to decide the start and stop times of the integrated emission. The onset time of the first block with the probability density greater than that of the background which is closer to the onset time of the burst and the end time of the last block after which the background continues are considered as the start and stop times of the time interval of integrated emission respectively (Fig. A1 in Appendix). In the next section, we describe the methodology of spectroscopy using 1-pixel and 2-pixel CZTI events and CsI-veto detected events followed by broadband spectroscopy results for the eleven bright GRBs detected in 2015–2016. For CZTI events, we utilize both the standard good pixels and the newly calibrated low-gain pixels to extend the energy spectra to sub-MeV region.

82

Page 8 of 20

J. Astrophys. Astr. (2021)42:82

Figure 8. GRB light curves from CZTI 1-pixel (red) and 2-pixel events (black points) for the 11 GRBs. The time intervals of the bursts are obtained from the Bayesian block analysis on the 1-pixel CZTI light curves as shown by the vertical dashed lines. The ‘zero’ denoted in the time-axis stands for the trigger time reported by the Fermi-GBM.

J. Astrophys. Astr. (2021)42:82

3. Methodology for spectroscopy 3.1 Spectroscopic response The 2D spectral responses for CZTI 1-pixel, 2-pixel and CsI/veto spectroscopy are generated using GEANT4 simulation. Here we outline the basic steps of response generation. Response is computed using GEANT4 mono-energy simulations of the full AstroSat mass model at specific h and / viewing angles for each GRB (h and / for a given GRB are provided by either Swift or Fermi). The mono-energetic lines for simulations were selected between 100 keV and 2 MeV at every 20 keV till 1 MeV and at every 100 keV in 1–2 MeV, totalling around 56 mono-energies. Simulation is done for a large number of photons (109 photons for each energy) in order to have a statistically significant energy distribution in CZTI for each mono-energetic line. The simulation file contains information of total seven interactions or steps for each incident photon (x, y, z-position of interactions in CZTI and deposited energy in each interaction, see Chattopadhyay et al. (2014)) in CZTI modules. We add up the energies from all the interactions happening within a pixel of 2.5 mm  2.5 mm in Interactive Data Language (IDL2) based routine outside the GEANT4. We apply the same CZTI pixel-level LLD (Lower Level Discriminator) values in the simulation data whereas the ULDs (Upper Level Discriminator) were computed from actual observational data for each module and is applied to simulation data accordingly. From this event list, the 1-pixel and 2-pixel events are separated and processed differently for final response generation. For 1-pixel events, the distribution of deposited energies is calculated at a bin size of 1 keV from 0 keV to 1000 keV (total 1000 bins) for each of the 56 mono-energies. It is to be noted that Geant4 simulation takes care of all types of interactions with appropriate probabilities including photoelectric, Compton, Rayleigh inside CZTI and photons scattered from the spacecraft or other surrounding payloads to CZTI. Because of these multiple interactions and scattered events from surrounding materials, the distribution of deposited energy in CZTI is broad and non-gaussian. However, the large number of photon simulation gives sufficient statistics to obtain the correct energy distribution in the full range of 100 bins for all 56 mono-energies. The 2D matrix (56  1000) of deposited energy 2

Research Systems Inc. (1995). IDL user’s guide: interactive data language version 4. Boulder, CO: Research Systems.

Page 9 of 20

82

distributions for the mono-energetic lines is then convolved with a Gaussian function of appropriate width to generate the 1-pixel spectral response or the Redistribution Matrix File (RMF). On the other hand, we apply the Compton kinematics criteria on the 2pixel events to select the valid Compton events. The energies of the two pixels are then added up to calculate the total deposited energy. The deposited energy distributions for the 56 mono-energies are then convolved with a Gaussian to obtain the 2-pixel response. The CsI (or veto) spectral responses are generated for each quadrant in the same fashion using the same AstroSat mass model simulation data where we only consider events and associated energies deposited in the CsI detectors to estimate the deposited energy distribution. It is to be noted that we use a ls and charge diffusion based line profile model (Chattopadhyay et al. 2016) for mask weighted response below 150 keV, whereas for this work, we use a simple Gaussian model for simplicity.

3.2 Single pixel (1-pixel) spectroscopy Because the CZTI surrounding structures and the collimators become increasingly transparent above 100 keV, the spectral analysis of the GRBs starts from 100 keV and extends to 900 keV after incorporating the low-gain pixels. Detection efficiency of a 5 mm CZT drops below 10% above 1 MeV resulting in a low signal-to-noise ratio at those energies. The single pixel events are selected such that there are no other events reported in 100 ls time window on either side of the event. Energies deposited in all such events in the full burst region (interval obtained from Bayesian block analysis) are used to generate the spectrum with a 10 keV binning. The 1-pixel light curves and the selected time intervals are shown in Fig. 8 (red solid lines). The background spectrum is constituted by selecting at least 300 seconds of time window from the pre and post-burst regions. We quantify the systematics in the 1-pixel spectral data arising due to the uncertainties and inaccuracies in the AstroSat mass model and the CZTI detector via the analysis of the spectral data of GRBs detected at different incoming orientations. We use Band model (Band et al. 1993) to fit the spectra while keeping the power law indices (a and b) and peak energy (Epeak ) frozen at the values reported by either the Konus Wind or Fermi spectral analysis, and the normalisation of the Band model is left free.

–1.11 0.41 11.95 –1.31 7.63 –0.72 –2.24 –0.36 113.47 3.27

GRB151006A GRB160106A GRB160131A GRB160325A GRB160509A GRB160607A GRB160703A GRB160802A GRB160821A GRB160910A

30.21 47.26 49.63 46.41 22.06 14.96 28.31 17.84 159.87 15.04

Tstop (s)

b 2:19þ0:03 0:04 2:33þ0:13 0:18 – 1:82þ0:03 0:04 2:16  0:01 1:57  0:05 2:17þ0:15 0:17 – 2:13  0:02 2:37þ0:04 0:04

a 1:31þ0:02 0:02 0:66  0:05 1:02  0:03 0:85þ0:07 0:05 0:75  0:01 0:80þ0:18 0:12 0:94þ0:07 0:06 0:82  0:02 0:98  0:005 0:61þ0:02 0:02

1000þ1008 139 246þ26 24 407þ52 39 156þ20 21 279þ7 6 108þ35 32 195þ45 34 263þ10 9 860þ22 23 222þ9 8

Epeak =Ecut (keV) 6:37  0:005 6:07  0:005 6:2þ0:004 0:005 6:37  0:004 5:09  0:001 5:55  0:003 6:28  0:004 5:61  0:002 5:01  0:0008 5:34þ0:002 0:002

log10 (Flux) (erg/cm2 /s) 664.70/772 520.04/697 297.04/436 717.06/768 896.15/709 250.34/420 264.04/423 743.78/690 1178.55/699 778.32/680

Chi-square/ DoF

Band Band Cutoff power law Band Band Band Band Cutoff power law Band  Highecut Band

Model fit

GBM ? LAT ? BAT GBM BAT GBM ? BAT GBM ? LAT BAT BAT GBM GBM ? LAT GBM

Other instruments

Page 10 of 20

The errors are reported for 68% confidence interval. The references for the burst detection in different instruments are provided. GRB151006A—GBM (Roberts & Meegan 2015), LAT (Ohno et al. 2015), BAT (Cummings et al. 2015); GRB160131A—Fermi-GBM trigger#473813134; GRB160325A—GBM (Roberts 2016), LAT (Axelsson et al. 2016), BAT (Lien et al. 2016a); GRB160509A—GBM (Roberts et al. 2016), LLE (Kocevski & Longo 2016); GRB160607A—BAT (Lien et al. 2016b); GRB160703A—BAT (Lien et al. 2016c); GRB160802A—GBM (Bissaldi 2016); GRB160821A—GBM (Stanbro & Meegan 2016), LAT (McEnery et al. 2016); GRB160910A—GBM (Veres & Meegan 2016).

Tstart (s)

GRB name

Table 1. Spectral fit results of the analysis of the time integrated emission of the bursts in the sample using the CZTI, Fermi and Niel Gehrel Swift BAT data.

82 J. Astrophys. Astr. (2021)42:82

J. Astrophys. Astr. (2021)42:82

Page 11 of 20

82

Figure 9. Example of count spectra and residuals obtained for the 1-pixel (red) and 2-pixel (black) CZTI events for one back side GRB (GRB 160623A detected at angle  140 , left panel) and one front side GRB (GRB 160802A, viewing angle  53 , right panel). We fit the spectra with band model keeping the spectral parameters frozen at the values reported in literature to check for the consistency in spectral shape with Fermi and Konus-Wind.

For the different GRBs, listed in Table 1, detected at different orientations including most of the incoming angles around the spacecraft, once obtaining a best fit,3 we find unresolvable discrepancies between calculations and the data, i.e., systematic errors. Without knowing or assuming the origin of these features, we characterize the effect by adding systematic errors in an incremental fashion until we achieve a uniform residual with a reduced v2 \ 2. We, therefore, add 10–15% systematic to the 1-pixel spectroscopic data for all the GRBs to take care of the inaccuracies in the AstroSat mass model. For example, the spectral fits with the respective residuals obtained for the GRB 160623A (left) and GRB 160802A (right) that are detected on either side of the spacecraft are shown in the Fig. 9 where 1-pixel spectra in 100–900 keV obtained from the full burst region are shown in red crosses.

window with an additional Compton kinematics criteria of ratio of two deposited energies between 1 and 6 (also discussed in Chattopadhyay et al. (2014)). The energies from the two events recorded are added up to get the total energy and therefrom the spectrum with a bin size of 10 keV. The systematics involved in the Compton spectral data are assessed using the same methodology adopted for 1-pixel spectral data (Section 3.2). The spectral fits and the residuals obtained for the Compton spectra of GRB 160623A and GRB 160802A in 100–700 keV are shown in red data points in Fig. 9. Similar to the 1-pixel spectra, we find reasonable fit to the data and agreement with Fermi norm by adding a systematic of 10–15% uniformly throughout the energy range, 100–700 keV. Therefore, this systematic is added to the 2-pixel spectral data of all the GRBs.

3.4 CsI (or Veto) spectroscopy 3.3 Compton (2-pixel) spectroscopy The Compton spectroscopy is carried out in the energy range of 100–700 keV since above 700 keV there is no sufficient Compton scattering efficiency of CZTI detectors. The 2-pixel Compton events are identified from adjacent pixel events within 20 ls coincidence 3

The CZTI spectral data fit is considered to be reasonable when (a) the obtained residuals are roughly randomly distributed around zero, (b) the reduced chi-square v2 \2 and the (c) normaliza-

tion of the band function is found to be consistent with what is obtained from Konus-wind and Fermi spectral analysis.

There are four CsI(Tl) scintillator detectors (each 167 mm  167 mm in size and 2 cm in thickness) below CZTI quadrants to veto the high energy particle induced background events reported in both CZTI and CsI detectors (Bhalerao et al. 2017; Rao et al. 2016). The veto detectors were initially not meant for spectroscopy. However since the detectors possess sufficient detection efficiency in the sub-MeV region, we explored the possibility of using them for spectroscopy to enhance the overall spectroscopic sensitivity. The existing CZTI pipeline provides the veto spectrum at every second. We employ the available data to generate spectrum for each

82

Page 12 of 20

J. Astrophys. Astr. (2021)42:82

Figure 10. Top left: The count spectra (upper panel) and their respective residuals (lower panel) obtained for the three quadrants of the Veto detectors (black: quadrant A, red: quadrant B and green: quadrant C) for GRB 160623A (detected from the back side of CZTI). We see a systematic trend in the residuals possibly due to lower detection probability by the scintillators around 100 keV which improves at higher energies. Top right: Same as the left figure but after implementing an energy-dependent correction (1  eEnergy=E0 ), where E01 ¼ 0:0045 keV1 (see text for more details). Bottom: same as the top figure but for GRB 160802A (detected from the front side) after implementing the energy-dependent correction with a higher value of E01 ¼ 0:01 keV1 .

Veto detector in a similar way that is used for CZTI single pixel events. However, we do not use the poorly calibrated 4-th Veto quadrant for spectroscopy. It is to be noted that the Veto spectrum consists of all interactions in the CsI detectors and different from the Veto tagged events where the both CZTI and Veto are triggered due to simultaneous events recorded in those detectors. For all the GRBs detected from the rear side of the spacecraft, we find the observed spectra to be flatter than the response folded model. An example is shown in the top-left spectral plot of Fig. 10 for GRB 160623A which is detected at h of  140 . We find an identical systematic trend in all the back side GRBs. However, we do not attribute the systematic to the mass model as CZTI 1-pixel and 2-pixel spectral fits

for back side GRBs do not show such systematic trend in the residuals. On the other hand, the trend is significantly lower in Veto detectors for the front side GRBs. Therefore we believe that this systematic is originated in the CsI detectors but primarily for detections from back side. CsI detectors are scintillator detectors where the scintillation light is collected by the PMTs (2 PMTs for each of the 4 CsI detectors). At lower energies (  100 keV), the number of scintillation photons generated is lower than that at higher energies. Given the fact that there are only two PMTs to collect the scintillation photons, the detection probability of the GRB photons at lower energies is expected to be relatively low. We also note that the detectors were initially not meant for spectroscopy

J. Astrophys. Astr. (2021)42:82

Page 13 of 20

82

Figure 11. The count spectra (upper panel) and the respective residuals (lower panel) obtained for the broad band joint spectral analysis consisting of Fermi ? CZTI data (? BAT data in cases where it is available) for GRB 151006A, GRB 160106A, GRB 160509A, GRB 160325A, GRB 160802A, GRB 160821A and GRB 160910A are shown.

82

Page 14 of 20

J. Astrophys. Astr. (2021)42:82

Figure 12. The count spectra (upper panel) and the respective residuals (lower panel) obtained for the joint spectral analysis consisting of Niel Gehrels Swift BAT ? CZTI data using the spectral model band function for the bursts GRB 160131A (top left), GRB 160607A (top right) and GRB 160703A (bottom) are shown. Here we demonstrate that for bursts without Fermi detections, the usage of CZTI data extending until 900 keV along with BAT, enables us to constrain the Epeak of the GRB spectrum.

and therefore the number of readout photo-multiplier tubes and optical coupling between the crystal and the photo-multiplier tubes (PMTs) were not optimized to enhance the detection probability. The light collection efficiency might be significantly compromised for events happening in the back side of CsI because of the absence of optical reflecting coating on the back surface and a relatively higher level of cover shielding on the back side near the PMTs (light collecting area is relatively lower on the back side). To take care of this, we multiply the photon detection probability (represented by an empirical term, 1  eEnergy=E0 ) to the model (same as multiplying to the CsI detector response) to mimic for an energy dependent systematic where the value of E01 depends on the location of transient observed with

respect to CZTI. For the front side GRBs (example shown in the bottom panel of Fig. 10 for GRB 160802A) i.e., h \ 60 the value of E01 is found to be around 0.01 keV1 which gives 90% detection probability at  200 keV, whereas for the orthogonal GRBs, i.e., 90 \ h \ 110 , the value of E01 comes out to be around 0.008 keV1 . For the back side GRBs, value of E01 is found to be around 0.0045 keV1 signifying poor detection probability (90% detection probability at  600 keV). Since we get similar values of E01 for front, back and orthogonal GRBs, we plan to incorporate the exponential feature observed in the Veto detectors in the response itself. We also include an additional 5% systematic in the data in case of back side GRBs.

J. Astrophys. Astr. (2021)42:82

Page 15 of 20

82

Figure 13. For GRB 151006A (top panel) and GRB 160325A (bottom panel), we demonstrate that the analysis of the BAT data alone (top left and bottom left) does not allow us to ascertain the spectral peak and b of the band function fit to the data. However, when using CZTI data along with BAT (middle plots of top and below panel), we find that we can constrain the Epeak of the spectrum which is reasonably consistent with that determined in solo Fermi GBM analysis (top right and bottom right). The fit parameter values are given in Table 2.

4. Results I: Broadband joint spectroscopy of GRBs With a fair assessment of the systematics present in the CZTI and Veto spectral data, we now conduct the broadband joint spectral analyses involving the spectral data from Fermi, Niel Gehrels Swift BAT, along with CZTI data including the single, Compton and Veto for the time integrated emission of different GRBs. We analyse the time integrated spectrum of 10 GRBs that were detected by CZTI in the first year of its operation (2015–2016). The time interval of the integrated emission is selected using Bayesian block binning technique and described in Section 2.3. The Fermi spectral data includes two bright sodium iodide (NaI) detectors with source angle less than (\60 ) and the brightest bismuth germanate (BGO) detector (Gruber et al. 2014). In case of GRB151006A, GRB160509A and GRB160821A, the low energy Large Area Telescope (LLE) data are also used. The Fermi spectral files are extracted using Fermi Burst Analysis GUI v. 02-03-00p33 (gtburst4).

The Swift BAT spectral files are prepared by the standard methodology.5 The spectral analyses are performed using the X-Ray Spectral Fitting Package (XSPEC, Arnaud 1996) version: 12.11.0 and have followed chi-square statistics. Both BAT and CZTI spectral files are compatible with Gaussian statistics, however, the GBM and LLE files are consistent with Pgstat wherein the background and signal are assumed to be Gaussian and Poissonian respectively. Therefore, using Heasoft Ftool GRPPHA,6 we rebinned both GBM and LLE spectral files such that each energy channel contains a minimum of 20 photons. The spectral fit results and the respective residuals obtained for the best fit empirical functions like Band function (Band) and cutoff power law (CPL) are reported in Table 1 and shown in Figures 11, 12 and 13 respectively. We find that residual obtained for CZTI spectral data are consistent with those obtained for Fermi. The residuals are found within 3r for CZTI data. The small energy window of Swift BAT (15–150 keV) generally does not allow us to constrain the Epeak of the spectrum in cases where there is only BAT detection. In 5

4

https://fermi.gsfc.nasa.gov/ssc/data/analysis/scitools/gtburst.html

https://swift.gsfc.nasa.gov/analysis/threads/bat_threads.html. https://heasarc.gsfc.nasa.gov/ftools/caldb/help/grppha.txt

6

82

Page 16 of 20

J. Astrophys. Astr. (2021)42:82

Figure 14. The effective area correction factors obtained for the different CZTI datasets: single (black square), Compton (blue diamond), Veto Q1 (yellow circle), Veto Q2 (purple triangle) and Veto Q2 (red star) with respect to the brightest NaI detector of Fermi for different GRBs in the sample except for GRB 160131A, GRB 160607A and GRB 160703A where the values are obtained with respect to Swift BAT detector are shown.

case of GRB 160607A, GRB 160703A and GRB 160131A, where Fermi detections were not available, we conducted the spectral analysis using Swift BAT along with CZTI data. We demonstrate that the usage of CZTI data extending until 900 keV allows us to well constrain the Epeak of the spectrum. The effective area correction

factor obtained between BAT and CZTI are shown in Fig. 14, where constant for BAT is frozen to unity. The energy flux estimated in the range of 10–1000 keV for the bursts are reported in the Table 1. During the joint spectral analysis using different detectors, we have tied the spectral parameters of all the detectors including the normalisation of the spectral model. The difference in count rates in different detectors are taken care of by including the effective area correction factor along with the spectral model that is used to analyse the data. To estimate the effective area correction factor between the different detectors, we multiply an energy independent constant factor to the spectral model during the fitting process. The effective area correction factor obtained between Fermi and the different datasets of CZTI except in GRB 160131A, GRB 160607A and GRB 160703A where the values are obtained with respect to the Swift BAT are shown in Fig. 14. On average, the normalization estimates of the empirical function fits done to the single, Compton, Veto 1, 2, and 3 data are found to vary around 20%, 55%, 55%, 40% and 40% of the normalization estimate of the brightest Fermi NaI detector respectively. While with respect to the BAT detector, the normalization estimates for the single, Compton, Veto 1, 2, and 3 events, vary around 40%, 140%, 85%, 20% and 70% respectively. In certain GRBs, we observe low normalizations for Compton and Veto data which results in an effective area correction factor [2. The cause of such cases are being studied. For GRB 151006A and GRB 160325A, both Fermi and BAT observations are available. So, in these

Table 2. The band model fit comparison between BAT ? CZTI and Fermi alone analysis of the bursts GRB 151006A and GRB 160325A. GRB name GRB151006A

GRB160325A

Band parameters

BAT

BAT ? CZTI

Fermi

a b Epeak ðkeVÞ Norm v2red a b Epeak ðkeVÞ Norm v2red

1:25þ0:07 0:14 9:37þ19 0:0 288þ257 117 0:007þ0:001 0:0009 0.68 0:87þ0:13 0:12 10þ1e15 0:0 137þ54 27 0:02þ0:003 0:002 0.55

1:23þ0:15 0:12 1:79þ0:18 0:17 262þ44 24 0:007þ0:002 0:001 0.69 0:82þ0:08 0:16 1:74þ0:06 0:09 124þ44 24 0:01þ0:002 0:003 0.91

1:08þ0:12 0:13 1:89þ0:11 0:20 350þ400 126 0:008þ0:002 0:001 1.02 0:77þ0:10 0:09 2:63þ0:42 2:36 214þ53 43 0:01þ0:002 0:001 0.81

The errors are reported for 90% confidence interval.

J. Astrophys. Astr. (2021)42:82

Figure 15. Above the CZTI flux estimates (Y) done for the spectral fits done to CZTI data alone versus the energy flux estimates done for Fermi (X) alone spectral fits are shown. The red (blue) shaded region marks the 1r (2r) scatter of the distribution of points around the Y ¼ X line shown in dotted red line.

Figure 16. Detection of GRB 160821A in the veto-tagged events. Different colours stand for different CZTI quadrants. The time-axis is plotted from AstroSat time 209507728 seconds (marked as zero) onward.

GRBs, we conduct a joint spectral analysis of BAT and CZTI data and then compare the spectral fit results with that obtained using Fermi GBM data alone. We are able to ascertain the a, Epeak and normalization values which are reasonably consistent with Fermi GBM results, within 90% error limits (Table 2 and Fig. 13). This further endorse the capability of CZTI as a sub-MeV spectrometer along with BAT to determine the GRB spectrum.

Page 17 of 20

82

We note here that being opaque below 100 keV, CZTI spectrum alone cannot measure the GRB spectral parameters fully. On the other hand, if we assume canonical values for the power law indices (a ¼ 1 and b ¼ 2:5) of the Band function, we can constrain the Epeak and normalisation of the spectrum. In certain cases, the Epeak estimates are found to lie close to the edge or outside the energy window of CZTI (e.g GRB160509A and GRB160821A). In Fig. 15, the energy fluxes estimated in the energy range 100 keV– 1000 keV for spectral fits of CZTI data alone (where the power law indices are frozen to the canonical values), are plotted against the respective energy flux estimated from the Fermi data only spectral fits (where all the fit parameters are left free) of the different bursts. We find the CZTI flux estimates are consistent within 2r scatter7 around the line denoting CZTI energy flux is equivalent to Fermi flux.

5. Summary and future plan CZT-Imager on board AstroSat has been a prolific GRB monitor with around detection of nearly 83 GRBs per year. In this article, we explored the spectroscopic sensitivity of CZTI in the sub-MeV region by attempting spectroscopic analysis for some of the bright GRBs detected in the first year (October 2015–Spetember 2016) of AstroSat operation. The improvement in the spectroscopic sensitivity has been possible because of (1) inclusion of the low-gain CZTI pixels after a thorough calibration which consists of around 20% of the CZTI detection area, and (2) identification and removal of 2-pixel noisy events. Both the methods improve the S/N of the bursts significantly and in particular the low-gain pixels enable the spectroscopy all the way up to 900 keV (1-pixel Compton spectroscopy: 100–700 keV, 1-pixel spectroscopy: 100–900 keV). We also utilize the CsI (or Veto) detectors for spectroscopy in 100–500 keV to enhance the overall sensitivity. In Section 4, we performed joint Fermi and AstroSat (and BAT wherever available) spectral analysis for 10 out of the eleven first year GRBs (except GRB 160623A where a concurrent observation with Fermi was not available) in the full burst region. We are able to obtain spectral fit parameter values that are in close agreement with those obtained in solo Fermi analysis. This provides 7

The scatter is the standard deviation of the Gaussian fit to the distribution of the displacement of the CZTI measured flux from the Fermi flux and is found to be r ¼ 0:21.

82

Page 18 of 20

an independent validation of the AstroSat mass model, thereby boosting the confidence in the spectral analysis of the CZTI GRBs. Spectral validation of the mass model and availability of CZTI spectra up to 900 keV also allows to explore spectral study of the GRBs detected only by Swift-BAT and CZTI but not by Fermi. This aspect has been particularly demonstrated in the case of GRB151006A and GRB160325A where we find reasonably consistent spectral fit values for BAT ? CZTI in comparison to solo Fermi data analysis of these bursts. Thus, the satisfactory spectral fits obtained in 15–900 keV (15–150 keV from BAT, 100–900 keV from CZTI) for GRB 160607A, GRB 160131A and GRB 160703A demonstrates the importance of CZTI sub-MeV spectroscopic capability particularly to characterize the GRBs that are not detected by Fermi. We also identify possible systematics involved in the mass model and attempt to quantify them (\15%) in the front and rear sides of the spacecraft. This paper primarily describes the new methods of sub-MeV spectroscopy with CZTI. We are continuing to refine these methods further, and will extensively test them against a much larger sample of bright GRBs detected by CZTI in the last five years. We also plan to explore the feasibility of using the CZT detectors and the CsI detectors in Compton camera configuration to enhance the spectroscopic sensitivity of the instrument. From a preliminary analysis, we could successfully detect the GRBs in the veto-tagged events (Compton scattered photons from CZT detectors which are absorbed by the CsI detectors) after applying Compton scattering kinematic conditions (see Fig. 16). We plan

J. Astrophys. Astr. (2021)42:82

to use the AstroSat mass model to generate response matrix for the veto-tagged events.

Acknowledgements This publication uses data from the AstroSat mission of the Indian Space Research Organization (ISRO), archived at the Indian Space Science Data Centre (ISSDC). CZT-Imager is built by a consortium of Institutes across India including Tata Institute of Fundamental Research, Mumbai, Vikram Sarabhai Space Centre, Thiruvananthapuram, ISRO Satellite Centre, Bengaluru, Inter University Centre for Astronomy and Astrophysics, Pune, Physical Research Laboratory, Ahmedabad, Space Application Centre, Ahmedabad: contributions from the vast technical team from all these institutes are gratefully acknowledged. We acknowledge the use of Vikram-100 HPC at the Physical Research Laboratory (PRL), Ahmedabad and Pegasus HPC at the Inter University Centre for Astronomy and Astrophysics (IUCAA), Pune. This research has also made use of data obtained through the High Energy Astrophysics Science Archive Research Center Online Service, provided by the NASA/Goddard Space Flight Center.

Appendix A Plots for the Bayesian block analysis conducted on single event data of the GRBs are shown in Fig. A1.

Figure A1. The Bayesian block binning of the single event CZTI light curve of the bursts are shown above in black solid lines. The time interval of the integrated emission of each burst is marked by the vertical dotted lines on the respective plots. The red dashed horizontal line marks the background level. The basic light curve is plotted in the background in pink colour. We note that here the 0 marks the start of the T90 region of the burst.

J. Astrophys. Astr. (2021)42:82

Page 19 of 20

82

Figure A1. Continued.

References Arnaud K. 1996, in Astronomical Data Analysis Software and Systems V, vol. 101, p. 17 Axelsson M., Bissaldi E., Desiante R., Longo F. 2016, GRB Coord. Netw., 19227 Band D., Matteson J., Ford L. et al. 1993, ApJ, 413, 281 Bhalerao V., Bhattacharya D., Vibhute A. et al. 2017, J. Astrophys. Astron., 38, 31 Bissaldi E. 2016, GRB Coord. Netw. 19754, 1 Burgess J. M. 2014, MNRAS, 445, 2589 Chand V., Chattopadhyay T., Oganesyan G. et al. 2019, ApJ, 874, 70 Chand V., Chattopadhyay T., Iyyani S. et al. 2018, ApJ, 862, 154 Chattopadhyay T., Vadawale S. V., Rao A. R. et al. 2016, in Proc. SPIE, vol. 9905, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, 99054D Chattopadhyay T., Vadawale S. V., Rao A. R., Sreekumar S., Bhattacharya D. 2014, Exp. Astron., 37, 555 Chattopadhyay T., Vadawale S. V., Aarthy E. et al. 2019, ApJ, 884, 123

Cummings J. R., Barthelmy S. D., Gehrels N. et al. 2015, GRB Coord. Netw., 18410, 1 Gruber D., Goldstein A., von Ahlefeld V. W. et al. 2014, Astrophys. J. Suppl. Ser., 211, 12 Kocevski D., Longo F. 2016, in Eighth Huntsville GammaRay Burst Symposium, vol. 1962, p. 4092 Lien A. Y., Barthelmy S. D., Cummings J. R. et al. 2016a, GRB Coord. Netw., 19234, 1 Lien A. Y., Barthelmy S. D., Cummings J. R. et al. 2016b, GRB Coord. Netw., 19506 Lien A. Y., Barthelmy S. D., Cenko S. B. et al. 2016c, GRB Coord. Netw., 19648 Lyne A., Pritchard R., Roberts M. E. 1999, https://www.jb. man.ac.uk/pulsar/crab.html McEnery J., Racusin J., Longo F. 2016, GRB Coord. Netw., 19831, 1 Odaka H., Asai M., Hagino K. et al. 2018, Nuclear Instrum. Methods Phys. Res. A, 891, 92 Ohno M., Bissaldi E., Vianello G., Kocevski D., Longo F. 2015, GRB Coord. Netw., 18406, 1 Paul B. 2013, Int. J. Mod. Phys. D, 22, 41009 Rao A. R., Chand V., Hingar M. K. et al. 2016, ApJ, 833, 86

82

Page 20 of 20

Roberts O. J. 2016, GRB Coord. Netw., 19224 Roberts O. J., Fitzpatrick G., Veres P. 2016, GRB Coord. Netw., 19411, 1 Roberts O. J., Meegan C. 2015, GRB Coord. Netw., 18404, 1 Scargle J. D. 1998, Astrophys. J., 504, 405 Scargle J. D., Norris J. P., Jackson B., Chiang J. 2013, ApJ, 764, 167 Sharma V., Iyyani S., Bhattacharya D. et al. 2019, ApJ Lett., 882, L10

J. Astrophys. Astr. (2021)42:82 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 9144, Society of PhotoOptical Instrumentation Engineers (SPIE) Conference Series Stanbro M., Meegan C. 2016, GRB Coord. Netw., 19835, 1 Vadawale S. V., Chattopadhyay T., Mithun N. P. S. et al. 2018, Nat. Astron., 2, 50 Veres P., Meegan C. 2016, GRB Coord. Netw., 19901, 1

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:64 https://doi.org/10.1007/s12036-021-09743-1

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

Using collimated CZTI as all-sky X-ray detector based on Earth occultation technique AKSHAT SINGHAL1,* , RAHUL SRINIVASAN1,2,4,6,7,8 , VARUN BHALERAO1 ,

DIPANKAR BHATTACHARYA3 , A. R. RAO4

and SANTOSH VADAWALE5

1

Indian Institute of Technology Bombay, Mumbai 400 076, India. Universite´ Coˆte d’Azur, Nice, France. 3 The Inter-University Centre for Astronomy and Astrophysics, Pune 411 007, India. 4 Tata Institute of Fundamental Research, Mumbai 400 005, India. 5 Physical Research Laboratory, Ahmedabad 380 009, India. 6 CNRS, Nice, France. 7 Laboratoire Lagrange, Nice, France. 8 Laboratoire Arte´mis, Nice, France. *Corresponding Author. E-mail: [email protected] 2

MS received 7 November 2020; accepted 8 March 2021 Abstract. All-sky monitors can measure the fluxes of astrophysical sources by measuring the changes in observed counts as the source is occulted by the Earth. Such measurements have typically been carried out by all-sky monitors like CGRO-BATSE and Fermi-GBM. We demonstrate for the first time the application of this technique to measure fluxes of sources using a collimated instrument: the Cadmium Zinc Telluride detector on AstroSat. Reliable flux measurements are obtained for the Crab nebula and pulsar, and for Cyg X–1 by carefully selecting the best occultation data sets. We demonstrate that CZTI can obtain such measurements for hard sources with intensities J1 Crab. Keywords. X-ray telescopes (1825)—occultation (1148)—astronomical techniques (1684).

1. Introduction Modern high energy instruments use a variety of techniques including collimators, coded aperture masks, and grazing incidence/multi-layer optics to characterise astrophysical sources and measure their position, flux, and spectra. Such instruments face an inherent trade-off between sky coverage and localisation accuracy. All-sky monitors like CGRO-BATSE and Fermi-GBM can characterise transient sources based on relative flux intensities on various detectors. But their methods rely on the fact that the effect of all other sources in the sky can be corrected for by using pre- and post-transient data. Such open-detector all-

This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

sky monitors have very limited abilities for studying persistent astrophysical sources. One such all-sky monitor is the Cadmium Zinc Telluride Imager (Bhalerao et al. 2017, CZTI) on board AstroSat (Singh et al. 2014). CZTI is a coded aperture mask instrument sensitive in the 20–200 keV band. At energies above  100 keV, the instrument and satellite structures become transparent to radiation, giving CZTI sensitivity to sources well outside the primary field of view. CZTI data have been used to detect over 400 Gamma Ray Bursts (GRBs) in the 5 years since launch (Sharma et al. 2021), and even study the Crab pulsar at angles from 5 to 70 from the principal axis (Anusree et al. 2021). However, CZTI faces the same limitations as other all-sky monitors for studying persistent astrophysical sources. A workaround for studying such sources which lack rapid intrinsic variability is to leverage any extrinsic

64

Page 2 of 9

variations in their flux. A popular method for this is to observe sources as they are occulted by the Earth or the Moon. The transition of the source behind the Earth’s limb (ingress) causes a rapid step-like decrease in the overall counts measured by the detector, and likewise the transition from behind the limb to visibility (egress) causes a rapid increase in the counts. Lunar occultation techniques have even been used in ground-based radio telescopes with great success (see for instance, Kapahi et al. 1973). Earth occultation studies have been successfully leveraged for studying high-energy sources by space-based detectors like CGRO-BATSE (Harmon et al. 2002) and Fermi-GBM (Wilson-Hodge et al. 2012). In this work, we demonstrate the application of the Earth Occultation Technique (EOT) for measuring source fluxes using data from the CZTI.

2. Overview of the Earth occultation technique We begin with a brief overview of the formulations of EOT. For more details, we refer the readers to Harmon et al. (2002) and Zhang (2013). For a satellite in say, a 650-km Low Earth Orbit (LEO), the Earth subtends an angle of  130 on the sky. The inclination and precession of the satellite orbit determine which astrophysical sources can get occulted by the Earth: for instance, for an equatorial orbit, objects near the celestial poles are never obscured by the Earth. For the 6 orbital inclination of AstroSat, sources with declination up to approximately 70 can get occulted at some point, while objects with declination in the range of about 60 will get occulted in each orbit. In each orbit around the Earth, the satellite can observe one ingress and one egress of these sources (Fig. 1). For very bright objects, occultations can be used to measure flux, spectra, and to measure the location of the source in the direction perpendicular to the Earth’s limb. In this work, we assume that the source position and approximate spectral shape are known, and attempt to measure the source flux (spectral normalisation).

2.1 Time and duration of occultation Using the known source coordinates and orbital parameters of the satellite, we can calculate the time of occultation (t0 ) of each ingress and egress of the source. At high altitudes (J100 km) the Earth’s

J. Astrophys. Astr.

(2021) 42:64

atmosphere is largely transparent to X-rays, and becomes increasingly opaque closer to the surface. This has two consequences for occultation studies. First, the nominal ingress and egress do not occur at the geometric limb of Earth, but some distance further out. Second, the decrease/increase in source flux are not sharp step functions, but more gradual transitions. To incorporate these effects in our work, we define an effective height hocc at which the source flux decreases to 50% of its unobstructed value. For the 20–200 keV energy range of CZTI, we adopt hocc = 70 km. It is to be noted that shape of the earth and energy dependent hocc can affect the calculations. However we used a simpler approach of spherical earth with constant hocc for its simplicity. In comparison to alternate approach of adding more degrees of freedom or removing the data near the transition resulted in more errors. To account for the gradual transition, we define a characteristic timescale s over which the source flux in an ingress drops from 90% to 10% of its unobstructed value. The orbital inclination of AstroSat and the declination of the source also play a role in determining the transition timescale. Due to projection effects, sources further away from the orbital plane of the satellite appears to pass through oblique layers of the atmosphere, extending the duration of the transition. This is accounted for by defining an ‘‘orbital latitude’’ b (Harmon et al. 2002), the projected angle between orbital plane and the line joining the source to the centre of the Earth (Fig. 2). The transition timescale is then given by s ¼ s0 sec b, where s0 is the minimum transition timescale, observed for sources in the orbital plane.

2.2 Background and observed flux The background for CZTI comprises of various components including cosmic X-ray background, Earth albedo, charged particle induced background, and electronic noise. In addition to these, for the purpose of our EOT analysis, photons from the onaxis source also contribute to background. The background level is not constant: it varies slowly through the orbit (dominated by proximity to the South Atlantic Anomaly), and also shows variations across orbits (dominated by orbital precession and solar activity). It is known that these background variations for CZTI occur on timescales of several hundreds to thousands of seconds, and is wellapproximated by a quadratic function (Sharma et al. 2021;

J. Astrophys. Astr.

(2021) 42:64

Page 3 of 9

64

Anumarlapudi et al. 2020). Hence we too model background as a quadratic function, bg ðtÞ ¼ aðt  t0 Þ2 þbðt  t0 Þ þ c. Various models have been used for the smooth source intensity transition. We adopt a simplistic error function, such that the transition light curve TE for an egress is given by    1 1 1:812ðt  t0 Þ ð1Þ TE ðt  t0 Þ ¼ h þ erf 2 2 s h ¼ pffiffiffi 2 p

Z

1:812ðtt0 Þ s

2

ex dx;

ð2Þ

1

where erf is the error function, s is the transition timescale discussed in Section 2.1, t0 is the nominal transition time when the source cross the altitude hocc , and h is the average count rate from the source being occulted. We note that h is negative for source ingress. The net model M for the occultation is thus given by Mðt  t0 Þ ¼ bg ðt  t0 Þ þ TE ðt  t0 Þ: Figure 1. A schematic of a source egress and the corresponding observed light curve. The top part shows the Earth (dark blue circle) and its atmosphere (slight blue annulus). AstroSat (green rectangle) is orbiting the Earth in a counter-clockwise sense, as shown by the violet curve. Vertical yellow arrows denote the parallel rays of light coming from a distant astrophysical source. The bottom part shows the corresponding light curve (solid black line) observed by CZTI, which we model as Mðt  t0 Þ (Eq. (3)). The dashed blue line marks the background-only counts curve (bg ) expected if there was no source egress. Vertical dashed lines mark various times of interest: the dashed black line nominal occultation time t0 when the source is at a height hocc above the Earth’s limb, dashed orange lines show the window of duration s when the source flux rises from 10% to 90% level, and dashed green lines show the complete window of interest used in our fitting procedures.

Figure 2. Schematic diagram for relative orbital declination. This diagram will be replaced by our own. This is from Fermi.

ð3Þ

Lastly, we note that we perform fits to observed data in the range t0  thw , where thw stands for the halfwindow time.

3. Methodology In this section, we discuss the specific details for EOT analysis of CZTI data. The overall process flow chart is shown in Fig. 3.

3.1 Preparing the data CZTI consists of four identical independent quadrants (labeled A, B, C, D), each containing sixteen detectors with 256 pixels each (Bhalerao et al., 2017). All observed data are downloaded in event mode, noting the time, energy, and position of interaction for every incident photon. We reduce the Level 1 data using the default CZTI pipeline,1 with one tweak. By default, the pipeline discards data in intervals when the on-axis source is close to or occulted by the Earth. Since we are not interested in the analysis of on-axis sources, we disable this flag in the pipeline. We use cztbindata to bin the data at t bin seconds, creating livetime-corrected light curves complete with count rate uncertainties for each bin. We limit ourselves to using only ‘‘grade 0’’ pixels. 1

CZTI pipeline: http://astrosat-ssc.iucaa.in/?q=cztiData.

64

Page 4 of 9

J. Astrophys. Astr.

(2021) 42:64

We calculate the source occultation time t0 based on the source coordinates and the location of the satellite with respect to the Earth. The latter is obtained from the CZTI Level 1 MKF file. Our window of interest is t0  thw as discussed above. Like all high energy instruments, CZTI stops collecting data when the satellite is passing through the South Atlantic Anomaly (SAA). We identify the start and end of the SAA as points in the light curve where the count rate drops to and rises from zero respectively. Data obtained very close to SAA are often very noisy. Hence, we add another 10 s ‘‘buffer’’ time on either side of the SAA. If any part of our window of interest overlaps with the SAA passage or the buffer time, we discard the overlapping part. After this cut, we require that at least tmin ¼ 0:417thw seconds of data must be present on either side of t0 , else the entire event is rejected.2

quadratic parameters a, b, c for the background. Of these, we keep s and t0 fixed, and solve for the remaining four using the least-squares solver curve_fit implemented in the scipy.optimize python package. A sample fit for the ingress of Crab is shown in Fig. 4. The blue lines denote the light curve, while the solid orange line is the best-fit result. The vertical black dashed line marks t0 , the expected time of the occultation. We find that these fits can be skewed by presence of outliers in the light curve, and outlier rejection is crucial for getting good fits. We implement this process iteratively: first, the model Mðt  t0 Þ is fit to the light curve in the t0  thw range. All data points are weighted by the uncertainties calculated by the CZTI pipeline. We select a threshold k such that kr outliers in this fit are rejected. The uncertainties on individual data points are ignored in this step. The fitting procedure is repeated, and outliers from the new fit are rejected again. We repeat this procedure 10 times or until there are no more kr outliers. Experimenting with our data, we found that the best results are obtained when k ¼ 2, corresponding to on average of 15% of data rejected. In addition to the uncertainty in fit parameters returned by the solver, we use another method to assess the significance of our fit. We exclude the central t0  s region of our window of interest, and evaluate the residuals obtained by subtracting our best-fit model from the data. We calculate the standard deviation (rcounts ) of these residuals for the pre-t0 and post-t0 regions. Each of these regions consists of Nbin ¼ ðthw  sÞ=tbin independent bins, so the uncertainty in estimating the background is approximately pffiffiffiffiffiffiffiffi rcounts = Nbin . Since h roughly corresponds to difference between the two background levels, the typical uncertainty of a detection of an occultation will be given by pffiffiffi rcounts 2 ð4Þ hmin  pffiffiffiffiffiffiffiffi : Nbin

3.2 Signal modelling

3.3 Selecting parameters

We now fit Eq. (3) to the light curves in the t0  thw time range (Fig. 4). The source model M has six parameters: the source counts h, the occultation timescale s, the time of occultation t0 , and the three

The algorithm described above still leaves several free parameters to be selected: the energy range, tbin , and thw . In most of our work, we use the full CZTI data range from 20–200 keV, with the exception of specific test discussed in Section 4.3. We explored various values for tbin from 1 s to 10 s. We observed that as tbin increases, the flux posterior

Discard event

Generate light curve at tbin binning

CZTI Level 1 Data

Get occultation time t0

CZTI Level 1 MKF

Sample data in t0 ± thw window

No Data window > tmin

Apply SAA veto

Yes Perform fit, measure h

Convert to flux using mass model

Figure 3. Flowchart for the Earth occultation pipeline for observing X-ray flux using AstroSat-CZTI data.

2

The fraction 0.417 = 5/12 comes from backward compatibility during code development.

J. Astrophys. Astr.

(2021) 42:64

Page 5 of 9

64

The flux as a function of energy is given by FðEÞ ¼ E  NðEÞ. Thus, the total expected counts in a quadrant are given by Z Emax AðEÞNðEÞdE: ð5Þ h¼ Emin

We define the net effective area for each quadrant by weighting the area at each energy with the source spectrum: R Emax AðEÞFðEÞdE : ð6Þ Aeff ¼ Emin R Emax Emin FðEÞdE Figure 4. Fitting of Eq. (3) to measure the count rate from Crab in CZTI Quadrant A.

broadens leading to progressively imprecise measurements. One possible explanation is that since the transition period is Oð10 s), comparable values of tbin smear out this transition period, and the fitting function — which is simply evaluated at bin centres — cannot accurately recover the change in count rates. As a result, we fix tbin = 1 s for our analysis. We tested various values of thw from 100 s to 1000 s. We observe that if thw is small, the estimation of background suffers, increasing hmin (Eq. (4)), equivalently increasing the uncertainty in h. On the other hand, if the window is too wide, the background can no longer be accurately modelled by a quadratic. We adopt thw = 500 s for our analysis. Based on these values, we find the typical value of hmin is around 0.9 counts for 20–200 keV energy range.

3.4 Effective area and source flux The sensitivity of CZTI to off-axis sources is a strong function of direction, owing to varying degrees of obscuration by various satellite elements. This has been calculated in detail using the CZTI Mass Model (Mate et al. 2021). Here, we use an older version of the mass model with minor differences. For each direction in the sky, the mass model gives us the A(E), the effective area for each quadrant as a function of energy. AstroSat has a geometric area of 976 cm2 and the all-sky median effective area at 180 keV is 190 cm2 (Bhalerao et al. 2017). We ensure that we add up the areas from ‘‘grade 0’’ pixels only. The expected count rate depends on the source spectrum, and its direction in satellite coordinates. Let the source photon flux as a function of energy be N(E).

Despite the physical proximity of the four quadrants, the effective area in the same direction can be quite different, especially when the collimators of one quadrant casts a shadow on another. We show the all-sky net effective area for all four quadrants for a Crab-like spectrum in Fig. 5. It is clear that effective area varies by more than an order of magnitude over the entire sky. In practice, we assume that the source spectrum is a power law of the form NðEÞ ¼ N0 EC , and use Eq. (5) to calculate N0 from the measured values of h. We then calculate the total source flux as R Emax F ¼ Emin AðEÞN0 ECþ1 dE.

3.5 Outlier rejection and vetos Low quality data with high noise could lead to false or inaccurate measurements. Hence, we define various quality cuts on the data to discard potentially problematic parts of the data. Our pipeline employs five vetos: the first (b-veto) is applied on all quadrants jointly, while others are applied to individual quadrant data. If only a single quadrant survives the vetos, we discard the entire event. (1) Events with b [ 60 have high s values, corresponding to very long transitions which are difficult to effectively decouple from the background. Hence, such events are not processed. (2) Based on our estimates of the minimum detectable transit amplitude (hmin ), we conclude that very small values of h are indicative of a failed fit. These are typically caused by noisy backgrounds or low effective area in the direction of the source. Guided by our calculation that hmin  0:9, we veto out any occultation where jhj\1. (3) Some data sets are particularly noisy, and the bestfit value of h has a wrong sign: showing an increase in counts at ingress or a decrease in counts

64

Page 6 of 9

J. Astrophys. Astr.

(2021) 42:64

Figure 5. Net effective area for Crab spectrum for each quadrant (A, B, D, C, respectively) as a function of relative h; / in satellite frame. The top of each plot corresponds to the pointing axis of the satellite. The low effective area regions correspond to lines of sight going through the entire satellite.

at egress. These are discarded by comparing the obtained sign of h with the expected sign. (4) The effective area of the detector to the source has a direct correlation with observed source photon count rate from the source. Only allowing events with good effective area ensures we can always expect a reasonable source photon count rate in our data and hence, a more confident fit. We accept only those events where the 4-quadrant effective area is [10 cm2 . (5) Since the four quadrants are independent, so are their noise properties. If only one quadrant is noisy, data from the others may still prove to be useful. We cannot directly compare the h values of different quadrants owing to differences in their effective areas. We convert the h to fluxes, then calculate the mean and standard deviation of the flux values for all four quadrants. We discard any quadrants where the flux is outside a 1-sigma interval centred on the mean.

4. Results We apply this method to two bright astrophysical sources: First, we test the methods on the Crab to measure its flux. Second, we select the variable source Cyg X–1 and measure its flux as a function of time.

4.1 Crab pulsar and nebula We use a simplified power-law spectral model for the Crab pulsar and nebula, FðEÞ ¼ F0 ECþ1 with photon power law index C ¼ 2:1 and normalisation F0 ¼ 9:7 keV cm2 s1 keV1 ¼ 1:5  108 erg cm2 s1 keV1 at 1 keV (Madsen et al. 2017). With this model, the Crab flux in the 20–200 keV band is 2:4  108 erg cm2 s1 . We note that this simplistic treatment does not account for the hard X-ray spectral breaks present in the Crab spectrum (see for instance, Kouzu et al. 2013), but suffices for our purposes. We run our codes on on readily available data across five years of CZTI observation, which span about ten thousand orbits. Owing to the low declination of Crab (d ¼ 22 ), this results in over 54,000 ingress/egress events in individual quadrants (about 25% of total possible events). Only 5559 events pass our veto filters. By treating all quadrants independently, we calculate the 20–200 keV Crab flux to be ð2:40  0:02Þ  108 erg cm2 s1 . If we instead add another intermediate step by combining h values and effective areas of all quadrants that passed the veto for any given event, and then combine results across all events, we measure the flux as ð2:65  0:02Þ 108 erg cm2 s1 , where 0:02  108 erg cm2 s1 in both measurements is the weighted error rx, using individual error ri computed as

J. Astrophys. Astr.

(2021) 42:64

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 rx ¼ Pn 2 : i¼1 ri

Page 7 of 9

ð7Þ

4.2 Higher effective area The amplitude of the measured signal (h) is directly related to the effective area in the direction of the source. Through the course of routine CZTI observations, Crab occultations have occurred at a wide variety of angles in the CZTI reference frame, corresponding to a wide range of effective areas. Figure 6 shows a histogram single-quadrant effective areas calculated from all observable occultations. We then start raising the effective area threshold, by selecting only events with higher effective areas. If we select the top 10% events with the highest effective area (corresponding to Aeff J41 cm2 ), we measure the Crab flux to be ð2:25  0:02Þ  108 erg cm2 s1 . Values for other percentile selections along with weighted error rx and median error of ri are given in Table 1.

4.3 Higher energy ranges While these results are consistent with the expected flux, they have large uncertainties. CZTI data have relatively higher electronic noise contribution below 50 keV. The energy range from 50–70 keV is dominated by fluorescence emission from the Tantalum present in the collimators, hence it is difficult to discern any source spectral

64

properties in this range. In order to improve our measurements, we repeat the analysis by restricting ourselves to the 70–200 keV energy range. The expected Crab flux in this range is 1:0  108 erg cm2 s1 . We re-calculate the effective area for each event in this energy range, and select the top 10% events with the highest effective area. This corresponds to a 70–200 keV effective area cut of 33 cm2 , and gives us 622 ingress or egress events. The measured flux from this sample is ð1:38  0:02Þ  108 erg cm2 s1 and median individual error of 1:17  108 erg cm2 s1 .

4.4 Cyg X–1 We then use our EOT pipeline to measure the flux of Cyg X–1. To limit ourselves to data with high effective areas, we select two on-axis observations of Cyg X–1 conducted in June 2019, and analyse the 70 ingress or egress events that occurred in this duration. We note that despite the source being on-axis, we do not utilise any coded aperture mask data processing for measuring fluxes. Cyg X–1 is highly variable, and expected to be visible to us only in the bright hard state. Following Lubin´ski et al. (2020), we model the ‘‘pure hard’’ state Cyg X–1 spectrum as a power-law with a photon index C ¼ 1:7. For comparison of our measurements, we obtain Cyg X–1 data from the Swift/BAT Hard X-ray Transient Monitor3 (Krimm et al. 2013). In Fig. 7, we plot our flux measurements alongside Swift-BAT count rates. Lubin´ski et al. (2020) show that a BAT count rate of 0.2 counts s1 corresponds to an Integral-ISGRI flux of 2  108 erg cm2 s1 in the 22–100 keV band. Extrapolating this value to our 20–200 keV band, we expect the corresponding CZTI flux to be 2  108 erg cm2 s1 from independent and individual qaudrants. As seen in Fig. 7, there is very good correspondence between our individual fluxes and BAT count rates, further validating our method.

5. Discussion The primary purpose of CZTI is the Hard X-ray imaging and spectroscopy of on-axis sources by using a coded aperture mask. The field of view is limited to a full-width at half maximum of 4:6  4:6 by using collimators. Despite this, the reduced opacity of the instrument and satellite structure at energies above Figure 6. Effective area for the Crab’s spectrum across all events.

3

https://swift.gsfc.nasa.gov/results/transients/BAT_current.html# anchor-CygX-1.

64

Page 8 of 9

J. Astrophys. Astr.

(2021) 42:64

Table 1. Refining Crab flux measurements based on combining all the quadrants by selecting a progressively smaller subsets of events with the highest effective area. The first column gives the selection cut applied above the default veto cut that requires the quadrant effective area to be at least 10 cm2 . The second column gives the corresponding effective area cutoff for our sample, in cm2 . The third column lists the flux measured from this sample, in units of 108 erg cm2 s1 . Fourth and the fifth columns gives nominal error measured in individual occultation event and weighted error as defined in Eq. (7) in the same unit of 108 erg cm2 s1 . The true flux in this energy range in 2:38  108 erg cm2 s1 . % selected All Top 50% Top 20% Top 10%

Area cut (cm2 )

Flux

[10 J25 J35 J41

2.65 2.41 2.29 2.25

Median error (108 erg cm2 s1 ) 1.20 0.95 0.81 0.75

Weighted error 0.02 0.01 0.01 0.02

Figure 7. Left: Shows direct comparison between the CZTI fluxes with BAT count rates. Here we have assumed zero errors for BAT values. Right: Comparing Cyg X–1 measurements from Swift-BAT and CZTI. Blue filled circles with error bars denote flux measurements by CZTI using EOT. Orange crosses are BAT count rates. BAT rates are scaled to CZTI fluxes as discussed in Section 4.4.

 100 keV makes it an effective all-sky monitor, capable of detecting transients at large off-axis angles. In this work, we have extended this capability to the study of persistent sources at large off-axis angles. We demonstrate a proof-of-concept by measuring the flux of the Crab nebula and pulsar, and measuring the variability of Cyg X–1. To the best of our knowledge, this is the first demonstration of measuring source fluxes by using a limited-field collimated instrument. We conclude that the relatively low off-axis effective area limits the applicability of the Earth Occultation Technique to the brightest sources, with fluxes J1 Crab. This method will be highly effective with future open-detector missions like Daksha which have a high effective area over the entire sky.

Acknowledgements We thank Vedant Shenoy and Akash Anumarlapudi for their assistance in data analysis. CZT–Imager is

built by a consortium of Institutes across India. The Tata Institute of Fundamental Research, Mumbai, led the effort with instrument design and development. Vikram Sarabhai Space Centre, Thiruvananthapuram provided the electronic design, assembly and testing. ISRO Satellite Centre (ISAC), Bengaluru provided the mechanical design, quality consultation and project management. The Inter University Centre for Astronomy and Astrophysics (IUCAA), Pune did the Coded Mask design, instrument calibration, and Payload Operation Centre. Space Application Centre (SAC) at Ahmedabad provided the analysis software. Physical Research Laboratory (PRL) Ahmedabad, provided the polarisation detection algorithm and ground calibration. A vast number of industries participated in the fabrication and the University sector pitched in by participating in the test and evaluation of the payload. The Indian Space Research Organisation funded, managed and facilitated the project. This work utilised various software including Python, AstroPy (Robitaille et al. 2013),

J. Astrophys. Astr.

(2021) 42:64

NumPy (van der Walt et al. 2011), and Matplotlib (Hunter 2007). References Anumarlapudi A., Bhalerao V., Tendulkar S. P., Balasubramanian A. 2020, AJ, 888, 40 Anusree K. G., Bhattacharya D., Rao A. R. et al. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036021-09707-5 Bhalerao V., Bhattacharya D., Vibhute A. et al. 2017, J. Astrophys. Astron., 38, 31 Harmon B., Fishman G., Wilson C. et al. 2002, AJ Suppl. Ser., 138, 149 Hunter J. D. 2007, Comput. Sci. Eng., 9, 90 Kapahi V. K., Joshi M. N., Subrahmanya C. R., Krishna G. 1973, AJ, 78, 673 Kouzu T., Tashiro M. S., Terada Y. et al. 2013, PASJ, 65, 74 Krimm H. A., Holland S. T., Corbet R. H. D. et al. 2013, AJ Suppl. Ser., 209, 14

Page 9 of 9

64

Lubin´ski P., Filothodoros A., Zdziarski A. A., Pooley G. 2020, AJ, 896, 101 Madsen K. K., Forster K., Grefenstette B. W., Harrison F. A., Stern D. 2017, AJ, 841, 56 Mate S., Chattopadhyay T., Bhalerao V. et al. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036021-09763-x Robitaille T. P., Tollerud E. J., Greenfield P. et al. 2013, A&A, 558, A33 Sharma Y., Marathe A., Bhalerao V. et al. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036021-09714-6 Singh K. P., Tandon S., Agrawal P. et al. 2014, in Space Telescopes and Instrumentation 2014: Ultraviolet to Gamma Ray, Volume 9144, International Society for Optics and Photonics, 91441S van der Walt S., Colbert S. C., Varoquaux G. 2011, Comput. Sci. Eng., 13, 22 Wilson-Hodge C. A., Case G. L., Cherry M. L. et al. 2012, AJ Suppl. Ser., 201, 33 Zhang Y. 2013, Ph.D. Thesis, LSU Doctoral Dissertations

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:66 https://doi.org/10.1007/s12036-020-09679-y

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

UV photometry of spotted stars in the horizontal branch of the globular cluster NGC 2808 using AstroSat DEEPTHI S. PRABHU1,2,* , ANNAPURNI SUBRAMANIAM1

and SNEHALATA SAHU1 1

Indian Institute of Astrophysics, Bangalore 560 034, India. Pondicherry University, R.V. Nagar, Kalapet 605 014, India. *Corresponding Author. E-mail: [email protected] 2

MS received 7 November 2020; accepted 29 November 2020 Abstract. A recent study of hot (20,000 to 30,000 K) extreme horizontal branch (EHB) stars in globular clusters (GCs) has led to the discovery of their variability. It is suggested that this variability is driven by the projected rotation of magnetic spots on the stellar surfaces and is expected to have higher amplitudes at shorter wavelengths. Here, we present the analysis of such hot stars in the massive GC NGC 2808 using the UltraViolet Imaging Telescope (UVIT), aboard AstroSat. We use the UVIT data in combination with the Hubble Space Telescope UV globular cluster survey (HUGS) data for the central region (within  2:70  2:70 ) and ground-based optical photometry for the outer parts of the cluster. We generate the Far-UV (FUV)– optical colour-magnitude diagrams (CMDs) and in these we find a population of hot EHB stars fainter than the zero-age horizontal branch (ZAHB) model. A comparison of our FUV magnitudes of the already reported variable EHB stars (vEHBs) shows that the longest period vEHBs are the faintest, along with a tentative correlation between rotation period and UV magnitude of spotted stars. In order to firmly establish any correlation, further study is essential. Keywords. Galaxy: globular clusters: individual (NGC 2808)—stars: Hertzsprung–Russell and C–M diagrams—stars: horizontal-branch—stars: variables: general—ultraviolet: stars.

1. Introduction Horizontal branch (HB) stars are low-mass helium (He) core burning stars consisting of a hydrogen-rich envelope. Globular clusters (GCs) are the best test beds to explore low-mass stars in various stages of evolution, including the HB stars. The morphology of the HB in GCs shows many peculiarities, one of which is the well-known ‘‘second parameter’’ problem (Sandage & Wallerstein 1960; Sandage & Wildey 1967; van den Bergh 1967). This refers to an observation that the colour distribution of HB stars is affected by parameters (eg., age, He abundance) other than metallicity (Catelan 2009). At a given metallicity, the colour distribution of the HB is a function of the envelope masses of stars which This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

in turn depends on the mass-loss on the red giant branch (RGB) phase (Iben & Rood 1970; Rood 1973). The less massive the envelope is, the hotter the resulting star is. The HB is thus made up of various sub-populations having similar core masses (  0.5M ) with differences in envelope masses. These include red HB (RHB), blue HB (BHB), extreme HB (EHB) and blue hook (BHk) stars or late hot flashers (D’Cruz et al. 1996, 2000; Sweigart 1997), in the increasing order of effective temperatures (Teff ). The hot HB stars along with post-HB (pHB) stars, white dwarfs (WDs) and the exotic blue straggler stars (BSSs) emit copious amount of flux in the ultraviolet (UV) wavebands. A recent study by Momany et al. (2020) utilizing near-UV (NUV) and optical data, led to the discovery of variable stars among the EHB population (vEHB), in three Galactic GCs, namely: NGC 2808,

66

Page 2 of 9

NGC 6752 and NGC 5139 (x Cen.). The EHB stars have extremely thin radiative envelopes (.0.02M ) and Teff ranging from  20,000 to 30,000 K. Their counterparts in the Galactic field are sub-dwarf B-type (sdB) stars. Due to their thin envelopes, the EHB stars do not ascend the asymptotic giant branch (AGB) phase after core He exhaustion. Instead, they briefly brighten up as AGB-manque´ stars and then evolve directly towards the WD phase. Momany et al. (2020) reported two types of EHB variability, periodic (periodicity of  2 to 50 days with DUJohnson  0:04 to 0.22 mag) and aperiodic. This EHB variability was observed to be tightly connected around the photometric jump at Teff  22;500 K (the Momany jump, Momany et al. 2002), a feature observed in all GCs (Momany et al. 2004; Brown et al. 2016). Through appropriate analyses and arguments, binary evolution or pulsation were discarded as reasons for the variability. The origin of both modes of EHB variability was attributed to the a2 Canum Venaticorum (a2 CVn) phenomenon. In this scenario, chemical inhomogeneities on the stellar surface cause spatial variations in opacity resulting in spots which are stabilized for a long time (several decades) by an underlying magnetic field. The projected rotation of these magnetic spots causes photometric/spectroscopic variability. The uneven surface chemical distribution in EHB stars is caused due to atomic diffusion processes. The magnetic fields required to sustain the spots were argued to be resulting from the second ionization He convective zone (HeIICZ), lying just beneath the radiative envelope. The aperiodic variability was explained as a consequence of the turbulent nature of the HeIICZ. Their work established that magnetism plays a key role in the formation and evolution of EHB stars and their Galactic field counterparts. The magnetic spots caused by the a2 CVn phenomenon are expected to be dark in the far-UV (FUV) and bright in the optical wavebands (Mikula´sˇek et al. 2019), resulting in the vEHBs being fainter in FUV, depending on the spot properties and showing higher variability in the FUV. Understanding the FUV properties of EHB stars as a function of their rotation will help in throwing more light on our emerging understanding of EHB stars. Moreover, observing such hot stars in the UV is advantageous because the contribution from cooler stellar populations such as main sequence (MS) and RGB stars gets suppressed, thereby considerably reducing the stellar crowding in the central regions of GCs, where many of the EHB stars are located. In this study, we present the far-UV and near-UV photometric analysis and spectral energy distributions

J. Astrophys. Astr.

(2021) 42:66

(SEDs) of 12 vEHB stars and 3 vEHB candidates newly discovered in the GC NGC 2808 by Momany et al. (2020). This GC has an age ¼ 10:9  0:7 Gyrs (Massari et al. 2016), ½Fe=H ¼ 1:14 dex and is located at a distance of 9.6 kpc (Harris 1996, 2010 edition, H96). We make use of the archival UV data from the AstroSat/ UVIT along with UV-optical data from the Hubble Space Telescope UV Globular Cluster Survey (HUGS) catalogue (Nardiello et al. 2018; Piotto et al. 2015), optical data from ground-based telescopes (Stetson et al. 2019) and others available in literature. Jain et al. (2019) used the UVIT data for this cluster to study the photometric gaps in the colour-magnitude diagrams (CMDs) and the multiple stellar populations. Prabhu et al. (2021) presented a detailed analysis of the stars in the pHB phase in this cluster utilizing the UVIT data in combination with other multiwavelength datasets. In this work, we study the UV properties of the spotted EHB stars in the cluster. The paper is laid out as follows: Section 2 gives the details of the observations and data reduction procedure. The UV photometric analysis of vEHB stars is described in Section 3. The SEDs of these stars and the derived results are presented in Section 4. The results are discussed in Section 5 and a summary is presented in Section 6.

2. Observations and data reduction In this study, we utilized the archival AstroSat/UVIT data for NGC 2808. The UVIT is made up of two 38cm diameter telescopes, one for FUV (k ¼ 1300– ˚ band pass, and the other for NUV 1800 A) ˚ and VIS (k ¼ 3200–5500 A) ˚ (k ¼ 2000–3000 A) band passes. The data from the VIS channel are used mainly for the drift-correction of images. The Fieldof-View (FoV) of UVIT is circular with a diameter of 280 . For more details regarding the instrument and calibration, refer Tandon et al. (2017). We used the data in two FUV (F154W and F169M) and four NUV filters (N242W, N245M, N263M and N279N). The images in these filters were generated using the CCDLAB software package (Postma & Leahy 2017) by correcting for the spacecraft drift, geometrical distortions and flat-field illumination. The final science-ready images were created by aligning and merging the images corresponding to different orbits. We then executed crowded field photometry on these images using the DAOPHOT software package

J. Astrophys. Astr.

(2021) 42:66

of IRAF/NOAO (Stetson 1987). The detailed steps of the procedure can be found in Prabhu et al. (2021). The final magnitudes (AB system) in all the filters were corrected for extinction by choosing the reddening value of EðB  VÞ ¼ 0:22 mag (Harris 1996, 2010 edition) and the ratio of total to selective extinction, RV ¼ 3:1. To calculate the extinction coefficients in various filters, the reddening law of Cardelli et al. (1989) was employed. For identifying the different classes of stars detected in the UVIT images and to obtain their cluster membership probabilities, the stars within the central  2:70  2:70 region (inner region) were cross-matched with the HUGS catalogue as described in Prabhu et al. (2021). The HUGS catalogue consists of data in five filters, namely, the WFC3/UVIS F275W (NUV), F336W (U) and F438W (B) filters, along with the ACS/WFC F606W (V) and F814W (I) filters. The UVIT covers a much larger FoV as compared to the Hubble Space Telescope (HST). Thus, in order to identify stars in the outer region (region outside HUGS coverage) and to select the cluster members among them, the UVIT data were combined with other datasets. The membership probabilities of the stars were estimated based on the Gaia DR2 proper-motion data using the technique of Singh et al. (2020). The JohnsonCousins UBVRI photometry of cluster members thus selected, were obtained by a cross-match with the data from Stetson et al. (2019). In order to plot the stars in the inner and outer regions in the same colour and magnitude plane, we transformed the Johnson V magnitudes into the equivalent HST ACS/WFC filter using the transformation equations of Sirianni et al. (2005).

3. Spotted EHB stars The 12 vEHB stars and the 3 vEHB candidates (vEHB-C) from Momany et al. (2020) were crossidentified in our UV catalogue. We note that some of these vEHB stars did not meet the membershipprobability cut-off. Nevertheless, we include them in our study, as they are considered by Momany et al. (2020) and this is a follow up study of their objects. Figure 1 shows the spatial locations of these vEHBs (including candidates) marked over the UVIT F154W image. The stars within the inner region, covered by the HST WFC3/UVIS, are encircled in blue and the other stars in red. We would also like to point out that only one of the vEHBs located in the inner region (vEHB-7) is affected by neighbour contamination in

Page 3 of 9

66

the UVIT images. Hence, we do not use the UVIT data for further analysis of this star. 3.1 vEHBs in FUV-optical CMD Figure 2 shows the mF154W  mF606W vs. mF154W CMD for the cross-matched stars in our catalogue. To create this CMD, the F606W magnitudes are converted from the VEGA magnitude system to the AB system using appropriate conversion factors.1 The HB sequence of the CMDs are overlaid with the updated BaSTI (a Bag of Stellar Tracks and Isochrones) theoretical zero-age HB (ZAHB) and terminal-age HB (TAHB; end of He burning phase) models from Hidalgo et al. (2018). These models correspond to metallicity ½Fe=H ¼ 0:9 dex, He abundance = 0.249, solar scaled ½a=Fe ¼ 0:0, and no convective overshoot. The models also account for the effects due to atomic diffusion. The different classes of HB stars (BHk, EHB, BHB), B gap objects, the pHB stars and the BSSs are indicated with distinct colours. In the figure, the vEHB stars are shown with black star symbols and the candidates with red diamonds. We do not include vEHB-7 in this plot for the reason mentioned earlier. One vEHB (vEHB-10) is seen to be located just above the TAHB sequence. This star along with vEHB-7 have been previously identified as pHB candidates by Prabhu et al. (2021) with vEHB-7 corresponding to Star 21 and vEHB-10 corresponding to Star 33 in their catalogue. One of these stars (vEHB10) is also listed as an AGB-manque´ star (NGC 2808594) in the study by Schiavon et al. (2012). In the CMD, several EHB stars are observed to be sub-luminous with respect to the ZAHB. About half of the vEHB stars belong to this sub-luminous EHB population. 3.2 Correlation between UV magnitudes and periods of vEHBs To check whether any correlation exists between the UV magnitudes and periods of these vEHBs, we plotted them against each other in Fig. 3. The periods of these stars are obtained from Momany et al. (2020). We note the following from the figure: in general, stars with the longest periods are fainter in the FUV and NUV. There is a tentative correlation between period and FUV magnitudes (F154W and F169M), and also within errors in the NUV magnitudes (N242W, N245M, N263M and N279N) such that 1

http://waps.cfa.harvard.edu/MIST/BC_tables/zeropoints.txt.

66

Page 4 of 9

J. Astrophys. Astr.

(2021) 42:66

from the figures, even though we notice a tentative correlation, we are unable to make a definite correlation. In order to further understand this aspect and to explore the correlations (if any) between the fundamental parameters of vEHBs and their periods, we generated their SEDs as explained in the next section.

4. Spectral energy distributions

Figure 1. The vEHB stars marked over the UVIT F154W image of NGC 2808. The magenta region marks the central  2:70  2:70 region (inner region) covered by HST and the stars within this region are encircled in blue. The stars lying in the outer region are encircled in red.

longer period stars are fainter in the UV, except for two stars, which deviate from this trend. That is, among the 4 stars that are UV faint and found below the ZAHB (vEHB-3, vEHB-6, vEHB-9 and vEHB11), vEHB-3 and vEHB-6 have longer periods, whereas the other two have short periods. This is noticed in all the sub-plots shown in Fig. 3. Therefore,

We exploited the available multiwavelength photometric data for the vEHB stars to construct their SEDs and derive parameters such as Teff , luminosity (L), and radius (R). We used a virtual observatory tool, VOSA (VO SED Analyser; Bayo et al. 2008), for this purpose. VOSA produces synthetic photometric points for the selected theoretical models using the response curves of the required photometric filters. In order to estimate the best-fit parameters for the SEDs, the observed and synthetic photometric points are compared using the v2 minimization technique. The v2red value is found using the relation ( ) N 1 X ðFo;i  Md Fm;i Þ2 2 ; ð1Þ vred ¼ N  Nf i¼1 r2o;i where N is the number of photometric points, Nf is the number of fitted parameters for the model, Fo;i is the observed flux, Fm;i is the flux predicted by the theoretical model, Md ¼ ðDR Þ2 is the multiplicative dilution factor (where R is the radius of the star and

Figure 2. The mF154W  mF606W vs. mF154W CMD for the cluster member stars common in F154W UVIT filter and other catalogues (HUGS and ground-based optical data). The black solid and dashed lines represents the BaSTI ZAHB and TAHB models with ½Fe=H ¼ 0:9 dex, respectively. The photometric errors in magnitude and colour are also shown along the left side of the plot. The four faintest vEHB stars are annotated.

J. Astrophys. Astr.

(2021) 42:66

Page 5 of 9

66

Figure 3. The magnitudes (extinction corrected) of vEHB stars along with their photometric errors in different UVIT filters, plotted against their periods from Momany et al. (2020). In the filters N263M and N279N, not all vEHBs are detected. The star vEHB-7 is excluded because it is affected by neighbour contamination. The IDs of the four stars with the faintest UV magnitudes are marked.

66

Page 6 of 9

D is the distance to the star) and ro;i is the error in the observed flux. For all the vEHB stars, we assumed D ¼ 9:6 kpc and EðB  VÞ ¼ 0:22 mag (Harris 1996, 2010 edition). For accounting the extinction in the observed photometric data points, VOSA uses the Fitzpatrick reddening relation (Fitzpatrick 1999). For the vEHB stars located within the central  2:70  2:70 region of the cluster, we used the photometric data from UVIT and HUGS catalogues to generate the SEDs. In the case of stars located outside this region, we used the UVIT data in combination with UBVRI photometry from Stetson et al. (2019), and other available VO photometric data from VOSA. The Kurucz stellar atmospheric models (Castelli et al. 1997; Castelli & Kurucz 2003) were used to fit the SEDs. The free parameters in these models are: log g with a range 0.0 to 5.0, [Fe/H] with a range -2.5 to 0.5 and Teff ranging from 3500 to 50000 K. In order to fit the SEDs with these models, the value of [Fe/H] was fixed at -1.0 dex, close to the value for the cluster. The best-fit parameters derived from the SED analysis for these stars are tabulated in Table 1. The parameter errors quoted in the table were estimated as half the grid step, around the best-fit value. We refrain from tabulating the best-fit log g values as SED analysis does not give accurate values of this parameter. Figure 4 shows examples of SED fits for vEHB-1 and vEHB-6 with fit residuals. The residuals were calculated for each photometric point as follows: Fo  Fm ; ð2Þ Residual ¼ Fo where Fo and Fm are the observed and the theoretical model fluxes corresponding to the photometric points. We now plot the derived fundamental parameters, L and Teff , against the periods of the vEHB stars in Fig. 5. From the right panel, we hardly see any correlation between the Teff and period. The left panel shows the same tentative trend that was observed earlier, that the vEHBs with the longest periods have the smallest bolometric luminosities. The two stars which slightly deviate from the trend are also the same. Therefore, we conclude there is a tentative correlation between the luminosity and period among the EHB stars.

5. Discussion The UV follow-up study of the newly discovered spotted EHB stars in the GC NGC 2808 is presented here. We utilize the UVIT data complemented by the

J. Astrophys. Astr.

(2021) 42:66

data from HUGS catalogue for the inner region of the cluster and the ground-based optical photometry catalogue of Stetson et al. (2019) for the outer region. From the mF154W  mF606W versus mF154W CMD, we find that about half of the vEHB stars belong to the EHB population which is sub-luminous with reference to the theoretical ZAHB model. The faintest among these stars (vEHB-6 and vEHB-3) are those with the longest periods. However, the location of these stars in the CMD may not be an effect of their rotation as we detect two other stars with shorter periods which have similar F154W magnitudes within errors. The SED analysis of vEHB stars leads to an inference that the slowest rotators also have the smallest bolometric luminosities. We also find that two of the vEHBs have very high luminosities (vEHB-7 and vEHB-10). These stars have been classified as AGBmanque´ stars in the works of Schiavon et al. (2012) and Prabhu et al. (2021). Also, since the amplitude of variability of all the vEHB stars is very small (DUJohnson  0:04–0.22 mag), it highly improbable that the results obtained in this study are significantly affected due to the variability. Recio-Blanco et al. (2002) obtained the rotational velocities of hot HB stars in this GC. Their analysis showed that about 20% of the HB stars hotter than Teff  11; 500 K are slow rotators (v sin i\2 km s1 ). We plan to obtain the catalogue of these slow rotating hot HB stars from the authors and locate them in our UV study. This might also give some clues regarding these stars and their UV properties. In future, we also plan to look into the NUV and FUV variability of these spotted stars using the UVIT data by generating light curves. 6. Summary (1) We present the AstroSat/UVIT photometric analysis of the recently discovered vEHB stars in the GC NGC 2808. The UVIT data is used in combination with the data from HUGS, and groundbased optical photometry catalogues. The UV study of these objects is crucial because the a2 CVn mechanism which is responsible for the detected variability in EHB stars results in fainter UV magnitude depending on spot properties and higher variability in the UV wavebands than in the optical. (2) We present the FUV-optical CMD of the UVbright cluster members along with all the reported

J. Astrophys. Astr.

(2021) 42:66

Page 7 of 9

66

Table 1. The results of SED fitting procedure for the 15 vEHBs in the cluster. Columns 1 and 2 show the star IDs and periods of the vEHBs from Momany et al. (2020). The estimated parameters such as effective temperature, luminosity (in solar units) and radius (in solar units) of the stars and the errors in these parameters are tabulated in columns 3 to 5. The errors are obtained as half the grid step, around the best-fit value. Columns 6 and 7, respectively, show the reduced chi square value corresponding to the fit and the number of photometric points used for fitting. Column 8 shows the spatial location (SL) of the stars where ‘inner’ stands for the star lying within the HST FoV and ‘outer’ stands for star lying outside this region. The membership probability (MP) of each star is indicated in the last column. ID

Perioda (days)

Teff (K)

L L

R R

v2red

Nfit

S.L.

M.P. (%)

vEHB-1 vEHB-2 vEHB-3 vEHB-4 vEHB-5 vEHB-6 vEHB-7* vEHB-8 vEHB-9 vEHB-10* vEHB-11 vEHB-12

3.38683880 5.47704961 26.01823489 1.97628738 3.23990003 50.10395158 3.02583807 2.89086189 6.90595098 3.58179363 3.19599005 4.26386105

21; 000  500 22; 000  500 21; 000  500 23; 000  500 22; 000  500 19; 000  500 24; 000  500 21; 000  500 18; 000  500 26000  500 17; 000  500 21; 000  500

20:56  0:19 22:11  0:11 13:44  0:21 22:15  0:22 18:92  0:11 12:54  0:21 49:30  0:08 28:77  0:17 14:96  0:25 53:26  0:51 16:35  0:16 25:24  0:48

0:34  0:02 0:32  0:01 0:28  0:01 0:30  0:01 0:30  0:01 0:32  0:02 0:41  0:02 0:41  0:02 0:40  0:02 0:36  0:01 0:47  0:03 0:38  0:02

7.05 2.09 4.09 9.57 5.10 4.20 3.75 3.06 2.57 2.39 1.75 8.99

8 8 12 10 8 12 5 9 8 11 7 15

Inner Inner Outer Outer Outer Outer Inner Inner Outer Outer Inner Outer

97.3 97.3 0 0 – 98.9 96.6 98 0 92.6 59.4 99.3

Candidate vEHBs vEHB-13 0.80305948 vEHB-14 11.37292900 vEHB-15 2.32221267

21; 000  500 24; 000  500 25; 000  500

27:57  0:17 18:75  0:22 21:94  0:16

0:40  0:02 0:25  0:01 0:25  0:01

8.53 4.00 3.00

9 11 8

Inner Outer Outer

97.9 0 96.1

From Momany et al. (2020). These stars are classified as pHB stars by Prabhu et al. (2021). vEHB-10 is categorized as a pHB star also by Schiavon et al. (2012).

a

Figure 4. The SEDs for two of the vEHBs, namely vEHB-1 and vEHB-6, after extinction correction. The gray line shows the model spectrum. The residuals of SED fit are shown in the bottom panels of both plots.

vEHB stars. The location of the HB sequence is compared with the theoretical ZAHB and TAHB models. We find that about half of the vEHB stars belong to the EHB population that is sub-luminous in comparison with the theoretical ZAHB.

(3) A plot of the UV magnitudes versus the periods of the vEHB stars shows that the two longest period variables (vEHB-6 and vEHB-3) are the faintest in the two FUV wavebands and within errors in the four NUV wavebands.

66

Page 8 of 9

J. Astrophys. Astr.

(2021) 42:66

Figure 5. The L=L and Teff derived from SED fitting technique, plotted against the periods of all the vEHB stars. The longest period variables are marked in the plots.

(4) SED fitting technique was adopted to estimate parameters such as Teff , R=R and L=L of these stars. The plot of L=L against the periods shows that the two longest period vEHBs also have the smallest bolometric luminosities among the sample. (5) A detailed study of the UV variability of these spotted stars using UVIT data will be carried out in the near future.

Acknowledgements The authors gratefully acknowledge Y. Momany for the useful discussions. This publication utilizes the data from AstroSat mission the Indian Space Science Data Centre (ISSDC).This publication uses of UVIT data processed by the payload operations center by the IIAIIA. The unit is built in collaboration between IIA, IUCAA, TIFR, ISRO, and CSA. This research made use of VOSA, developed under the Spanish Virtual Observatory project supported by the Spanish MINECO through grant AyA2017-84089. This research also made use of Topcat (Taylor 2005), ‘‘Aladin sky atlas’’ developed at CDS, Strasbourg Observatory, France (Bonnarel et al. 2000; Boch & Fernique 2014), Matplotlib, NumPy, SciPy and pandas.

References Bayo A., Rodrigo C., Barrado Y Navascue´s D. et al. 2008, A&A, 492, 277 Boch T. Fernique P. 2014, in Manset N., Forshay P., eds, Astronomical Society of the Pacific Conference Series,

Vol. 485, Astronomical Data Analysis Software and Systems XXIII, 277 Bonnarel F., Fernique P., Bienayme´ O. et al. 2000, A&As, 143, 33 Brown T. M., Cassisi S., D’Antona F. et al. 2016, ApJ, 822, 44 Cardelli J. A., Clayton G. C., Mathis J. S. 1989,ApJ, 345, 245 Castelli F., Gratton R. G., Kurucz R. L. 1997, A&A, 318, 841 Castelli F., Kurucz R. L. 2003, in Piskunov N., Weiss W. W., Gray D. F., eds, IAU Symposium, Vol. 210, Modelling of Stellar Atmospheres, A20 Catelan M. 2009, Astrophys Space Sci, 320, 261 D’Cruz N. L., Dorman B., Rood R. T. O’Connell R. W. 1996, ApJ, 466, 359 D’Cruz N. L., O’Connell R. W., Rood R. T. et al. 2000, ApJ, 530, 352 Fitzpatrick E. L. 1999, PASP, 111, 63 Harris W. E. 1996, AJ, 112, 1487 Hidalgo S. L., Pietrinferni A., Cassisi S. et al. 2018, ApJ, 856, 125 Iben Icko J., Rood R. T. 1970, ApJ, 161, 587 Jain R., Vig S., Ghosh S. K. 2019, MNRAS, 485, 2877 Massari D., Fiorentino G., McConnachie A. et al. 2016, A&A, 586, A51 Mikula´sˇek Z., Paunzen E., Hu¨mmerich S. et al. 2019, in ed. Kudryavtsev D. O., Romanyuk I. I., Yakunin, Astronomical Society of the Pacific Conference Series, Vol. 518, Physics of Magnetic Stars, 117 Momany Y., Bedin L. R., Cassisi S. et al. 2004, A&A, 420, 605 Momany Y., Piotto G., Recio-Blanco A. et al. 2002, ApJL, 576, L65 Momany Y., Zaggia S., Montalto M. et al. 2020, Nature Astron., arXiv:2006.02308 Nardiello D., Libralato M., Piotto G. et al. 2018, MNRAS, 481, 3382 Piotto G., Milone A. P., Bedin L. R. et al. 2015, AJ, 149, 91 Postma J. E., Leahy D. 2017, PASP, 129, 115002

J. Astrophys. Astr.

(2021) 42:66

Prabhu D. S., Subramaniam A., Sahu S. 2021, Astrophys. J. 908, 66 Recio-Blanco A., Piotto G., Aparicio A., Renzini A. 2002, ApJ, 572, L71 Rood R. T. 1973, Metal-Poor Stars. V. ApJ, 184, 815 Sandage A., Wallerstein G. 1960, ApJ, 131, 598 Sandage A., Wildey R. 1967, ApJ, 150, 469 Schiavon R. P., Dalessandro E., Sohn, S. T. et al. 2012, AJ, 143, 121 Singh G., Sahu S., Subramaniam A., Yadav R. K. S. 2020, Peculiarities in the horizontal branch stars of globular cluster NGC 1851: Discovery of a Blue straggler companion to an EHB star, arXiv:2010.06979

Page 9 of 9

66

Sirianni M., Jee M. J., Benı´tez N. et al. 2005, PASP, 117, 1049 Stetson P. B. 1987, PASP, 99, 191 Stetson P. B., Pancino E., Zocchi A., Sanna N., Monelli M. 2019, MNRAS, 485, 3042 Sweigart A. V. 1997, ApJL, 474, L23 Tandon S. N., Subramaniam A., Girish V. et al. 2017, AJ, 154, 128 Taylor M. B. 2005, in Shopbell P., Britton M., Ebert R., eds, Astronomical Society of the Pacific Conference Series, Vol. 347, Astronomical Data Analysis Software and Systems XIV, 29 van den Bergh S. 1967, AJ, 72, 70

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:69 https://doi.org/10.1007/s12036-021-09759-7

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

Revisiting the Earth’s atmospheric scattering of X-ray/c-rays and its effect on space observation: Implication for GRB spectral analysis SOURAV PALIT1,* , AKASH ANUMARLAPUDI1,2

and VARUN BHALERAO1

1

Department of Physics, Indian Institute of Technology Bombay, Mumbai 400 076, India. Department of Physics, University of Wisconsin-Milwaukee, Milwaukee, WI, USA. *Corresponding Author. E-mail: [email protected] 2

MS received 7 November 2020; accepted 19 April 2021 Abstract. A considerable fraction of incident high energy photons from astrophysical transients such as Gamma Ray Bursts (GRBs) is Compton scattered by the Earth’s atmosphere. These photons, sometimes referred to as the ‘‘reflection component’’, contribute to the signal detected by space-borne X-ray/c-ray instruments. The effectiveness and reliability of source parameters such as position, flux, spectra and polarization, inferred by these instruments are therefore highly dependent on the accurate estimation of this scattered component. Current missions use dedicated response matrices to account for these effects. However, these databases are not readily adaptable for other missions, including many upcoming transient search and gravitational wave high-energy electromagnetic counter part detectors. Furthermore, possible systematic effects in these complex simulations have not been thoroughly examined and verified in literature. We are in the process of investigation of the effect with a detailed Monte Carlo simulations in GEANT4 for a Low Earth Orbit (LEO) X-ray detector. Here, we discuss the outcome of our simulation in form of Atmospheric Response Matrix (ARM) and its implications of any systematic errors in the determination of source spectral characteristics. We intend to apply our results in data processing and analysis for AstroSat-CZTI observation of such sources in near future. Our simulation output and source codes will be made publicly available for use by the large number of upcoming high energy transient missions, as well as for scrutiny and systematic comparisons with other missions. Keywords. Gamma ray bursts—atmospheric scattering—X-ray—GEANT4.

1. Introduction X-ray/c-ray astrophysical observations are primarily conducted from space-based platforms as the Earth’s atmosphere does not transmit these high energy photons. While most photons are absorbed, a fraction of the X-ray/c-ray photons from any extra-terrestrial sources undergoes Compton scattering in the upper layers of the atmosphere. In the Earth’s atmosphere, which has a mass column density of *103 g cm2 at sea level, some of the scattered photons travel upwards having an appearance of reflection of X-ray/ c-ray by the atmosphere. Any space borne This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

measurement of those high energy photons from astrophysical sources will be affected by simultaneous detection of the reflected photons. The topic of interaction of extra-terrestrial high energy photons with the atmosphere is not new. Various studies have been performed to determine the spectrum of reflected Cosmic X-ray Background (CXB)1 photons from the Earth’s atmosphere and their effects on observations by various space-based high-energy instruments (e.g. Churazov et al. 2008). Such works are often done in conjugation with studies determining the X-ray/c-ray spectrum arising from the interaction of cosmic ray particles (e.g. Sazonov et al. 1

Alternately termed as Cosmic Diffused Gamma Ray Background (CDGRB).

69

Page 2 of 15

2007) with the molecules in Earth’s atmosphere. Together these two components constitute the X-ray/ c-ray albedo contribution and add up to the total background noise in high energy space detectors. Apart from such studies in the context of Earth’s atmosphere, the main motivation of research related to the Compton reflection of high energy photons has been of purely astrophysical nature. This includes mainly the study of the scattering during the passage and interaction of cool electrons in relatively cooled plasma, such as the corona above a relatively cool, optically thick accretion disk (e.g. Bisnovatyi-Kogan & Blinnikov 1977). In such studies one usually calculates a Green’s function describing the reflection of monochromatic radiation (White et al. 1988), and convolves the incident spectrum with this function to calculate the reflected spectrum. From the studies of Earth’s atmospheric scattering of high energy photons it is estimated that a moderate fraction of the incoming X-ray/c-ray photons is reflected or back-scattered in the upper atmospheric layers of the Earth. Some studies estimate that in the *30 to 300 keV energy range (hereafter referred to as ‘‘intermediate energy’’) of X-rays/c-rays, as many as *30% of the incident photons can be scattered back (see for instance Churazov et al. 2008). The scattered fraction decreases outside this energy range, but remains non-negligible over a much broader X-ray/cray band.This suggests that Low Earth Orbit (LEO) satellites observing energetic photons from Gamma Ray Bursts (GRBs) and other astrophysical transients detect significant flux reflected from the Earth’s atmosphere as well. The magnitude and spectrum of the contribution should vary largely based on the relative angular position (viewing angle) of the GRB with respect to the instrument, as well as to the line joining the instrument to the geo-center, relative orientation of the instrument with the ground, and the source spectral properties. Whether this is considered as an interpretable addition to the source signal or merely as an addition to the background noise detected by the instrument like the CXB and albedo photons, any analysis of properties of the transients should correctly account for the contribution. Till date, the contribution of Earth’s reflection of X-rays and c-rays in prompt GRB observation has mainly been studied from the standpoint of its application in study of firstly, polarization in those wavelengths (e.g. Willis et al. 2005) and secondly in the process of localization of transient source (e.g. Pendleton et al. 1999). The fact that the angular

J. Astrophys. Astr.

(2021) 42:69

distribution of Compton scattered flux from Earth’s atmosphere should be dependent on the direction of polarization of the incoming photon beam from a distant GRB sources, has been exploited to find the polarization characteristics from measurement of the reflected X-ray flux. There has been rigorous investigation (Pendleton et al. 1999) for estimating the atmospheric scattering contribution to accurately determine transient source location by CGRO/BATSE and Fermi missions. They mainly concentrate on Monte Carlo simulation studies and some observations with solar flares. We investigated how various missions have incorporated the Earth’s reflection effect in their data processing. We have been particularly interested in digging into the CGRO/BATSE and Fermi/GBM spectral data processing and distribution, as these two have been the main workhorses in providing the spectral observation of GRB prompt emission from hard X-rays up to soft c-ray part of the spectrum. In case of CGRO/BATSE the atmospheric contribution is evaluated during burst localization using EGS electromagnetic cascade Monte Carlo code with a spherical geometry of concentric shells representing an atmosphere with exponential density. Subsequent iterative v2 minimization technique is followed (Pendleton et al. 1999) to differentiate the scattering contribution from the total observed spectra consisting of photons both directly detected from the source and those scattered from atmosphere. The analyses presented by Pendleton et al. (1999) show us the extent to which the atmospheric scattering of source photons can contaminate observation of transients. In Fig. 3 of the article demonstrating the distribution of number of GRBs detected by the detectors of different viewing angles (with respect to the burst) as function of the ratio of the detected atmospheric scattered flux to that of directly detected flux, it seems that for the detector with large viewing angles for a significant number of cases the detected scattered counts are well above the direct counts (by as much as an order). For moderate viewing angles the scattering rate is between 20 and 50% of that of the direct count rate. Figure 4 of the same article is also significant, depicting how the detected scattered to direct photon ratio can vary with zenith angle (i.e., the angle between the direction from instrument to the geo-center and the direction to the GRB), the viewing angle (w.r.t. the burst) of detector and the orientation of the detector normal respective to the ground normal. The bottom-left plot showing the variation of the

J. Astrophys. Astr.

(2021) 42:69

ratio with the source zenith angle for large viewing angle of the downward-facing detectors demonstrates the highest contribution of atmospheric scattering and its sharp dependence on photon incidence angle. For Fermi/GBM, according to Meegan et al. (2009), the atmospheric scattering response estimation has been performed following the same method as described for BATSE in the GEANT4 based GBM Response Simulation System (GRESS; Hoover et al. 2005; Kippen et al. 2007) and validated with BATSE simulation results. This is then used to form a sourcespecific total response function for a particular set of earth-spacecraft viewing conditions, to be calculated in the IODA DRMGEN software in stand alone mode. Whereas, Connaughton et al. (2015) directly use the atmospheric response database established for BATSE (Pendleton et al. 1999) in their study. The problem of unwanted contribution of atmospheric scattering of X-ray/c-ray is well addressed and steps have been followed to mitigate it, albeit, seemingly, there has been no quantitative, comprehensive and generalized description on how this should manifest in the estimation of the source characteristics, such as photon spectrum and polarization. These studies have been highly focused on specific instruments, and the final products are often deeply integrated with the instrument response function. Instead, in this work we take a generalized approach by concentrating on the Monte Carlo simulation of the interactions of energetic astrophysical source photons at atmosphere, completely devoid of any specific instrument interactions and characteristics. Thus, our goal is to calculate the distribution of atmospheric Compton scattered photons for a plane-parallel photon beam incident upon the Earth’s atmosphere and how it should manifest on the outcome of the spectral and polarization characteristics of the cumulative photon detection. Here we are presenting our findings only on the spectral front and intend to address the same on polarization elsewhere. The broader goal of this work is two-fold. The first is to replicate the work that has been performed for other missions, and apply it directly to analysis of sources in AstroSat-CZTI (Singh et al. 2014; Bhalerao et al. 2017b; Rao et al. 2017). Since its launch 5 years ago CZTI has detected more than 400 GRBs (Sharma et al. 2021). CZTI has been used for some GRB localization studies (Bhalerao et al. 2017a), and has also been very successful in detection of polarization in them (see Chattopadhyay et al. 2019). Inclusion of the process of evaluation and

Page 3 of 15

69

deduction of atmospheric scattered contribution from prompt GRB spectra will be a significant step towards accurate analysis of all the above characteristics. The second goal of the work is to publicly release all source codes and simulation data in a format that will be easy to adapt to any other mission. This goal stems from the plurality of upcoming and proposed missions to study high energy transients and gravitational wave high-energy electromagnetic counterpart detection, such as large instruments like SVOM (Cordier et al. 2015), GECAM (Zhang et al. 2019); as well as cubesat class instruments like BurstCube (Racusin et al. 2017), HERMES (Fuschino et al. 2019), BlackCAT (Chattopadhyay et al. 2018), and Camelot (Werner et al. 2018). There are also proposed missions, which are scientifically oriented towards hard X-ray/c ray polarimetry, such as POLAR-2 (Hulsman et al. 2020) on-board Chinese space station and a Large Area GRB Polarimeter (LEAP) (McConnell et al. 2017) on-board ISS. Rather than each of these missions developing their own atmospheric scattering databases, a single open database will serve the purpose well. It will also make it possible for comparisons between any other atmospheric scattering simulations, thus arriving at a quantitative estimate of systematic errors that may be present in the responses. The organization of the paper is as follows. In Section 2, we discuss our simulation setup including detector geometry of Earth’s atmosphere and photon collecting detector, the physics of interaction and photon generation and run execution process. In Section 3, we presents our simulation outcome for mono-energetic photons and a hypothetical GRB spectrum, represented by Band function (Band et al. 1993). In this section, we will also investigate how the scattered photon contribution may change the spectral characteristics and possibly contribute artefacts in the exact evaluation of the spectral slope, if they are not accounted for correctly. We conclude in Section 4 by discussing about the outcome of our study and describing our future plan of work.

2. Monte Carlo simulation with GEANT4 We have used GEometry ANd Tracking 4 (GEANT4) for simulating the Compton reflection of X-rays from Earth’s atmosphere. GEANT4 is a well known Monte Carlo detector simulation package (Agostinelli et al.

69

Page 4 of 15

2003) written in C??, initially developed for the simulation of high energy physics and gradually got enhanced, in order to be applied to lower energies also. Currently, it is used extensively in high energy detector and instrument simulation (mass model) of most of the astronomical missions including AstroSatCZTI (Mate et al. 2021). Below we describe the construction of the simulation geometry including the Earth’s atmosphere, the generic detector at LEO height, the physics models used in the simulation, and the methods for particle generation, data extraction, and analysis.

2.1 Geometry and Earth’s atmosphere The Monte Carlo code developed for this study primarily imitates the interactions of incident parallel mono-energetic beams of X-ray and c-ray photons with Earth’s atmosphere and find the detection responses at an imaginary flat and fully efficient detector placed at LEO, at height of 650 km. The attenuation of incoming X-ray to c-ray photons in the atmosphere occurs predominantly from few 100s of km down to *10 km (see Palit et al. 2018). About 99.7% of Compton back-scattering events occur below 125 km (Willis et al. 2005), with the majority of the interactions occurring in the 20– 50 km range. Considering this, we restrict the atmospheric heights from *10 to 500 km in the simulation. The lower part of this range is of significantly varying density and concentration of neutral components. Those are mainly nitrogen molecules (N2 ), oxygen molecules (O2 ) and some other less abundant elements like helium and argon, which play an important role in the interaction process of high energy photons. Above *100 km the mass column density of the atmosphere is below 1:4  103 gcm2 and the optical depth is so low that the variation in such concentration has negligible effect. Other than the inhomogeneity along vertical direction, the atmosphere also has inhomogeneities along latitude and longitude, and is dynamic in nature. For example, the atmosphere is thinner over the poles and thicker over the equator. Since our focus is on satellites in Equatorial orbits and we aim to calculate the nominal scattering effect, these are secondary concerns and can be neglected safely. In view of this, we incorporated only vertical stratification in the atmosphere with the necessary data on density, and various neutral concentrations.

J. Astrophys. Astr.

(2021) 42:69

GEANT4 detector or geometry construction supports only concentric spherical shells. We model the Earth’s atmosphere using multiple layers in the detector construction class. The distribution of layers is as follows: The first ninety layers starting from 10 km above the Earth’s surface have a thickness of 1 km each, followed by 20 layers of 5 km each, and lastly the 12 outermost layers are of 25 km each. They are formed with average molecular densities and other atmospheric parameters at those heights. Corresponding neutral atmospheric data are obtained from NASA-MSISE-90 atmospheric model (Hedin 1991). This atmospheric structure is directly imported from previous studies (Palit et al. 2013, 2018), which already have successfully evaluated the X-ray interaction process though in the context of solar flares and their effect on radio propagation. A NaI detector disc of radius of 100 km and thickness of 10 meter (ensuring 100% absorption of received photons) is placed at a height of 650 km from the Earth’s surface at the zenith (direction, vertically upward from Earth’s centre). A schematic of the simulation geometry along with the photon generation and propagation through atmosphere has been presented in Fig. 1.

2.2 Physics of interaction The main interaction processes for energetic photons corresponding to soft to hard X-rays and soft c-rays with the atmospheric molecules are the primary photoionization and subsequent secondary electron ionization. The photoionization can be due to photoelectric effect and Compton scattering. Electron–positron pair creation takes place for photon energies exceeding *1000 keV and has very small contribution in the overall interaction statistics. Due to the presence of a small amount of helium and few other heavier elements in the Earth’s atmosphere The photoelectric cross section exceeds the scattering one up to energies *30 keV and inelastic (Rayleigh) scattering in atmospheric composition dominates the total scattering cross section at energies below 10–20 keV (Churazov et al. 2008). Each of the Physics processes in GEANT4 uses several models. Models for all electromagnetic processes can be subdivided into four general physics scenario categories: standard, low-energy, polarized and adjoint. For unpolarized low energy photons, most Compton scattering models corresponding to ‘lowenergy’ and the ‘standard’ categories use the Klein–

J. Astrophys. Astr.

(2021) 42:69

Figure 1. A schematic of geometry set up for Monte Carlo simulation consisting of Earth’s atmosphere, the ideal detector (green) at LEO height, parallel beam of photons originating from a flat circular disc of radius 7200 km touching the surface of an Earth-centric imaginary sphere of same radius and falling on the atmosphere of Earth is shown.

Nishina approach (Klein & Nishina 1929) for crosssection. The other two efficient models (of ‘lowenergy’ category), namely, Livermore and Penelope based Compton scattering models use cross section calculation based on the EPDL data library (Cullen 1995) and Penelope code (Sempau et al. 2003). All these models, namely, the G4Livermore ComptonModel, G4PenelopeComptonModel, G4LowEPComptonModel etc. produce very similar outcome for our simulations. In the subsequent sections, we present the results computed with the G4PenelopeComptonModel and corresponding other Penelope low energy physics models.

2.3 Source photons, interactions and simulation steps

During source photon generation within primary generation class of GEANT4, an imaginary hollow sphere of large radius (7200 km) is assumed concentric to the Earth’s layered atmosphere. The photons are generated randomly on the surface of an imaginary flat disc (Fig. 1) of same radius, touching the outer

Page 5 of 15

69

surface of the sphere at a point of intersection of the incoming photon direction with it and placed tangentially. This ensures that the photon incidence corresponds to plane wave coming from virtually infinite distance. During the simulation the detector is always kept at the vertically up-word position (zenith) that is on an imaginary vertical line extending upward from the center of the Earth. The angular position of the source (h) i.e., the incidence angle is always set with respect to the vertical line (Fig. 1). h ¼ 0 corresponds to normal incidence on the atmosphere, with the detector directly between the source and the Earth. The detector is insensitive to direct source photons. The effect of the small shadow of the detector is corrected for by scaling up the observed counts to the unobstructed cross-section of the Earth. We run each of our simulations with a mono-energetic photon beam having sufficient number (108 ) of photons, so that the detector can collect *10000 scattered photons from the whole of the atmosphere. Each run comprises of primary photons at a single energy. Our simulations span the energy range from 5 keV to 5 MeV as follows: 2 keV intervals from 10 to 50 keV, 5 keV intervals from 50 to 200 keV, 20 keV intervals from 200 to 400 keV, 100 keV intervals from 400 to 1000 keV, and simulations at 2000, 3000, 4000 and 5000 keV. This division is carefully chosen to get a smooth Atmospheric Response Matrix (ARM) (described in Section 3) by interpolation of the resultant mono-energetic distribution functions over incident energy values. Simulations are carried out for varying incidence angles (h), and here we focus on a representative subset of these incidence angles. Details of the simulation steps and subsequent processing and analysis are covered in the Section 3.

3. Results As discussed in the previous section, we undertake simulations only for mono-energetic beams with different incidence angles. The outcome, with suitable interpolation can easily be converted into response matrix and the spectrum for any realistic astrophysical source has to be synthesized by using a weighted sum of the simulated mono-energetic responses. The weights are to be determined by the number of source photons expected at that incident energy. The exercise of starting with mono-energetic incident beams has two main advantages. Firstly, it is

69

Page 6 of 15

helpful in better understanding of the exact nature of the atmospheric scattering response and how it varies over the incident photons at different ranges of X-ray/ c-ray energies. Secondly, this allows us to re-use our simulation results for any incident spectrum, without resorting to computationally expensive runs for each of them.

3.1 Mono-energetic response and Atmospheric Response Matrix (ARM) Figure 2 shows the detected (simulated) atmospheric scattered spectra for few of the incident energies, normalized for a single incident photon at that energy. The incidence angle (h) for this plot is 0 . Due to inelastic interactions, many source photons get scattered to lower energies, creating a broadband spectrum for each of the mono-energetic incident photons. The maximum energy of scattered photons, as well as the energy at which the scattered spectrum peaks, are both below the incident energy. Both the parameters increase with the incidence beam energy up to few hundreds of keVs, but as incidence energy goes higher and crosses 1 MeV the maximum and peak energies of scattered photons stop increasing. We found that whatever higher the incident energy values are used,

Figure 2. Calculated response functions (spectra) are shown in the plot for 0 incidence angles of monochromatic photon beams. Here, the horizontal axis corresponds to the detected atmospheric scattered photon energy. The vertical axis represents the number of reflected (scattered) photons from Earth’s atmosphere detected at 1 cm2 of detector area at the detector at LEO height due to 1 incidence photon per square of centimeter of whole of the Earth’s atmosphere at the mentioned energies and angle. The dashed black line denotes the overall envelope of the responses.

J. Astrophys. Astr.

(2021) 42:69

the energies of scattered photons remain well below *1 MeV. The maximum value of the scattering response peak occurs for incidence photon energy of *40 keV and lies at detected energy of *35 keV. Note that this maximum refers only to the peak in the observed spectrum, and not the total number of detected photons. The response diminishes as one goes from lower incidence angles to higher ones. In Fig. 3, the detected atmospheric responses to incidence photons of different energies are demonstrated for two different angles of incidence with values represented by the color bars. The Y-axis corresponds to the energy of the incident photons, while the X-axis denotes the energy of scattered photons. The color coding shows the intensity at each energy pair. It is obvious that the detected energy is never greater than the incident energy, leaving the upper left part of these figures blank. These plots represent the ARM for various incidence angles, and the curves shown in Fig. 2 correspond to horizontal slices through the plot in top panel of Fig. 3. Values for incident energies other than those used in our simulation grid are obtained by suitable interpolation. For each energy of incidence, we calculate the area under the scattering response curve (Fig. 2) to obtain the atmospheric scattering efficiency, i.e., the total number of scattered photons (at any possible scattered energy) for a single incident photon on Earth’s atmosphere. We converted this in percentage and plot in Fig. 4 as function of incidence energy for various incidence angles. The plots show that the scattering efficiency is maximum for 0 incidence of photons (black curve) and diminishes as the angle of incidence increases. For normal incidence the highest efficiency is at *120 keV, and as Fig. 2 shows, these photons are spread over a broad band of lower energies. The energy at which the atmospheric scattering efficiency is highest tends to gradually increase until the efficiency profile becomes nearly uniform in all energies above *100 keV for large incidence angles. 3.2 Spectral response for GRBs We now discuss the scattering response of the Earth’s atmosphere to an incident GRB spectrum, approximated by a typical Band function (Band et al. 1993). The parameters of the Band function given by Wanderman and Piran (2015) corresponding to a typical short GRB, viz, a ¼ 0:5, b ¼ 2:25, and Epeak ¼ 800 keV are chosen for the calculation. The

J. Astrophys. Astr.

(2021) 42:69

Page 7 of 15

Figure 3. ARM for two angles of incidence. Simulations are carried out at various incidence energies as discussed in Section 2.3, and are interpolated for other energies. The top panel is a two-dimensional version of the plots in Fig. 2, for photons incident at h ¼ 0 , while the bottom panel is the ARM for h ¼ 60 . 40 0 deg 20 deg 40 deg 60 deg 80 deg

Efficiency (%)

30

20

10

0 10

100

1000

Energy (keV)

Figure 4. Efficiency (ordinate) of Compton reflection of the incident photons as function of incidence energy (abscissa) for various incidence angles.

69

scattered component corresponding to the spectrum for 0 incidence is estimated by the convolution of the spectrum with the corresponding response matrix (ARM representing the atmospheric scattering in top of the Fig. 3). We pick the norm for the band function such that the fluence of the GRB is 106 erg cm2 (20–200 keV) over 1 s interval of prominence. In order to disentangle the effects of the ARM from any instrument response effects, we assume that the instrument is an ideal omnidirectional detector: it measures the exact energy of each incident photon, but cannot discern photon incidence directions. In actual analysis of course, the direction- and energydependent response of the instrument would have to be folded into the calculation. Simulation results for the atmospheric scattering component is shown in the upper panel of Fig. 5. The dashed blue curve is the incident band spectrum, that would be detected directly, and the dashed red curve shows the scattered component from the atmosphere. The solid orange curve represents the net detected spectrum by a detector in LEO. We can see that a significant amount of photons scattered by the atmosphere are detected at the intermediate energy range of *30 to 300 keV. This reflected component produces a broad hump around *60 keV on top of the direct Band spectrum. In Table 1, we show the ratio of the detected scattered flux from atmosphere to that of the incident flux integrated over various energy ranges for the Band function corresponding to the GRB used in our study. We see that the number of reflected photons in the 50–100 keV range is as high as 66% of the incident photons, which corresponds to *40% of the total observed photon counts in this energy range. At a glance, the scattered component mimics a blackbody component (Fig. 5: top panel), and would have to be modelled so, if atmospheric scattering effects were not accounted for. As an illustration, we show such a naive analysis where the net observed spectrum can indeed be fit well by a Band ? blackbody spectral model (Fig. 5: bottom panel). To undertake formal fits we simulate an ‘‘observed’’ spectrum consisting of the direct incident and reflected spectra. The GEANT4 output is binned at 1 keV resolution, and multiplied by the effective area of the detector to obtain the expected count rate kE (in units of counts keV1 ) as a function of energy. To account for Poisson noise, we calculate the final observed number of photons in each 1 keV by using a Poisson distribution with rate equal to kE . The uncertainty for

69

Page 8 of 15

J. Astrophys. Astr.

(2021) 42:69

Table 1. Ratio of detected atmospheric scattered flux to that of incidence flux integrated over various energy ranges. Energy range (keV) 50–100 100–200 200–400 400–1000 20–2000

Figure 5. Simulations of reflected spectra (top panel) and xspec fits to the data (bottom panel) for the GRB considered are demonstrated. The dashed blue line in the top panel shows the spectrum of a short GRB with parameters a ¼ 0:5, b ¼ 2:25, and Epeak ¼ 800 keV, which is incident at 0 and detected directly. The dashed red line shows the scattered spectrum. The solid orange line denotes the net spectrum incident on the detector. The bottom panel shows the simulated photon data and the bestfit model obtained from xspec. Black symbols with error bars show the simulated data from the grouped .pha files. Dashed blue and red lines show the best-fit Band (grbm) and blackbody (bbody) components used in the fitting, respectively. The solid orange line is the sum of these two components, which agrees well with data giving residuals consistent with zero. The best-fit parameter values are given in Table 2.

pffiffiffiffiffi each bin is set to kE . We note that in practice, the uncertainties will be higher due to the presence of background counts. Both the spectrum and uncertainty are written into a .pha file with 1 keV channels, spanning the energy range from 10 keV–2 MeV. We

Flux ratio 0.66 0.46 0.08 0.001 0.25

assume an ideal detector of 150 cm2 collecting area, comparable to several operational GRB instruments. We assume that the effective area of the detector is independent of energy, and that the detector reports the true energy of each incident photon without any redistribution. The detector response (.rsp file) is then modeled simply as a diagonal matrix, with diagonal elements equal to the effective area. We then follow usual X-ray data analysis procedures. We run the grppha FTOOL on the .pha file to create channel bins having at least 30 photons each. We then attempt to fit the grbm spectral model to this simulated data in xspec. As expected, the fits are of poor quality due to the presence of the reflection hump in the data. We then model the data with grbm ? bbody, and obtain acceptable fits (Table 2, Fig. 5: bottom panel). In this naive analysis, where the fitting process ignores the atmospheric scattering effects, we find that the best-fit blackbody temperature is kT  40 keV. The spectral parameters a and b are satisfactorily recovered. The xspec grbm model parameterizes the Band function in terms of the characteristic energy Ec , which is related to the peak energy as Our simulation thus uses Ep ¼ Ec ð2 þ aÞ. Ec ¼ 533 keV. We find that the best-fit value is consistent with the simulation, albeit with large error bars. We attribute these uncertainties to small number statistics: with our selected parameter values, we have *1000 photons, with only a small fraction of those incident at energies above Ec . To put these results in perspective by comparing to a real detector, we proceed to measure the possible contribution of both components to the flux in the AstroSat-CZTI 20–200 keV band. The flux of individual components is calculated by using the cflux model in xspec. We find that the best-fit black body contributes about a third of the total flux in the energy range of interest (Table 3).

J. Astrophys. Astr.

(2021) 42:69

Page 9 of 15

Table 2. Best-fit values for all the parameters with their 1-r error and the reduced v2 of the fit presented in bottom panel of Fig. 5. Description of the parameters: a and b are Band spectral indices, Ec is the characteristic energy; related to peak energy Ep by Ep = Ec ð2 þ aÞ. T represents the temperature of the black body where k is the Boltzmann constant. Both Ec and kT are expressed in keV. norm is the spectral norm for each of the components of the model and is expressed in photons cm2 s1 keV1 . Model name

Parameter

Band

a b Ec ðkeVÞ norm=103 kT ðkeVÞ norm

Black body v2 (DoF)

69

Table 3. Comparing the contributions of incident and scattered components in the total GRB flux measured by a detector like AstroSat-CZTI. Flux values are reported in the 20–200 keV range, in units of 107 erg cm2 s1 . Numbers in parenthesis indicate the percentage contribution to the total flux. Fluxes of individual components are measured by using the cflux model component in xspec. Error bars denote and 90% confidence intervals. The relative contributions of the individual components to the total flux are mentioned as (% total).

Value - 0.51 ± 0.06 - 2.4 ± 0.5 560 ± 83 40.3 ± 3.6 39.7 ± 3.4 8.6 ± 1.6 81 (76)

In practice, of course, analyses of GRB spectra account for the atmospheric contribution – often transparently to the end users. For example BATSE and Fermi have introduced the correction in their analysis step by convolving ARM to the Detector Response Matrix (DRM) (Pendleton et al. 1999; Meegan et al. 2009). The ARM is obtained primarily by performing extensive Monte Carlo simulation of interaction of incident photons with the atmosphere and then improving on it gradually with some observation of the same for X-rays from solar flares in an iterative manner. However, It is obvious from above exercise, that the unchecked atmospheric contribution can be misinterpreted as an added thermal component during fitting of observed counts, though the odd of it to actually happen is small, as most, if not all the missions already or will have the provision of mitigating the effect through the inclusion of ARM. the similarity of this effect to blackbody spectra raises the question of the susceptibility of models to errors in the ARM. We explore this further in Section 3.3 by testing the effects of *10% errors in the ARM on the spectral analysis. We note that throughout this work, we consider the ideal detector case. In practice, the effect of such errors on the final fit will depend on the directiondependent sensitivity of the instrument. In a case where the instrument has higher on-axis sensitivity as compared to off-axis sensitivity, the reflected component will be weaker if the instrument was pointing

20–200 keV flux

Flux (107 erg cm2 s1 ) Fraction of total

Band

Black body

Total

9:9þ1:0 1:0 65%

5:4þ1:4 1:3 35%

þ1:78 15:31:6 100%

to the GRB, and stronger if the instrument was pointing towards the Earth. An illustration of the latter case is when the primary target of CZTI is almost occulted by the Earth, when the boresight points straight to the atmosphere. In such ‘‘strong reflection’’ cases, even smaller errors in the ARM can cause large changes in the inferred parameters.

3.3 Effects of imcorrect ARM In order to demonstrate how systemic errors in the ARM can affect the modelled spectrum, we synthesise three scenarios. In the first one, we show that we can recover the true incident spectrum by accurately accounting for the ARM (100%). In the other two, we introduce a 10% error to the ARM and then use these incorrect response matrices as the atmospheric responses to demonstrate the effect of improper modelling of atmospheric response. We examine the three scenarios for two types of sources (incident at 0 angle): a GRB without a blackbody component (Section 3.3.1), and a GRB with an intrinsic blackbody component (Section 3.3.2). In order to get enough photon statistics, we consider a GRB with 50 s duration. The spectral parameters for the Band component are same as in Section 3.2. We simulate observations using the same methods as discussed in Section 3.2, hence the observed spectrum consists of the incident GRB spectrum as well as the photons scattered from the atmosphere. While analysing the data for both source models, we include the ARM as per the three cases: the exact

69

Page 10 of 15

J. Astrophys. Astr.

(2021) 42:69

Figure 6. Simulations of the total GRB spectrum (incident?scattered) and the corresponding xspec fits to the simulated data for Band incidence. Each column corresponds to a different response matrix (Atmospheric Response Matrix (ARM) convolved with the detector response matrix (DRM)). The first column represents the accurate ARM and the next two columns correspond to ARM which is under and over estimated by 10%. The incident spectrum is that of a GRB modelled by a band function with parameters a ¼  0:5, b ¼  2:25, and Epeak ¼ 800 keV (Ec ¼ 533 keV) and norm = 0.0427. The simulated data are marked in black (with error bars overlaid). The solid orange line denotes the xspec fit for the simulated data. Two models are used to fit the data. Model 1 (m1) is the band function (grbm) and model 2 (m2) is band function (grbm) with an additive blackbody (bb) component. The top row shows the xspec fits to the simulated data with m1 represented in blue and m2 depicted by red colour (changes in the band component are difficult to observe directly when plotted in the absolute scale and hence successive residual plots are presented to depict the effect of the two models considered). The second row from top shows the residual plot (blue) when the data is fit with m1. The third row from top shows the residuals (red) when m2 is used to fit the data. The bottom row shows the residuals (green) of the grbm component of model m2. These are obtained by fitting the whole model m2 to the data and then deleting the bb component, thus highlighting the presence of a residual black body (if any), with the black body component itself overlaid marked as ‘trend’. The best-fit parameter values are given in Table 4.

ARM, an underestimated ARM (corresponding to 90% reflection strength), and an overestimated ARM (corresponding to 110% reflection strength): for simulating the 10% systematic errors.

3.3.1 Band spectral model. In the ‘‘Pure Band’’ incidence case, the source spectrum is represented by a Band function, with the parameters a ¼ 0:5, b ¼  2:25, Epeak ¼ 800 keV (Ec ¼ 533 keV) and

J. Astrophys. Astr.

(2021) 42:69

Page 11 of 15

69

Table 4. Best-fit values for all the parameters with their 1-r errors and the reduced v2 of the fits as described in Section 3.3.1 (pure band incidence) for all the three cases mentioned in the Section 3.3. The two models considered here are grbm and bb. Model m1 corresponds to a pure grbm spectrum while model m2 reflects an additive bb component in addition to the grbm (consistent with the Fig. 6). Description of the parameters: a and b are rising and decaying spectral indices, Ec is the characteristic energy; related to peak energy Ep by Ep ¼ Ec ð2 þ aÞ. T represents the temperature of the black body where k is the Boltzmann constant. Both Ec and kT are expressed in keV. norm is the spectral norm for each of the components of the model and is expressed in photons cm2 s1 keV1 . 100% ARM

90% ARM

110% ARM

Best-fit values Model

Parameter

m1

m2

m1

Band

alpha beta Ec (keV) norm/103 kT (keV) norm

–0.5 ± 0.01 –2.27 ± 0.04 542 ± 12 42.5 ± 0.25 – – 560 (550)

–0.52 ± 0.02 –2.32 ± 0.05 575 ± 28 40.3 ± 0.9 67.2 ± 15.1 1.3 ± 0.7 554 (548)

–0.54 ± 0.12 –2.27 ± 0.04 553 ± 12.5 44.1 ± 0.25 – – 566 (550)

Blackbody v2 (DoF)

norm = 0.0427. Analysis results are shown in Figure 6, with best-fit parameters given in Table 4. The topmost panels show the simulated data in gray, along with the folded model in orange. The three columns show the cases where the analysis uses the correct ARM, an underestimated ARM and an overestimated ARM from left to right respectively. In real analysis, a user would not have prior knowledge of the exact nature of the source spectrum. Hence, we fit two models to the data in each case: model m1, the ‘‘pure Band’’ model (‘‘grbm’’), and model m2 (‘‘grbm?bb’’), which incorrectly assumes the presence of a blackbody component. The second row shows residuals (blue) after fitting the spectrum with model m1, while the third row shows residuals (red) after fitting the spectrum with model m2. The bottom row (green) is the residuals corresponding to the grbm component of m2, obtained after fitting the spectrum with the full model m2, then deleting the bb part (similar to plots used to show line features in X-ray analyses). It can be seen from Table 4 (and Fig. 6) that when 100% ARM is considered, we accurately recover the incident band parameters. Further if we try to include a bb component in this case, the v2 value decreases only marginally by 6 for 2 additional Degrees of Freedom (DoF) from model m1 to model m2. On the other hand, when we use an underestimated ARM, the best-fit (with m1) band component appears softer (a decreases) as compared to the incident pure band

m2 –0.52 –2.32 572 41.0 52.3 1.43 542

± 0.02 ± 0.05 ± 22 ± 0.8 ± 7.8 ± 0.4 (548)

m1

m2

–0.47 ± 0.01 –2.28 ± 0.04 531 ± 11 40.9 ± 0.24 – – 560 (550)

–0.53 ± 0.04 –2.33 ± 0.06 601 ± 48 38.2 ± 1.35 94.3 ± 13.7 2.5 ± 1.6 555 (548)

spectrum. Addition of a black body component (using model m2) improves the quality of the fit with prominent blackbody component being ‘‘detected’’ in data, though the source model is a pure band spectrum (Fig. 6, middle column). In this case, the v2 value decreases by 24 for two additional DoFs. This additive model component is shown by the solid red line in the top panel, and also by the green ‘‘grbm residuals (m2)’’ sub-plot. The error estimate on the blackbody norm indicates a statistically significant fit — arising out of ignored/unknown systematic errors in the ARM. Lastly, when we use overestimated ARM, the best-fit (with m1) band component appears harder (a increases) as compared to the injected source spectrum. If we directly compared this with the source spectrum, we would have seen negative residuals (not shown here) as the ARM used in fitting over-predicts the observed spectrum. Adding a blackbody component to this should imply that we are trying to fit a negative one. Even then, if we try to force-fit such a model to the data, we get a fit with the band component becoming much softer and unrealistic realizations for the blackbody parameters are obtained (evident from the unusually large kT value and large error in the norm of the balckbody in Table 4). 3.3.2 Band and blackbody spectral model. We now consider a case where the GRB intrinsically has a blackbody component. We simulate a source spectrum with the same band function as before, and an additive

69

Page 12 of 15

J. Astrophys. Astr.

(2021) 42:69

Figure 7. Simulations of the total GRB spectrum (incident?scattered) and the corresponding xspec fits to the simulated data for Band ? Blackbody incidence. Each column represents a different response matrix (Atmospheric Response Matrix (ARM) convolved with the detector response matrix (DRM)). The first column represents the accurate ARM and the next two columns correspond to ARM which is under and over estimated by 10%. The incident spectrum is a short GRB modelled as a band function with parameters a ¼ 0:5, b ¼ 2:25, and Epeak ¼ 800 keV (Ec ¼ 533 keV) and norm = 0.0427 and an additive black body component with kT = 40 keV and norm = 5. The simulated data are marked in black (with error bars overlaid). The solid orange line denotes the xspec fit for the simulated data. The data is modelled by a band function (grbm) with an additive blackbody (bb) component. The top row shows the xspec fits to the simulated data with the fitting model (grbm ? bb) shown in red. The second row (from top) shows the residual plots (red) when grbm ? bb model is used to fit the data. Although all the residuals look similar they do exhibit differences and the differences are small compared to the value of the residuals themselves. The bottom row shows the residuals (green) of the additional grbm component of the model. These are obtained by fitting the model to the data and then deleting the incident spectrum, thus highlighting the presence of an additional residual black body (if any), with the additional contribution from the black body component itself overlaid marked as ‘trend’. The best-fit parameter values are given in Table 5.

blackbody characterised by kT ¼ 40 keV and norm ¼ 5. The temperature and relative intensity of this component are consistent with typical blackbody components detected in some GRBs. Since the source has an intrinsic blackbody, fits with a pure band function always give poor results and we do not disucuss those. Instead, we focus our attention on modelling the spectrum with the m2 model comprising of the band spectrum and a blackbody (Fig. 7, Table 5). As before, we find that if the

ARM is correct, then the parameters are recovered well. However, any errors in the ARM will alter the values: for instance, if the ARM is underestimated, we find a brighter blackbody (see Table 5 and bottom row of the column at the middle of Fig. 7), and conversely if the ARM used in the analysis overestimates the Earth reflection component, the inferred blackbody is fainter than the real value (shown by the plot at bottom-right of Fig. 7). The best-fit values for the model parameters are shown in Table 5. It can be seen from Table 5

J. Astrophys. Astr.

(2021) 42:69

Page 13 of 15

69

Table 5. Best-fit values for all the parameters with their 1-r errors and the reduced v2 of the fit as described in Section 3.3.2 (Band and bb incidence) for all the three cases mentioned in the Section 3.2. The model considered here is grbm with an additive bb (consistent with the Fig. 7). Description of the parameters: a and b are rising and decaying spectral indices, Ec is the characteristic energy; related to peak energy Ep by Ep = Ec (2 ? a) of the Band function. T represents the temperature of the black body where k is the Boltzmann constant. norm is the spectral norm for each of the components of the model and is expressed in photons cm2 s1 keV1 . Model

Parameter

Incident

Band

alpha beta Ec (keV) norm/103 kT (keV) norm

-0.5 -2.25 533 42.7 40 5 –

Blackbody v2 (DoF)

100% ARM

90% ARM Best-fit values

110% ARM

-0.54 ± 0.04 -2.26 ± 0.04 562 ± 23 43.4 ± 1.1 40.8 ± 1.5 4.7 ± 0.4 528 (548)

-0.55 ± 0.04 -2.26 ± 0.05 567 ± 24 43.6 ± 1.0 40.2 ± 1.3 5.31 ± 0.37 530 (548)

-0.54 ± 0.04 -2.26 ± 0.04 557 ± 23 43.3 ± 1.0 41.4 ± 1.8 4.08 ± 0.37 528 (548)

that the best-fit values of the band spectrum are apparently the same in all the cases, with the error in ARM reflected mostly in the bb component (can be seen from the norm of the blackbody component).

4. Conclusion and future work In this study, we have presented some of our simulation outcomes helpful in better understanding the nature of the Earth’s atmospheric reflection of X-ray/ c-ray photons from transient space objects and how can it affect the observation. We are in the process of developing a comprehensive and open-source simulation framework for estimating the atmospheric scattered component to be used in the construction of a publicly available database on Atmospheric Response Matrix (ARM). This particular study concentrates on contemplating the significance of proper evaluation of the atmospheric reflection component in terms of its effect on prompt GRB spectral analysis. In some upcoming studies we intend to perform simulations to find out the influence of Earth’s atmospheric scattering on the determination of polarization of high energy photons from such sources. We find that the Earth’s atmosphere is quite effective at scattering incident X-rays and c-rays, creating an overall reflected spectrum that is usually strong in the 30–300 keV region. The GRB fireball model (Paczynski 1986; Goodman 1986; Lundman et al. 2012) predicts thermal components in GRB spectra. Such components have been identified in a multitude of GRBs (for instance, Agostinelli et al.

2003; Ryde 2005; Guiriec et al. 2011, 2013; Iyyani et al. 2013, 2015; Axelsson et al. 2012; Burgess et al. 2014; Nappo et al. 2017), and their temperatures are similar to the ones obtained in our fits above. The simplified example in Section 3.2 demonstrates how Earth’s reflection can mimic such a thermal component. While this component is typically modelled in analysis, Section 3.3 shows how even small systematic errors in the estimation of atmospheric reflection can alter our interpretations of GRB data. The problem may be further exacerbated when accounting for the direction-dependent sensitivity of the detector instrument. For instance, the effective area of AstroSat-CZTI varies by more than two orders of magnitude over the entire sky (Mate et al. 2021). If the GRB were to be incident from a lowsensitivity direction while the scattered radiation arrives from a high-sensitivity direction for the satellite, the scattered component can be an order of magnitude larger than the incident component — further magnifying the effect of any uncertainties in the scattering model. For typical observations for which one needs smaller contribution of thermal (blackbody) flux to account for the hump in the intermediate range of prompt GRB spectra, extra care should be taken to exclude any contribution from atmospheric scattering. For example, for GRB110721A (Axelsson et al. 2012; Iyyani et al. 2013), blackbody is found to contribute maximum of *10% of total flux, for GRB1000724B (Guiriec et al. 2011) this is found to be *4%. In these cases a slight over/under estimation of the scattering contribution may impose considerable artefacts in the

69

Page 14 of 15

analysis leading to wrong estimation of the thermal component. This underscores the need for having robust ARM calculations that can be used in any spectral analysis. Likewise, such calculations are also needed for localisation and polarisation analyses. The calculation of the reflected spectrum from the Earth involves several complexities like creating an accurate description of the atmosphere, accounting for all physical effects, and generating a large number of incident photons to get good statistics. Approximations have to be made at each stage to make the problem tractable, and each approximation may introduce systematic errors in the final answer, which can have significant impacts on the scientific outcomes. It is not feasible for each mission to dedicate significant resources and undertake simulations to calculate the ARM, nor should it be necessary. Instead, it is important that a few groups independently develop response matrices, and compare their results to quantify accuracy and reliability of the answers. Further, end users should be made aware of these limitations, and inclusion of appropriate systematic errors in analysis may be recommended. Going further, there are several improvements we intend to undertake. Firstly, there is a direct tradeoff between using a large detector for collecting sufficient number of photons, and the resultant large differences in the scattering angles of photons that are incident on opposite ends of the detector. We will explore more simulation geometries, including multiple-detector scenarios, that will allow us to more effectively utilise all data generated in a simulation. Second, the current simulations were carried out over a relatively coarse energy grid, at a certain incidence angle, with a modest number of photons. We plan to undertake convergence studies to determine the step sizes required in energies and angles to ensure reliable simulation results. We will also quantify the relationship between the number of photons in each monochromatic beam and the statistical uncertainty in the final reflected spectrum. Once these requirements are clearly defined, we will undertake simulations to create a database that can be re-used for all further calculations. The final source codes and outputs of our simulations will be made available publicly for scrutiny and reuse, in a format that may be easily repurposed for other missions.

J. Astrophys. Astr.

(2021) 42:69

Acknowledgements CZT–Imager is built by a consortium of Institutes across India. The Tata Institute of Fundamental Research, Mumbai, led the effort with instrument design and development. Vikram Sarabhai Space Centre, Thiruvananthapuram provided the electronic design, assembly and testing. ISRO Satellite Centre (ISAC), Bengaluru provided the mechanical design, quality consultation and project management. The Inter University Centre for Astronomy and Astrophysics (IUCAA), Pune did the Coded Mask design, instrument calibration, and Payload Operation Centre. Space Application Centre (SAC) at Ahmedabad provided the analysis software. Physical Research Laboratory (PRL) Ahmedabad, provided the polarisation detection algorithm and ground calibration. A vast number of industries participated in the fabrication and the University sector pitched in by participating in the test and evaluation of the payload. The Indian Space Research Organisation funded, managed and facilitated the project. Sourav Palit wants to thank Indian Institute of Technology Bombay (IITB) for providing the scholarship and necessary resources to be able to perform this study.

References Agostinelli S. et al. 2003, Nucl. Instrum. Methods. A, 506, 250 Axelsson M., Baldini L., Barbiellini G. et al. 2012, The Astrophys. J., 757, L31 Band D., Matteson J., Ford L. et al. 1993, ApJ, 413, 281 Bhalerao V., Kasliwal M., Bhattacharya D. et al. 2017a, Astrophys. J., 845, http://arxiv.org/abs/1706.00024 Bhalerao V., Bhattacharya D., Vibhute A. et al. 2017b, J. Astrophys. Astr., 38, 31 Bisnovatyi-Kogan G. S., Blinnikov S. I. 1977, A&A, 59, 111 Burgess J. M., Preece R. D., Connaughton V. et al. 2014, The Astrophys. J., 784, 17 Chattopadhyay T., Falcone A. D., Burrows D. N., Fox D. B., Palmer D. 2018, arXiv:1807.03333 Chattopadhyay T., Vadawale S. V., Aarthy E. et al. 2019, ApJ, 884, 123 Churazov E., Sazonov S., Sunyaev R., Revnivtsev M. 2008, MNRAS, 385, 719 Connaughton V., Briggs M. S., Goldstein A. et al. 2015, The Astrophys. J. Suppl. Ser., 216, 32 Cordier B., Wei J., Atteia J. L. et al. 2015, arXiv e-prints, arXiv:1512.03323 Cullen D. E. 1995, Nuclear Instrum. Methods in Phys. Res. Section B, 101, 499

J. Astrophys. Astr.

(2021) 42:69

Fuschino F., Campana R., Labanti C. et al. 2019, Nuclear Instrum. Methods in Phys. Res. A, 936, 199 Goodman J. 1986, ApJ, 308, L47 Guiriec S., Connaughton V., Briggs M. S. et al. 2011, The Astrophys. J., 727, L33 Guiriec S., Daigne F., Hascoe¨t R. et al. 2013, The Astrophys. J., 770, 32 Hedin A. E. 1991, J. Geophys. Res.: Space Phys., 96, 1159 Hoover A. S., Kippen R. M., Meegan C. A. et al. 2005, Nuovo Cimento C Geophys. Space Phys. C, 28, 797 Hulsman J. et al. 2020, Proc. SPIE Int. Soc. Opt. Eng., 11444, 1 Iyyani S., Ryde F., Axelsson M. et al. 2013, MNRAS, 433, 2739 Iyyani S., Ryde F., Ahlgren B. et al. 2015, MNRAS, 450, 1651 Kippen R. M., Hoover A. S., Wallace M. S. et al. 2007, AIP Conference Proceedings, 921, 590 Klein O., Nishina T. 1929, Zeitschrift fur Physik, 52, 853 Lundman C., Pe’er A., Ryde F. 2012, MNRAS, 428, 2430 Mate S., Chattopadhyay T., Bhalerao V. et al. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036-02109763-x McConnell M. L., Baring M. G., Bloser P. F. et al. 2017, in AAS/High Energy Astrophysics Division, Vol. 16, 103.20 Meegan C., Lichti G., Bhat P. N. et al. 2009, The Astrophys. J., 702, 791 Nappo F., Pescalli A., Oganesyan G. et al. 2017, A&A, 598, A23 Paczynski B. 1986, ApJ, 308, L43 Palit S., Basak T., Mondal S. K., Pal S., Chakrabarti S. K. 2013, Atmos. Chem. Phys., 13, 9159 Palit S., Raulin J., Szpigel S. 2018, J. Geophys. Res.: Space Phys., 123, 10, 224

Page 15 of 15

69

Pendleton G. N., Briggs M. S., Kippen R. M. et al. 1999, ApJ, 512, 362 Racusin J., Perkins J. S., Briggs M. S. et al. 2017, arXiv e-prints, arXiv:1708.09292 Rao A., Bhattacharya D., Bhalerao V., Vadawale S., Sreekumar S. 2017, Curr. Sci., 113, https://doi.org/10. 18520/cs/v113/i04/595-598 Ryde F. 2005, The Astrophys. J., 625, L95 Sazonov S., Churazov E., Sunyaev R., Revnivtsev M. 2007, MNRAS, 377, 1726 Sempau J., Ferna¨ndez-Varea J., Acosta E., Salvat F. 2003, Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms, 207, 107 Sharma Y., Marathe A., Bhalerao V. et al. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036021-09714-6 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, in Proc. SPIE, Vol. 9144, Space Telescopes and Instrumentation 2014: Ultraviolet to Gamma Ray, 91441S Wanderman D., Piran T. 2015, MNRAS, 448, 3026 Werner N., Pa´l A., Ohno M. et al. 2018, in den Herder J.W. A., Nakazawa K., Nikzad S., eds, Space Telescopes and Instrumentation 2018: Ultraviolet to Gamma Ray, (SPIE), 96 White T. R., Lightman A. P., Zdziarski A. A. 1988, ApJ, 331, 939 Willis D. R., Barlow E. J., Bird A. J. et al. 2005, A&A, 439, 245 Zhang D., Li X., S. Xiong et al. 2019, Nuclear Instrum. Methods in Phys. Res., Section A, 921, 8

 Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:68 https://doi.org/10.1007/s12036-021-09750-2

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

Characterisation of cosmic ray induced noise events in AstroSat-CZT imager D. PAUL1,2, A. R. RAO1,3,* , A. RATHEESH1,4,5, N. P. S. MITHUN6,

S. V. VADAWALE6, A. VIBHUTE3, D. BHATTACHARYA3, P. PRADEEP7 and S. SREEKUMAR7 1

Tata Institute of Fundamental Research, Homi Bhabha Road, Colaba, Mumbai 400 005, India. Gubbi Labs LLP, No: 2-182, 2nd Cross, Extension, Gubbi, Tumkur 572 216, India. 3 Inter University Centre for Astronomy and Astrophysics, Post Bag 4, Ganeshkhind, Pune 411 007, India. 4 Department of Physics, Tor Vergata University of Rome, Via della Ricerca Scientifica 1, 00133 Rome, Italy. 5 INAF-IAPS, Via Fosso del Cavaliere 100, 00133 Rome, Italy. 6 Physical Research Laboratory, Navrangpura, Ahmedabad 380 009, India. 7 Vikram Sarabhai Space Centre, Kochuveli, Thiruvananthapuram 695 022, India. *Corresponding Author. E-mail: [email protected] 2

MS received 7 November 2020; accepted 12 April 2021 Abstract. The Cadmium Zinc Telluride (CZT) Imager onboard AstroSat consists of pixelated CZT detectors, which are sensitive to hard X-rays above 20 keV. The individual pixels are triggered by ionising events occurring in them, and the detectors operate in a self-triggered mode, recording each event separately with information about its time of incidence, detector co-ordinates, and channel that scales with the amount of ionisation. The detectors are sensitive not only to photons from astrophysical sources of interest, but also prone to a number of other events like background X-rays, cosmic rays, and noise in detectors or the electronics. In this work, a detailed analysis of the effect of cosmic rays on the detectors is made and it is found that cosmic rays can trigger multiple events which are closely packed in time (called ‘bunches’). Higher energy cosmic rays, however, can also generate delayed emissions, a signature previously seen in the PICsIT detector on-board INTEGRAL. An algorithm to automatically detect them based on their spatial clustering properties is presented. Residual noise events are examined using examples of Gamma Ray Bursts as target sources. Keywords. Space vehicles: instruments—instrumentation: detectors—X-rays: detectors—X-rays: analysis.

1. Introduction AstroSat is a broad band high energy Indian mission covering UV, soft X-rays and hard X-rays (Singh et al. 2014; Rao et al. 2016a). It comprises four co-aligned instruments to cover a wide bandwidth: Ultra Violet Imaging Telescope/UVIT (Subramaniam et al. 2016; Tandon et al. 2017a, b; Rahna et al. 2017), Soft X-ray Telescope/SXT (Singh et al. 2016, 2017), Large Area

This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

X-ray Proportional Counter/LAXPC (Agrawal et al. 2017; Antia et al. 2017; Yadav et al. 2017) and Cadmium Zinc Telluride (CZT) Imager/CZTI (Bhalerao et al. 2017). CZTI is a hard X-ray instrument sensitive in the energy range 20–200 keV, consisting of an array of CZT detectors. Each detector module consists of 256 independent detectors, called pixels, of nominal size 2.5 mm  2:5 mm and 5 mm thickness. The CZT plane consists of four quadrants, each with 16 detector modules, comprising an effective area of 1024 cm2 . CZTI has imaging capabilities below 100 keV, using the Coded Aperture Mask (CAM) placed above collimator slats that surround the detector modules. The collimator

68

Page 2 of 14

is made of tantalum of size 4 cm  4 cm, which allows for a field of view (FoV) of 4:6  4:6 . In addition to spectroscopic, timing, and localization capabilities (Rao et al. 2016b), CZTI can measure polarization in the hard X-rays with exposure time an order of magnitude smaller than previously existing instruments (Chattopadhyay et al. 2019). Further, CZTI acts as an all sky monitor above  100 keV enabling it to do spectro-polarimetric studies of gamma-ray bursts (Rao et al. 2016b), making it a unique instrument at these energies. The details of CZTI including overall instrument configuration, the detectors and electronics, the data characteristics, processing pipeline and default products have been given in Bhalerao et al. (2017). In CZTI terminology, an ‘event’ is a trigger in any of the pixels, characterized by a unique time-stamp, the pixel co-ordinates, and the ‘pulse height amplitude’ or PHA, which is linearly related to the amount of ionisation in the triggered pixel. The cause of ionisation is mainly due to X-ray photons from the targeted X-ray sources along with the background X-rays (cosmic X-ray background as well as locally generated by cosmic rays). In addition to this, ionisation is also caused by interactions of charged particles in the detector. As CZTI is a hard X-ray instrument, for most sources including the Crab Nebula, the number of background events are much more than the source events. These background events are mostly induced by the charged particle environment of the detector, which, other than showing predictable variation with the satellite location, should show a uniform random distribution in spatial and temporal dimensions. Direct interaction of charge particles in the detector can cause a heavy ionisation resulting in multiple events and events may also be generated by peculiarities in the detector. These additional events (other than those caused by the interaction of X-ray photons) are called ‘noise’ events and they need to be identified and removed before doing any scientific investigation of the events from the target source, both in the temporal and energy domains. The presence of cosmic ray induced noise events was deduced during the first few days of the mission, and an algorithm is implemented in the ground analysis software (hereafter called the CZTI pipeline) to eliminate them from the science data. They are easily distinguished from science events due to their temporal characteristics: they all ‘bunch’ together within the smallest interval of time resolvable by the CZT detector electronics, 20 ls (Bhalerao et al. 2017). A steady stream of these ‘bunches’, defined as three or more events with a time separation between successive

J. Astrophys. Astr. (2021)42:68

events no more than one time stamp (20 ls), are observed in the data and they are understood as due to cosmic rays continuously bombarding the detector plane. The number of such bunches that should occur from background X-rays by chance is 103 bunches per second whereas about 70 bunches per second are seen in the data. These bunches temporally track the variation of the cosmic rays bombarding the entire satellite, independently measured by the Charged Particle Monitor (CPM) on-board AstroSat (Rao et al. 2017). It was also known, before the launch of the satellite, that some of the pixels could be prone to electronic noises. The CZTI pipeline identifies these pixels, removes the events from these pixels from further analysis, keeping track of the effects on live time and effective area. If any pixel shows a consistently noisy behaviour, they are permanently disabled by a ground command. The subtle effects of a pixel becoming electronically noisy immediately after the bombardment of a high energy charge particle in that pixel is also taken care of in the current CZTI pipeline. Hence the pipeline removes both types of noise viz., charge particle induced and pixel misbehaviour: the former by removing the bunches and the post-bunch noise events and the latter by identifying the misbehaving pixels in a given observation and completely eliminating data from these pixels. In this work, we present the results of an investigation of the noise events in CZTI. In particular, we have examined whether all cosmic ray induced events could be identified by temporal bunching alone. It was found that the heavy deposition of charge in the detector modules by very high energetic particles can be identified via patterns on the detector modules that are characteristic of pixelated detectors collecting data at such high time resolutions. A new algorithm is developed which automatically identifies and removes these events from the data. The manifestation of cosmic rays as spatial structures in the data is investigated in the next section, via an algorithm detailed in the Appendix. In Section 3, we discuss these spatial structures as manifestations of higher energy cosmic rays. In Section 4, concluding remarks are given along with some test results.

2. Examination of spatial structures in the data For the analysis presented here, we use the astronomer-friendly ‘Level 2’ (L2) FITS files created by the Payload Operation Centre (POC) of CZTI, located in the Inter-University Centre for Astronomy &

J. Astrophys. Astr. (2021)42:68

Astrophysics (IUCAA) in Pune, India. The analysis employs the standard procedure outlined in the CZTI pipeline User Guide available at the AstroSat Science Support Cell (http://astrosat-ssc.iucaa.in). However, we have also used the ‘Level 1’ FITS files to repeat the exercise for some of the data sets. The details of the data structure and this ground data analysis pipeline can be found in Ratheesh et al. (2021).

2.1 Data preparation The CZTI pipeline uses several parameters to clean the data. The major thrust of cleaning the cosmic ray induced noise events in the data is based on the observation that long lingering noise events can occur in certain pixels after the incidence of cosmic rays, particularly for the higher energy cosmic rays. Further, events from certain pixels that show abnormal behaviour throughout the observations (gross noisy pixels) or at specific times (flickering pixels) are typically discarded for the length of the observation or for specific durations, respectively. For the present work, we make a conservative use of the event removal so that we can attempt a spatial identification of the residual noise. We therefore remove only the events from the gross noisy pixels and retain the rest of the data. The understanding behind the idea of these temporal bunches is that each bunch is created by one cosmic ray particle generating a series of electronic events within timescales shorter than the instrumental resolution of 20 ls. It is to be noted that the 20 ls time bin is digitally generated but the onboard analogue electronics is capable of recording events at a much faster time scale (  1 ls) and then serially transmit the events. If such is the case, then the time of occurrence of respective bunches is independent of each other, and the interval between one bunch and the next, DT, is expected to follow a flat temporal distribution. When plotting the histogram of DT using the bunch data, it is clearly seen that such is not the case (Fig. 1(a)). This observation leads one to assume that the effect of cosmic rays sometimes can last longer than the instrumental time resolution of 20 ls, used for defining bunches. Hence, we redefine bunches such that, if the interval between one bunch and the next is less than a certain duration t2 , then these two bunches are understood to be created by the same cosmic ray particle, and all data of these two bunches as well as data within the two are considered to belong to a single bunch. Empirically, it is seen that the sharp

Page 3 of 14

68

spike observed in Fig. 1(a) is removed by choosing t2 ¼ 60 ls (see Fig. 1(b)). It is observed that about 10% of the bunches are added together by this method. To remove the effects of cosmic-ray particles that persist for some amount of time post-bunch after initially triggering a series of events in the detector, we also discard events up to time t3 after the end of a bunch. In Fig. 2, a we have plotted the co-added light curves of events after the bunches. That is, we generate a lightcurve of events from the end of a bunch to the beginning of the next bunch. We then repeat this for all the bunches, and redefine the start time for each lightcurve as the end of the first bunch. Then, we add all these data, which we refer to as co-adding. Although each bunch lightcurve can be of different lengths, the process of co-adding takes care of the different durations. This is because the resultant lightcurve for each bunch is renormalised by the amount of live time available for the bin. In Fig. 2(a), we show this re-normalised post bunch light curve. If the data is fully Poissonian, then it should vary randomly about unity, for all times. In Fig. 2(a), we see a sharp rise and fall at the smallest time interval, which is indicative of remnant effects even after co-adding the bunches. On choosing t3 ¼ 60 ls and removing data for t3 after each bunch, the co-added light curve becomes flat around unity (see Fig. 2(b)). Hence, all gross features of the cosmic ray induced events can be removed by assuming that the post bunch effect lasts for t3 , although subtle after-effects of bunches at longer time scales cannot be completely ruled out.

2.2 DPHclean In the light curves made from the data after the procedure described above, strong temporary features, on a time scale of hundred milliseconds, are observed. To investigate this further, we examine the detector plane histograms (DPHs) of the events that create these features. DPH, for a given length of time, is created by adding together all counts registered in a pixel and showing all pixels in a geometric plane, with appropriate colour coding for the number of events. DPH created for a full observation is used to identify noisy pixels and a series of DPH, created every 100 ls, shown as a movie (such a movie can be seen in the CZTI website1), highlights the impact of cosmic rays 1

http://astrosat.iucaa.in/czti/?q=czti_movie.

68

Page 4 of 14

J. Astrophys. Astr. (2021)42:68

a)

b) 1000

10000

800

8000

6000

600

4000

400

2000

200

0

10

20

30

40

50

60

70

80

0

90

10

20

30

40

50

60

70

80

90

ΔT [20 μs]

ΔT [20 μs]

Figure 1. A histogram of the number of bunches within time DT of a preceding bunch. (a) The histogram peaks at small DT showing that some bunches are connected to the subsequent bunches. The bunches are redefined as a single bunch if the time separation is less than t2 . (b) With t2 ¼ 60 ls, this initial peak is removed. a)

b)

1.7 1.12

Coadded Data

Coadded Data

1.6

1.5

1.4

1.3

1.10

1.08

1.06

1.04

1.2

10

20

30

40

50

60

70

80

90

τ [20 μs]

10

20

30

40

50

60

70

80

90

τ [20 μs]

Figure 2. (a) The co-added light curve of events immediately after a bunch, normalised to the exposure time. A sharp spike is seen at timescales less than 60 ls. (b) After removing all data events after time t3 ¼ 60 ls, post bunches, the coadded light curve looks flat, all the way up to 2 ms. All errors assume that the data are Poissonian.

in CZTI data. The DPH created for 100 ms, when a spike in the light curve is observed (see for example Fig. A4 in the Appendix), shows that the excess events cluster in some parts of the detector plane rather selectively. The timescale for detecting such clustering is examined, parametrized by tlook , the bin size for the light curve generation. Initially, such spatial clustering is observed to be present for 5-r outliers in light curves binned at 100 ms. The events that contribute to the clustering are spread over timescales less than 100 ms, and only very rarely involve two consecutive bins of 100 ms. We have developed an

algorithm called ‘DPHclean’ for the automatic identification of such spatial clustering in the DPHs, henceforth called ‘DPHstructures’. This algorithm is specifically suited for CZTI data where (a) the disturbance due to cosmic ray interaction is a few pixels wide, (b) a limited number of pixels are affected by the interaction (c) the geometric regions where the disturbance occur could be a few and (d) the time scale of disturbance could be several tens of milliseconds. We should note here that while the particle interaction time scale is in nanoseconds, the electronic data reading time scale is in micro-seconds, and such

J. Astrophys. Astr. (2021)42:68

Page 5 of 14

fast disturbances are already seen as bunches. As discussed earlier, the 20 ls time bin is digitally generated but the onboard analogue electronics has a much faster time scale (  1 ls). The examination of the data at longer time scales is motivated by the existence of such structures in the data. The details are given in the Appendix. Since this algorithm is independent of the total number of events in the DPH, the only constraint on tlook (optimised to 100 ms) is that it should be more than the duration of such events. We also note here that the DPHclean specifically caters to the need of the simultaneous two-event data indicative of Compton scattering by keeping aside all events satisfying the Compton criteria: simultaneous double events from neighbouring pixels. A large fraction of the events clustered in DPHstructures occupy regions only a few pixels wide. A histogram of DPHstructures with respect to the number of pixels or points (Npoints ) in them (for an observation lasting for an orbit) is shown in Fig. 3. Some of these DPHstructures include pixels which register counts 4 or higher in 100 ms bins. Note that in case a DPH shows clustering, only those events in the DPH that are responsible for the clusters are removed from the data by DPHclean. The algorithm thus has the ability to identify the clusters, and even during bright GRBs algorithm selectively picks out only the events in the cluster and removes them, without affecting the GRB photons. The advantage of such selective identification is evident. It is noticed that running this algorithm on all DPHs made from a given 200

Q0 Q1 Q2 Q3

175 150

set of data reduces the noise significantly more than selectively running it on (say 5-r) outliers in the light curve. We have examined the light curves of bunches and DPHstructures and we see an overall rising trend, in both the light curves, when the satellite enters the South Atlantic Anomaly (SAA). This similarity points to the origin of both kinds of events in charged particles. The quadrant-averaged orbit-averaged rate of DPHstructures is 0:25 s1 . We have investigated any possible temporal correlation of bunches with DPHstructures, to understand whether DPHstructures could be caused by bunches. The DPHstructure rate is 0:25 s1 and their duration is in tens of milliseconds, whereas the bunches are more numerous (70 s1 ) and are of very short duration (  0.1 ms), hence, there will be a large number of chance associations in millisecond timescales. Therefore, we have examined the occurrence of the start of a DPHstructure after a ‘super bunch’, defined as a bunch with the participating number of events more than 100 (Ratheesh et al. 2021). The rate of such super bunches is about 1:5 s1 . In Fig. 4, we show a histogram of the start time of DPHstructures plotted with respect to the end time of an immediately preceding super bunch, for DPHstructures observed during one orbit of data for one quadrant. The expected numbers purely by chance are shown as a dashed line in the figure. The association is extremely significant: the probability of so many DPHstructures clustering

100 Number

125

68

100 75

50

50 25 0 0

10

20

30

40

50

60

70

80

90

Npoints Figure 3. A histogram of the number of noise structures, called DPHstructures, obtained from the algorithm DPHclean, as a function of the number of unique pixels in them (Npoints ), for an observation using CZTI lasting for an orbit, given separately for each quadrants (Q0 to Q3).

0 0

50

Figure 4. Histogram of the number of DPHstructures as a function of the time difference between the start of a super bunch just prior to the DPHstructures and the start of the DPHstructure. The horizontal dashed line is the expected number of chance coincidences.

68

Page 6 of 14

J. Astrophys. Astr. (2021)42:68

within 25 ms of the super bunches, purely by chance, is negligibly small. We note here that a similar distribution is obtained for the other quadrants too and the distribution of super bunches occurring after a DHPstructure is as expected by chance. It can be seen that there are two types of associations. Firstly, we find that a substantial number ( 20%) of DPHstructures are closely associated with the super bunches: they occur within 5 ms of a super bunch (see Fig. 4), indicating that these DPHstructures could be post bunch noises caused by super bunches. Secondly, there is a peak between 5 and 25 ms in the delay histogram (for about 30% of the DPHstructures – see Fig. 4) along with a very significant dip at around 10 ms. We have examined the lightcurves of the events participating in the DPHstrucures. We make a lightcurve for each DPHstructure, renormalise the start time with respect to the end time of a super bunch just preceding the concerned DPHstructure and co-add all the light curves together. Such lightcurves are made separately for the DPHstructures which have a delay from a super bunch less than 5 ms (the first type discussed above – Fig. 5(a)) and for DPHstructures having a delay between 5 ms and 25 ms (the second type – Fig. 5(b)). For the first type, we see a sharp fall in the events initially and then a slow profile. The

initial surge of events perhaps indicates the post bunch noise. It is interesting to note the clear morphological differences in the temporal structure of the events between these two types of associations. For DPHstructures with a delay less than 5 ms from a super bunch, there is a surge of events initially showing that the concerned DPHstructures could be post bunch noise in the affected pixels. On the other hand, for the second type, there is an interesting trend in the events participating in the DPHstructures. They have a typical time scale of 50 ms to 100 ms and they start generating events 10 ms after the occurrence of the super bunch. Since we identify each bunch as due to cosmic rays and since each cosmic ray is expected to be independent of other cosmic rays, about 50% of DPHstructures are causally connected to super bunches. Some of them are post bunch noises while the majority of them are delayed emissions. The remaining 50% of the DPHstructures could be a combination of these two: post bunch noises from other bunches or separate emissions with a slow rise morphology. A cursory glance of the co-added light curves of these DPHstructures indicate that they may be dominated by the delayed emission morphology.

3. DPHstructures as high energy cosmic ray events 200 (a)

Counts

150 100 50 0

Counts

150

(b)

100

50

0 0

50

100

Figure 5. Light curve of the events in a DPHstructure as a function of the time difference between the start of a super bunch just prior to these DPHstructures, (a) for DPHstructures with a delay from the preceding super bunch less than 5 ms, and (b) for DPHstructures with a delay from the preceding super bunch between 5 ms and 25 ms.

Segreto et al. (2003) studied the detector characteristics of the PICsIT detector plane of the IBIS instrument on board INTEGRAL, and found events similar to DPHstructures. To investigate their cause, they plotted detector delay histograms (DDHs) corresponding to each DPHstructure. DDHs are histograms, on the detector plane, of the delay of the events contributing to a particular DPHstructure, with respect to the first event. They found evidence of two kinds of events: linear tracks, and a particular kind of delay pattern – a gradual increase of the delay towards the centre of the elliptical patterns seen in the DDHs. They explained these by the bombardment of the detector plane by charged particles or cosmic ray showers generated by hadronic and leptonic processes very close to the detector. They demonstrated that the delay in the first and last events in a particular DPHstructure being  100 ms could be explained by the saturation of the pixels by the extreme high energies of the charged particles, the pattern on the detector tracing the density of the cosmic ray showers in the logarithmic scale. Inspired by these findings, we plotted the DDHs for our DPHstructures, some

J. Astrophys. Astr. (2021)42:68

Page 7 of 14

examples of which are given in Fig. 6. We see two kinds of events: (1) Those tracing linear tracks indicating trajectories of physical entities along them (Fig. 6(a), (b)). This points to the origin being charged particles which deposit their energy over multiple pixels that fall on their trajectory of motion through the detector. (2) Those with the delay being more in the inside of a cluster compared to its spatial boundaries (Fig. 6(c), (d)). Both kinds of events are in striking similarity with the DDHs observed in PICsIT on board INTEGRAL, and naturally leads one to the hypothesis that DPHstructures in CZTI are also created by the bombardment of high-energy charged particles or cosmic ray showers.

a) 64

150

68

Based on the similarity of the DDH found in CZTI, for DPHstructures, with those in PICsIT onboard the INTEGRAL satellite, we can conclude that some cosmic rays, being more energetic than their counterparts that create bunches, might deposit their energies over multiple pixels instead of a few, thus resulting in DPHstructures. They appear to saturate these pixels, and, when the pixel output current drops below a threshold, the affected pixels start registering events. The saturation timescale observed in the detectors then corresponds to the delay from the onset of the events created by these cosmic rays on the detector plane. The fact that DPHstructures are much less frequent than bunches also corroborates such a hypothesis. Further, it naturally explains the delay pattern of the second kind, i.e., those with

b) 64

100

140 130

56

90

56

120 48

80

48

110

70

100 40

90 80

32

70

dety

dety

40

60

24

40 30

40

16

30

20

20

8

50

24

50

16

60

32

8

10

10 0 0

8

16

24

32

40

48

0

detx

c)

0

56

8

16

24

48

56

130

64

130

120

120

56

40

detx

d)

64

32

56

110

110 100

48

100

48

90

90

70

32

60 50

24

80

40

80

dety

dety

40

70 32

60 50

24

40

40 16

30 20

8

16

30 20

8

10

10 0 0

8

16

24

32

detx

40

48

56

0 0

8

16

24

32

40

48

56

detx

Figure 6. Detector delay histograms: Plotted in colour are the delay of the particular event from the first event in the cluster, in milliseconds, as a function of the position in the detector plane. These examples last unusually long, covering two consecutive time bins (100 ms). In the top two panels (a) and (b), we observe linear tracks, the delay increasing along the track in (b). The delay patterns in the bottom two panels (c) and (d) are remarkably similar to those seen in PICsIT on INTEGRAL due to phosphorescence-decays from events triggered by cosmic ray showers.

68

Page 8 of 14

J. Astrophys. Astr. (2021)42:68

102

Rate [s−1]

101

The energy of the cosmic rays cannot be calculated directly. However, we place constraints on the energy of both bunches and DPHstructures from the observed rate of the events, assuming a standard spectrum of cosmic rays (Longair 2011):  2:7 dN nucleons E : ð1Þ ¼ 1:8  104 2 dE s m sr GeV 1 GeV

Bunches

100

10−1

10−2

DPHstructures

10−3

10−4

0

200

400

600

800

1000

Energy [GeV]

Figure 7. The limits of the energies of three kinds of events: bunches, high-frequency DPHstructures, and lowfrequency DPHstructures, shown in pink, grey and red colours, respectively. The frequency cut of DPHstructures is based on the number of unique points in the DPHstructures, and is put roughly at the start of the tail of this curve, see Fig. 3. The higher end of the low-frequency cosmic rays is used as 100 TeV, but shown here only until 1 TeV for representational purposes. The arrow figuratively shows this extension of the energy limit.

progressively higher delays towards the centre of the spatial pattern (Segreto et al. 2003). When a cosmic ray shower hits the detector, the density of the particles in the shower are traced by the delay in the DDH. The delay timescales in the detector pixels are thus deduced to be a few 100 milliseconds. We can thus formulate the following scenario for the effect of the interaction of cosmic rays in CZTI: (1) Low energy cosmic rays interact in a few pixels and the heavy charge deposition is cleared out by the fast electronics in a series of events called bunches. (2) When the charge deposition is high, it has a paralysing effect on the concerned pixels, and these pixels give out events in a slow time scale, manifested as DPHstructures. (3) In some cases, multiple interaction is also possible, giving rise to multiple centres in the DPHstructure as well as a DPHstructure existing together with a bunch. We emphasise again that Compton scattering giving simultaneous interactions is treated separately as Compton events and does not affect the bunch and DPHstructure analysis presented here. (4) Additionally, some bunches can also generate post-bunch electronic noise events.

By taking the total area of the detector and integrating over all solid angles, between energy limits Emin and Emax , we match the resultant counts with the observed rates of bunches and DPHstructures to determine Emin and Emax . For the purpose of continuity, we divide the DPHstructures into two kinds of events on the basis of their frequency, with the criterion being the number of unique points in the DPHstructure 710. The orbitaveraged, quadrant averaged, rate of bunches is 70 s1 . Similar rates for the high-frequency and lowfrequency DPHstructures are, respectively, 0:200 s1 , and 0:044 s1 . Assuming an upper limit of the lowfrequency DPHstructures as 100 TeV, the energylimits thus derived are shown in Fig. 7. What is the physical mechanism that creates the  100 ms timescale saturation effect in the pixels? For PICsIT, the timescale was explained by fluorescence states of the CsI detectors, which is not possible for the CZT detectors in CZTI. In CZTI, the high voltage (HV) to the detectors is provided through an RC network (with a time constant of  100 ms). The extreme high energy of the cosmic rays that hit the individual detectors can momentarily reduce the HV such that the charges are not collected and the events are not registered. The time required for the HV to stabilise and for further events to be registered is  100 ms.

4. Discussion and conclusions The long time scale (up to 100 ms) of the disturbance from some cosmic rays is quite difficult to handle in the conventional dead time method: ignoring all data for these intervals would have increased the dead time considerably. The method described here, DPHclean, however, removes the affected events selectively. Since only a small number of pixels are affected, the effective area reduction is extremely negligible. We show here the efficacy of DPHclean by showing a light curve near a GRB in Fig. 8. Lightcurves binned at 100 ms (tlook ) centred around the time toffset , corresponding to the trigger time of bright GRB160802A,

J. Astrophys. Astr. (2021)42:68

Page 9 of 14

68

Figure 8. (a)–(d) Lightcurves binned at tlook ¼ 100 ms before (left – top and bottom panels) and after (right – top and bottom panels) DPHclean near the bright GRB160802A, red points showing 2-r outliers. In the top two panels, longer stretches of data from Q1 are shown whereas the bottom two panels show the zoomed data around the GRB from Q0 (hence scale is different).

are shown along with 2-r outliers highlighted in red. A longer stretch of data from Quadrant 1 (Q1) is shown in the top panel (Fig. 8(a), (b)) and a zoomed picture (for Q0, with a different scale), is shown in the bottom panel (Fig 8(c), (d)). The left panels (Fig 8(a), (c)) are the light curves before applying DPHclean and the right panels (Fig. 8(b), (d)) are after discarding all events responsible for DPHstructures. It can be seen that noisy features in the light curve are removed by DPHclean, but GRB photons are not flagged. The second feature within 20 seconds of the start of the prompt emission is also a part of the GRB, and is seen distinctly in the zoomed light curves of all quadrants with a similar profile, and it should not be mistaken as due to noise. In conclusion, we have understood the various effects of cosmic rays on the CZTI data. The new method, DPHclean, effectively removes the events from the affected pixels and this algorithm can be easily included in the ground analysis software. A new event cleaning method based on these and other

considerations has been developed and it is described in detail elsewhere (Ratheesh et al. 2021). Acknowledgements This publication uses data from the AstroSat mission of the Indian Space Research Organization (ISRO), archived at the Indian Space Science Data Centre (ISSDC). The CZT Imager instrument was built by a TIFR-led consortium of institutes across India, including VSSC, ISAC, IUCAA, SAC, and PRL. The Indian Space Research Organisation funded, managed, and facilitated the project. Appendix A: Algorithm for detecting ‘DPHstructures’

The aim of such an algorithm is to consider a DPH, and numerically decide whether the DPH shows any

68

Page 10 of 14

J. Astrophys. Astr. (2021)42:68

(a)

(b) 2500

3500

threshold threshold threshold threshold threshold

3000

= = = = =

0.50 0.55 0.60 0.65 0.70

threshold threshold threshold threshold threshold

2000

= = = = =

0.50 0.55 0.60 0.65 0.70

2500 1500

2000

1500

1000

1000 500

500

0

0

4

6

8

10

Npoints

2

12

4

6

Msum

8

10

(c) threshold threshold threshold threshold threshold

2500

2000

= = = = =

0.50 0.55 0.60 0.65 0.70

1500

1000

500

0

2

4

6

Npairs

8

10

12

Figure A1. Histograms of (a) Npoints , (b) Msum , and (c) Npairs with different values of ‘threshold’. Since most of the pairs are non-identical, Npoints is more likely to be even than odd, hence it shows regular dips at odd integers (as can be seen as slight dips in (a)). It is observed that these parameters are insensitive to the value of ‘threshold’ chosen. Hence, the ‘threshold’ is fixed it at its most conservative upper limit throughout the work.

spatial clustering or not. This algorithm will output a flag 0 if clustering is detected or 1 if it is not. A requirement of such an algorithm is that it should be independent of the total number of events in the DPH, since it is to be run on DPHs made during average count-rates as well as during GRBs. The basic assumption made here is that genuine X-ray photons (from the astrophysical sources of interest as well as background X-ray photons extraneous to the instrument) are independent events and are randomly distributed in the detector plane, whereas the ‘noise’ events, ether induced by cosmic rays or from local electronic effects, would be spatially clustered in the detector plane.

The algorithm is detailed below: • Consider only those pixels in the DPH which register non-zero counts. If there are n such pixels, there are n C2 pairs. For each pair, calculate a measure of ‘hotness’, ci  cj ; mij ¼ Dij where ci is the count in the i-th pixel and Dij is the distance between the pixels (in units of detx/ dety, which is unity). This quantity is large if count in either pixel is large and/or if the distance between the pixels constituting the pair is small.

J. Astrophys. Astr. (2021)42:68

Page 11 of 14

(a)

68

(b)

8 25

7 20

6 5

15

4 10

3 2

5

1 0

0

0.5

1.0

1.5

2.0

2.5

4

3.0

9 14 19 24 29 34 39 44 49 54 59 64 69 74 79 84 89 94 99

Msum/Npoints

Msum/Npoints

Figure A2. (a) Histogram of the parameter Msum =Npoints for the DPHs which are not flagged. This ratio rarely goes close to 3, which is the value of ‘allowable’ used for flagging. (b) Same for the ones that are flagged. Note that although the total number of flagged DPHs in a typical dataset is much smaller than the number of DPHs that are flagged, most of them do not have any hot pairs, hence both Msum and Npoints are zero. Figure (a) includes only those within finite values of Msum and Npoints , explaining why the total number is smaller than in Fig. (b). The sharp increase at values close to 4 in (a), compared to the rarity of those in (b) below 3, demonstrates that the distinction is real. The reality of this distinction is also verified by manual examination of each DPH for long stretches of data which preferentially include weak as well as bright GRBs, examples of which are given in Fig. A4.

• If

which is the case for two diagonally-located neighbouring pixels registering 1 count each. If ‘threshold’ is larger than this, we will miss these hot pairs, thus defeating the purpose. • Construct the set of all pixels which contribute to any such hot pair. • Modify the choice of hot pairs: If a pair is such that it consists of two neighbouring pixels only, each registering one count, and there is no hot pixel in its immediate neighbourhood, then do not consider the pair for the following steps. This ensures that actual double events are not considered whereas neighbouring pixels with one count each in the neighbourhood of a cluster are retained. • Calculate the ‘gross’ parameters: (1) total number of non-identical points contributing to the identified hot pairs: Npoints ;

Flagged percentage

call it a ‘hot pair’. It is to be noted that the maximum allowable value for threshold is 11 thresholdmax ¼ pffiffiffi ’ 0:707; 2

average counts high counts

0.10

mij [ threshold;

0.08

0.06

0.04

0.02

0.00 1.50

1.75

2.00

2.25

2.50

2.75

3.00

3.25

Msum/Npoints

Figure A3. Random DPHs are simulated with the eventrate as input. The parameter ‘allowable’ is allowed to vary, and flagging is carried out on the random DPHs based on these variable values. The plot show the resulting number of DPHs flagged as a percentage of the total number of DPHs simulated (5400), as a function of the variable. An average count-rate of 150 per second, and a high count-rate during bright GRBs, of 1500 per second, are considered. The flagged percentage is 0 for Msum =Npoints [ 3, proving the robustness of the algorithm for both the average data and during GRBs. In fact, it is more robust when the countrates are high as seen during GRBs, i.e. DPHs during bright GRBs have  0 probability of getting flagged.

68

Page 12 of 14

J. Astrophys. Astr. (2021)42:68

64

64

56

56

48

48

40

40

32

32

24

24

16

16

8

8 0

8

0

16

24

1

32

2

40

48

3

56

4

0

5 64

56

56

48

48

40

40

32

32

24

24

16

16

8

8 8

16

24

32

40

48

56

0

0

1

16

6

64

0

8

8

24

32

7

16

24

40

8

32

48

56

9

40

48

10

56

2

Figure A4. Instructive examples of Detector Plane Histograms (DPH), plots of counts shown at positions det x vs. det y. Top: On the left is a DPH which shows clustering (the color-bar is of counts), with the identified clustered events shown in the right. Bottom: On the left is a DPH that does not show clustering. The pairs that are used to test the clustering are explicitly shown in the right to demonstrate that the presence of such random pairs are not enough to flag this DPH.

(2) the sum of the measures of the hotness for each such hot pairs: X mij ; Msum ¼ fall pairsg

(3) the number of hot pairs detected (note that even if one pixel contributes to two/more hot pairs, all these pairs are counted): Npairs . • Construct a parameter based on these gross parameters as a proxy for the randomness in the

DPH. When the value of this proxy exceeds a certain cutoff, parametrized by ‘allowable’, then the DPH is flagged, i.e. deemed to show clustering; otherwise not. • If the DPH shows clustering, identify only those events in it that contribute to this flagging. Particularly, remove any lingering isolated single or double event that may have correlated with a pixel registering multiple counts, owing only to their proximity.

J. Astrophys. Astr. (2021)42:68

To optimize the values of ‘threshold’ and ‘allowable’, we resort to simulations of random DPHs, with mean count-rate of single and double events as inputs. The mean count-rate is typically 90 events s1 for single and 60 events s1 for double events, so in a 100 ms timescale, they are 9 and 6 events respectively. First, the number of single and double events to be chosen for a particular DPH to be simulated are drawn from Poisson distributions with the given means. Then, these many values of detx and dety are drawn from a uniform random distribution of all possible detx and dety values (0 to 63). For double events, one of the neighbouring events is first chosen randomly and the other is drawn randomly from the neighbouring coordinates, taking due care of corners and edges. For the case of GRBs, the mean count-rates input into the simulation process are increased, as discussed below (see Fig. A3). For each such simulated DPH, the gross parameters Npoints , Msum , and Npairs are calculated, and this is done on multiple DPHs (typically 5400 for one full orbit) with different inputs to the parameter ‘threshold’. The identification of hot pairs based on ‘threshold’ is insensitive to the value of this parameter, as demonstrated in Fig. A1. Hence it is safe to keep it fixed at its most conservative maximum value, i.e. 0.70, which will detect diagonally-placed neighbouring pixels each registering a count. Next, we experiment the construction of ‘allowable’ based on the three gross parameters, and flag random DPHs based on the different experimental values of these parameters. It turns out that both allowable ¼ Msum ¼ 8 and allowable ¼ Npairs ¼ 8 flag less than 1% random DPHs, but this conclusion is seen to break down in the presence of bright GRBs like GRB160802A, since the number of photons in the DPH are  10 times greater than the usual, resulting in random pixels getting paired and marked as hot pairs. Normalizing any of the parameters by the total number of photons does not help because extremely bright DPHstructures have total number of counts comparable to the total counts in random DPHs during GRBs, simply because the clustering illuminates its neighbourhood very brightly. Hence, we define allowable ¼ Msum =Npoints , which normalizes for the additional Msum contribution from the pairs that are created due to chance co-incidence of a larger number of random events during GRBs. This simple modification fantastically discriminates clustered DPHs

Page 13 of 14

68

from random ones, as shown in Fig. A2. The reason is that, although the total number of counts in a clustered DPH is large, the clustering is spread over a few pixels, and the same pixels register many events; on the other hand, random DPHs with increased total counts, where the Msum is increased by co-incidental pairing of random events, have many such pairs which are themselves randomly distributed over the entire quadrant. In comparison, allowable ¼ Msum =Npairs does not do a better job because the small number of neighbouring pixels in a cluster tend to pair up with most of the other pixels in the cluster. Random DPHs from GRBs and during average count-rates are examined along with DPHs that show clustering: it is seen that allowable ¼ Msum =Npoints ¼ 3 distinctly separates clustered DPHs from random ones, whether they are during a GRB or otherwise. This is verified first visually by looking at a significant number of DPHs manually, and also demonstrated in Figures A2 and A3. Examples of detected DPHstructures and DPHs with non-detections are shown in Fig. A4.

References Agrawal P. C., Yadav J. S., Antia H. M. et al. 2017, J. Astrophys. Astr., 38, 30 Antia H. M., Yadav J. S., Agrawal P. C. et al. 2017, ApJS, 231, 10 Bhalerao V., Bhattacharya D., Vibhute A. et al. 2017, J. Astrophys. Astr., 38, 31 Chattopadhyay T., Vadawale S. V., Aarthy E. et al. 2019, ApJ, 884, 123 Longair M. S. 2011, High Energy Astrophysics, Cambridge University Press, Cambridge, UK Rahna P. T., Murthy J., Safonova M. et al. 2017, MNRAS, 471, 3028 Rao A. R., Singh K. P., Bhattacharya D. 2016a, ArXiv e-prints, arXiv:1608.06051 Rao A. R., Chand V., Hingar M. K. et al. 2016b, ApJ, 833, 86 Rao A. R., Patil M. H., Bhargava Y. et al. 2017, J. Astrophys. Astr., 38, 33 Ratheesh A., Rao A. R., Mithun N. P. S. et al. 2021, J. Astrophys. Astr., 42, https://doi.org/10.1007/s12036021-09716-4 Segreto A., Labanti C., Bazzano A. et al. 2003, A&A, 411, L215 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, in Takahashi T., den Herder J.-W. A., Bautz M., eds,

68

Page 14 of 14

Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 9144, Space Telescopes and Instrumentation 2014: Ultraviolet to Gamma Ray, 91441S Singh K. P., Stewart G. C., Chandra S. et al. 2016, in SPIE, Vol. 9905, Space Telescopes and Instrumentation 2016: Ultraviolet to Gamma Ray, 99051E Singh K. P., Stewart G. C., Westergaard N. J. et al. 2017, J. Astrophys. Astr., 38, 29

J. Astrophys. Astr. (2021)42:68 Subramaniam A., Tandon S. N., Hutchings J. et al. 2016, in SPIE, Vol. 9905, Space Telescopes and Instrumentation 2016: Ultraviolet to Gamma Ray, 99051F Tandon S. N., Subramaniam A., Girish V. et al. 2017a, AJ, 154, 128 Tandon S. N., Hutchings J. B., Ghosh S. K. et al. 2017b, J. Astrophys. Astr., 38, 28 Yadav J. S., Agrawal P. C., Antia H. M. et al. 2017, Curr. Sci., 113, 591

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:67 https://doi.org/10.1007/s12036-021-09711-9

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

Exploring sub-MeV sensitivity of AstroSat–CZTI for ON-axis bright sources ABHAY KUMAR1,2,* , TANMOY CHATTOPADHYAY3, SANTOSH V. VADAWALE1,

A. R. RAO5,6, SOUMYA GUPTA4, N. P. S. MITHUN1, VARUN BHALERAO6 and DIPANKAR BHATTACHARYA4 1

Physical Research Laboratory, Navrangpura, Ahmedabad 380 009, India. Indian Institute of Technology, Gandhinagar 382 355, India. 3 Kavli Institute of Astrophysics and Cosmology, 452 Lomita Mall, Stanford, CA 94305, USA. 4 The Inter-University Centre for Astronomy and Astrophysics, Pune 411 007, India. 5 Tata Institute of Fundamental Research, Mumbai 400 005, India. 6 Indian Institute of Technology Bombay, Mumbai 400 076, India. *Corresponding author. E-mail: [email protected] 2

MS received 7 November 2020; accepted 13 January 2021 Abstract. The Cadmium–Zinc–Telluride Imager (CZTI) onboard AstroSat is designed for hard X-ray imaging and spectroscopy in the energy range of 20–100 keV. The CZT detectors are of 5-mm thickness and hence have good efficiency for Compton interactions beyond 100 keV. The polarisation analysis using CZTI relies on such Compton events and have been verified experimentally. The same Compton events can also be used to extend the spectroscopy up to 380 keV. Further, it has been observed that about 20% pixels of the CZTI detector plane have low gain, and they are excluded from the primary spectroscopy. If these pixels are included, then the spectroscopic capability of CZTI can be extended up to 500 keV and further up to 700 keV with a better gain calibration in the future. Here we explore the possibility of using the Compton events as well as the low gain pixels to extend the spectroscopic energy range of CZTI for ON-axis bright X-ray sources. We demonstrate this technique using Crab observations and explore its sensitivity. Keyword. AstroSat—CZT-Imager—sub-MeV spectroscopy—Crab.

1. Introduction The Cadmium–Zinc–Telluride Imager (hereafter CZTI) onboard AstroSat, India’s first dedicated astronomy satellite (Paul 2013; Singh et al. 2014), is primarily designed for hard X-ray coded mask imaging and spectroscopy in the energy range of 20–100 keV (Bhalerao et al. 2017). It consists of 64 Cadmium–Zinc–Telluride (CZT) detector modules having 5-mm thickness and 4 cm  4 cm in dimension. Each module is further segmented spatially in 16  16 array of pixels (2.5 mm  2.5 mm of pixel size). After the launch of AstroSat, about 20% of the CZTI pixels

This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

(therefore  20% of the total 924 cm2 geometric area) were found to have a relatively lower gain than that of the spectroscopically good pixels, which makes these pixels sensitive to higher energy photons (  70–1000 keV for a gain shift [4). The 5-mm thick CZT detector provides sufficient detection efficiency up to 1 MeV. Motivated by this, we tried to explore the possibility of including these low gain pixels in the analysis to enhance the spectroscopic sensitivity of CZTI up to the sub-MeV region. The Coded mask spectrum generated using the standard pipeline is restricted up to 100 keV, where the background is simultaneously obtained from the coded mask imaging. Above 100 keV, the 0.5 mmthick tantalum mask becomes increasingly transparent, along with the collimators and CZTI support structures. To obtain the high energy spectra above

67

Page 2 of 9

100 keV in the absence of simultaneous background measurements requires a careful selection of blank sky observations. Background flux depends on multiple factors like the spacecraft’s geometric location in orbit, orbital precession of the satellite, and the time spent within the high background South Atlantic Anomaly (SAA) region in an orbit. These contribute to a systematic modulation in the flux along the satellite’s orbit, hence making the background subtraction quite challenging. Another challenge is to calibrate these pixels in the absence of any mono energetic lines at energies above 100 keV to estimate their gains. A careful calibration of these pixels have been attempted in a companion paper by Chattopadhyay et al. (2021) in this issue, and spectroscopy up to 900 keV is explored for Gamma Ray Bursts (GRBs). Compared to the ON-axis sources, spectroscopy of GRBs is relatively easy because of the availability of background (from the immediate pre-GRB and post-GRB observations) and the significantly higher signal to noise ratio for the GRBs. On the other hand, the ON-axis bright astrophysical sources are fainter, and hence longer exposure observations are required for sufficient detection, which leads to more instrumental, charged particle, and cosmic X-ray background contributions. This paper outlines the methodology of sub-MeV spectroscopy with CZTI for bright ON-axis sources. We utilize the 2-pixel Compton events, which are used to extract polarimetry information of the X-ray photons (Chattopadhyay et al. 2014, 2019; Vadawale et al. 2015, 2018) to enhance the spectroscopic capability. The selection of background observations and the background subtraction used for polarimetric measurements is described in detail here, with an emphasis on spectroscopy. We utilize the low gain pixels to extend the spectroscopic energy range of the instrument well beyond the standard limit. With the inclusion of the low gain pixels, Compton spectroscopy was also extended to 500 keV. We use the AstroSat mass model in GEANT4 (Agostinelli et al. 2003) to generate the spectral response. Here we carry out broadband spectroscopy of Crab using standard spectroscopic events (30–100 keV), 2-pixel Compton events including low gain pixels (100–500 keV), and also explored the 1-pixel events including low gain pixels (100–700 keV) to establish the sub-MeV spectroscopy methods and at the same time try to constrain the spectral parameters of Crab at higher energies. The X-ray emission coming from Crab Nebula can be divided into three parts: pulsed point source from the neutron star, synchrotron emission

J. Astrophys. Astr.

(2021) 42:67

powered by charged particles coming from the centrally located pulsar, and the large diffused emission region in the Nebula. Toor and Seward (1974) fitted the Crab spectra in 2–60 keV using a power law and concluded that it could be used as a standard calibration source due to its steady nature. It is believed that it appears to be steady on a time scale of a few years because most of the emission is from the diffused expanding ejecta, which is extended. There is, however, no theoretical basis that the pulsed emission will be steady. Different models have been tried to explain the stable behavior and emission mechanism of Crab. It is often described by a power law model (Kirsch et al. 2005; Kuiper et al. 2001). A broken power law is also used to describe the slope evolution, with a break around 100 keV (Strickman et al. 1979; Ling & Wheaton 2003). Massaro et al. (2000) and Mineo et al. (2006) used a single curved power law with a variation of the slope with log(E) to describe the spectra. The INTEGRAL/SPI data were fit with a broken power law (Jourdain & Roques 2008) and also by the Band model generally used for GRBs (Band et al. 1993; Jourdain & Roques 2020). There is no general agreement on the best overall spectral model so far. Further, the absolute flux, emission mechanism, and the cause of Crab’s stability are also not well known. The extended bandwidth of CZTI using the Compton and single spectrum, including low gain pixel events, can help us understand the spectra of the Crab. It is to be noted that CZTI is also sensitive to polarisation, and hence a simultaneous measurement of polarisation along with spectroscopy can help us in a better understanding of the Crab emission mechanism and the cause of the stability in its emitted flux. In Section 2, a brief description of the observations and analysis procedure is given. The results obtained are presented in Section 3. Finally, in Section 4, we discuss the sub-MeV sensitivity of CZTI and future plans.

2. Observations and analysis procedure There are many observations of Crab over the past five years of operation of AstroSat. We have selected those observations with sufficient exposure and also having a suitable background observation. The selection of appropriate background observation and subtraction is an important part of the analysis. The CZTI support structure becomes increasingly transparent above 100 keV. Therefore, background measurements can be affected due to the presence of bright X-ray sources within 70 of the pointing direction of CZTI. The Crab

J. Astrophys. Astr.

(2021) 42:67

Page 3 of 9

67

Table 1. Summary of Crab and blank sky observations. Crab ObsID 9000000096 9000000252 9000000406 9000000964 9000000970

Blank Sky

Date (yyyy/mm/dd)

Exposure (ks)

ObsID

Date (yyyy/mm/dd)

Exposure (ks)

RA (deg)

2015/11/12 2016/01/07 2016/03/31 2017/01/14 2017/01/18

41 60 114 78 123

9000000276 9000000276 9000000404 9000000974 9000000974

2016/01/16 2016/01/16 2016/03/29 2017/01/22 2017/01/22

64 64 64 51 51

183.48 183.48 228.21 183.48 183.48

DEC (deg) 22.8 22.8 –9.09 22.8 22.8

Crab RA: 83:63 and DEC: 22:01 .

and Cygnus X-1 are two bright sources which should be avoided during the background observation. Since both the sources are located almost opposite to each other in the sky, it is possible to find a good region, away from these two sources, for the background observations. The background observation should also be close to the source observation time to avoid the error in some long-term secular variations in the background behavior. There are a few such observations that satisfy the criterion of the background selection. Based on these considerations, we have finally selected five observations between 2015 and 2017 for further analysis. Details of the Crab observations and the corresponding background observations selected for the present analysis are given in Table 1.

2.1 Single and Compton event selection CZTI is a pixelated detector. The scientific data analysis of CZTI is done with two types of CZTI events: 1-pixel or single pixel events and 2-pixel events. The single pixel events registered in the CZTI are considered as the true 1-pixel events if there is no event registered within 100 ls time window on either side of the single event. The events in the CZTI are time stamped at every 20 ls (Bhalerao et al. 2017) and any two events occurring within 20 ls coincidence time window in two pixels are considered as true 2-pixel event (Chattopadhyay et al. 2014). The standard single pixel mask-weighted spectra in 30–100 keV (hereafter EB1) is generated following standard pipeline software available at the AstroSat science support cell (ASSC).1 1

http://astrosat-ssc.iucaa.in/.

After the launch of AstroSat, it is observed that CZTI had low gain pixels of about 20% of the detector plane. These pixels are found to have a shift in gain (gs) by a factor of 1.5-5 compared to that calculated during the ground calibration such that a mono-energetic line of energy ‘E’ now appears at a lower energy PI bin (E/gs); hence the name low gain pixels. Because these pixels now record higher energies, the spectroscopic range of CZTI with single pixel events can be extended up to 700 keV with the inclusion of low gain pixels. Detailed characterisation methods of low gain pixels have been discussed in the companion paper by Chattopadhyay et al. (2021). Above 100 keV single event spectra, including the low gain pixel events, need to be analysed outside of the standard pipeline because they are not included in the standard pipeline of CZTI. The initial processing is to remove the intervals of high background during the passage of South Atlantic Anomaly from each observation and removal of the noisy (pixel having counts more than five sigma above mean value) or spectroscopically bad pixels (energy resolution is poor compared to the normal pixels). This procedure is valid for both the single and 2-pixel event spectra. After selecting the single pixel events, energy deposited in each pixel is used to generate the single pixel spectrum including low gain pixels in the 100–700 keV (hereafter EB2). We have binned the events into 60 channels with 10 keV energy bin sizes in the 100–700 keV energy range. Above 100 keV, the 5-mm thick CZTs have sufficient efficiency for Compton interactions. Polarisation analysis in the 100–380 keV range depends upon such Compton events (Chattopadhyay et al. 2014; Vadawale et al. 2015). These Compton events are used here to do spectroscopy in the 100–380 keV range. After incorporating the low gain pixels, Compton

67

Page 4 of 9

spectroscopy can be further extended to 500 keV. For the Compton event selection, we follow the 2-pixel Compton event selection criteria, as discussed in Chattopadhyay et al. (2014). The readout logic in the CZTI is such that it reads events from one module at a time. If two events are registered in two different pixels in the same module, then it is possible that two events would get two different time stamps. Hence, all the events occurring within the coincidence time window of 40 ls are selected for the analysis. These events are further filtered through Compton kinematics criteria to enhance the signal-to-noise ratio. After the selection of the Compton events, the sum of the energy deposited in the scattering and the absorption pixel is used to generate the 2-pixel Compton events spectrum including low gain pixels in the 100–500 keV (hereafter EB3). We have binned the data at 10 keV energy bin size in the 40 channels ranging from 100 keV to 500 keV.

2.2 Background subtraction: Phase match method The primary sources of background in the CZT detector are the Cosmic X-ray background, the Earth’s X-ray Albedo, and the locally produced X-rays due to Cosmic ray interactions. The Compton background in the detector is due to the Compton scattering of these background X-rays. In addition to this, a small part of the Compton background consists of chance coincidence events within 40 ls coincidence time window. The background events from the blank sky observations are filtered through the same selection criteria as the source, as discussed in Section 2.1. The observed background counts show a prominent orbital variation as well as a diurnal variation depending on the geometric location of the spacecraft in the orbit, orbital precession, and the location and duration of the SAA region in orbit. All these contribute to a systematic modulation (see the top panel of Fig. 1) in the flux along the orbit of the satellite within the duration of the observation. Because of the modulating flux, it is important to select similar portions of the Crab and background orbits based on the spacecraft’s ground tracks (latitude and longitude) and use them for further analysis. But this puts a strict condition on the selection of the orbits and leaves a short usable exposure of Crab and background observations. An alternate method developed for background subtraction is to match the phase of the background and Crab light curves (the ‘phase match method’), used for the

J. Astrophys. Astr.

(2021) 42:67

polarisation measurements of Crab (Vadawale et al. 2018). In the phase match method, first, the Crab and background light curves are fitted with an appropriate higher-order (here 5th order) polynomial. Then the Crab and background light curves are matched by sliding the Crab polynomial over the background polynomial every 10 seconds and estimating the best match by minimising the v2 . Within the matched region of the two polynomials, the background is taken only for those time regions for which there is a source observation (see bottom right of Fig. 1). From the background’s phase-matched region, we calculate the ratio of the average background count rate to the count rate in the phase matched region (‘correction factor’) and multiply that to the total background exposure to calculate an effective background exposure. The use of effective exposure automatically takes care of the different phases of source and background observations. For example, if the source and backgrounds are observed in the same phase, the multiplication factor will be close to 1. For longer source observations (exposure  background exposure), it is divided into multiple parts and for each part of the source observations, we calculate the correction factor in the way described above and then the final correction factor is calculated as the weighted average. It is to be noted that the phase match method only ensures that the source (Crab in this case) and blank sky background data are taken for the same orbital phase (spacecraft orbit). It is not for the background spectral modeling. The background scaling is done to the phase matched background region to best mimic the actual background during the source observation. Because the background spectra are obtained from observations of ‘blank sky’ with no other hard X-ray sources in the FoV (confirmed from BAT catalog), the spectra for the used backgrounds are expected to be the same. To demonstrate this, we selected a CZTI observation from UVIT catalog (ObsID 1008) such that there are no other bright X-ray sources in the FOV of CZTI (call this ‘CU’). We then used the polynomial method to do phase match between CU and background data ObsID 974 (call this ‘B’), which is used in the analysis and correct for total flux in B for the phase of CU. Because CU is essentially a blank sky background for CZTI, similar flux and spectra for both B and CU after phase correction is expected. We found identical flux for both B and CU (see Fig. 2), signifying that the polynomial method is capable of finding the common phase and scale the

(2021) 42:67

80

80

60

60

Counts

Counts

J. Astrophys. Astr.

40 20

Page 5 of 9

40 20

0

0 0

5×103

1×104

0

Time (s)

5×103

1×104

Time (s) 80

350 60

Counts

χ2

300 250 200

40 20

150 100

0 0

2000

4000

6000

Time (s)

0

5×103

67

obtained from the actual observational data are applied to the simulation data. For 2-pixel Compton events, the sums of the energies in the corresponding two pixels are used to obtain the total deposited energy, while for single event response, total absorbed energy for a given incident photon is used. We applied the same criterion of event selection, as discussed in Section 2.1. We then convolve the 1 keV bins by a Gaussian of 8 keV Full Width at Half Maximum (FWHM). We have not noticed any significant increase in FWHM with energy for CZTI pixels during ground calibration. Therefore, FWHM is kept constant across the energy. It is to be noted that the response for EB1 is generated using ls and charge diffusion based line profile model (Chattopadhyay et al. 2016).

1×104

Time (s)

Figure 1. Top panel shows the light curves of Crab (ObsId 96) on the left and background (ObsId 276) on the right, both fitted with fifth degree polynomials (red curve for Crab and green curve for background). Bottom left figure shows the variation of v2 as a function of shift in the Crab light curve with respect to the background, to determine the time for ‘phase matching’. Bottom right figure shows the ‘phase matched’ background and Crab polynomials, in the green curve and red, respectively. The blue curve represents the background where Crab observation is available.

flux accordingly. The spectra are also found to be identical, justifying the underlying assumption that the blank sky observations for CZTI with the predefined selection criteria yield a similar photon energy distribution.

2.3 Spectral response The response for EB1 is generated using the standard pipeline of the AstroSat and for both EB2 and the EB3 using the GEANT4 simulation of the AstroSat mass model. Details of the AstroSat mass model and its validation has been discussed in Chattopadhyay et al. (2019) and Mate et al. (2021). We simulated the mass model for 56 mono-energies ranging from 100 keV to 2 MeV (at every 20 keV up to 1 MeV and 200 keV till 2 MeV) for 109 photons. For each energy, the distribution of deposited energy in CZTI pixels is computed at 1 keV binning for each pixel. The CZTI pixel-level LLD (Lower Level Discriminator), the ULDs (Upper Level Discriminator), list of noisy and dead pixels

3. Results 3.1 Single and compton event spectra (EB1 and EB3) We use the broken power law model to fit the Crab curved spectra in 30–500 keV. It has been long used to explain the Crab spectra with break energy at 100 keV (Strickman et al. 1979; Ling & Wheaton 2003). INTEGRAL/SPI has also shown the spectral fitting using broken power law up to sub-MeV region (20 keV–1 MeV) (Jourdain & Roques 2008). We have analysed all the selected observations and then fitted the resultant spectra (EB1 and EB3) simultaneously using constbknpower in XSPEC (Arnaud 1996) freezing break energy at 100 keV while the other parameters (photon indices) are tied across the spectra. To account for the cross calibration and differences between the different spectra, a constant was multiplied to the model. It was fixed to one for EB1 and left free to vary for others. The EB1 below 30 keV and above 100 keV is ignored due to calibration issues. No systematic has been added to the EB1 and EB3. Spectral fitting for one of the five observations (ObsID 406, 114 ks) is shown in Fig. 3. The values of the fitted parameters for all the five observations along with the INTEGRAL/SPI results are given in Table 2. The low energy slope (PhoIndx1) and the higher energy slope (PhoIndx2) are well constrained and consistent with the INTEGRAL/SPI (Jourdain & Roques 2008) values within errors (see Fig. 4). The contour plots of PhoIndx2 vs. Norm for all the five observations are shown in Fig. 5. The inputs for the contour plot are generated using the chain command

67

Page 6 of 9

J. Astrophys. Astr.

(2021) 42:67

show that the value of PhoIndx2 is 2:04þ0:48 0:44 and that þ0:71 of Norm is 6:390:63 , which is  31% smaller than the Norm for INTEGRAL/SPI. CZTI sensitivity is also studied by looking at the flux variation of Crab with time. Figure 6 compares the Crab flux covering the period between 2015 Nov. 12 and 2017 Jan. 18. Each data point corresponds to the flux obtained in each observation, and the dotted line represents the flux measured by INTEGRAL/SPI. The flux from ObsID 96 is much lower than the values found for the other ObsIDs. If we exclude the data from ObsID 96, the remaining four measurements are within 5% of the mean value. We note here that the background for this ObsID is measured more than two months before the Crab observation, whereas for all the other ObsIDs, the background is measured within ten days of the respective Crab observation. The effect of secular variations in the background on the flux measurements needs to be investigated further.

3.2 Low gain spectra

normalized counts s−1 keV−1

Figure 2. The top figure shows the phase matched 2-event Compton spectra (EB3) of the blank sky observation CU (red) and background B (blue) from 100 keV to 500 keV. The ratio of the two spectra is shown in the middle panel and the residuals of the two are shown in the bottom panel. The bottom figure shows similar plots for the low gain pixel spectra (EB2) from 100 keV to 700 keV. 0.01 10−3 10−4 10−5 4

Δχ2

2 0 −2 50

100 200 Energy (keV)

500

Figure 3. Broadband spectra of Crab (ObsID 406, 114 ks) fitted with a broken power law. The black, red colors are used for EB1, and EB3 respectively.

in XSPEC after getting the best fit parameters, and the corner plots are generated using the python corner module (Foreman-Mackey 2017). The corner plots

As discussed earlier, 20% of the pixels in CZTI are found to have lower gains as compared to the spectroscopically good pixels, which makes these pixels sensitive to higher energy photons (  70–1000 keV for a gain shift [4). Initially, these pixels were considered as bad pixels and removed from the analysis. Here, we have explored the spectroscopic capability of CZTI up to 700 keV, including these pixels. We have analysed all the selected observations and then fitted the resultant spectra (EB1, EB2, and EB3) simultaneously using the same constraints as described in Section 3.1. An 8% systematic error, however, have been added to EB2 to take into account of the uncertainties in the background estimation and gain calibration. The systematic error is added to the data till the residuals are uniformly distributed across zero and the spectral fit is acceptable. Spectral fitting for one of the five observations (ObsID 406, 114 ks) is shown in Fig. 7. The fitted parameters are given in Table 2. The low energy slope (PhoIndx1) is well constrained and close to INTEGRAL/SPI (Jourdain & Roques 2008) value. The higher energy slope (PhoIndx2), however, is  38% lower than the INTEGRAL/SPI reported values, though the errors in the parameter are quite large. The influence of subtle background spectral variations and possible gain nonlinearity at higher energies (the CZT detectors are not calibrated outside the energy range of 10 to 150 keV)

J. Astrophys. Astr.

(2021) 42:67

Page 7 of 9

67

Table 2. Comparison of fitted parameters between INTEGRAL/SPI and AstroSat for broken power law. The errors are reported for 90% confidence interval. The mean value of the parameters of all the observations for different combinations of data is mentioned in the bottom row of each block. Instrument INTEGRAL CZTI

Norm

Flux (30–100 keV) 108 (erg/cm2 s)

ObsID

PhoIndx1

PhoIndx2

Ebreak (keV)

EB1 and EB3

96 252 406 964 970 Mean

2:08þ0:01 0:01 2:18þ0:05 0:05 2:08þ0:03 0:03 2:12þ0:03 0:03 2:08þ0:04 0:04 2:10þ0:03 0:03 2:11þ0:02 0:02

2:23þ0:05 0:05 1:80þ0:48 0:47 1:72þ0:44 0:48 1:63þ0:39 0:36 2:65þ0:90 0:74 1:98þ0:48 0:44 1:95þ0:18 0:18

100 100 100 100 100 100 100

9:3þ0:14 0:14 5:98þ1:27 1:04 6:00þ0:84 0:73 7:5þ0:78 0:70 5:73þ0:90 0:77 6:44þ0:72 0:64 6:33þ0:32 0:32

1.30 0:56þ0:10 0:09 0:85þ0:12 0:10 0:89þ0:09 0:08 0:82þ0:11 0:12 0:82þ0:09 0:08 0:79þ0:06 0:06

EB1 and EB2

96 252 406 964 970 Mean

2:21þ0:04 0:04 2:1þ0:03 0:03 2:13þ0:02 0:02 2:13þ0:03 0:03 2:14þ0:02 0:02 2:14þ0:02 0:02

1:20þ0:24 0:18 1:33þ0:20 0:16 1:46þ0:17 0:14 1:5þ0:22 0:17 1:34þ0:17 0:14 1:36þ0:05 0:05

100 100 100 100 100 100

6:75þ1:29 1:09 6:48þ0:73 0:66 7:76þ0:59 0:55 6:98þ0:69 0:63 7:26þ0:57 0:54 7:05þ0:22 0:22

0:55þ0:10 0:09 0:84þ0:09 0:09 0:89þ0:07 0:06 0:80þ0:08 0:07 0:81þ0:06 0:06 0:78þ0:06 0:06

225/195 220/195 274/195 225/195 317/195

EB1, EB2 and EB3

96 252 406 964 970 Mean

2:21þ0:04 0:04 2:1þ0:03 0:03 2:13þ0:02 0:02 2:13þ0:02 0:02 2:13þ0:02 0:02 2:14þ0:02 0:02

1:24þ0:24 0:18 1:36þ0:2 0:16 1:48þ0:17 0:14 1:5þ0:23 0:18 1:38þ0:18 0:15 1:39þ0:05 0:05

100 100 100 100 100 100

6:68þ1:28 1:07 6:45þ0:72 0:66 7:71þ0:58 0:55 6:94þ0:69 0:64 7:21þ0:57 0:54 7:0þ0:22 0:22

0:55þ0:10 0:09 0:84þ0:09 0:09 0:89þ0:07 0:06 0:80þ0:08 0:07 0:81þ0:06 0:06 0:78þ0:06 0:06

244/234 321/234 309/234 243/207 384/234

Spectra

v2 /dof 232/175 238/172 288/175 244/172 328/172

INTEGRAL/SPI flux in the 30–100 keV is estimated using the model parameters given in (Jourdain & Roques 2008, 2020).  is for fixed parameters.

perhaps can lead to the somewhat flatter spectra above 100 keV. Though we are getting a flatter spectrum, we get consistent flux within 10–20% of the mean value (see Fig. 8). This gives enough confidence that there is a possibility of extending CZTI bandwidth up to 700 keV with better energy calibration of the low gain pixels.

4. Discussion and conclusions In this article, we have attempted to explore the sensitivity of CZTI in the sub-MeV region and have outlined a methodology of sub-MeV spectroscopy using Crab observations. For this purpose, we have used the single pixel mask-weighted spectral data in the 30–100 keV energy range (EB1), 2-pixel Compton events including low gain pixels in the 100–500 keV energy range (EB3), and the single pixel events including low gain pixels in the 100–700 keV energy range (EB2).

For this work, we used the calibration parameters obtained by Chattopadhyay et al. (2021) for similar work to explore the sub-MeV spectroscopic sensitivity for Gamma Ray Bursts (GRBs). The advantage in the case of GRBs is the higher signal strength and, in particular, the availability of simultaneous background events before and after the burst. In the case of persistent sources, the unavailability of simultaneous background spectra makes the selection of proper blank sky flux and its subtraction extremely important (Section 2.2), particularly when the signal to noise ratio is relatively low. For this work, the spectral response was generated using a simple Gaussian energy distribution for simplicity (Section 2.3), which we plan to improve later with the use of a more physical line profile model based on charge trapping and diffusion (Chattopadhyay et al. 2016). We applied these techniques for spectral analysis of the Crab, where we used a broken power law (bknpower in XSPEC) for the spectral fitting. The spectral fits show sufficient flux sensitivity of CZTI to carry

Page 8 of 9

Figure 4. Best-fit parameter of all the selected observations obtained after simultaneous fitting of EB1 and EB3 using broken power law model. Red diamonds represent higher energy slope (PhoIndx2) and the blue squares represent low energy slope (PhoIndx1). The dashed-dot horizontal lines represent the INTEGRAL value of PhoIndx1 (blue) and PhoIndx2 (red) in the corresponding energy regime.

J. Astrophys. Astr.

(2021) 42:67

Figure 6. Estimated flux of the selected observations after fitting EB1 and EB3 simultaneously using broken power law. The blue diamond represents the CZTI value of the Crab flux. The dashed-dot horizontal line represent the INTEGRAL value of the Crab flux in the corresponding energy band.

normalized counts s−1 keV−1

67

0.1 0.01 10−3 10−4 10−5 4

Δχ2

2 0 −2 50

100 200 Energy (keV)

500

Figure 7. Broadband spectra of the Crab (ObsID 406, 114 ks) fitted with a broken power law. The black, green and red colors are used for EB1, EB2 and EB3 respectively.

Figure 5. Corner plot of the phoIndx2 verses Norm for all the five ObsIDs plotted together for EB1 and EB3. Different colors of contours represents different observations as shown in legend. The peak value of the PhoIndx2 is þ0:71 2:04þ0:48 0:44 and the Norm is 6:390:63 .

out spectroscopy for ON-axis bright sources (see Table 2) up to 500 keV, and it can be extended up to 700 keV with better gain calibration. We find that the overall flux measured by CZTI in 100–700 keV band (EB2) is consistent throughout the observations within 20% from the mean value. For the single event and Compton event spectra in 30–500 keV (EB1 and EB3), the low energy slope (PhoIndx1) and higher

energy slope (PhoIndx2) agree reasonably with the INTEGRAL/SPI result. However, for the combined spectral fit over all three bands (EB1, EB2, EB3), the spectral parameters are not fully consistent with the results reported by INTEGRAL/SPI with the high energy slope (PhoIndx2) showing a possible flattening. This could be either due to spectral calibration or incorrect background estimation at higher energies above 400 keV. We note that the measured flux by CZTI is lower by  31% than that estimated by INTEGRAL/SPI. However, it should also be noted that in general, it is quite difficult to make a comparison of flux measurements of the two instruments, particularly in hard X-rays, due to the difficulty in measurement of the absolute effective areas of various instruments.

J. Astrophys. Astr.

(2021) 42:67

Page 9 of 9

67

bright sources and add to the better understanding of the emission mechanisms in these sources. Acknowledgements

Figure 8. Estimated flux of the selected observations after fitting EB1, EB2, and EB3 simultaneously using broken power law. The blue diamond corresponds to Crab flux in the 100–700 keV energy band (EB2). The red dashed-dot horizontal lines represent the mean value (0:78þ0:05 0:05 ) of the Crab flux of all the five selected observations added together in the corresponding energy range.

Considering the errors in our flux measurements (  15%) and the 5% accuracy claimed for Crab flux measurements (Jourdain & Roques 2020), there appears to be a  2 sigma difference in the flux measurements of Crab made by CZTI as compared to INTEGRAL/SPI. We plan to do a cross-calibration with simultaneous NuSTAR observations and try to understand the effective area calibration of CZTI in a future work. To summarize, we find that CZTI has sufficient spectral sensitivity in the sub-MeV region for the ONaxis sources. The parameters are well constrained in the 30–500 keV range, but the spectra become flatter above 500 keV. In future, we plan to develop better background subtraction methods, and at the same time, we attempt to develop a detailed background model using 5-years of CZTI data. Moreover, we plan to investigate the gain of the low gain pixels in more detail; in particular, we look for various high energy background lines by proton induced radio-activation as shown by Odaka et al. (2018). With the better characterization of the low gain pixels and background subtraction methods, CZTI is expected to provide sensitive spectroscopic information for various hard X-rays sources such as Cygnus X-1 and other

This research is supported by the Physical Research Laboratory, Ahmedabad, Department of Space, Government of India. We acknowledge the ISRO Science Data Archive for AstroSat Mission, Indian Space Science Data Centre (ISSDC) located at Bylalu for providing the required data for this publication and Payload Operation Center (POC) for CZTI located at Inter University Centre for Astronomy & Astrophysics (IUCAA) at Pune for providing the data reduction software.

References Agostinelli S. et al. 2003, NIMPA, 506, 250 Arnaud K. A. 1996, ASPC, 101, 17 Band D. et al. 1993, ApJ, 413, 281 Bhalerao V. et al. 2017, J. Astrophys. Astr., 38, 31 Chattopadhyay T. et al. 2016, SPIE, 9905, 99054D Chattopadhyay T. et al. 2014, ExA, 37, 555 Chattopadhyay T. et al. 2019, ApJ, 884, 123 Chattopadhyay T. et al. 2021, J. Astrophys. Astr., 42. https://doi.org/10.1007/s12036-021-09718-2 Foreman-Mackey D. 2017, ascl.soft, ascl:1702.002 Jourdain E., Roques J. P. 2008, Proceedings of the 7th Integral Workshop (Proceedings of Science), 143 Mate S. et al. 2021, J. Astrophys. Astr., 42. https://doi.org/ 10.1007/s12036-021-09763-x Jourdain E., Roques J. P. 2020, ApJ, 899, 131 Kirsch M. G. et al. 2005, SPIE, 5898, 22 Kuiper L. et al. 2001, A&A, 378, 918 Ling J. C., Wheaton W. A. 2003, ApJ, 598, 334 Massaro E. et al. 2000, A&A, 361, 695 Mineo T. et al. 2006, A&A, 450, 617 Odaka H. et al. 2018, NIMPA, 891, 92 Paul B. 2013, IJMPD, 22, 1341009 Singh K. P. et al. 2014, SPIE, 9144, 91441S Strickman M. S. et al. 1979, ApJL, 230, L15 Toor A., Seward F. D. 1974, AJ, 79, 995 Vadawale S. V. et al. 2015, A&A, 578, A73 Vadawale S. V. et al. 2018, NatAs, 2, 50

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:63 https://doi.org/10.1007/s12036-021-09707-5

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

AstroSat-CZTI as a hard X-ray pulsar monitor K. G. ANUSREE1,* , D. BHATTACHARYA2, A. R. RAO2,5, S. VADAWALE3,

V. BHALERAO4 and A. VIBHUTE2 1

School of Pure and Applied Physics, Mahatma Gandhi University, Kottayam 686 560, India. Inter-University Centre for Astronomy and Astrophysics, Pune 411 007, India. 3 Physical Research Laboratory, Ahmedabad 380 009, India. 4 Indian Institute of Technology Bombay, Mumbai 400 076, India. 5 Tata Institute of Fundamental Research, Mumbai 400 005, India. *Corresponding Author. E-mail: [email protected] 2

MS received 30 October 2020; accepted 18 January 2021 Abstract. The Cadmium–Zinc–Telluride Imager (CZTI) is an imaging instrument onboard AstroSat. This instrument operates as a nearly open all-sky detector above 60 keV, making possible long integrations irrespective of the spacecraft pointing. We present a technique based on the AstroSat-CZTI data to explore the hard X-ray characteristics of the c-ray pulsar population. We report highly significant (  30r) detection of hard X-ray (60–380 keV) pulse profile of the Crab pulsar using  5000 ks of CZTI observations within 5 to 70  of Crab position in the sky, using a custom algorithm developed by us. Using Crab as our test source, we estimate the off-axis sensitivity of the instrument and establish AstroSat-CZTI as a prospective tool in investigating hard X-ray characteristics of c-ray pulsars as faint as 10 mCrab. Keywords. Pulsars: individual (Crab—PSR J0534?2200)—calibration—hard X-ray—piggyback data— LAT pulsars.

1. Introduction Pulse profiles of rotation powered pulsars are shaped by the geometry of the emission region as well as the radiation processes at work in pulsar magnetosphere (Watters & Romani 2011; Pierbattista 2014). The emission is usually broadband, covering radio through c-rays. However, emission at different wavebands typically arises in different regions of the magnetosphere, giving rise to pulse profiles that are wavelength dependent. There exist a variety of magnetospheric models (e.g. Cheng et al. 1986; Harding & Muslimov 2004; Pe´tri 2011; Cerutti 2017) which differ in their prediction of the shape and distribution of acceleration and radiation zones. Broadband Spectral Energy Distribution (SED) and the energy dependence of the pulse shape and arrival time provide clues to the This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

distribution of these zones and have been used to constrain the magnetospheric geometry and to discriminate between theoretical models (e.g. Romani & Yadigaroglu 1995; Cheng 2000; Abdo 2009, 2010b; Bai 2010; Du et al. 2012; Pierbattista 2012, 2014). The number of known c-ray pulsars stood at just seven until the launch of NASA’s Fermi Large Area Telescope (LAT) in 2008. Since then, the continuous accumulation of data, together with highly efficient searching algorithm, has resulted in the number of pulsars detected by LAT in the c-ray band to mount to 253 (Ray & Smith 2020), of which 71 have no radio counterpart. The modelling of c-ray and radio emission together can provide important constraints on the global magnetospheric properties (e.g. see Pe´tri & Mitra 2020). For radio-quiet c-ray pulsars, the magnetospheric X-ray emission provides the only additional clue to the emission process. The radiation energy output of a rotation powered pulsar typically peaks in the X-ray/c-ray region. However, only 18

63

Page 2 of 8

pulsars have been detected in the X-ray band till date (Caraveo 2014; Kuiper 2015) and the SED of many of them are sparsely sampled. Using ephemerides determined from LAT data, four of the LAT radioquiet pulsars have been observed at photon energies \20 keV (Lin et al. 2010, 2013, 2014; Caraveo 2010; Marelli 2014). These fall in the category of ‘Gemingalike’ pulsars, for which the profile and spectra are known at soft X-rays (i.e.,\ 20 keV) and c-rays alone (PSR B0633?17; Halpern & Holt 1992; Bertsch 1992). Unfortunately, at energies above 20 keV, the major X-ray observatories have had relatively low sensitivity. Additional deep observations in the hard X-ray band would therefore augment the existing information, providing a better estimate of spectral shape and bolometric luminosity of these and other pulsars hitherto undetected in X-rays. Three major emission mechanisms operate in pulsar magnetospheres, namely synchrotron radiation, curvature radiation and inverse-Compton scattering. A wellmeasured SED can distinguish the relative contributions of these components, leading to a model of the particle energy distribution in the emission zones. To increase the sample of pulsar detections in the hard X-ray/soft c-ray bands, and to investigate how their properties fit in the general picture emerging from the theoretical studies of the Fermi’s young gamma-ray pulsars, we need ‘open all-sky X-ray detectors’. The pulsar spectra are steep at high energies. In general, photon flux of the young/middle-aged LAT c-ray pulsars can be represented by a power-law with a simple exponential cutoff, i.e. Fc ¼ k  ðEc =E0 ÞC  expððEc =Ec Þb Þ, where b  1 and Ec is the photon energy, E0 a normalisation energy, Ec the cutoff energy and k is the normalization. The photon index C has been found to lie in the range 0:4 to 1:7 (Kuiper & Hermsen 2015). Even with a highly sensitive detector with sub-second timing resolution, it takes extremely long exposures to detect a significant number of photons from a typical c-ray pulsar. Such long integrations are not affordable by any of the missions at present. Open all-sky detectors, on the other hand, can collect photons during other observations, making it possible to search for these pulsars. The first Indian multi-wavelength Satellite AstroSat was launched in 2015 with five instruments onboard (Singh et al. 2014). One of them, the Cadmium–Zinc– Telluride Imager (CZTI; Bhalerao et al. 2017), can detect photons in 20–380 keV energy range. Its housing and collimators are made of aluminium alloy and thin Tantalum shields that define its low-energy field of view

J. Astrophys. Astr.

(2021) 42:63

but allow sufficient uncollimated penetration above  60 keV to make CZTI an excellent wide-angle monitor at higher energies, covering roughly one-third of the sky at all times. This monitoring capability has been leveraged to detect many transients, including over 300 Gamma-Ray Bursts (Sharma et al. 2021). A more detailed description of CZTI can be found in Bhalerao et al. (2017). The aim of this paper is to assess the suitability of CZTI for the detection and study of pulsars in the hard X-ray band, using its off-axis detection capability. During CZTI pointing observations, photons from candidate hard X-ray sources that shine in through the walls will come to piggyback on any ongoing observation, with varying sensitivity depending on the pointing. Arrival times of the photons from a pulsar carry the signature of its spin period, enabling us to search the data for hard X-ray pulsars with known ephemeris. This presents the detection of Crab Pulsar in the energy range 60–380 keV from off-axis CZTI observations, using a custom algorithm developed by us. The Crab pulsar (PSRJ0534?2200) is well-studied at all wavelengths. Using Crab as our calibration source, we determine the off-axis sensitivity of the instrument. This paper is structured as follows. Section 2 describes the instruments used in this work and the analysis of data, followed by a discussion of results in Section 3 which includes the comparison of hard Xray pulse profiles of the Crab pulsar obtained from CZTI with c-ray profiles from LAT data. Finally, we assess the potential of AstroSat-CZTI in the investigation of hard X-ray counterparts of c-ray pulsars.

2. Data and analysis This work is based on data from AstroSat-CZTI pointing observations that were released for public use on or before 30th April 2019. We have also used publicly available archival data from NASA’s FermiLAT mission. All material informations about the instruments and data, along with the characteristics of analysis methods, are described in this section. Moreover, we also discuss some of our checks against the vulnerabilities anticipated during long integration.

2.1 AstroSat-CZTI CZTI is the AstroSat instrument primarily designed for simultaneous hard X-ray imaging and spectroscopy of

J. Astrophys. Astr.

(2021) 42:63

celestial X-ray sources in the energy band 20–150 keV. Its functioning employs the technique of coded mask imaging. The CZTI instrument bears a two-dimensional coded aperture mask (CAM) above its pixellated, 5-mm thick solid-state CZT detector modules spread over four quadrants. Passive collimators are placed in each quadrant of CZTI that support the coded mask. The set up defines a 4.6  4.6 deg field of view in  20–100 keV range with an angular resolution of  8 arcmins. A critical aspect of the collimators and mask is that these are designed to be effective up to  100 keV above which they become transparent. The transparency being a function of energy and angle of incidence, appreciable sensitivity for off-axis sources extends down to  60 keV (see Bhattacharya et al. 2018). The 976 cm2 of the detector’s total geometric area is distributed over 16384 pixels, with 4096 pixels in each independent quadrant. After the launch, about 15 percent of the pixels were disabled for having shown excessive electronic noise. Nearly 25 percent of the remaining pixels seemed to have an inadequate spectroscopic response. Considering that the coded mask has a  50% open fraction, the total effective area at normal incidence is  420 cm2 in all active pixels at energies below 100 keV. The detected events are recorded with a time resolution of 20 ls. The absolute timestamp assigned to CZTI events is estimated to have a jitter of  3 ls RMS (Bhattacharya 2017) and a fixed offset with respect to Fermi of 650  70 ls (Basu et al. 2018). 2.1.1 Analysis of CZTI data. All available CZTI pointing observations during MJD 57366-58362, with the Crab pulsar within 5–80 degrees from normal incidence, were selected for this work, resulting in a total exposure of 5096ks. The merged Level-1 data of all the selected CZTI observations were reduced to Level-2 using standard CZTI analysis software. Details and the sequence of analysis modules can be found in the latest version of the CZTI user-guide available at the ASSC website1. There are intervals during pointing observation where data is absent due to SAA passage and data transmission loss. There are also intervals when the earth occults the target source. To generate science products from such observations, identifying such intervals and removing the data for that duration by adequately accounting for the gaps is essential. This task is performed by the module cztgtigen. It generates Good Time Interval

Page 3 of 8

(GTI) files based on various threshold parameters. We generated custom GTI files to take into consideration the earth occultation of the off-axis source and filtered the original event file accordingly. These filtered, clean event files were used in the subsequent steps of analysis. In this work, we have used AstroSat CZTI observations with on-axis pointing of the Crab pulsar as well as others when the pulsar is off-axis at angles between 5 to 70 deg (see below for the choice of this angle range). We have also used Fermi-LAT observations of the Crab pulsar spanning a similar time range for comparison. A brief description of the data sets used is provided in Table 1. The arrival times of the CZTI events at the spacecraft were converted to those at the Solar-system Barycenter using the JPL DE200 solar-system ephemeris. This was done using the tool as1bary, a version of NASA/HEASOFT AXBARY package, customised for AstroSat, using the well-known astrometric position of the Crab pulsar. We developed a custom code to fold the barycentered event data spanning many months but including frequent gaps. This code assigns an absolute phase to each recorded event using the polynomial model timing solutions known as SSB polycos generated using tempo2 (Hobbs et al. 2006). Polycos predict a pulsar’s parameters at a particular epoch. A polyco file contains pulsar ephemerides over a short period, typically hours, in simple polynomial expansion. All the polycos were generated at an epoch in the centre of each 6-hour interval in this work. These 6-hour ephemerides were then used for folding the LAT data as well as the CZTI data. As an initial test of the custom code, Fermi-LAT c-ray events for a few pulsars were folded and compared with their published timing models. The SSB polycos were generated from LAT timing models available publicly at Fermi’s website2. Finally, the code was tested on AstroSatCZTI data. For this, publicly available AstroSat-CZTI data from six pointing observations of Crab pulsar (Table 1, first row) were selected and reduced using the default CZTI data analysis pipeline. The 30–60 keV pulse profile, thus obtained, is shown in Fig. 2(a). The Crab pulsar ephemeris derived from Fermi observations and used in this work are given in Table 2 along with the reference epochs. To compare the pulse profiles between hard X-rays and c-rays, we folded all the events extracted from 2

1

http://astrosat-ssc.iucaa.in/.

63

https://confluence.slac.stanford.edu/display/GLAMCOG/LAT? Gamma-ray?Pulsar?Timing?Models.

63

Page 4 of 8

J. Astrophys. Astr.

(2021) 42:63

Table 1. Brief summary of the observations. Participating telescopes and their instruments, the energy range chosen for the work, range of MJDs for which the data have been collected for this work and the total exposure achieved are listed. Telescope

Energy range (keV)

Start MJD

Stop MJD

Exposure (ks)

30–60 60–380

57290 57290

58232 58232

519 4752

0.1–300

57290

58500

5962

AstroSat-CZTI (Crab at nominal pointing) AstroSat-CZTI (Crab within 5–70 degree of nominal pointing) Fermi-LAT (within 1 degree of c-ray position of Crab pulsar)

...

AstroSat-CZTI observations and those obtained from Fermi observation within MJD 57290-58500 using the c ray ephemeris mentioned above. For all CZTI data analysis, we have directly combined the data of all four quadrants, as they run on a synchronised time reference (see Fig. 1). We initially folded the off-axis data separately for every 10-degree angle interval and found that when the source is located beyond 70 degrees from the pointing axis, the signal-to-noise ratio in a given integration time drops significantly below those for smaller off-axis angles. We, therefore, decided to restrict the accumulation of off-axis data to the angle range of 5 to 70 deg (below 5 deg, the source would appear in the main FoV). With the phase reference as presented in Table 3, we consider the ‘off-pulse’ region of the profile, phase 0.5 through 0.8, as the background. The signal-to-

104

counts/bin

(9.2842295751e-32) 55555 50739 55107.807158553 56730.15526586924 DE405 TDB

Second derivative, f€ð1020 s3 Þ

5000

PEPOCH POSEPOCH DMEPOCH TZRMJD Solar system ephemeris model Time system

First derivative, f_ð1010 s2 Þ

0

Third derivative, f ð1028 s4 Þ

05:34:31.94 ?22:00:52.1 54686-58767 29.7169027333 (0.0000148379) 3.71184342371 (4.6591493867e-17) 3.3226958153 (1.1931513864e-24) 1.2860657170

1.5×104

Parameters Right ascension, a Declination,d Valid MJD range Pulse frequency, f ðs1 Þ

Q0 Q1 Q2 Q3

2×104

Table 2. Fermi-LAT c-ray timing solution of PSR J0534?2200 by Fermi Timing Observers. The numbers in parentheses denote 1r errors in the parameters.

0

0.5

1 pulse phase

1.5

2

Figure 1. Phase histograms of PSR J0534?2200 in 60-380 keV band obtained from 5–70 degree off-angle data (this work) from the four different quadrants (Q0, Q1, Q2 and Q3) of CZTI operating as independent detectors. All the quadrants are found to be time aligned with no measurable relative delay Table 3. Phase component definitions for the Crab pulsar (Abdo 2010) adopted in this study Component Peak 1 Peak 2 Bridge Off pulse

Abbreviation

Phase interval

Width

P1 P2 Bridge OP

0.87–1.07 0.27–0.47 0.098–0.26 0.52–0.87

0.20 0.20 0.162 0.35

noise ratio is calculated as the ratio of the peak count to the standard deviation of the counts in the off-pulse region. We also divided the obtained X-ray events into several different energy bands to check consistency, as shown in Fig. 2(b).

Page 5 of 8

150−380 keV

CZTI

100−150 keV

CZTI

60−100 keV

CZTI

0

0

0.5

0.5

1

CZTI 30−60 keV

0.5

1

0

0.5

normalised intensity

CZTI (this work) 60−380 keV

0

1

0

0

0.5

0.5

Fermi−LAT 0.1−300 GeV

1

normalised intensity

63

1

(2021) 42:63 1

J. Astrophys. Astr.

0

0.5

1 pulse phase

1.5

0

2

(a)

0.5

1 pulse phase

1.5

2

(b)

Figure 2. The pulse profile of the Crab pulsar observed with Fermi-LAT (black) and the AstroSat-CZTI (purple). The integration time in the 0.1–300 GeV LAT profile is  6000 ks, while that in the 60–380 keV CZTI profiles is  5000 ks. The bottom panel of Fig. 2(a) shows the 30–60 keV profile obtained by folding  600 ks of on-axis pointing observations of the Crab pulsar with AstroSat-CZTI. In CZTI profiles, data from multiple observations during MJD 57290-58232 with PSR J0534?2200 within 5–70 degree from pointing axis have been accumulated. Two cycles in 512 phase bins are plotted for clarity. The energy dependence of the profile is apparent here: below 60 keV, the first peak P1 (near phase 1.0) dominates, and emission in the bridge phase interval is moderate, while above 60 keV the second peak dominates with significant bridge emission. At MeV energies, P1 starts to dominate again with strongly reduced bridge emission. A detailed reference to the energy dependence of Crab pulse profile is available in Kuiper et al. (2001)

2.2 Fermi-LAT The LAT instrument is described by Atwood et al. (2009). We have used the already publicly available data from LAT. A unique value of the LAT data is that a pulsar’s discovery in c-rays often enables the immediate measurement of the pulsar parameters over the ten-year span in which the LAT has been operating. LAT data have been used to find precise timing solutions for many pulsars, including radio-quiet and radio-faint pulsars (Ray et al. 2011; Kerr 2015; Clark 2017). 2.2.1 Analysis of Fermi-LAT data. In order to get a significant c-ray profile to compare with the CZTI results, we took all available photon data for the LAT source PSRJ0534?2200 from MJD 57290 through 58500, bracketing the CZTI data used in this work. It leads to a total effective exposure time of 5962-kilo

seconds, as the observatory scans the entire sky once every three hours. We used the HEADAS-FTOOLS3 on HEAsoft-ver 6.27. (Blackburn 1995) to perform the data reduction. We obtained the Fermi-LAT data in the energy range of 0.1–300 GeV within a circular region of interest (ROI) with a 1-degree radius from the decided c-ray position of PSR J0534?2200. We used Pass 8 data and selected events in the ‘Source’ class (i.e. event class 2). We also excluded the events with zenith angles larger than 105 degrees to reduce the contamination by the Earth albedo c-rays. We used gtbary tool in Fermi science tools to apply barycentric corrections to photon arrival times in LAT event files using corresponding Fermi orbit files. After barycentering the events using Fermi science tool4 3

http://heasarc.gsfc.nasa.gov/ftools. https://fermi.gsfc.nasa.gov/ssc/data/analysistools/overview.html

4

63

Page 6 of 8

gtbary, the absolute phase of each event in 0.1–300 GeV was determined by the custom code developed by us, with 6-hour SSB polycos generated using the timing parameters in Table 2. 3. Discussion of results Figure 2 shows the folded pulse profile of the Crab pulsar (period:  33 ms) using the LAT and the CZTI instruments. The CZTI profiles (purple in colour) accumulate data from multiple observations spread over several months, with Crab position within 5 to 70 degrees away from the nominal pointing direction. The total integration time in these CZTI profiles is about 5000 ks, giving a signal to noise ratio of  30 in the energy integrated profile and over  15 in the energyresolved ones. The long-known energy dependence of the Crab pulse profile can be seen when compared with the softer bands (e.g. 30–60 keV in CZTI) in Fig. 2(a). The left peak is taller at lower energies while the right peak dominates at high energies. The bridge emission connecting the two peaks is also seen relatively stronger at higher X-ray energies. The normalised light curves in three separate hard X-ray bands are shown in Fig. 2(b). The right peak(P2) grows, but the left peak (P1) again starts to dominate at very high energies, beyond (  10 MeV) in the c-regime, as seen in Fig. 2(a). For a more quantitative comment on the morphology change observed, we determined the intensity ratios P2/ P1 and Bridge/P1 in three separate hard X-ray bands shown in Fig. 2(b), adopting the phase interval definitions of Abdo (2010) as shown in Table 3. The values obtained for P2/P1 ratios are 1:301  0:002, 1:340  0:002, 1:385  0:002 and that for Bridge/P1 are 0:536  0:002, 0:574  0:002, 0:609  0:003 in 60–100 keV, 100–150 keV and 150–380 keV bands respectively. The gradual increase of both the ratios is consistent with that reported by Kuiper et al. (2001). This validates the accuracy and stability of AstroSatCZTI clocks and the robustness of our custom algorithm for long-term phase-connected analysis with available accurate c-ray/radio timing models. Integrating the well established broken power-law model spectrum for Crab (Ulmer et al. 1995), we calculated the hard X-ray flux in 60-380 keV band to be 0.011 ph cm2 s1 . The observed CZTI detection count rates of 0:876  0:01 cps in 60–380 keV band translates to an average CZTI off-axis effective area of  80 cm2 , averaged over 5–70 degree off-axis. This is about 20 percent of the on-axis effective area at lower energies.

J. Astrophys. Astr.

(2021) 42:63

Assuming a Crab-like spectrum and based on the observed count rates mentioned above, we estimate the possible integration time required for a 5r detection of a 10 mCrab off-axis pulsed source with CZTI to be  14 Mega seconds. With the continuous accumulation of CZTI data since launch, one can foresee the detection of more than 90 per cent of the c-ray pulsars from the second Fermi/LAT pulsar catalog in the CZTI hard X-ray (60–380 keV) band, as shown in Fig. 3. 4. Conclusion We have presented a successful attempt to detect the Crab pulsar by the AstroSat CZT Imager from pointings where the pulsar was off-axis, at angles ranging

Figure 3. Detectability of Fermi-LAT pulsars with the CZTI. The green horizontal line marks the estimated flux of the faintest detectable hard X-ray pulsar using five years of archival AstroSat-CZTI data. The red diamonds represent the hard X-ray pulsar candidates discovered/observed in the soft X-ray band (\20 keV). The brightest c-ray pulsars Vela (PSRJ0835-4510) and Geminga (PSRJ0633?1746), with fluxes of 7 and 3 Crab respectively, have been omitted from the figure to clearly depict the rest of the population

J. Astrophys. Astr.

(2021) 42:63

from 5 to 70 degrees. This is possible because the Collimator and the housing of the CZTI become gradually transparent at energies above 60 keV, turning it into an all-sky detector at higher energies. Accumulating an off-axis exposure of  5 Ms, we obtain a  30r pulse profile of the Crab pulsar in the 60–380 keV energy band. Energy-resolved pulse profiles constructed at multiple sub-bands reproduce the known energy dependence of the profile shape. This demonstrates the capability of AstroSat-CZTI to act as a hard Xray pulsar monitor, in 60–380 keV band, with an average off-axis effective area of  80 cm2 . Our results establish that the CZTI time stamps possess sufficient long-term stability to carry out phase connected timing spanning many years. We estimate that with continued accumulation of data, it will become possible for the CZTI to detect pulsars with hard X-ray fluxes down to  10 mCrab, thus making a large majority of Fermi/LAT pulsars accessible for study in the 60–380 keV energy band. Such a survey is currently ongoing with several successful detections already made. These results will be reported elsewhere.

Acknowledgements This publication makes use of data from the CZTI onboard Indian astronomy mission AstroSat, archived at the Indian Space Science Data Centre (ISSDC). The CZT Imager instrument was built by a TIFR-led consortium of institutes across India, including VSSC, ISAC, IUCAA, SAC, and PRL. The Indian Space Research Organisation funded, managed and facilitated the project. We extend our gratitude to CZTI POC team members at IUCAA for helping with the augmentation of data. We thank Fermi Timing Observers Paul Ray and Kerr Mattew for their timely and favourable response in providing LAT ephemeris for Crab pulsar and helping with queries related to SSB polyco generation using tempo2. We thank the anonymous referee for his/ her valuable suggestions to improve the paper. We would like to thank Avishek Basu, Karthik Rajeev and Atul Mohan for useful discussions. We thank IUCAA HPC facility where we carried out all the analysis. Anusree K. G. acknowledges support for this work from DST-INSPIRE Fellowship grant, IF170239, under Ministry of Science and Technology, India.

Page 7 of 8

63

References Abdo A. A., Ackermann M., Atwood W. B., Bagagli R. et al. 2009, Astrophys. J. 696, 1084 Abdo A. A. et al. 2010a, Astrophys. J. Suppl. Series, 208, 17 Abdo A. A., Ackermann M., Ajello M. et al. 2010b, Astrophys. J., 708, 1254 Atwood W. B., Abdo A. A., Ackermann M. et al. 2009, Astrophys. J., 697, 1071 Bai X. N., Spitkovsky A. 2010, Astrophys. J., 715, 1282 Basu A., Joshi B. C., Bhattacharya D. et al. 2018, A&A, 617, A22 Bertsch D. L., Brazier K. T. S., Fichtel C. E. et al. 1992, Nature, 357, 306 Bhalerao V., Bhattacharya D., Vibhute A. et al. 2017, J. Astrophys. Astr., 38, 31 Bhattacharya D. 2017, J. Astrophys. Astr., 38, 51 Bhattacharya D., Dewangan G. C., Antia H. M. et al. 2018, AstroSat Handbook, http://www.iucaa.in/astrosat/ AstroSat\_handbook.pdf Blackburn J. K. 1995, Astronomical Data Analysis Software and Systems IV, 77, 367 Caraveo P. A. 2014, Gamma-ray pulsar revolution, Annual Review of Astronomy and Astrophysics, 52 Caraveo P. A., De Luca A., Marelli M. et al. 2010, Astrophys. J. Lett., 725, L6 Cerutti B., Beloborodov A. M. 2017, Space Science Rev., 207, 111 Cheng K. S., Ho C., Ruderman M. 1986, Astrophys. J., 300, 522 Cheng K. S., Ruderman M., Zhang L. 2000, Astrophys. J., 537, 964 Clark C. J., Wu J., Pletsch H. J. et al. 2017, Astrophys. J., 834, 106 Du Y., Qiao G., Wang W. 2012, Astrophys. J., 748, 84 Halpern J. P., Holt S. S. 1992, Nature, 357, 222 Harding A., Muslimov A. 2004, 35th COSPAR Scientic Assembly, 35, 562. Hobbs G. B., Edwards R. T., Manchester R. N. 2006, MNRAS, 369, 655 Kerr M., Ray P. S., Johnston S. 2015, Astrophys. J., 814, 128 Kuiper L., Hermsen W., Cusumano G. et al. 2001, A&A, 378, 918 Kuiper L., Hermsen W. 2015, MNRAS, 449, 3827 Lin L. C., Huang R. H., Takata J. et al. 2010, Astrophys. J. Lett., 725, L1 Lin L. C. C., Hui C. Y., Hu C. P. et al. 2013, Astrophys. J. Lett., 770, L9 Lin L. C. C., Hui C. Y., Li K. T. et al. 2014, Astrophys. J. Lett., 793, L8 Marelli M., Harding A., Pizzocaro D. et al. 2014, Astrophys. J., 795, 168 Muslimov A. G., Harding A. K. 2004, Astrophys. J., 606, 1143 Pe´tri J., 2011, MNRAS, 412, 1870 Pe´tri J., Mitra D. 2020, MNRAS, 491, 80

63

Page 8 of 8

Pierbattista M., Grenier I. A., Harding A. K., Gonthier P. L. 2012, A&A, 545, A42 Pierbattista M., Grenier I., Harding A., Gonthier P. 2014, Radio and c-ray light-curve morphology of young pulsars: Comparing Fermi observations and simulated populations, Saint-Petersburg, Russia, July 28–August 1, 2014, 99 Ray P. S., Kerr M., Parent D. et al. 2011, Astrophys. J. Suppl. Series, 194, 17

J. Astrophys. Astr.

(2021) 42:63

Ray P. S., Smith D. A. et al. 2020, https://confluence.slac.stanford.edu/display/GLAMCOG/ Public?List?of?LAT-Detected?Gamma-Ray?Pulsars/ Romani R., Yadigaroglu I.-A. 1995, Astrophys. J., 438, 314 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, SPIE, 9144E, 1S Ulmer M. P., Matz S. M., Grabelsky D. A. et al. 1995, Astrophys. J., 448, 356 Watters K. P., Romani R. W. 2011, Astrophys. J., 727, 123

 Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:89 https://doi.org/10.1007/s12036-021-09746-y

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

UOCS. IV. Discovery of diverse hot companions to blue stragglers in the old open cluster King 2 VIKRANT V. JADHAV1,2,* , SINDHU PANDEY3 , ANNAPURNI SUBRAMANIAM1

and RAM SAGAR1,3 1

Indian Institute of Astrophysics, Sarjapur Road, Koramangala, Bangalore 560 034, India. Joint Astronomy Programme and Department of Physics, Indian Institute of Science, Bangalore 560 012, India. 3 Aryabhatta Research Institute of Observational Sciences, Manora Peak, Nainital 263 002, India. *Correspondence author. E-mail: [email protected] 2

MS received 7 November 2020; accepted 15 February 2021 Abstract. King 2, one of the oldest clusters in the Milky Way, with an age of  6 Gyr and distance of  5700 pc, has been observed with UVIT payload on the AstroSat. With membership information derived from Gaia EDR3, the cluster is found to have 39 blue straggler stars (BSSs). We created multi-wavelength spectra-energy distributions (SED) of all the BSSs. Out of 10 UV detected BSSs, 6 bright ones were fitted with double component SEDs and were found to have hotter companions with properties similar to extreme horizontal branch (EHB)/subdwarf B (sdB) stars, with a range in luminosity and temperature, suggesting a diversity among the hot companions. We suggest that at least 15% of BSSs in this cluster are formed via mass-transfer pathway. When we compared their properties to EHBs and hotter companions to BSS in open and globular clusters, we suggest that EHB/sdBs like companions can form in binaries of open clusters as young as 6 Gyr. Keywords. Open star clusters (1160)—blue straggler stars (168)—extreme horizontal branch stars (513)— B subdwarf stars (129)—ultraviolet astronomy (1736)—spectral energy distribution (2129)—binary stars (154).

1. Introduction The evolution of binary systems strongly depends on the initial orbital parameters and its further evolution, where any change in their orbits can lead to a widely different evolution. If one of the stars evolves and fills its Roche lobe, the system will undergo mass transfer. Details such as duration and rate of mass transfer will depend on the orbits and masses of the binary stars. If such a binary is present in a star cluster, and the secondary of the binary has mass similar to the main sequence turnoff (MSTO) mass, then the secondary will become brighter than the MSTO and appear as a This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

blue straggler star (BSS). Otherwise, the secondary will be fainter than the MSTO and can be classified as a blue lurker. The blue lurkers are identified by their high stellar rotation (Leiner et al. 2019) or evidence of extremely low-mass (ELM) white dwarf (WD) companion (Jadhav et al. 2019). Depending on the evolutionary status of binary components, the binary system can be observed as main sequence (MS)?MS, contact binaries (Rucinski 1998), common envelop, MS?horizontal branch (HB; Subramaniam et al. 2016), MS?extreme HB (EHB; Singh et al. 2020), MS? subdwarf-B (sdB; Han et al. 2002), MS?WD (Jadhav et al. 2019), WD?WD (Marsh et al. 1995) and many more combinations. The binary evolution also depends on external factors such as collisions in a high-density environment which can decouple the binary (Heggie 1975) and

89

Page 2 of 14

J. Astrophys. Astr. (2021)42:89

tertiary star which can expedite the mass transfer/ merger by reducing the orbital separation (Kozai 1962). We are carrying out a long term project1 of characterising products of binary stars such as BSSs in open clusters (OCs). Ultraviolet imaging of binary systems reveals the presence of hotter companions in the binary system, given that the hotter companion is luminous in UV. Old OCs such as NGC 188 and NGC 2682 are rich with BSSs, binary stars and contain many such optically sub-luminous UV-bright companions (Subramaniam et al. 2016; Sindhu et al. 2019; Jadhav et al. 2019). Similar companions have been identified to BSSs in the outskirts of GCs (Sahu et al. 2019; Singh et al. 2020). Subramaniam et al. (2020) provides a summary of the BSSs and post mass transfer systems in star clusters. King 2 is one of the oldest clusters in the Milky Way, with an age of  6 Gyr and distance of  5700 pc (Table 1). However, it has been poorly studied due to its considerable distance and unknown membership information. For identifying and characterising hot BSSs and their possible companions, we obtained Ultra Violet Imaging Telescope (UVIT)/AstroSat observations of rich OC King 2 (a2000 ¼ 12: 75; d2000 ¼ þ58: 183; l ¼ 122: 9 and b ¼ 4: 7) under AstroSat proposal A02_170. Kaluzny (1989) presented the first optical colour-magnitude diagram (CMD) study of this distant cluster using BV CCD photometric data. This yielded a range of plausible ages and distances for different assumed reddenings and metallicities. The galactocentric distance of the cluster was estimated to be  14 kpc. Aparicio et al. (1990) (A90 hereafter) did a comprehensive study on the cluster using UBVR photometry and derived an age of 6 Gyr and a distance of 5.7 kpc for solar metallicity. They also indicated the presence of a good fraction of binaries in the MS. Tadross (2001) estimated a value of [Fe/H] ¼ 0:32 using the (U  B) colour excess

from the literature data, while Warren and Cole (2009; WC09 hereafter) derived a value of [Fe/H] ¼ 0:42  0:09 using spectroscopic data. These metallicity estimates are significantly sub-solar and inconsistent with the finding of A90. WC09 found a distance of 6.5 kpc and a slightly younger age,  4 Gyr, better fitted the optical CMD and 2MASS Ks, red clump if the reddening is adopted as EðB  VÞ ¼ 0:31 mag. This distance puts King 2 at RGC = 13 kpc, where its metallicity falls close to the trend of the galactic abundance gradients derived in Friel et al. (2002). There has been no proper motion study available for this cluster till Gaia DR2 (Gaia Collaboration et al. 2018). Cantat-Gaudin et al. (2018) provided a membership catalogue of King 2 with 128 members with Gaia DR2, and Jadhav et al. (2021) provided kinematic membership of 1072 stars (and 340 probable members) using kinematic data taken from Gaia EDR3. Above discussed optical photometric studies indicate a good number of post-MS hot stars in King 2. In fact, Ahumada and Lapasset (2007) have identified 30 BSS candidates based on the location of these stars in the cluster. We present the UVIT and the archival data used in this study in the next section, followed by analyses, results and discussion. 2. UVIT and archival data We observed King 2 with UVIT, which is one of the payloads on the first Indian multiwavelength space observatory AstroSat, launched on 28 September 2015. The observation was carried out by UVIT on 17 December 2016, simultaneously in two filters. The telescope has three channels with a set of filters in them: Far-UV (FUV; 130–180 nm), near-UV (NUV; 200–300 nm) and visible (VIS; 350–550 nm), where the VIS channel is intended to correct the drift of the

Table 1. Age, distance, reddening ðEðB  VÞÞ and metallicity of King 2 estimated by various investigators are listed.

1

Age (Gyr)

Distance (pc)

EðB  VÞ (mag)

Metallicity

References

6.02 6

5750 5690  65

0.31 0:31  0:02

–0.42 –0.5 to –2.2 –0.32

4 to 6

 7000

0.23 to 0.5

Dias et al. (2002) Aparicio et al. (1990) Tadross (2001) Kaluzny (1989)

UOCS: UVIT Open Cluster Study

J. Astrophys. Astr. (2021)42:89

Page 3 of 14

89

Figure 1. Two-component SED of BSS1. (a) Composite SED (green curve) is shown along with the observed flux (as black error-bars). The unfitted point (in this case: CAHA.R) is shown as orange dot. The cooler (BSS, orange dot-dashed curve) and hotter (blue dashed curve) component are also shown with their Teff and log g. The model, B component and residuals of noisy iterations are also shown as light coloured lines. (b) The fractional residual is shown for single component fit (orange dot-dashed curve) and composite fit (green solid curve). The fractional observational errors are also indicated on X-axis. (c) The v2i of each data point. (d) H–R diagram of the two components along with the isochrone for reference. The density distribution of the noisy B component fits is plotted in blue. (e) Teff  v2 distribution for the noisy B component fits coloured according to their radii. The dashed lines are the quoted limits of the temperature.

spacecraft (see Kumar et al. 2012; Tandon et al. 2017 for more details). The cluster was observed in one FUV (F148W, limiting magnitude 23 mag) and one NUV (N219M, limiting magnitude 22 mag) filter for exposure time of  2.7 ks. The FWHM of PSF in F148W and F219N images is 1:00 33 and 1:00 35 respectively. The data reduction was done using CCDLAB (Postma & Leahy 2017) and PSF photometry was performed using DAOPHOT package of IRAF (Tody 1993). More details of the reduction process are presented in Jadhav et al. (2021). We have detected ten member stars in either F148W and/or N219M filter. We obtained archival optical (UBVR) photometry data from A90 catalogue (Calar Alto Observatory; CAHA) and cross-matched with UVIT data using TOPCAT (Taylor 2005). The cluster was observed with GALEX under All-sky Imaging Survey (AIS) in NUV filter (exp. time  100 s). All the detected member stars were further cross-matched with photometric data from UV to IR wavelength bands obtained from GALEX (Bianchi et al. 2000), PANSTARRS PS1 (Chambers et al. 2016), Gaia EDR3 (Gaia Collaboration et al. 2020), 2MASS (Skrutskie et al. 2006), WISE (Wright et al. 2010) using virtual observatory tools in VOSA (Bayo et al. 2008).

3. SED fitting and colour magnitude diagrams The data were corrected for reddening ðEðB  VÞ ¼ 0:31  0:02Þ using Fitzpatrick (1999) and Indebetouw et al. (2005) and calibrated with the cluster distance of 5750±100 pc (we have overestimated the error to cover distance estimates from Dias et al. (2002) and A90). We have adopted the metallicity of [Fe/H]= -0.5 for all the stars and used Kurucz model spectrum (Castelli et al. 1997) for comparison. The SED fitting was done as follows: (1) We constructed the observed spectral energy distribution (SED) for all stars using the data from UV to IR wavelength, as mentioned above. The fluxes of all stars are given in Table 3. (2) Kurucz models (Castelli et al. 1997) of log g 2 ð3:0; 5:0Þ were fitted to optical and IR points ˚ ) using VOSA2 (Bayo et al. 2008). (above 3000 A There were some sources which showed UV excess in multiple UV points compared to the model fit. We selected such stars to be fitted with a two component SED. Otherwise, the single component fits are deemed satisfactory and are stated in the lower part of Table 2. 2

http://svo2.cab.inta-csic.es/theory/vosa/index.php

89

Page 4 of 14

(3) We used the cooler component parameters from above fits and then fitted a hotter component to the residual using Binary_SED_Fitting3. In preliminary double component fits, the hotter components were found to be compact objects, hence they were fitted with log g ¼ 5 Kurucz models. (4) Very small errors in PAN-STARRS PS1, Gaia EDR3 and A90 photometry led to ignoring relatively high error UV data-points, hence they were replaced with mean errors for better residual across all wavelengths. A few data points were removed to achieve better fits and lower v2 (see Fig. 1(a)). The error values given in Table 3 are original unmodified errors. The detailed SED fitting method is explained in Section 3.3 of Jadhav et al. (2019). (5) The best-fit parameters for single stars or cooler components are taken from the VOSA fits. The hotter component parameters are taken from the least v2 model in the two component fitting. (6) The errors in cooler component parameters are fairly low and are taken as the grid values. To derive errors in the hotter component parameters, we used a statistical approach. We first generated 100 iterations of observed SEDs with added Gaussian noise in each data point. These 100 SEDs were then fitted with double components. However, not all the double fits converged, hence we only kept hotter components with 6000 K\Teff \37000 K. Logarithmic distributions of the parameters from the noisy and converging iterations were then fitted with Gaussian distributions. FWHM of these Gaussian distributions are defined as the upper and lower limits of the fitting parameters (temperature, radius and luminosity). Figure 2 shows the CMD of 1412 cluster candidates identified from Gaia EDR3 with probability of over 50% (Jadhav et al. 2021) and are marked as grey points. Among these stars, we have selected 39 member stars brighter (G\17:5 mag) and bluer (GBP  GRP \1:1 mag) than the MSTO as BSSs. Seven of them were detected in F148W, and seven were detected in N219M (four in both filters). An isochrone of log age ¼ 9.7 is over-plotted on the CMD. 20 of the BSSs are detected in the NUV filter of GALEX. We fitted Kurucz model SEDs to all BSSs and found excess UV flux in 15 BSS (BSS 1, 2, 3, 4, 5, 7, 8, 9, 10, 19, 26, 28, 29, 33 and 36). Among these, 3

https://github.com/jikrant3/Binary_SED_Fitting Binary_SED_ Fitting is a python code which uses v2 minimisation

technique to fit two component SEDs.

J. Astrophys. Astr. (2021)42:89

(1990)

Figure 2. CMD of King 2 cluster candidates using Gaia EDR3 data. All BSS members are shown as blue circles. The UV bright BSS (see Section 3) are shown as red squares, and other UVIT detected BSS are shown as red X’s. The Gaia members are shown as black dots along with the PARSEC isochrone of log age = 9.7, [M/H] = –0.4, DM = 13.8 and EðB  VÞ ¼ 0:45.

BSS1, 2, 3, 4, 5 and 7 have multiple UV data points from UVIT or GALEX or both. Only these six were fitted with double component SEDs, because a hot component fit can be reliable if the number of UV data points is more than one. Hereafter, these six BSS will be referred to as ‘UV bright BSSs’ and others will be referred to as ‘UV faint BSSs’. The four BSS detected in UVIT but not fitted with hotter component are shown as red X’s in the CMD. We have shown an example of a double component SED fit of BSS1 in Fig. 1(a). The BSS1-A component is a BSS with 7750 K, while the BSS1-B component has Teff of 22000 K. The reduction in residual after including the hotter component is visible in Fig. 1(b). The v2i for individual points is shown in Fig. 1(c) with v2r of 0.53. Although, we note that the v2r need not be  1, for a non-linear model fitting (Andrae et al. 2010). One has to look at residuals and v2 both to determine the goodness of fit. Figure 1(d) shows the Hertzsprung–Russell (H–R) diagram of A and B components. The density distribution of noisy and converging iterations is also shown to get an idea of degeneracy in temperature and luminosity. Figure 1(e) panel shows the best fit and the noisy and converging

J. Astrophys. Astr. (2021)42:89

Page 5 of 14

89

Table 2. Fitting parameters of the best fit of the double and single component fits of BSSs with the hotter component. Scaling factor is the value by which the model has to be multiplied to fit the data, Nfit is the number of data points fitted and v2r is the reduced v2 for the composite fit. The v2r values of single fits of the cooler components are given in brackets. Note: the log g values are imprecise due to the insensitivity of the SED to log g. Name

Comp.

log g

Teff [K]

R [R ]

L [L ]

Scaling factor

7750 ± 125 22000 þ4842 7269 8250 ± 125 24000 þ6802 4996 7250 ± 125 24000 þ3229 10186 8000 ± 125 26000 þ1912 13323 6500 ± 125 14000 þ12479 5493 8500 ± 125 19000 þ8088 2901

2.44 ± 0.04 0.122 þ0:150 0:037 3.72 ± 0.06 0.234 þ0:131 0:089 3.56 ± 0.06 0.094 þ0:191 0:016 2.16 ± 0.04 0.089 þ0:304 0:011 2.29 ± 0.04 0.237 þ0:376 0:150 2.90 ± 0.05 0.270 þ0:102 0:134

19.3 ± 1.8 3.1 þ0:7 0:4 57.6 ± 3.4 16.4 þ1:5 1:2 31.6 ± 2.1 2.7 þ0:8 0:3 17.2 ± 1.5 3.3 þ0:8 0:3 8.4 ± 0.8 1.9 þ2:1 0:5 39.6 ± 1.4 8.6 þ1:2 0:8

8.99E-23 2.23E-25 2.09E-22 8.27E-25 1.92E-22 1.34E-25 7.06E-23 1.20E-25 7.90E-23 8.47E-25 1.27E-22 1.11E-24

v2r

Nfit

(v2r; single )

Double fits BSS1 BSS2 BSS3 BSS4 BSS5 BSS7 Name

A B A B A B A B A B A B

4.5 5 3.5 5 4.5 5 5 5 3.5 5 3 5

16

0.5 (10.5)

16

3.9 (1.2)

16

1.0 (16.1)

17

2.3 (9.6)

12

0.4 (19.9)

11

0.6 (1.4)

log g

Teff [K]

e Teff [K]

R [R ]

e R ½R 

L [L ]

e L [L ]

Scaling factor

Nfit

v2r

4 3 3.5 3 4 3.5 3.5 3 3 3 3.5 4 4.5 5 5 3.5 3.5 3 5 3 4.5 4 4 4.5 5 4 4 4

6500 7000 7500 6250 6750 7000 6750 6750 6500 6750 7000 6500 7250 6500 6500 6250 6500 7250 6250 7250 6250 7250 6500 6250 8250 6250 6500 6500

125 125 125 125 125 125 125 125 125 125 125 125 125 125 125 125 125 125 125 125 125 125 130 125 127 125 125 125

2.32 4.19 2.81 2.91 5.28 1.97 2.32 2.38 2.44 1.99 2.32 2.22 2.89 2.23 3.08 3.28 2.84 2.12 2.64 2.28 2.17 2.23 2.27 2.45 1.40 2.57 2.04 3.80

0.04 0.07 0.05 0.05 0.09 0.03 0.04 0.04 0.04 0.03 0.04 0.04 0.05 0.04 0.05 0.06 0.05 0.04 0.05 0.04 0.04 0.04 0.04 0.04 0.02 0.04 0.04 0.07

8.74 38.17 22.58 11.67 52.55 8.66 10.17 10.63 9.71 7.42 11.60 8.08 20.92 8.05 15.32 14.88 13.00 11.24 9.73 13.06 6.47 12.57 8.40 8.35 8.09 9.27 6.79 23.42

0.41 2.05 1.14 0.54 2.64 0.43 0.51 0.48 0.49 0.37 0.55 0.34 1.17 0.32 0.61 0.62 0.59 0.59 0.39 0.67 0.26 0.68 0.41 0.34 0.35 0.39 0.28 1.01

8.15E-23 2.66E-22 1.20E-22 1.28E-22 4.22E-22 5.85E-23 8.16E-23 8.54E-23 9.02E-23 6.00E-23 8.12E-23 7.48E-23 1.26E-22 7.53E-23 1.43E-22 1.63E-22 1.22E-22 6.77E-23 1.06E-22 7.89E-23 7.10E-23 7.54E-23 7.81E-23 9.08E-23 2.98E-23 9.98E-23 6.29E-23 2.19E-22

11 11 11 11 11 12 11 11 11 11 11 15 8 11 11 11 11 11 15 15 11 15 5 11 15 15 15 11

3.4 3.2 4.2 3.9 1.6 47.9 3.2 2.5 3.6 3.1 4.3 12.3 13.3 3.4 4.9 1.5 3.7 2.1 11.7 11.3 4.3 9.8 125.1 5.3 9.0 19.5 10.9 4.2

Single fits BSS6 BSS8 BSS9 BSS10 BSS11 BSS12 BSS13 BSS14 BSS15 BSS16 BSS17 BSS18 BSS19 BSS20 BSS21 BSS22 BSS23 BSS24 BSS25 BSS26 BSS27 BSS28 BSS29 BSS30 BSS31 BSS32 BSS33 BSS34

89

Page 6 of 14

J. Astrophys. Astr. (2021)42:89

Table 2. Continued. Name

log g

Teff [K]

e Teff [K]

R [R ]

e R ½R 

L [L ]

e L [L ]

Scaling factor

Nfit

v2r

BSS35 BSS36 BSS37 BSS38 BSS39

3 3.5 5 5 3.5

6250 6500 6500 6500 6500

125 125 125 125 125

2.41 2.37 2.25 2.68 3.31

0.04 0.04 0.04 0.05 0.06

8.07 9.13 8.23 11.70 17.67

0.38 0.44 0.34 0.48 0.80

8.75E-23 8.51E-23 7.65E-23 1.09E-22 1.65E-22

11 11 11 11 11

4.4 3.3 3.6 3.4 3.9

iterations in Teff –v2 phase-plane. The double component fits of BSS2, 3, 4, 5 and 7, and single-component fits of BSS10 and BSS15 are shown in Fig. 3. The fitting parameters are mentioned in Table 2. The H–R diagram of the BSSs detected in King 2 is shown in Fig. 4. We have shown the UV faint BSSs as blue dots, UV bright BSSs are represented as blue diamonds, and the hotter components of UV bright BSSs as filler circles. We have taken the parameters of the hotter companions detected along with the BSSs in NGC 188 from Subramaniam et al. 2016 and NGC 2682 from Sindhu et al. (2019), Jadhav et al. (2019) and Sindhu et al. (in prep.). They are plotted in the figure as orange cross and triangles respectively. The parameters of the EHB stars in NGC 1851 are taken from Singh et al. (2020) and are shown as orange stars. The PARSEC4 isochrone of log age ¼ 9:7 is over-plotted and shown in black (Bressan et al. 2012), along with the WD cooling curves5 (thick grey curves, Tremblay & Bergeron 2009) and BaSTI6 zero age HB (ZAHB; dashed black curve; Hidalgo et al. 2018).

(Sindhu et al. 2019) are known to have evolved companions. The companions were classified as WD, ELM WDs, post-AGB/HB according to their luminosity and temperature. BSS2-A, BSS3-A and BSS7-A lie above/on the ZAHB in Fig. 3. There is a degeneracy in this region of the H–R diagram where one could find both massive BSS as well as ZAHB stars. Stars in these two evolutionary phases will have different masses (HB mass \ MSTO; BSS mass [ MSTO), that could be used to lift the degeneracy. Bond and Perry (1971) measured the masses of stars in this region of the NGC 2682 CMD and determined that they are indeed high mass BSSs. One star is found in this region of the NGC 188 CMD and it is classified as a BHB (Rani et al. 2021), this star is significantly brighter than the rest of the BSSs. In the case of King 2, the BSSs show a continuous distribution up to the brightest BSS, hence BSS2-A/BSS3-A are most likely normal BSSs. However, their mass estimations (via log g measurements or asteroseismology) are required before confirming their evolutionary status.

4. Results and discussion What are the hotter companions? BSS and their companions in literature The BSSs have Teff range of 5750 to 8500 K and radii of 1.4 to 5.21R . By comparison to isochrones, they have mass in the range of 1.2 to 1.9M , the brightest BSS being 3 mag brighter than the MSTO. Majority of BSSs in King 2 have Teff similar to the older NGC 188 (6100–6800 K; Gosnell et al. 2015), but are cooler than NGC 2682 (6250–9000 K; Sindhu et al. in prep.), which is expected due to its slightly younger age. The BSSs in NGC 188 (Geller & Mathieu 2011; Gosnell et al. 2014; Subramaniam et al. 2016) and NGC 2682 4

http://stev.oapd.inaf.it/cgi-bin/cmd www.astro.umontreal.ca/*bergeron/CoolingModels/ 6 http://basti-iac.oa-abruzzo.inaf.it/isocs.html 5

The hotter companions in UV bright BSSs have Teff of 14000 to 26000 K (spectral type B) and radii of 0.09 to 0.27R . Figure 3 shows the density distributions of the best 100 fits for the hotter companions. Figure 3 also shows the location of various companions to BSS in NGC 188 and NGC 2682, and EHB stars in NGC 1851, one of which has a BSS as its companion (Singh et al. 2020). The hotter companion to BSSs in NGC 2682 are all fainter and near the WD cooling curves. While those for NGC 1851 and NGC 188 lie closer to the ZAHB region. In King 2, the limiting magnitude of UVIT observations is 23 and 22 mag in F148W and N219M, respectively. According to WD cooling models (corrected for distance and extinction; Tremblay & Bergeron 2009), only WDs younger than 0.7 and 16

R.A. (J2016)

12.735411 12.851678 12.748154 12.741835 12.676827 12.852115 12.497494 12.475572 12.562694 12.951969 12.601162 12.718310 12.668926 12.677840 12.630396 12.578670 12.637498 12.729170 12.585603 12.703872 12.916705 12.792388 12.755911 12.600595 12.792363 12.744375 12.669486 12.751046 12.651172 12.944214 12.741104 12.736065 12.759065 12.516954 12.507227 12.551578 12.533187 12.530339 12.532937

Name

BSS1 BSS2 BSS3 BSS4 BSS5 BSS6 BSS7 BSS8 BSS9 BSS10 BSS11 BSS12 BSS13 BSS14 BSS15 BSS16 BSS17 BSS18 BSS19 BSS20 BSS21 BSS22 BSS23 BSS24 BSS25 BSS26 BSS27 BSS28 BSS29 BSS30 BSS31 BSS32 BSS33 BSS34 BSS35 BSS36 BSS37 BSS38 BSS39

58.185675 58.152155 58.207693 58.196669 58.114184 58.202338 58.135206 58.232940 58.306732 58.156780 58.118577 58.184925 58.042374 58.058054 58.055428 58.094139 58.089431 58.183177 58.116081 58.117638 58.156035 58.127478 58.107444 58.148060 58.189219 58.209324 58.259244 58.197227 58.264862 58.230622 58.182103 58.182505 58.182083 58.129105 58.180810 58.191653 58.207971 58.248740 58.265730

Dec. (J2016) 424416887106018688 424415066039931520 424418398934484480 424418364574755328 424416234271059072 424415684515168768 425074017100947968 425168433361005568 425170361807707392 424414207046475904 424323394258415744 424416917160868480 424321809405171968 424322226027437440 424322569624703744 424322874557099008 424323016301297792 424416921465758976 424323291174262400 424416268630796672 424414928600979712 424416436124233344 424416096832121088 424417265062966656 424418188471143040 424418394634254720 424419395366848640 424418360269944320 424419601519284096 424421456951191040 424416887106023168 424416887106023040 424416887106024704 425073944081350784 425074188899601024 425168094064999808 425168162784475008 425168609461065344 425168918698706688

Gaia EDR3 source_id 1.87e-16±2.18e-17 1.08e-15±5.07e-17 1.57e-16±2.16e-17 2.03e-16±2.17e-17 – – 6.42e-16±3.65e-17 1.96e-16±2.09e-17 7.94e-17±1.43e-17 – – – – – – – – – – – – – – – – – – – – – – – – – – – – – –

UVIT.F148W±err – 4.67e-16±3.57e-17 – 1.07e-16±2.9e-17 4.98e-17±1.53e-17 – 3.96e-16±3.82e-17 1.31e-16±2.8e-17 – 5.57e-17±1.51e-17 1.14e-16±2.18e-17 – – – – – – – – – – – – – – – – – – – – – – – – – – – –

UVIT.N219M±err 2.74e-16±6.08e-17 6.86e-16±6.07e-17 2.51e-16±4.47e-17 2.30e-16±4.14e-17 4.93e-17±3.16e-17 – 4.86e-16±4.40e-17 – 1.76e-16±4.34e-17 – 1.23e-16±2.83e-17 8.45e-17±3.9e-17 3.74e-17±1.55e-17 – 3.68e-17±1.72e-17 – 6.69e-17±3.16e-17 – 1.13e-15±6.10e-17 6.26e-17±2.11e-17 – 4.66e-17±2.28e-17 – – – 3.23e-16±5.68e-17 – 1.73e-16±4.38e-17 3.58e-16±4.71e-17 – – – 7.26e-17±3.01e-17 – – 8.47e-17±2.75e-17 – – –

GALEX.NUV±err 8.05e-16±1.29e-17 – 1.21e-15±1.94e-17 7.21e-16±9.45e-18 – – – – – – – 3.00e-16±1.16e-17 – – – – – 2.52e-16±5.75e-18 – – – – – – 2.71e-16±3.56e-18 5.17e-16±8.32e-18 – 4.46e-16±0.e?00 – – 3.32e-16±5.35e-18 2.64e-16±6.02e-18 2.12e-16±3.40e-18 – – – – – –

CAHA.U±err

1.11e-15±1.45e-17 – 1.68e-15±2.20e-17 9.75e-16±9.03e-18 – – – – – – – 4.42e-16±1.49e-17 – – – – – 3.74e-16±7.79e-18 – – – – – – 3.99e-16±3.69e-18 7.68e-16±1.01e-17 – 6.75e-16±0.e?00 – – 4.67e-16±6.12e-18 4.07e-16±8.46e-18 3.14e-16±4.12e-18 – – – – – –

CAHA.B±err

˚ 1 . GALEX, Gaia (EDR3), PanTable 3. Coordinates, Gaia EDR3 source IDs, flux and flux errors of the stars in all used filters. All flux are given in erg s1 cm2 A Starrs (PS1) and 2MASS photometry is taken from archives. CAHA photometry is taken from A90.

J. Astrophys. Astr. (2021)42:89 Page 7 of 14 89

CAHA.V±err

1.04e-15±9.6e-18 – 1.66e-15±1.53e-17 9.37e-16±0.e?00 – – – – – – – 4.57e-16±8.49e-18 – – – – – 4.32e-16±4.e-18 – – – – – – 5.25e-16±0.e?00 7.72e-16±7.15e-18 – 7.04e-16±0.e?00 – – 4.40e-16±4.07e-18 5.01e-16±4.64e-18 3.7e-16±3.42e-18 – – – – – –

Name

BSS1 BSS2 BSS3 BSS4 BSS5 BSS6 BSS7 BSS8 BSS9 BSS10 BSS11 BSS12 BSS13 BSS14 BSS15 BSS16 BSS17 BSS18 BSS19 BSS20 BSS21 BSS22 BSS23 BSS24 BSS25 BSS26 BSS27 BSS28 BSS29 BSS30 BSS31 BSS32 BSS33 BSS34 BSS35 BSS36 BSS37 BSS38 BSS39

Table 3. Continued.

7.48e-16±1.56e-17 – 1.17e-15±1.54e-17 6.64e-16±6.14e-18 – – – – – – – 3.3e-16±8.7e-18 – – – – – 3.48e-16±7.25e-18 – – – – – – 4.42e-16±4.09e-18 5.47e-16±1.14e-17 – 5.13e-16±0.e?00 – – 3.03e-16±6.31e-18 4.19e-16±8.71e-18 2.92e-16±6.08e-18 – – – – – –

CAHA.R±err 1.05e-15±3.77e-18 3.31e-15±9.33e-18 1.74e-15±1.18e-17 9.13e-16±4.15e-18 3.95e-16±8.03e-18 4.35e-16±2.74e-18 2.22e-15±6.52e-18 2.17e-15±9.38e-18 1.24e-15±3.85e-18 5.81e-16±2.82e-18 2.71e-15±8.32e-18 5.41e-16±5.55e-18 5.22e-16±2.54e-18 5.58e-16±2.88e-18 5.08e-16±2.38e-18 3.89e-16±2.6e-18 6.09e-16±2.68e-18 4.26e-16±2.36e-18 1.11e-15±1.67e-17 3.91e-16±2.34e-18 7.4e-16±3.48e-18 7.48e-16±3.48e-18 6.58e-16±3.64e-18 6.26e-16±2.97e-18 4.83e-16±2.45e-18 7.67e-16±3.92e-18 3.13e-16±2.65e-18 7.09e-16±3.49e-18 4.93e-16±1.33e-17 3.99e-16±2.37e-18 4.42e-16±1.99e-18 4.91e-16±6.23e-18 3.58e-16±1.94e-18 1.14e-15±4.32e-18 4.02e-16±2.9e-18 4.58e-16±2.33e-18 3.98e-16±2.75e-18 5.62e-16±2.29e-18 8.86e-16±3.28e-18

GAIA3.Gbp±err 9.17e-16±2.42e-18 2.73e-15±7.e-18 1.44e-15±3.74e-18 7.77e-16±2.13e-18 3.82e-16±2.38e-18 4.31e-16±1.12e-18 1.81e-15±4.63e-18 1.78e-15±5.63e-18 1.10e-15±2.83e-18 5.97e-16±1.54e-18 2.54e-15±6.92e-18 4.12e-16±1.11e-18 5.1e-16±1.32e-18 5.40e-16±1.4e-18 4.93e-16±1.28e-18 3.80e-16±1.00e-18 5.79e-16±1.49e-18 4.14e-16±1.11e-18 9.60e-16±2.47e-18 3.95e-16±1.03e-18 7.54e-16±1.94e-18 6.97e-16±1.83e-18 6.49e-16±1.68e-18 5.66e-16±1.46e-18 4.99e-16±1.29e-18 6.56e-16±1.7e-18 3.22e-16±8.48e-19 6.11e-16±1.58e-18 3.24e-16±1.01e-18 4.11e-16±1.07e-18 3.77e-16±9.87e-19 4.95e-16±2.30e-18 3.44e-16±9.05e-19 1.16e-15±2.97e-18 4.09e-16±1.07e-18 4.61e-16±1.19e-18 4.02e-16±1.04e-18 5.72e-16±1.48e-18 8.86e-16±2.27e-18

GAIA3.G±err 6.33e-16±2.98e-18 1.76e-15±6.5e-18 1.14e-15±8.38e-18 5.48e-16±3.35e-18 2.96e-16±5.61e-18 3.47e-16±1.73e-18 1.15e-15±4.43e-18 1.44e-15±6.71e-18 7.85e-16±2.92e-18 4.86e-16±2.03e-18 2.06e-15±7.59e-18 3.8e-16±5.22e-18 3.91e-16±1.81e-18 4.18e-16±1.93e-18 3.91e-16±1.87e-18 2.93e-16±1.67e-18 4.37e-16±2.e-18 3.21e-16±1.65e-18 7.52e-16±5.66e-18 3.22e-16±1.65e-18 6.12e-16±2.69e-18 6.42e-16±2.76e-18 5.25e-16±2.48e-18 4.01e-16±1.65e-18 4.11e-16±1.73e-18 4.77e-16±2.45e-18 2.68e-16±1.53e-18 4.58e-16±3.21e-18 3.96e-16±5.41e-18 3.42e-16±1.91e-18 2.48e-16±1.33e-18 3.89e-16±4.94e-18 2.7e-16±1.42e-18 9.36e-16±3.78e-18 3.32e-16±1.81e-18 3.65e-16±1.9e-18 3.25e-16±1.65e-18 4.63e-16±2.08e-18 7.11e-16±2.90e-18

GAIA3.Grp±err 8.98e-16±2.18e-18 2.63e-15±4.42e-18 1.47e-15±1.16e-17 7.69e-16±4.1e-18 4.26e-16±1.86e-18 4.34e-16±1.30e-18 1.75e-15±2.87e-18 1.94e-15±4.52e-18 1.09e-15±2.63e-18 5.96e-16±1.29e-18 2.62e-15±3.41e-18 4.71e-16±2.73e-18 5.17e-16±1.98e-18 5.41e-16±1.41e-18 4.93e-16±4.97e-18 3.82e-16±1.94e-18 5.8e-16±1.76e-18 4.14e-16±1.49e-18 1.11e-15±6.75e-17 3.92e-16±1.55e-18 7.54e-16±2.82e-18 7.62e-16±2.52e-18 6.58e-16±3.49e-18 5.70e-16±1.94e-18 4.99e-16±1.66e-18 6.63e-16±1.22e-18 3.21e-16±1.05e-18 6.28e-16±3.32e-18 – 4.11e-16±1.73e-18 3.69e-16±1.07e-18 5.35e-16±3.38e-18 3.48e-16±1.96e-18 1.16e-15±2.23e-18 4.04e-16±2.08e-18 4.63e-16±1.08e-18 4.04e-16±2.1e-18 5.77e-16±2.91e-18 9.05e-16±5.27e-18

PS1.r±err 1.17e-15±6.58e-18 3.73e-15±3.04e-18 1.82e-15±6.67e-18 9.92e-16±5.18e-18 4.86e-16±1.02e-17 4.47e-16±6.11e-19 2.52e-15±4.96e-18 2.38e-15±2.27e-17 1.36e-15±5.10e-18 5.98e-16±1.66e-18 2.79e-15±1.59e-17 5.4e-16±1.11e-17 5.58e-16±1.13e-18 5.94e-16±3.12e-18 5.24e-16±2.41e-18 4.24e-16±1.89e-18 6.61e-16±2.81e-18 4.59e-16±1.32e-18 1.09e-15±4.76e-18 4.03e-16±1.12e-18 7.69e-16±1.61e-18 7.44e-16±1.84e-18 6.84e-16±1.84e-18 6.87e-16±3.e-18 4.92e-16±2.24e-18 8.35e-16±3.78e-18 3.14e-16±1.08e-18 7.66e-16±6.66e-18 – 4.04e-16±1.69e-18 5.e-16±2.37e-18 4.50e-16±3.41e-18 3.74e-16±1.76e-18 1.2e-15±4.82e-18 4.21e-16±2.32e-18 4.76e-16±1.54e-18 4.04e-16±2.03e-18 5.80e-16±2.53e-18 9.20e-16±2.39e-18

PS1.g±err

6.57e-16±3.91e-18 1.81e-15±5.46e-18 1.10e-15±1.27e-17 5.56e-16±3.92e-18 3.60e-16±9.59e-19 3.57e-16±1.66e-18 1.19e-15±1.7e-18 1.51e-15±6.08e-18 8.17e-16±1.91e-18 5.06e-16±1.76e-18 2.10e-15±4.01e-18 3.12e-16±3.80e-19 4.1e-16±2.38e-18 4.32e-16±1.15e-18 3.99e-16±9.95e-19 3.04e-16±1.66e-18 4.55e-16±2.08e-18 3.29e-16±2.25e-18 7.48e-16±3.34e-18 3.34e-16±1.13e-18 6.36e-16±1.77e-18 6.41e-16±2.33e-18 5.47e-16±1.15e-18 4.21e-16±1.53e-18 4.20e-16±1.15e-18 4.83e-16±2.19e-18 2.76e-16±1.51e-18 4.64e-16±4.53e-18 3.53e-16±1.56e-18 3.53e-16±2.26e-18 2.6e-16±7.87e-19 3.73e-16±1.09e-17 2.77e-16±1.17e-18 9.64e-16±3.07e-18 3.41e-16±1.22e-18 3.74e-16±1.76e-18 3.36e-16±3.62e-18 4.79e-16±2.68e-18 7.36e-16±1.62e-18

PS1.i±err

89 Page 8 of 14 J. Astrophys. Astr. (2021)42:89

PS1.z±err

5.25e-16±1.08e-18 1.39e-15±4.31e-18 9.14e-16±1.53e-17 4.4e-16±3.34e-18 2.43e-16±5.71e-18 2.99e-16±9.55e-19 9.12e-16±3.76e-18 1.18e-15±9.73e-18 6.41e-16±1.70e-18 4.28e-16±1.07e-18 1.71e-15±9.50e-19 2.62e-16±7.86e-18 3.31e-16±1.17e-18 3.52e-16±1.72e-18 3.27e-16±8.77e-19 2.48e-16±1.38e-18 3.69e-16±9.43e-19 2.72e-16±1.06e-18 6.05e-16±2.66e-18 2.76e-16±1.32e-18 5.32e-16±2.70e-18 5.30e-16±3.77e-18 4.51e-16±1.68e-18 3.34e-16±1.49e-18 3.57e-16±1.07e-18 3.92e-16±2.16e-18 2.34e-16±8.07e-19 3.72e-16±2.29e-18 2.83e-16±3.36e-19 2.97e-16±2.38e-18 1.97e-16±8.58e-19 3.29e-16±7.91e-18 2.3e-16±9.57e-19 8.05e-16±1.36e-18 2.88e-16±9.61e-19 3.17e-16±4.78e-19 2.8e-16±2.77e-18 3.99e-16±2.09e-18 6.08e-16±1.66e-18

Name

BSS1 BSS2 BSS3 BSS4 BSS5 BSS6 BSS7 BSS8 BSS9 BSS10 BSS11 BSS12 BSS13 BSS14 BSS15 BSS16 BSS17 BSS18 BSS19 BSS20 BSS21 BSS22 BSS23 BSS24 BSS25 BSS26 BSS27 BSS28 BSS29 BSS30 BSS31 BSS32 BSS33 BSS34 BSS35 BSS36 BSS37 BSS38 BSS39

Table 3. Continued.

4.74e-16±2.72e-18 1.22e-15±3.58e-18 8.26e-16±5.47e-18 3.93e-16±3.24e-18 2.58e-16±1.69e-18 2.69e-16±2.31e-18 8.00e-16±3.15e-18 1.04e-15±7.33e-18 5.95e-16±4.39e-18 3.83e-16±2.33e-18 1.49e-15±4.28e-18 2.29e-16±3.82e-18 2.94e-16±1.54e-18 3.15e-16±9.87e-19 3.e-16±2.43e-18 2.17e-16±1.34e-18 3.35e-16±1.67e-18 2.48e-16±4.72e-18 5.45e-16±5.40e-18 2.49e-16±1.19e-18 4.8e-16±2.17e-18 4.75e-16±3.28e-18 4.12e-16±1.62e-18 2.97e-16±1.90e-18 3.24e-16±3.41e-18 3.78e-16±2.60e-18 2.15e-16±2.34e-18 3.35e-16±2.27e-18 – 2.73e-16±2.7e-18 1.76e-16±1.28e-18 2.68e-16±6.06e-18 2.1e-16±2.92e-18 7.30e-16±2.53e-18 2.6e-16±3.76e-18 2.78e-16±3.62e-18 2.51e-16±2.45e-18 3.57e-16±1.09e-18 5.54e-16±2.85e-18

PS1.y±err 2.62e-16±2.54e-17 6.45e-16±1.90e-17 5.06e-16±2.24e-17 2.49e-16±2.09e-17 1.72e-16±8.39e-18 1.68e-16±8.98e-18 4.13e-16±1.10e-17 6.38e-16±1.82e-17 3.28e-16±1.18e-17 2.35e-16±9.74e-18 8.75e-16±2.74e-17 – 1.90e-16±8.77e-18 1.81e-16±9.68e-18 1.76e-16±1.34e-17 1.22e-16±8.34e-18 1.95e-16±9.35e-18 1.40e-16±1.03e-17 – 1.40e-16±8.54e-18 2.52e-16±1.02e-17 2.99e-16±1.26e-17 2.41e-16±1.84e-17 1.63e-16±1.14e-17 1.80e-16±9.14e-18 2.06e-16±2.72e-17 1.29e-16±9.73e-18 1.84e-16±1.77e-17 – 1.75e-16±1.01e-17 8.13e-17±7.64e-18 1.60e-16±1.2e-17 1.16e-16±1.36e-17 4.40e-16±1.3e-17 1.80e-16±9.80e-18 1.68e-16±9.89e-18 1.60e-16±9.29e-18 2.24e-16±1.07e-17 3.34e-16±1.23e-17

2MASS.J±err 1.13e-16±1.34e-17 2.62e-16±9.67e-18 2.38e-16±1.18e-17 1.23e-16±1.08e-17 8.52e-17±8.16e-18 8.63e-17±7.15e-18 1.62e-16±8.08e-18 2.81e-16±1.14e-17 1.22e-16±7.77e-18 1.1e-16±7.49e-18 4.44e-16±1.55e-17 – 7.19e-17±8.28e-18 9.1e-17±8.05e-18 8.93e-17±1.00e-17 4.58e-17±7.26e-18 7.14e-17±8.22e-18 7.21e-17±7.51e-18 – 6.93e-17±7.35e-18 1.42e-16±7.47e-18 1.65e-16±8.21e-18 1.08e-16±1.24e-17 8.01e-17±8.78e-18 9.28e-17±7.43e-18 1.02e-16±1.42e-17 5.7e-17±8.24e-18 8.e-17±1.08e-17 – 7.47e-17±7.36e-18 5.44e-17±0.e?00 8.60e-17±1.11e-17 5.84e-17±9.30e-18 2.05e-16±8.11e-18 7.33e-17±6.75e-18 8.51e-17±6.97e-18 7.96e-17±6.89e-18 1.07e-16±7.31e-18 1.46e-16±8.77e-18

2MASS.H±err 5.08e-17±4.92e-18 1.11e-16±5.71e-18 9.34e-17±5.51e-18 5.51e-17±4.98e-18 3.49e-17±4.5e-18 2.51e-17±4.62e-18 7.57e-17±4.60e-18 1.35e-16±6.98e-18 6.14e-17±4.64e-18 4.01e-17±4.43e-18 1.79e-16±6.59e-18 – 3.57e-17±4.53e-18 3.39e-17±4.47e-18 3.82e-17±4.71e-18 3.06e-17±4.23e-18 2.53e-17±4.2e-18 2.63e-17±4.23e-18 – 3.35e-17±4.38e-18 5.78e-17±4.63e-18 6.69e-17±4.81e-18 3.85e-17±4.39e-18 3.33e-17±4.36e-18 3.86e-17±4.62e-18 2.97e-16±0.e?00 2.50e-17±4.08e-18 3.44e-17±5.70e-18 – 4.95e-17±4.47e-18 1.57e-17±0.e?00 3.44e-17±4.47e-18 3.00e-17±4.79e-18 9.78e-17±4.96e-18 3.7e-17±4.26e-18 3.27e-17±4.31e-18 3.00e-17±4.59e-18 4.86e-17±4.70e-18 6.36e-17±4.69e-18

2MASS.Ks±err

J. Astrophys. Astr. (2021)42:89 Page 9 of 14 89

89

Page 10 of 14

J. Astrophys. Astr. (2021)42:89

Figure 3. The descriptions of double component fits are same as Fig. 2. The single component fits of BSS10 and BSS15 are shown as an example with model fit (blue curve), fitted data points (red points) with 1r and 3r errors as solid and dashed lines. The theoretical spectra (in grey) is added for reference. The observed (reddening affected) SED is shown in grey below the corrected data-points. The title mentions the Teff , log g, metallicity and AV of the model fit.

J. Astrophys. Astr. (2021)42:89

Page 11 of 14

89

Figure 3. Continued.

Myr old would be detectable. As seen in Fig. 4, the hotter companions are well above the WD cooling curves, and these are not WDs. The hotter companions are likely to be hot HB stars which are also known as EHB stars or sdB stars, as inferred from their Teff , radii and luminosity. These are core-helium burning stars with Teff

in the range of 20,000–40,000 K and are compact (0.15–0.35R ; Heber 2016; Sahoo et al. 2020). As these stars are hot and not as small as WDs, they appear bright in the UV. These stars are thought to contribute to the UV upturn seen in elliptical or in early-type galaxies (Brown et al. 1997). The sdB stars have very thin hydrogen envelope and are

89

Page 12 of 14

J. Astrophys. Astr. (2021)42:89

thought to be the stripped core of a red giant star (Heber 2016). Maxted et al. (2001) found a good fraction of the sdB stars in detached, but short period binary systems. sdB stars are thought to provide important clues to common envelope evolution in tight binaries. BSS2-B lies on the blue end of ZAHB and very similar to the EHBs in the outskirts of GC NGC 1851. Similarly, hot and bright post-AGB/HB candidate was found as a companion to a BSS in NGC 188 (Subramaniam et al. 2016). Hence, BSS2-B could be an EHB star. BSS7-B is near a known EHB from NGC 1851 and is slightly fainter than ZAHB, hence, it is likely to be an EHB. BSS5-B lies slightly above the WDs. Hence, it can be a very young He-WD (Panei et al. 2007) or an sdB star. Rest of the King 2 BSS companions are slightly fainter than the ZAHB, and they are likely sdBs. Formation pathways of BSSs and EHBs/sdBs

Figure 4. H–R diagram of locations of components of binaries in King 2, NGC 2682 (Sindhu et al. 2019; Jadhav et al. (2019); Sindhu et al. in prep.), NGC 188 (Subramaniam et al. 2016) and EHB stars in NGC 1851 (Singh et al. 2020). In King 2, the UV faint BSS (blue dots), UV bright BSS (blue diamonds with their ID) and hotter components in BSSs (coloured filled circles with numbers and error bars) are shown. Hotter components in NGC 2682 and NGC 188 are shown as orange triangles and cross. EHB stars in NGC 1851 are shown as stars. The PARSEC isochrone (black curve), PARSEC zero age MS (dashed grey line), WD cooling curves (thick grey curves) and BaSTI ZAHB (dotted grey curve) are shown for reference.

BSS formation mechanism involves mass gain while the EHB/sdB formation involves the stripping of the envelope of a post-MS star. The detection of EHB/ sdBs confirms ‘binary mass transfer’ as the formation mechanism for BSS1 to BSS5 and BSS7. As the cooler companions are BSSs that are supposed to have gained mass, we can infer that the detected EHBs/sdBs have transferred mass to the BSSs companions. Therefore, the BSS?EHB/sdB systems in King 2 illustrate stars on both sides of the mass exchange. We see a range in their temperature and luminosity, suggesting a diversity among the hotter companions. The lifespan of sdB stars is expected to be between 100 to 200 Myr (Bloemen et al. 2014; Schindler et al. 2015 and references therein), after which they descend the WD cooling curve. When the current BSS expands, the mass transfer is expected to start again due to the short orbit, which already allowed the previous instance of mass transfer. The system can begin stable/unstable mass transfer and become a WD?WD system. Alternatively, it can merge through a common-envelop phase and become a massive WD. The exact evolution will depend on then orbital parameters, mass transfer efficiency and mass loss. King 2 is one of the oldest OC, lies in the outskirts of the galactic disk. It is metal-poor compared to the Galactic disc OCs. The environment is quite similar to outskirts of GCs, which are also metal-poor, old and

J. Astrophys. Astr. (2021)42:89

of comparable density. While most of the BSSs in GCs lie in core and are formed via mergers (Chatterjee et al. 2013), BSSs in outskirts of GCs can form through mass transfer as seen in EHB-4 of NGC 1851 (Singh et al. 2020). Our study suggests that the at least 15% of the BSSs in King 2 are formed via mass transfer pathway of formation. We have seen sdB companions to BSSs in NGC 188 and NGC 1851 (both are older systems), however none in NGC 2682 (which is younger). NGC 6791 of slightly younger age also has sdB stars (Kaluzny & Udalski 1992; Reed et al. 2012). This might suggest that there is an upper age limit of  5 Gyr for the formation of sdB stars in OCs.

5. Conclusions and summary • The old OC King 2 has a large population (39) of BSSs, spreading up to 3 mag brighter than the MSTO. We constructed SEDs of all the BSS using UV to IR data. The BSSs have Teff in the range of 5750–8250 K, luminosity in the range of 5.6–57.5L and mass in the range of 1.2– 1.9M . • Six of the UV bright BSS showed excess UV flux and were successfully fitted with double component SEDs. The hotter components have Teff of 14000–26000 K and R=R of 0.09–0.27, suggesting a range of properties. Two of the hotter companions to the BSS are likely EHB stars, while four are likely sdB stars. • EHB/sdB companions imply that these 6 (out of 39) BSSs have formed via binary mass transfer. The SED fits show that sdB stars can be created in old OCs such as King 2 (similar to old OC NGC 188 and GC NGC 1851). Spectroscopic time series and radial velocity variations can uncover the binary nature as well as properties of these systems and help in estimating the log g and mass of these stars. The mass and orbital estimations will expand our knowledge of BSS, EHB and sdB formation scenarios.

Acknowledgements We thank the anonymous referee for their valuable comments and inputs. RS would like to thank the National Academy of Sciences, India (NASI), Prayagraj for the award the NASI Honorary Scientist; Alexander von

Page 13 of 14

89

Humboldt Foundation, Germany for the award of longterm group research linkage program and Director, IIA for hosting and making available facilities of the institute. This publication uses the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). This publication uses UVIT data processed by the payload operations centre at Indian Institute of Astrophysics (IIA). The UVIT is built in collaboration between IIA, Inter-University Centre for Astronomy and Astrophysics (IUCAA), Tata Institute of Fundamental Research (TIFR), ISRO and Canadian Space Agency (CSA). This work has made use of data from the European Space Agency (ESA) mission Gaia (https:// www.cosmos.esa.int/gaia), processed by the Gaia Data Processing and Analysis Consortium (DPAC, https:// www.cosmos.esa.int/web/gaia/dpac/consortium). This publication makes use of VOSA, developed under the Spanish Virtual Observatory project supported by the Spanish MINECO through grant AyA2017-84089. References Ahumada J. A., Lapasset E. 2007, A&A, 463, 789 Andrae R., Schulze-Hartung T., Melchior P. 2010, arXiv e-prints, arXiv:1012.3754 Aparicio A., Bertelli G., Chiosi C., Garcia-Pelayo J. M. 1990, A&A, 240, 262 Bayo A., Rodrigo C., Barrado Y., Navascue´s D. et al. 2008, A&A, 492, 277 Bianchi L., GALEX Team 2000, Memorie della Societa` Astronomia Italiana, 71, 1123 Bloemen S., Hu H., Aerts C. et al. 2014, A&A, 569, A123 Bond H. E., Perry C. L. 1971, Pub. Astron. Soc. Pac., 83, 638 Bressan A., Marigo P., Girardi L., Salasnich B., Dal Cero C., Rubele S., Nanni A. 2012, MNRAS, 427, 127 Brown T. M., Ferguson H. C., Davidsen A. F., Dorman B. 1997, ApJ, 482, 685 Cantat-Gaudin T., Jordi C., Vallenari et al. 2018, A&A, 618, A93 Castelli F., Gratton R. G., Kurucz R. L. 1997, A&A, 318, 841 Chambers K. C., Magnier E. A., Metcalfe N. et al. 2016, arXiv e-prints, arXiv:1612.05560 Chatterjee S., Rasio F. A., Sills A., Glebbeek E. 2013, ApJ, 777, 106 Dias W. S., Alessi B. S., Moitinho A., Le´pine J. R. D. 2002, A&A, 389, 871 Fitzpatrick E. L. 1999, Pub. Astron. Soc. Pac., 111, 63 Friel E. D., Janes K. A., Tavarez M., Scott J., Katsani R., Lotz J., Hong L., Miller N. 2002, AJ, 124, 2693 Gaia Collaboration, Brown A. G. A., Vallenari A. et al. 2018, A&A, 616, A1

89

Page 14 of 14

Gaia Collaboration, Brown A. G. A., Vallenari A. et al. 2020, arXiv e-prints, arXiv:2012.01533 Geller A. M., Mathieu R. D. 2011, Nature, 478, 356 Gosnell N. M., Mathieu R. D., Geller A. M., Sills A., Leigh N., Knigge C. 2014, ApJL, 783, L8 Gosnell N. M., Mathieu R. D., Geller A. M., Sills A., Leigh N., Knigge C. 2015, ApJ, 814, 163 Han Z., Podsiadlowski P., Maxted P. F. L., Marsh T. R., Ivanova N. 2002, MNRAS, 336, 449 Heber U. 2016, Pub. Astron. Soc. Pac., 128, 082001 Heggie D. C. 1975, MNRAS, 173, 729 Hidalgo S. L., Pietrinferni A., Cassisi S. et al. 2018, ApJ, 856, 125 Indebetouw R., Mathis J. S., Babler B. L. et al. 2005, ApJ, 619, 931 Jadhav V. V., Sindhu N., Subramaniam A. 2019, ApJ, 886, 13 Jadhav V. V., Pennock C. M., Subramaniam A., Sagar R., Nayak P. K. 2021, MNRAS, 503, 236 Kaluzny J. 1989, Acta Astron., 39, 13 Kaluzny J., Udalski A. 1992, Acta Astron., 42, 29 Kozai Y. 1962, AJ, 67, 591 Kumar A., Ghosh S. K., Hutchings J. et al. 2012, SPIE, 8443, 84431N Leiner E., Mathieu R. D., Vanderburg A., Gosnell N. M., Smith J. C. 2019, ApJ, 881, 47 Marsh T. R., Dhillon V. S., Duck S. R. 1995, MNRAS, 275, 828 Maxted P. F. L., Heber U., Marsh T. R., North R. C. 2001, MNRAS, 326, 1391 Panei J. A., Althaus L. G., Chen X., Han Z. 2007, MNRAS, 382, 779 Postma J. E., Leahy D. 2017, Pub. Astron. Soc. Pac., 129, 115002

J. Astrophys. Astr. (2021)42:89 Rani S., Subramaniam A., Pandey S., Sahu S., Mondal C., Pandey G. 2021, J. Astrophys. Astr., 42, https://doi.org/ 10.1007/s12036-020-09683-2 Reed M. D., Baran A., Tensen R. H., Telting J., O’Toole S. J. 2012, MNRAS, 427, 1245 Rucinski S. M. 1998, AJ, 116, 2998 Sahoo S. K., Baran A. S., Heber U. et al. 2020, MNRAS, 495, 2844 Sahu S., Subramaniam A., Simunovic M. et al. 2019, ApJ, 876, 34 Schindler J.-T., Green E. M., Arnett W. D. 2015, ApJ, 806, 178 Sindhu N., Subramaniam A., Jadhav V. V. et al. 2019, ApJ, 882, 43 Singh G., Sahu S., Subramaniam A., Yadav R. K. S. 2020, ApJ, 905, 44 Skrutskie M. F., Cutri R. M., Stiening R. et al. 2006, AJ, 131, 1163 Subramaniam A., Sindhu N., Tandon S. N. et al. 2016, ApJL, 833, L27 Subramaniam A., Pandey S., Jadhav V. V., Sahu S. 2020, J. Astrophys. Astr., 41, 45 Tadross A. L. 2001, New Astronomy, 6, 293 Tandon S. N., Hutchings J. B., Ghosh S. K. et al. 2017, J. Astrophys. Astr., 38, 28 Taylor M. B. 2005, Astronomical Data Analysis Software and Systems XIV, 347, 29 Tody D. 1993, Astronomical Data Analysis Software and Systems II, 52, 173 Tremblay P.-E., Bergeron P. 2009, ApJ, 696, 1755 Warren S. R., Cole A. A. 2009, MNRAS, 393, 272 Wright E. L., Eisenhardt P. R. M., Mainzer et al. 2010, AJ, 140, 1868

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:85 https://doi.org/10.1007/s12036-021-09749-9

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

A comparison of the UV and HI properties of the extended UV (XUV) disk galaxies NGC 2541, NGC 5832 and ESO406-042 M. DAS1,* , J. YADAV1, N. PATRA2, K. S. DWARAKANATH2, S. S. MCGAUGH3,

J. SCHOMBERT4, P. T. RAHNA5 and J. MURTHY1 1

Indian Institute of Astrophysics, II Block, Koramangala, Bangalore 560 034, India. Raman Research Institute, Sadashivanagar, Bangalore 560 034, India. 3 Case Western Reserve University, Cleveland, OH 44106, USA. 4 University of Oregon, Eugene, OR 97403, USA. 5 CAS Key Laboratory for Research in Galaxies and Cosmology, Shanghai Astronomical Observatory, Shanghai 200030, China. *Corresponding Author. E-mail: [email protected] 2

MS received 7 November 2020; accepted 27 March 2021 Abstract. We present a UV study of 3 extended UV (XUV) galaxies that we have observed with the UVIT and the GMRT. XUV galaxies show filamentary or diffuse star formation well beyond their optical disks, in regions where the disk surface density lies below the threshold for star formation. GALEX observations found that surprisingly 30% of all the nearby spiral galaxies have XUV disks. The XUV galaxies can be broadly classified as Type 1 and Type 2 XUV disks. The Type 1 XUV disks have star formation that is linked to that in their main disk, and the UV emission appears as extended, filamentary spiral arms. The UV luminosity is associated with compact star forming regions along the extended spiral arms. The star formation is probably driven by slow gas accretion from nearby galaxies or the intergalactic medium (IGM). But the Type 2 XUV disks have star formation associated with an outer low luminosity stellar disk that is often truncated near the optical radius of the galaxy. The nature of the stellar disks in Type 2 XUV disks are similar to that of the diffuse stellar disks of low surface brightness galaxies. The star formation in Type 2 XUV disks is thought to be due to rapid gas accretion or gas infall from nearby high velocity clouds (HVCs), interacting galaxies or the IGM. In this paper, we investigate the star formation properties of the XUV regions of two Type 2 galaxies and one mixed XUV type galaxy and compare them with the neutral hydrogen (HI) emisison in their disks. We present preliminary results of our UVIT (FUV and NUV) observations of NGC 2541, NGC 5832 and ESO406-042, as well as GMRT observations of their HI emission. We describe the UV emission morphology, estimate the star formation rates and compare it with the HI distribution in these Type 2 and mixed XUV galaxies. Keywords. UV astronomy—neutral hydrogen—galaxies—star formation.

1. Introduction Star formation is one of the main processes driving galaxy evolution. It can be triggered externally by interactions or mergers with nearby galaxies, by gas infall from companion galaxies or by gas accretion from the intergalactic medium (IGM). Gas infall cools galaxy disks leading to disk instabilities and star This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

formation. Star formation is also internally driven by secular evolution, this includes global disk instabilities such as spiral arms and bars (Kataria & Das 2018). Most of the star formation activity in spirals is confined to the inner optical disk (\5 kpc), where the stellar and gas surface densitites are large enough for instabilities to set in (Kennicutt 1989). Earlier studies indicate that the disk surface brightness is exponential and has a sharp disk truncation radius, where the stellar mass surface density falls and the star

85

Page 2 of 10

formation rate declines (van der Kruit & Searle 1981). However, deep Ha studies have shown that there is star formation in the extreme outer disks of some nearby galaxies, such as NGC 628 and NGC 6946 (Ferguson et al. 1998), where Ha emission is detected beyond the R25 optical disk radii. The most definitive evidence, however, of outer disk star formation came from the ultraviolet (UV) observations of nearby galaxies by GALEX (Martin et al. 2005; Morrissey et al. 2007). The GALEX surveys showed that  30% of all spiral galaxies at distances of \40 Mpc have UV emission from HII regions in their outer disks (Gil de Paz et al. 2007). This is surprising as in these regions the stellar density declines. Such galaxy disks are called extended UV (XUV) disk galaxies (Thilker et al. 2007). Good examples are M83 and NGC 2090 that show extended spiral features in their extreme outer, low density disks (Thilker et al. 2005). Their existence supports models of inside out star formation, where star formation occurs in the extreme (optically) low surface brightness (LSB) zones of galaxy disks resulting in disk building and galaxy growth. XUV disks are also one of the best examples of star formation in low density environments (Krumholz & McKee 2008) and the stellar disk initial mass function (IMF) in such environments is known to be deficient in massive stars (Bruzzese et al. 2020). Hence, XUV disks are interesting to study–both for understanding disk growth as well as for studying star formation in underdense environments. The XUV disk galaxies have been classified earlier in the GALEX surveys into 3 types: Type 1, Type 2, and one mixed category (Thilker et al. 2007). The Type 1 XUV disks have UV bright complexes arising from spiral arms, arm segments or complexes in the faint, outer parts of stellar disks. Hence they can be associated with the disk spiral structure and are thought to arise from slow gas accretion in the outer disk regions (de Blok et al. 2014). They span over all Hubble types of spiral galaxies. The Type 2 XUV galaxies have a blue color (FUV-NIR) in the extreme outer parts of their galaxy disks. These regions appear as optically low in surface brightness and similar to the disks of LSB galaxies (Honey et al. 2016; Das 2013). The LSB zone lies outside most of the K band luminosity of the disk (  80%). Type 2 XUV galaxies are mainly gas rich, late type spiral galaxies that have low stellar masses. Their XUV disks are thought to arise from rapid gas accretion and the ensuing star formation. This is unlike Type I XUV disks where the gas accretion is thought to be slow. There are many

J. Astrophys. Astr. (2021)42:85

XUV disks that show both Type 1 and 2 properties, they are called mixed XUV disks. About 20% of the spirals in the GALEX atlas of nearby galaxies have Type I XUV disks and 10% have Type II XUV disks. A common property of all XUV galaxies and of those that show outer disk star formation is that they are rich in neutral hydrogen gas (HI) and their HI disks are usually more extended than normal galaxies (Bigiel et al. 2010). They are overall nearly twice as gas rich compared to non-XUV galaxies, but on the other hand there are many non-XUV galaxies that have extended gas disks but do not support XUV star formation (e.g. NGC 2915) (Thilker et al. 2007). So being gas rich in the outer disk is a necessary but not sufficient condition for XUV star formation. Some XUV disks are interacting with nearby companions, but not all. In such cases the HI is disturbed and there maybe HI tidal tails or bridges (Kaczmarek & Wilcots 2012). In this study we focus on the star formation in Type 2 XUV galaxies, and try to understand the nature of star formation in their extended LSB disks (McGaugh et al. 1995). We use both UVIT observations and HI observations to compare the distribution of star forming complexes (SFCs) within the gas distribution. The long term goal is to understand what drives XUV star formation and why only some extended HI disks show XUV star formation. In the following sections we describe our observations, our results and then discuss the implications of our study for understanding star formation in the outer disks of galaxies. 2. Galaxy sample All our targets have been observed by GALEX and form part of the surveys of XUV disk galaxies (Lemonias et al. 2011). Two targets are Type 2 XUV galaxies and one is a mixed type XUV galaxy. We have focussed largely on the Type 2 XUV disks as we want to understand the rapid gas accretion onto the outer disks of galaxies and see its relation to HI morphology. The galaxies are bright, nearby sources which means that the high spatial resolution of UVIT can resolve star forming regions in their XUV disks (Rahna et al. 2018). Below we discuss the general properties of the galaxies, which are also listed in Table 1. ESO406-042. This is a relatively small galaxy with a stellar disk of size 12.6 kpc (Table 1). The Spitzer 3.6 lm image shows that there is a small but bright bulge in the galaxy center, but the stellar disk appears

J. Astrophys. Astr. (2021)42:85

Page 3 of 10

85

Table 1. List of galaxies and parameters. Galaxy ESO406-042 NGC 2541 NGC 5832

RA, Dec (J2000)

Class

vsys (km/s)

Diameter (00 )

Distance (Mpc)

Scale (kpc/00 )

XUV type

23 02 14.2, 37 05 01.4 08 14 40.1, ?49 03 42.2 14 57 45.7, ?71 40 56.4

SAB(s) SA(s)cd SB(rs)b?

1365 548 447

138.0 396.4 222.9

18.6 11.9 9.5

0.090 0.058 0.046

Type 2 Type 2 Mixed type

Figure 1. The above panel shows the UV and optical images of ESO406-042. From top left clockwise: (i) the UVIT FUV image, (ii) the UVIT NUV image, (iii) the DSS IIIaj optical image which is based on the photographic data obtained using the U.K. Schmidt telescope and (iv) the Spitzer 3.6 lm image.

relatively diffuse and hence maybe low in surface density (Fig. 1). It is thus not surprising that the galaxy is classified as a Type 2 XUV galaxy, since the disk or outer XUV region is LSB in nature. The optical image shows that there are star forming knots distributed over the entire disk but there is no clear spiral structure. Two small arms can be seen in the optical image, but the overall appearance suggests a flocculant spiral structure rather than a strong one. NGC 2541. This is the largest XUV galaxy in our sample and it has a disk diameter of 21.19 kpc (Table 1). It has a bright oval shaped bulge and an extended LSB disk, which is clearly detected in the B

band image (Fig. 2). The stellar disk in the 3.6 lm image does not show any clear spiral structure, but faint spiral arms can be seen in the B band image. The outer disk shows surprisingly clear spiral arms in UV but they appear more flocculent in nature, and do not appear to be connected with the inner disk which is typical of the Type 1 XUV galaxies. NGC 5832. This is also a relatively small galaxy with a disk diameter of 10.25 kpc (Table 1). It is classified as a barred galaxy but the bar is difficult to trace in the Spitzer 3.6 lm image (Fig. 3) and it is relatively small compared to the disk size. The inner stellar disk is bright at 3.6 lm but is very diffuse in the outer parts.

85

Page 4 of 10

J. Astrophys. Astr. (2021)42:85

Figure 2. The above panel shows the UV and optical images of NGC 2541. From top left clockwise: (i) the UVIT FUV image, (ii) the UVIT NUV image, (iii) the SDSS g band optical image and (iv) the Spitzer 3.6 lm image.

This is similar to disks in Type 2 XUV galaxies. The outer disk also appears to have a different positions angle with repect to the inner one, this maybe due to the effect of the bar. There appears to be a faint spiral structure which is clearest in the B band image.

3. Observations and data analysis We performed FUV and NUV imaging observations of the galaxies listed in Table 1 using the UVIT on board the AstroSat Satellite (Kumar et al. 2012). The details of the filters and observational time are listed in Table 2. The instrument has two co-aligned Ritchey Chretien (RC) telescopes, one for FUV (1300–1800 ˚ ), and another for both the NUV (2000–3000 A ˚ ) and A visible bands. The UVIT is capable of simultaneously observing in all three bands, and the visible channel is used for drift correction. The UVIT has multiple photometric filters in both UV bands and the field-ofview of around 280 and a spatial resolution  1.500

which is much better than GALEX (Rahna et al. 2017). In our images the spatial resolution is  100 and the image resolution is determined from measuring the PSF of a background star (see Table 2 for our image values). A comparison of our UVIT and GALEX observing times are given in Table 3. Our UVIT FUV and NUV observing times are  2 Ks whereas the GALEX exposure times are of the order of 0.1 to 0.3 Ks. Hence our UVIT observations of these 3 XUV galaxies are far deeper than GALEX. We downloaded the UVIT Level 1 data of the 3 galaxies from the Indian Space Science Data Centre (ISSDC). We used CCDLAB (Postma & Leahy 2017) to reduce the Level 1 data. The CCDLAB has a graphical user interface to reduce the UVIT data and corrects for field distortions, and also does drift corrections. We did the astrometry on the UVIT images using GAIA data. We used a tool in CCDLAB which can match GAIA sources with UVIT sources and do the astrometry. We determined the background counts and then subtracted it from the image. We calculated

J. Astrophys. Astr. (2021)42:85

Page 5 of 10

85

Figure 3. The above panel shows the UV and optical images of NGC 2541. From top left clockwise: (i) the UVIT FUV image, (ii) the UVIT NUV image, (iii) the B band optical image from the Spitzer Legacy Optical Photometry Survey (Cook et al. 2014), and (iv) the Spitzer 3.6 lm image. Table 2. Summary of UVIT observations and the UV fluxes. Galaxy

Band

Filter

ESO406-042

FUV NUV FUV NUV FUV NUV

CaF2 Silica Sapphire N2 CaF2 NUVB13

NGC 2541 NGC 5832

Exposure time PSF (00 ) (image resolution) (Ks) 1.994 2.010 1.992 2.008 1.972 1.691

1.1 1.0 1.0 1.1 1.1 1.0

Cycle, date

Flux (erg cm2 s1 A1 )

A04-053, 14 Oct. 2017

2.58*1014 1.15*1014 1.32*1013 6.56*1014 6.56*1014 2.64*1014

A02-075, 21 Dec. 2016 A04-053, 03 Mar. 2018

The point spread function or PSF of the image is measured by fitting a Gaussian to a background bright star and taking the FWHM. For our UV images this is an approximate measure of the resolution of the image.

the counts per second (CPS) from a galaxy by summing over the CPS from a circular aperture of radius equal to the optical radius (R25 ) of the galaxy (see Table 2). To calculate the flux we used the formula: Fluxðerg s1 cm2 A1 Þ ¼ CPS  ðUCÞ;

ð1Þ

where UC is the unit conversion factor and is derived from the equation ZP ¼ ½2:5 logðUC  k2 Þ  2:407

(Tandon et al. 2017), and ZP is the zero point of the filter (Tandon et al. 2020). To calculate the flux in Jansky (Jy) which is listed in Table 3, we converrted the CPS to AB magnitude using the formula, mðABÞ ¼ 2:5 log 10ðCPSÞ þ ZP. After calculating the magnitude we corrected for the Milky way extiction for each filter and then converted the Milky Way extinction corrected magnitudes to fluxes in Jansky

85

Page 6 of 10

J. Astrophys. Astr. (2021)42:85

Table 3. Comparison of fluxes between UVIT and GALEX, and the Star Formation Rates (SFR) derived from UVIT.

Galaxy

UVIT exposure Band time (Ks)

ESO406-042 NGC 2541 NGC 5832

FUV FUV FUV

1.994 1.992 1.972

GALEX exposure time (Ks)

UVIT flux density (Jy)

GALEX F flux density (Jy)

0.281 0.106 0.111

2.10E-03±4.87E-04 1.55E-02±3.87E-03 5.08E-03±2.04E-03

2.05E-03±1.89E-05 1.09E-02±1.00E-04 5.35E-03±2.46E-04

SFR (UVIT)a M yr1 log(SFR) 0.089±0.021 0.317±0.056 0.056±0.022

–1.05 –0.50 –1.25

a

The SFR(UVIT) is calculated using the FUV and NUV intensities from Table 2 and using the formula from Salim et al. (2007) and the filter information from Tandon et al. (2020). For NGC 2541, since the FUV filter width was quite different compared to the GALEX filter, we normalised the CPS to GALEX CPS using background stars (see text). However, it should be noted that the SFRs have been corrected only for Galactic (Milky Way) extinction and not for the dust extinction arising from within the individual galaxies. Hence the true SFRs could be slightly higher.

using the formula, FluxðJyÞ ¼ 10ðMðABÞcorr 8:9Þ=2:5Þ , where MðABÞcorr is the corrected absolute AB magnitude. The star formation rate (SFR) was calculated using the FUV luminosities in Jy, using the formula from Salim et al. (2007) and the filter information from Tandon et al. (2020). For the galaxies ESO406042 and NGC5832, this was straight forward because the FUV fluxes were obtained using the CaF2 filter which has a wavelength range similar to GALEX. Hence the formula from Salim et al. (2007), which is applicable to GALEX observations, could be used. However, for NGC 2541 the filter used was sapphire. Hence, we calibrated the CPS in this filter with the GALEX observations using 7 stars in the field of the galaxy. It should be noted that since the UV fluxes have only been corrected for Milky Way extinction and not for the dust extinction arising from within the individual galaxies, i.e. without including the IR contribution, our SFR estimates are possibly lower than the true values. As part of this project, we observed all the three galaxies in our sample with the Giant Meter wave Radio Telescope (GMRT) during November–December, 2018 (Patra et al. 2019) (see Table 4 for details). All these galaxies were observed with the GMRT Software Backend (GSB) with either a 16.67 or 4.16 MHz bandwidth divided into 512 channels. This results in a spectral resolution of  32/8 kHz (  6.8/ 1.7 km s1 ); 8 h of telescope time was used for every observation. This led to an on-source time of  5 hours per source. Standard flux calibrator on the GMRT sky (3C48, 3C147, or 3C286) was observed for  10–15 min in the beginning and at the end of every observing run. Secondary calibrators near the target source (within 10 ) were observed for  6 min at  30 min cadence during the observations. The details of the observations are presented in Table 4.

The data were analyzed using the classic Astronomical Image Processing Software (AIPS). For every observing run, the data were first inspected for dead antennas, and visibilities coming from these antennas were removed. Next, the visibilities were further inspected and edited to exclude any bad data due to Radio Frequency Interference (RFI), sudden gain variation, antenna malfunction, etc. Flux and phase calibration was done using the AIPS task CALIB taking the primary and secondary calibrators as reference. A comparison of the recovered flux and the secondary calibrator’s reported flux in the NVSS catalog indicates consistency within  10  15%. After the flux and the phase calibrations were done, we performed a bandpass calibration to account for the gain variation in frequency. To do that, we used secondary calibrators, as they were observed with a higher cadence. We do not attempt to perform any self-calibration as the number of bright sources in the field is limited in L-band (1.4 GHz). We then applied all these calibrations to the target sources (our galaxies) and separated their data using the task SPLIT. This data contains signals from both the continuum and the HI line. However, as we are only interested in HI emission from these galaxies, we removed the continuum before we imaged the line emission. To do that, we first averaged all the visibilities from the line-free channels in a visibility cube. This average visibility data was then imaged to produce continuum maps of our galaxies. These maps only contain emission from the continuum sources in the galaxies. We used these maps to produce visibility cubes only for the line. We used the AIPS task UVSUB to subtract the continuum from the parent visibility cubes. However, in the presence of a strong continuum (e.g., for ESO 406-042), UVSUB may not be able to remove it completely. In such cases, we further

J. Astrophys. Astr. (2021)42:85

Page 7 of 10

85

Table 4. GMRT observation details.

Galaxy ESO406-042 NGC 2541 NGC 5832

Date of observation

On-source time (hr)

Observing BW (MHz)

Number of channels

Spectral width (km/s)

Flux Cal

Phase Cal

RMS/chan (mJy/bm)

November 25, 2018 December 23, 2018 December 23, 2018

4.5 5.5 5.5

16.66 4.16 4.16

512 512 512

6.9 1.8 1.8

3C48 3C147 3C286

0010–418 0713?438 1400?621

3.0 3.2 -

used the AIPS task UVLIN to fit the residual continuum as a first-order polynomial and subtracted it. We thus produced a visibility cube without having any continuum. The frequency axes of these visibility cubes were then shifted to a heliocentric velocity frame using the AIPS task CVEL. After the frequency standardization, all the continuum free visibility cubes were imaged channel by channel using the AIPS task IMAGR. To maximize the S/N and not resolve the diffused HI emission, we use a UV tapering of 5 kilolambda and a natural weighting scheme during imaging. For our observations, we find a typical single channel RMS of  3 mJy/beam. The image cubes are then further used to produce moment maps using the AIPS task MOMNT. We note that for NGC 5832, the data were heavily corrupted by RFI, so no usable spectral cube could be made. Hence, here we only present the results from NGC 2841 and ESO 406-042. In Fig. 4, we show the global HI spectra (blue dashed lines) for our sample galaxies, as obtained by our GMRT observations. For comparison, we also plot the same as obtained by single-dish observations (solid red lines).

4. Results (i) The UV emission and the stellar disks. The UV emission from galaxies traces hot, young star forming regions for at least 10 times longer timescales than Ha emission (Bianchi 2011). In general, UV traces star formation for  108 years (Thilker et al. 2007). A small but significant fraction also traces low mass, helium-core burning stars on the horizontal branch of the main sequence; this is often called the UV-upturn and is commonly seen in bright elliptical galaxies (Yi et al. 2011). There is also some UV emission from post asymptotic giant

branch stars (Montez et al. 2017) and white dwarfs (Sahu et al. 2019). The FUV emission arises from mainly hot stars, of which  40% is from stars younger than 10 Myr and  60% from stars that are older than 10 Myr (Calzetti 2013). The low sky background in FUV and low number of foreground stars that are bright in FUV ensures a high stellar contrast. This makes FUV emission an ideal wavelength to detect small star forming regions that are not detected in Ha, especially since O-type stars account for  90% of the FUV emission (Bianchi et al. 2014). However, FUV sources account for only  10% of the UV sources in GALEX surveys. Most of the UV emission from galaxies is composed of NUV emission from the redder population of stars (older and cooler). Hence, the NUV emission from galaxies is always more smoother and extended compared to the FUV emission. This is true for the 3 galaxies in our observations as well (Figures 1, 2 and 3), especially for ESO406-042 and NGC 5832. However, for NGC2541 the FUV emission appears brighter. This maybe because of a shorter exposure time, but we note that both FUV and NUV exposures were of similar time (Table 2), and the filters are different and hence have different wavelength coverage (Rahna et al. 2017). Some of the NUV data could be more noisy. Or there could be more young star formation in this galaxy compared to the other two galaxies. The fluxes are listed in Table 2. (ii) The nature of the underlying stellar disks. When we compare the UV emission with the near-infrared (NIR) emission, we find that the stellar disks are surprisingly diffuse. This is because the stellar disks in the XUV regions have low surface densities, or in other words the number density of stars in these regions is low. Even in regions supporting many compact star forming knots, the underlying stellar disk as traced by the NIR emission appears to be very low.

85

Page 8 of 10

J. Astrophys. Astr. (2021)42:85

Figure 4. The above panel shows the HI line emission for the 3 galaxies. They have been derived from our GMRT interferometric observations and also compared with single dish fluxes from other telescopes.

This is typical of Type 2 XUV disks as noted in earlier studies (Thilker et al. 2007). It is also seen in dwarfs where there is often a very low surface density disk underlying an extended HI gas disks (Patra et al. 2016; Das et al. 2019). When the mass surface density is plotted against the disk radius, it is seen to rapidly fall near the optical radius (e.g. NGC 628, NGC 3184) (Das et al. 2020). This has important implications for star formation in these regions as a lower stellar disk surface density means the disk has less self gravity and so cloud collapse maybe affected. (iii) Morphology of star formation. The UV morphology traces the star formation in galaxy disks. In all three galaxies the star formation is distributed over the diffuse stellar disks and does not follow any spiral arms, except for the northern arms of NGC 2541 where the star formation lies along elongated structures resembling arms. The starforming complexes are generally small and compact in nature (e.g. NGC 5832) or localized into large knots that are distributed over the entire disk (e.g. ESO406-042, NGC 2541). (iv) Comparing the UV and HI emission. The HI emission in all 3 galaxies is considerable and fairly extended (Figures 5 and 6). The HI appears to be more concentrated in the inner region around the UV emisison. However, surprisingly the HI is not as extended as observed in Type 1 XUV galaxies (e.g. NGC 628). This suggests that the cold gas is denser in the centers of Type 2 XUV galaxies compared to Type 1 XUVs, although the stellar disks in the XUV regions are very faint in both Type 1 and Type 2 XUVs. The dense HI is able to support the star formation. The molecular hydrogen (H2 ) is probably localized around the compact star-forming regions and hence not easy to detect (Bicalho et al. 2019). This is evident from previous

detections of very compact regions of molecular gas in the XUV disk of M63 (Dessauges-Zavadsky et al. 2014).

5. Discussion It is clear from the comparison of the FUV and NUV images with the 3.6 lm images that the underlying stellar disk in these star forming regions is very diffuse i.e. low in surface density. This is surprising as the star formation is widespread in all the galaxies. It is also seen in smaller dwarf galaxies as well (Das et al. 2019), where an outer blue disk extends well beyond an inner old stellar disk which supports most of the star formation. It is known that the main triggers for outer disk star formation in XUV galaxies are (i) galaxy interactions (e.g. NGC 4625), (ii) gas accretion from high velocity clouds (HVCs) (e.g. NGC 891) and (iii) cold gas accretion by galaxies from the cosmic web (e.g. NGC 2403) (de Blok et al. 2014). However, not all galaxies with extended HI disks have XUV disk star formation nor do they all show signs of gas accretion from HVCs or the intergalactic medium (IGM). Hence, it is still unclear why some galaxies show extended disk star formation and some do not, despite being rich in HI gas. One possibility is that disk dark matter plays an important role in both supporting the HI disk in regions of low stellar mass density as well as increasing the disk mass surface density so that the Toomre instability factor Q reduces and instabilities can form in the diffuse stellar disks. The onset of local instabilities leads to star formation (Das et al. 2020). This can explain the widespread star formation in Type 2 XUV disks and the compact nature of the star formation as well. The dark matter associated with the

J. Astrophys. Astr. (2021)42:85

Page 9 of 10

85

Figure 5. The panel on the left shows the HI emission for the galaxy NGC 2541 observed with the GMRT. The plot on the left is the HI intensity image or moment0 map. The contours are ð1; 1:4; 2; 2:8; . . .Þ  8  1019 atoms/cm2 , i.e., the pffiffi contours start from 8  1019 and successive contours are separated by ð2Þ. The one in the middle is the HI velocity field or the moment1 map with contours from 470 km/s to 640 km/s. Contour separation is by 10 km/s. The figure on the right is the moment2 map of NGC 2541 with minimum = 0 km/s, maximum = 22 km/s. Note that the HI emission has an arm-like structure west of the galaxy center. This appears as an extended spiral structure in the deeper Westerbrock telescope images.

Figure 6. The figure on the left shows the HI emission for the galaxy NGC 5832 observed with the Westerbrock Synthesis Radio Telescope (WSRT) (van der Hulst et al. 2001). The plot on the right is the FUV image of NGC 5832 with contours of HI emission overlaid. The contours are 1/5, 1/3 and 1/2 of the peak intensity which is 9.7 mJy/beam and the beam is 10:800  10:100 . The contours follow the FUV emission quite well but also extend slightly beyond the emisison.

disk could be due to the presence of a very oblate halo (Das et al. in preparation) or it may have resulted from the accretion of dark matter from several episodes of minor mergers during the evolution of the galaxy. 6. Conclusion (1) We present UVIT FUV and NUV images of 3 XUV type galaxies, ESO406-042, NGC 2541 and NGC 5832. The first two are Type 2 XUV galaxies and the third one is a mixed type XUV galaxy. (2) We also present the GMRT HI maps of NGC 2541, both the intensity map and the velocity field. We also compare the UV emission from NGC 5832 with the Westerbrock archival HI map. All three galaxies are rich in HI and the star formation is

associated with the cold gas emission. However, the HI emission is not as extended as in Type 1 XUV galaxies. (3) The star formation is distributed over the diffuse stellar disks of the galaxies and is comprised of compact knots of star-forming complexes. This suggests that the star formation originated in local instabilities in the galaxy disks rather than global instabilities such as spiral arms or bars.

Acknowledgements The authors gratefully acknowledge the IUSSTF grant JC-014/2017, which enabled the authors MD, NNP, and KSD to visit CWRU and develop the science

85

Page 10 of 10

presented in this paper. This publication uses the data from the UVIT which is part of the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). This publication uses UVIT data processed by the payload operations centre at IIA. The UVIT is built in collaboration between IIA, IUCAA, TIFR, ISRO and CSA. The HI observations were done using the GMRT. We thank the staff of the GMRT that made these observations possible. The GMRT is run by the National Centre for Radio Astrophysics of the Tata Institute of Fundamental Research. This research has used Spitzer 3.6 micron images. This research has also made use of the NASA/IPAC Extragalactic Database (NED), which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration. The facilities used are Astrosat (UVIT), GALEX, GMRT, WSRT, Spitzer, SDSS, GBT, Parkes. References Bianchi L. 2011, Astrophys. Space Sci., 335, 51 Bianchi L., Conti A., Shiao B. 2014, Adv. Space Res., 53, 900 Bicalho I. C., Combes F., Rubio, M., Verdugo C., Salome P. 2019, A&A, 623, A66 Bigiel F., Leroy A., Seibert M. et al. 2010, ApJL, 720, L31 Bruzzese S. M., Thilker D. A., Meurer G. R. et al. 2020, MNRAS, 491, 2366 Calzetti D. 2013, in Falco´n-Barroso J., Knapen J.-H., eds, Star Formation Rate Indicators, 419 Cook D. O., Dale D. A., Johnson B. D. et al. 2014, MNRAS, 445, 881 Das M. 2013, J. Astrophys. Astr., 34, 19 Das M., McGaugh S. S., Ianjamasimanana R., Schombert J., Dwarakanath K. S. 2020, ApJ, 889, 10 Das M., Sengupta C., Honey M. 2019, ApJ, 871, 197 de Blok W. J. G., Keating K. M., Pisano D. J. et al. 2014, A&A, 569, A68 Dessauges-Zavadsky M., Verdugo C., Combes F., Pfenniger D. 2014, A&A, 566, A147 Ferguson A. M. N., Wyse R. F. G., Gallagher J. S., Hunter D. A. 1998, ApJL, 506, L19 Gil de Paz A., Boissier S., Madore B. F. et al. 2007, ApJS, 173, 185

J. Astrophys. Astr. (2021)42:85 Honey M., Das M., Ninan J. P., Manoj P. 2016, MNRAS, 462, 2099 Kaczmarek J. F., Wilcots E. M. 2012, AJ, 144, 67 Kataria S. K., Das M. 2018, MNRAS, 475, 1653 Kennicutt Robert C. Jr. 1989, ApJ, 344, 685 Krumholz M. R., McKee C. F. 2008, Nature, 451, 1082 Kumar A., Ghosh S. K., Hutchings J. et al. 2012, in Takahashi T., Murray S. S., den Herder J.-W. A. eds., Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 8443, Space Telescopes and Instrumentation 2012: Ultraviolet to Gamma Ray, 84431N Lemonias J. J., Schiminovich D., Thilker D. et al. 2011, ApJ, 733, 74 Martin D. C., Fanson J., Schiminovich D. et al. 2005, ApJL, 619, L1 McGaugh S. S., Schombert J. M., Bothun G. D. 1995, AJ, 109, 2019 Montez Rodolfo J., Ramstedt S., Kastner J. H., Vlemmings W., Sanchez E. 2017, ApJ, 841, 33 Morrissey P., Conrow T., Barlow T. A. et al. 2007, ApJS, 173, 682 Patra N. N., Chengalur J. N., Karachentsev I. D., Kaisin S. S., Begum A. 2016, MNRAS, 456, 2467 Patra N. N., Kanekar N., Chengalur J. N. et al. 2019, MNRAS, 483, 3007 Postma J. E., Leahy D. 2017, PASP, 129, 115002 Rahna P. T., Das M., Murthy J., Gudennavar S. B., Bubbly S. G. 2018, MNRAS, 481, 1212 Rahna P. T., Murthy J., Safonova M. et al. 2017, MNRAS, 471, 3028 Sahu S., Subramaniam A., Simunovic M. et al. 2019, ApJ, 876, 34 Salim S., Rich R. M., Charlot S. et al. 2007, ApJS, 173, 267 Tandon S. N., Subramaniam A., Girish V. et al. 2017, AJ, 154, 128 Tandon S. N., Postma J., Joseph P. et al. 2020, AJ, 159, 158 Thilker D. A., Bianchi L., Boissier S. et al. 2005, ApJL, 619, L79 Thilker D. A., Bianchi L., Meurer G. et al. 2007, ApJS, 173, 538 van der Hulst J. M., van Albada T. S., Sancisi R. 2001, in Hibbard J. E., Rupen M., van Gorkom J. H., eds, Astronomical Society of the Pacific Conference Series, Vol. 240, Gas and Galaxy Evolution, 451 van der Kruit P. C., Searle L. 1981, A&A, 95, 105 Yi S. K., Lee J., Sheen Y.-K. et al. 2011, ApJS, 195, 22

J. Astrophys. Astr. (2021) 42:86 https://doi.org/10.1007/s12036-021-09764-w

Ó Indian Academy of Sciences Sadhana(0123456789().,-volV)F T3](0123456789().,-volV)

SCIENCE RESULTS

The sharpest ultraviolet view of the star formation in an extreme environment of the nearest Jellyfish Galaxy IC 3418 ANANDA HOTA1,* , ASHISH DEVARAJ2 , ANANTA C. PRADHAN3,

C. S. STALIN2, KOSHY GEORGE4, ABHISEK MOHAPATRA3, SOO-CHANG REY5, YOUICHI OHYAMA6, SRAVANI VADDI7, RENUKA PECHETTI8, RAMYA SETHURAM2, JESSY JOSE9, JAYASHREE ROY10 and CHIRANJIB KONAR11 1

#eAstroLab, UM-DAE Centre for Excellence in Basic Sciences, University of Mumbai, Mumbai 400 098, India. 2 Indian Institute of Astrophysics, Koramangala II Block, Bangalore 560 034, India. 3 Department of Physics and Astronomy, National Institute of Technology, Rourkela 769 008, India. 4 Faculty of Physics, Ludwig-Maximilians-Universita¨t, Scheinerstr 1, 81679 Munich, Germany. 5 Department of Astronomy and Space Science, Chungnam National University, Daejeon 34134, Republic of Korea. 6 Institute of Astronomy and Astrophysics, Academia Sinica, No. 1, Sec. 4, Roosevelt Rd, Taipei 10617, Taiwan. 7 Arecibo Observatory, NAIC, HC3 Box 53995, Arecibo, PR 00612, USA. 8 Astrophysics Research Institute, Liverpool John Moores University, 146 Brownlow Hill, Liverpool L3 5RF, UK. 9 Indian Institute of Science Education and Research (IISER) Tirupati, Tirupati 517 507, India. 10 Inter-University Centre for Astronomy and Astrophysics (IUCAA), Pune 411 007, India. 11 Amity Institute of Applied Sciences, Amity University Uttar Pradesh, Sector-125, Noida 201 303, India. *Corresponding Author. E-mail: [email protected] MS received 7 November 2020; accepted 29 April 2021 Abstract. We present the far ultraviolet (FUV) imaging of the nearest Jellyfish or Fireball galaxy IC3418/ VCC 1217, in the Virgo cluster of galaxies, using Ultraviolet Imaging Telescope (UVIT) onboard the AstroSat satellite. The young star formation observed here in the 17 kpc long turbulent wake of IC3418, due to ram pressure stripping of cold gas surrounded by hot intra-cluster medium, is a unique laboratory that is unavailable in the Milky Way. We have tried to resolve star forming clumps, seen compact to GALEX UV images, using better resolution available with the UVIT and incorporated UV-optical images from Hubble Space Telescope archive. For the first time, we resolve the compact star forming clumps (fireballs) into sub-clumps and subsequently into a possibly dozen isolated stars. We speculate that many of them could be blue supergiant stars which are cousins of SDSS J122952.66?112227.8, the farthest star (*17 Mpc) we had found earlier surrounding one of these compact clumps. We found evidence of star formation rate (4–7.4 9 10–4M yr–1) in these fireballs, estimated from UVIT flux densities, to be increasing with the distance from the parent galaxy. We propose a new dynamical model in which the stripped gas may be developing vortex street where the vortices grow to compact star forming clumps due to self-gravity. Gravity winning over turbulent force with time or length along the trail can explain the puzzling trend of higher star formation rate and bluer/younger stars observed in fireballs farther away from the parent galaxy. Keywords. Galaxies: star formation—galaxies: formation—galaxies: evolution—galaxies: IC3418— ultraviolet: galaxies. This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

86

Page 2 of 12

1. Introduction The basic understanding of the star formation process has still been a puzzle (McKee & Ostriker 2007; Zinnecker & Yorke 2007; Kennicutt & Evans 2012; Krumholz 2014). The high resolution optical observations with the Hubble Space Telescope (HST), Ultraviolet (UV) observations with the GALEX, infrared observations with the Spitzer and recently cold molecular gas and dust observations with the Atacama Large Millimeter/submillimeter Array (ALMA) have been providing crucial missing pieces towards resolving this age-old puzzle. Further this has been assisted with modern magneto-hydrodynamic numerical simulation studies to achieve a correct theoretical understanding of star formation (e.g. Wu et al. 2017). The nearest dwarf galaxies, Small Magellanic Cloud (SMC) and Large Magellanic Cloud (LMC), having little dust and gas or almost transparent, have been providing unique opportunity to investigate star formation at high resolution (Oey & Massey 1995; Indu & Subramaniam 2011). Triggered star formation has been commonly observed in our Milky Way galaxy and, LMC and SMC where gaseous flow of the interstellar medium (spiral arm density wave or supernova shell expansion) is triggering further young star formation (Elmegreen & Elmegreen 1986; Oey & Massey 1995). Galaxy merger has also been a well-documented triggering mechanism as observed in interacting, merging and Ultra-Luminous Infra-Red Galaxies (ULIRGs) with massive central starbursts (Jog & Solomon 1992; Sanders & Mirabel 1996). Role of various instabilities, turbulence (both radiation and hydrodynamically driven) and magnetic field are certainly critical to star formation though it complicates the matter significantly (Federrath & Klessen 2012; Wu et al. 2017; Mu¨ller et al. 2021). We bring attention to triggered star formation in external galaxies in two types of extreme but opposite environments which are not only rare but also can not be observed in the Milky Way or LMC/SMC. They are so unique opportunities that it is almost like our understanding of star formation is put to extreme tests. First case is when the relativistic radio jet of nonthermal plasma, emitting synchrotron radiation, from the accretion on to the supermassive black hole, hits the outer cold gas (seen in HI and CO emission lines) in a galaxy and triggers young star formation as seen in Ha line or blue optical light and UV emission. This case, also referred to as ‘positive feedback’, is seen in

J. Astrophys. Astr.

(2021) 42:86

the nearest radio galaxy Cen A, Death-star galaxy, Minkowski Object etc. (Salome et al. 2016; Santoro et al. 2015; Croft et al. 2006; Evans et al. 2008; Zovaro et al. 2020). This is in contrast to the wellknown negative feedback that is required for AGNfeedback models to explain black hole galaxy coevolution (e.g., Springel et al. 2005; Croton et al. 2006; Hopkins et al. 2006; Fabian 2012; Hota et al. 2012, 2016; Hardcastle & Croston 2020). The second unique case, which is the focus of this manuscript, is in the ‘Jellyfish’ or ‘Fireball’ galaxy where the triggering mechanism is almost opposite. Here the cold InterStellar Medium (ISM) of the galaxy has been ram pressure stripped by the million degree hot, X-ray emitting, thermal plasma of the Intra-Cluster Medium (ICM) through which the galaxy has been moving with nearly a thousand km s-1 velocity (Gunn & Gott 1972; Abadi et al. 1999; Oosterloo & van Gorkom 2005; Vollmer et al. 2001). In some rare cases, the ongoing star formation in the striped gas tail of these systems along with the disk of the galaxy gives the appearance of a Jellyfish in the sky. Here the cold gas and the surrounding thermal plasma are similarly in high relative velocity and in extreme contrast of physical composition and state of matter. The number of jet-triggered star formation can be observed is nearly six radio galaxies and cases of Jellyfish galaxies could be nearly a dozen (GASP survey; Poggianti et al. 2017). We had proposed to observe two such targets NGC3801 with radio-lobe feedback (Hota et al. 2012) and IC3418 the nearest Jellyfish galaxy with the AstroSat for best possible angular and spatial resolution. IC3418 was observed with UVIT onboard the AstroSat satellite and we report our study briefly in this manuscript focusing on the compact star forming clump or fireballs.

2. The Jellyfish or Fireball galaxy IC3418 Numerous examples of ram pressure-stripped gas tails with HI and Ha emission, extending up to 100 kpc, have been discovered. However, only recently they have been confirmed to continue forming young stars in clumps of these tails. As the clumps become bright in UV, blue optical light and Ha, they create stunning ‘‘Jellyfish’’ galaxy structures with clumps having ‘‘Fireball’’ like impression where the small tails of the fireballs are seen oppositely directed to the large tail of the galaxy (Cortese et al. 2007; Sun et al. 2007; Yoshida et al. 2008; Smith et al. 2010; Hester et al.

J. Astrophys. Astr.

(2021) 42:86

Page 3 of 12

86

Figure 1. Both the GALEX NUV image (top right panel) and AstroSat UVIT far UV (F148W) image in both logarithmic grey scale and in green contours (top left panel) are shown. The images also label all the five compact star forming clumps (fireballs) that have been analysed in detail. The bottom panels show the FUV image from AstroSat with F148W on the right and F154W on the left. Both the images are matched for scale, limit and colour bar for precise comparison.

2010; Yagi et al. 2013; Poggianti et al. 2017; Gullieuszik et al. 2017; George et al. 2018). As seen in Fig. 1(top right panel), IC3418 has a clumpy star forming 17 kpc long tail which is bright in GALEX UV images (Hester et al. 2010). IC3418 being the closest (100 * 80 pc for 16.7 Mpc to Virgo cluster (Kenney et al. 2014)), it has enabled a detailed star formation study of its tail (Hester et al. 2010; Fumagalli et al. 2011; Kenney et al. 2014). We had independently discovered this in 2007 and have been following up with Lulin optical imaging, Subaru multi-slit spectroscopy and stellar population synthesis analyses. We had reported the discovery of the farthest single star SDSS J122952.66?112227.8, a

blue supergiant star in its tail using Subaru spectroscopy and CFHT imaging (Ohyama & Hota 2013). Several puzzles of the star forming clumps in the tail or ‘‘fireballs’’ are yet to be resolved. UV and optical colour of the fireballs farther away from the parent galaxy are found to be relatively bluer than the nearby fireballs as well as the parent galaxy (Hester et al. 2010; Yoshida et al. 2008; Fumagalli et al. 2011; Kenney et al. 2014). From Ha emission and colour observed over 80 kpc long tail of another target RB199 in Coma cluster, Yoshida et al. (2008) argued that the optically blue fireballs farther away contain a younger stellar population compared to fireballs near the parent galaxy. This requires further observational

86

Page 4 of 12

confirmation and careful interpretation as this is in contradiction to the natural expectation. The gasclumps freshly stripped should have formed stars only recently and on the other hand clumps farther away, stripped earlier, should have relatively old stars as they have been forming stars for a longer time. It is interesting to note that such a 20–80 kpc long tail is like a ‘‘chart recorder’’ of star formation parameters, gradually changing along the length of the tail as the parent galaxy moves with a few thousand km s–1 through the ICM. With the time since decoupling from the parent galaxy, the relative velocity of the stripped gas clouds and ambient medium should change from nearly a thousand km s-1 to nearly zero, at the end of the tail where it joins the ICM, a point of no return. Possibly due to the detection of the HI gas cloud near the end of the tail, Kenney et al. (2014) had postulated that once stars form in the striped gas cloud, those stars no longer feel the ram pressure and are left on the trail to age. On the other hand, the gas cloud continues to be accelerated away from the parent galaxy due to ram pressure and continues to form stars. This led to the observation of stellar trails behind each fireball and younger stars farther away from the parent galaxy. However, such a simple-minded explanation is not satisfactory to us, as very good HI and Ha imaging of other ram pressure stripped galaxies show continuous stripping of gas clouds from the parent galaxy. Each fireball (gas clump) may show this age/colour gradient on kpc-scale but on a larger 10–100 kpc scale this should be washed out. The most detailed investigation for 80-kpc long tail and its fireballs exists for the Coma cluster galaxy RB199 (Yoshida et al. 2008). Contamination from unresolved background sources (also discussed by Kenney et al. 2014) and the interfireball light may contribute to the confusing trend of farther away fireballs being bluer. It is also known that the inter-fireball filamentary or diffuse emission is redder than the fireballs (Hester et al. 2010; Yoshida et al. 2008). Hence, a higher angular resolution imaging than achieved by GALEX is going to provide more accurate colour, star formation rate and age estimation of fireballs. We may expect the fireballs to remain compact, be resolved to sub-clumps, show head-tail structure or even show bow-shock at the head as seen in the case of Mira-A star with the GALEX (Martin et al. 2007). With the new result obtained from our high resolution AstroSat study, we propose an alternate hypothesis on the dynamical evolution of these Jellyfish galaxies with young star forming fireballs.

J. Astrophys. Astr.

(2021) 42:86

3. Observation and data analysis ˚) IC3418 was observed in the Far-UV (1300–1800 A using the Ultra-Violet Imaging Telescope (UVIT) onboard AstroSat (Agrawal 2006). The observations were carried out during May 2018. The galaxy was observed in two FUV filters F148W and F154W ˚ and 1541 A ˚ (Tandon et al. 2020), centred at 1481 A respectively. The log of observations are given in Table 1. The images were processed at the UVIT-payload operation center (UVIT-POC) and made available to the investigators through the Indian Space Science Data Center (ISSDC). We carried out our analysis directly on the science-ready images made available to the ISSDC. The FUV images obtained are shown in Fig. 1. The images revealed five clumps of star formation or fireballs in the ram pressure-stripped tail of this Jellyfish galaxy. The AB magnitude of these clumps was measured using the latest photometric calibrations of UVIT given in Tandon et al. (2020). An aperture of 300 (*7 UVIT sub-pixels) centred at the local maxima were used for this purpose. This will contain more than 80% of the energy from the source (Tandon et al. 2020). The position and UV magnitudes of the clumps, in both the filters, are given in Table 2.

4. Results IC3418: Figure 1(top panel) shows the comparative high and low resolution structure of IC3418 as seen in AstroSat FUV (left) and GALEX NUV (right) images, respectively. In the bottom panel far-UV images from AstroSat, in both the filters, are presented. The FUV images, matched for scale, limit and colour bar, clearly show that higher significance of F148W image which has been used in all subsequent images. Here it is interesting to note that unlike the NUV GALEX image, the AstroSat FUV emission from the disk of the galaxy is getting fainter/resolved out at higher resolution compared to the compact clumps being analysed here. We have chosen to ignore the parent galaxy in our current study. Further, we have chosen not to present analysis on the diffuse kpc-scale star formation seen in this 17 kpc long tail which is detected clearly in the GALEX images along with these five compact clumps. However, note that the diffuse emission is redder in FUV-NUV colour compared to the compact clumps. This is clear from the

J. Astrophys. Astr.

(2021) 42:86

Page 5 of 12

86

Table 1. UVIT Filters used in observation. Filter F148W F154W

˚) Mean wavelength (A

˚) Bandwidth (A

Exposure time (s)

1481 1541

500 380

2927.940 1757.087

Table 2. Photometry results of individual clumps. Clump 1 2 3 4 5

R.A.

Dec.

Mag. in F148W

Error

Mag. in F154W

Error

12:29:47.56 12:29:49.66 12:29:50.73 12:29:52.73 12:29:52.03

11:23:35.67 11:23:08.58 11:22:35.01 11:22:49.21 11:22:04.74

22.035 21.812 21.553 21.368 21.644

0.283 0.256 0.227 0.208 0.237

21.812 21.778 21.458 21.321 21.444

0.384 0.378 0.326 0.306 0.324

Figure 2. Images presented here are for the Clump-1 (fireball-1). Left panel shows the UVIT FUV (F148W) image both in greyscale (logarithmic scale) and contours. The region-circle has a radius of 500 and centered on the R.A. and Dec. for which photometry has been performed as described in Table 1. Middle panel presents the same contour map along with UV images from HST obtained from Hubble Legacy Archive (Proposal ID: 12756). Right panel represents a colour image of a slightly larger region around the centre of the clump. The red ‘?’ mark in the colour image represents the centre of the left panel. The colour image has been created from the same HST data with blue, green and red showing F275W (effective wavelength = 2707.94 A), F475W (4746.16 A) and F814W (8044.99 A) filter images, respectively. The exposure times for the blue, green and red filter images are 2646, 2913 and 1356 seconds, respectively. Data presented in Fig. 2 to Fig 6 are identical except that they are for different clumps starting from Clump-1 to Clump-5.

photometric measurements presented in Table 1 by Hester et al. (2010). They find the average FUV-NUV colors for the diffuse emission is 0.35 ± 0.06 and for the fireballs to be –0.02 ± 0.03. The F148W-F154W colour as measured from AstroSat UVIT for fireballs 1 to 5 are 0.223, 0.034, 0.095, 0.047, 0.200 magnitudes, respectively (see Table 1). Hence, though the wavelengths are very close, the colour gradient has not changed its sign in any case, suggesting that even at the highest angular resolution possible they are young star-forming clumps and no other shock-related unusual process (Martin et al. 2007). We did not attempt photometric measurements of the diffuse emission but

focus on the comparative analysis of the new structures that can be seen in the compact clumps (mentioned in the literature differently as knots or fireball) and have compared that with UV-optical colour images from HST archive. Clump-1: Figure 2 presents zoomed in view of the UVIT FUV images of Clump-1 (Fireball-1), the closest to the parent galaxy, along with the UV image from HST. The contours from UVIT FUV (F148W) images are superposed on HST UV images for better comparison. The circles are centred on the compact clumps for which photometric measurements have

86

Page 6 of 12

been done on UVIT images. Note that the circles have a radius of 5 arcseconds which is about the angular resolution of GALEX images. The UVIT image showing compact central emission and diffuse emission seen on the south and south-west are consistent with the GALEX image. Note that our higher resolution images did not reveal any head-tail or bowshock structure which were expected in certain models for how stars are formed in this exotic dynamical system. The bright compact clump in UVIT has multiple components as demonstrated in UV image from HST. Nearly 8 point sources can be noticed in the HST image which are located in the unresolved UVIT peak. The faint resolved point sources seen in the HST image located to the south and south-west of the core of the clump explain the faint UV extension seen in the UVIT image. The bottom panel with colour image from HST clearly shows further diffuse star formation. It is unclear if the red point sources belong to the tail of IC3418 or background galaxies. The extended red sources are definitely background galaxies. As the diffuse UV emission observed in GALEX is redder, some red point sources may be contributing to it. Thus Clump-1 is compact in GALEX, and continues to be compact in UVIT but HST resolves that to a similar bright core consisting of nearly 8 surrounding point sources and scattered point sources to the south and south-west. Notably, the UVIT study did not show the head-tail or the shockshell-like extended structure as expected in some models for fireballs. However, HST resolves them to point sources, likely to be single resolved stars. Clump-2: In this compact clump UVIT seems to be resolving to multiple components with roughly equal brightness. HST resolves the structure further to a dozen of point sources that too conserve the overall structure. As this clump really lacks a prominent core, it is likely that all the point sources seen in HST images are single stars (similar to the claimed case of

J. Astrophys. Astr.

(2021) 42:86

SDSS J122952.66 ?112227.8, a blue supergiant star by Ohyama and Hota (2013) presented in a later section). The colour HST image did not show any red point source in the compact clump but only farther away which may belong to the diffuse star formation which got possibly decoupled from the clump due to ram pressure (see fireball model Fig. 16 of Kenney et al. 2014). In summary, Clump-2 (fireball-2) lacks a compact core, head-tail or bow-shock structure (Fig. 3). Clump-3: This compact clump in the GALEX image is resolved to a central peak and a tail to the north-west in the UVIT image. Here the tail extends towards the parent galaxy in the north-west. HST image finds bright point sources at the core with a sub-clump and a single point source making the tail. Also notice the diffuse emission seen in the clump. Within 5 arcsecond radius, a fainter but isolated point source could also be observed. Hence the compact core remains compact with close-enough point sources and a tail of sub-clump emission. Diffuse emission seen in the clump, in both UV and optical bands, may be scattering and/or unresolved stars (Fig. 4). Clump-4: This compact clump has been resolved by UVIT to be an elongated structure, roughly in the eastwest direction. The HST UV image does not find any sub-clumps in the elongation but the core has closelypacked multiple point sources lacking any dominant member. The east-west elongation of the Clump-4 can be resolved to point sources in the HST image (Fig. 5). Clump-5: This is the farthest clump that we have analysed. Here again the central peak is prominent with possible extension in the direction of the parent galaxy. The HST UV images resolve the peak to be closely-spaced multiple point sources with a dominant member at the peak. The fainter point sources located on the north and west makes the UV tail observed in UVIT. Hence the compact core of GALEX remains a

Figure 3. Imaging data presented here in all three panels are exactly the same as in Fig. 2 but for the Clump-2.

J. Astrophys. Astr.

(2021) 42:86

Page 7 of 12

86

Figure 4. Imaging data presented here in all three panels are exactly the same as in Fig. 2 but for the Clump-3 (Fireball-3).

Figure 5. Imaging data presented here in all three panels are exactly the same as in Fig. 2 but for the Clump-4 (Fireball-4). It is important to note here bad pixels running roughly north-south in a band. They appear in red and green channels only and HST UV image (blue) is not affected (see middle panel).

Figure 6. Imaging data presented here in all three panels are exactly the same as in Fig. 2 but for the Clump-5 (Fireball-5).

compact core in UVIT as well as HST images and the head-tail structure is also confirmed. Red optical point sources visible close to the peak and extended diffuse sources, either background or foreground, must be contaminating all measurements estimating colour and stellar populations (Fig. 6). Blue Supergiant Star (BSG): Finally, after these clumps/fireballs we focus our analyses with the star

SDSS J122952.66?112227.8 which we had discovered in the diffuse star forming region between Clump-4 and Clump-5 and presented it to be a possibly single blue supergiant star (Ohyama & Hota 2013). The precise location of this BSG derived from UV images taken with the HST is R.A. 12:29:52.695 Dec. ?11:22:27.937. We had noted a point source 100 away on the south-east and that is now resolved into two point sources with some diffuse emission. As

86

Page 8 of 12

J. Astrophys. Astr.

(2021) 42:86

Figure 7. Right panel shows the zoomed in UV image of the region around the blue supergiant (BSG) star SDSS J122952.66?112227.8 (topmost point source) from the UV HST data which is presented in grey logarithmic scale. The left panel is the same as in previous colour images from HST. As described in Fig. 5, bad pixels affect the HST colour map. The BSG has been labelled with a cyan circle. The blue point source (left panel), also seen as the bottommost point source in the (right panel), is likely an image artifact which has no counterpart in green and red filter images. Rest all three point sources in the UV image have optical counterparts in HST data.

presented in Fig. 7(right panel) in the UV image, the BSG indeed is seen isolated without any sub-structure and strongly supports our claim of a single BSG star. Except the bottom one which is seen only in UV and thus likely an artifact, all the point sources have counterparts in both green and red optical filter images with the HST. Unfortunately due to the band of bad pixels HST source catalogue does not list its magnitudes. Comparing the pixel value (0.075) of the peak for BSG seen in UV image taken with the HST we notice that several of the point sources seen in each of these five compact clumps (fireballs) have similar or higher than this pixel value. Hence, we speculate that there would possibly be dozens of BSGs in this 17 kpc long tail of IC3418 which is at *16.7 Mpc away. Since most of these compact fireball clumps are not having red point sources or diffuse/unresolved emission but resolved point sources they are likely wellseparated O and B type stars. As ram pressure does not affect stars but only gas in the clumps, stars formed at different times may appear at different locations in the tail, acting like a chart-recorder of evolution of star formation. Some red point sources seen in HST images may indeed be such decoupled stars from a compact clump on the left. It is important to note a caveat in this picture that the radial velocity measured for this BSG is –99 km s-1 which is different from radial velocities measured from the clumps in the tail. Though Ohyama & Hota (2013) have argued the physical association of the BSG with the tail of IC3418, based on luminosity and line ratio arguments, Kenney et al. (2014) doubted it based on

these velocity measurements. With the HST imaging presented here where the BSG lacks any surrounding diffuse emission and many such point sources can be seen in the fireballs, the circumstantial evidence of the association seems strong compared to the radial velocity argument.

5. Discussion Similar to the five compact clumps or fireballs that we have defined here, nearly six regions of diffuse star formation can be seen. Hester et al. (2010) defined 9 clumps (knots K1 to K9) and 3 diffuse regions (D1, D2, D3). Note that diffuse regions are in general redder and possibly contain old stars likely because they are already decoupled from star forming gas clouds due to difference in ram pressure that a molecular gas cloud and a compact star would experience. Such red colour in optical is also seen in RB199, the next nearest Jellyfish galaxy in the Coma cluster (Yoshida et al. 2008). Although the tail is expected to have a lower surface density of gas than the galaxy disk itself, the higher resolution of AstroSat shows the fireballs to be brighter in FUV than the disk (Fig. 1). It is unclear if higher extinction in the parent galaxy is the reason. However, neutral HI gas cloud is detected at the end of the tail and not from the parent galaxy (Kenney et al. 2014). A suitable mechanism of creating high density molecular gas clouds must be existing in these compact clumps to emit bright UV emission from young star formation. On the contrary,

J. Astrophys. Astr.

(2021) 42:86

Page 9 of 12

86

Table 3. Properties of the fireball clumps. Clump No. Fireball-1 Fireball-2 Fireball-3 Fireball-4 Fireball-5

Distance from galaxy in arcmin

FUV SFR (10–4M yr–1)

Error SFR (10–4M yr–1)

Fireball-galaxy velocity difference (km s–1)

Stellar age (M yr–1)

1.13 1.83 2.39 2.55 3.00

4.008 4.921 6.247 7.408 5.745

1.045 1.160 1.306 1.419 1.254

60 42 44 117 49

130 80 390 170 740

ionisation/heating and dispersion of the stripped diffuse cold neutral gas cloud would also be unavoidable coming in contact with the hot ICM plasma in high relative velocity. Fireball Evolutionary Sequence: We looked for possible gradual change in the measured physical parametres along the tail, with the distance from the parent galaxy. The star formation rate (SFR) is estimated following the prescription for nearby galaxies as in Karachentsev and Kaisina (2013). The FUV (F148W) magnitudes from the clumps/fireballs, as listed in the Table 2, has been used. Galactic extinction coefficient (A = 0.248 mag) was applied following Cardelli et al. (1989) and no correction for internal extinction was assumed as fireballs are far away from the dwarf parent galaxy, which has already been stripped of its HI ISM. We find that the SFR in the clumps/fireballs ranges from 4 to 7.4 9 10–4M yr–1 with the error ranging 1–4 9 10–4M yr–1 (see Table 3). This is comparable to the values found by Fumagalli et al. (2011) estimated using Ha flux densities. This is an order of magnitude less than that for fireballs in the 80 kpc long tail of RB199 (Yoshida et al. 2008). It can be noticed clearly that the FUV SFR of fireballs increases with the distance from the parent galaxy (Table 3 and Fig. 8). From the figure, we can see an increase of *1 9 10–4M yr–1 of SFR per arcmin (4.8 kpc) projected distance. We further expressed it in optical image with other measured parametres of the fireballs (Fig. 9). This increase in SFR can explain the observation of fireballs farther away showing relatively bluer optical and UV colour in various observations (Hester et al. 2010; Yoshida et al. 2008; Fumagalli et al. 2011; Kenney et al. 2014). As discussed earlier, the actual reason behind this is younger age of stars at the tail end or is due to higher extinction close to the parent galaxy or the difference in metallicity of stripped gas spread on to the tail could not be disentangled. However, at least in the case of IC3418, HI gas cloud is found near the tail end and Ha emission is seen at the tail end, ruling out

Figure 8. Plot of the star formation rate, along with the error, has been plotted with distance of the fireballs from the parent galaxy.

extinction as a major reason. Furthermore, IC3418 is a dwarf galaxy and a proper nucleus could not be figured out, which would disfavour the difference in metallicity. No gradient of metallicity was observed in the tail (Kenney et al. 2014). Hence, higher SFR of fireballs farther down the stream naturally explains the bluer colour of fireballs farther from the galaxy. This was not identified by previous studies possibly because they could not focus on the compact fireballs. We have further correlated the observed radial velocities of fireballs available in the Kenney et al. (2014). Direct measurements are not available for Fireball-1 and Fireball-2, hence, the velocity of the nearest region has been assumed. The radial velocity of the parent galaxy (176 km s–1) has been subtracted from the fireball velocity. Higher the fireball-galaxy velocity difference, lower would be the fireball-ICM velocity difference. Hence, as expected, the Fireball-4 with the highest SFR has the highest fireball-galaxy velocity difference (117 km s–1) or least fireball-ICM velocity difference. We further have collected the stellar population age of the fireballs from Fumagalli et al. (2011). Any obvious trend is not seen here. If fireballs continue to form stars and they indeed get

86

Page 10 of 12

J. Astrophys. Astr.

(2021) 42:86

Figure 9. The SDSS r-band image shows yellow circles around the fireballs with size proportional to FUV SFR and are labelled with velocity difference (age of stellar population). See Table 3 and Fig. 8 for more data.

decoupled after forming, a large-scale age gradient is likely to be not observed that easily. Hence SFR gradient seems to be the only explanation of the UVoptical colour gradient observed. Vortex street model: Various simulations of ram pressure stripping of cluster infalling galaxies have shown turbulent gas in the wake of the parent galaxy (Kapferer et al. 2009; Tonnesen & Bryan 2012; Steinhauser et al. 2016; Muller et al. 2021). The cold neutral atomic/molecular gas that is stripped from the galaxy experiences the million degree hot plasma of the ICM with significantly less density than the stripped gas. The Kelvin–Helmholtz instabilities that would be developing due to the velocity shear in between these two fluids are going to create vortices. Under self-gravity these vortices should be forming dense gas clouds collapsing to form stars. Furthermore, on large-scale, gas streams/clouds closer to the galaxy would have higher relative velocity with the ICM in the immediate vicinity of the parent galaxy than gas located farther down the stream of the tail which ideally should diminish to zero difference and eventually decouple from the galaxy. On large 10 to 100 kpc scales, such structures can be compared with the von Karman vortex street that forms in hydrodynamic flows around an obstacle (in relative terms, ICM is flowing past the galaxy with ISM as the dye injected to visualise the vortex). Such vortices grow with distance from the obstacle and gradually disperses in the ambient medium (Etling 1989; Horva´th et al. 2020). Unlike terrestrial fluid flow experiments,

here the vortices formed would be trying to nucleate or move radially inward under self-gravity and turbulence trying to compete with and disperse it down the stream. Hence with the decrease of relative velocity downstream, gravity can take over turbulence and the vortices can continue to collapse gas clouds and form young stars. This competing process can be seen on a larger 10–100 kpc scale along the tail and can explain the puzzling trend in Jellyfish galaxies of why Ha emission and optically-blue or UV-bright star formation is seen in clumps/fireballs downstream than close to the parent galaxy. Although molecular gas (CO emission line) has been detected, thanks to the amazing sensitivity of ALMA, from yet another stunning Jellyfish galaxy (Jachym et al. 2019), it is yet to image a full velocity field to confirm/inform dynamical model of vortices for fireballs that we hypothesise here. Cold molecular gas observations have detected the galaxy but not from the tail (Ja´chym et al. 2013). Here we propose that fireballs are the vortices where with the passage of time (with distance from parent galaxy), the self-gravity takes over turbulent forces (proportional to fireball-ICM velocity difference) and cold gas-density grows leading to higher SFR (expressed as blue fireballs). As seen in Fig. 8, SFR may increase to a certain distance and then decline. Further, recent magnetic field observations with the Karl G. Jansky Very Large Array (VLA) have revealed highly ordered magnetic fields in the tails of several Jellyfish galaxies and resolution to investigate vortices in fireballs/clumps are yet to be achieved (Muller et al. 2021). IC3418 has very faint

J. Astrophys. Astr.

(2021) 42:86

radio continuum detection and reasonable radio continuum imaging with polarisation information may require the next generation of radio telescopes like the Square Kilometre Array (SKA).

6. Conclusion (1) With AstroSat UVIT far-UV image resolution higher than the previously available GALEX UV images of this nearest Jellyfish galaxy we have resolved the compact star forming clumps (fireballs) to sub-clumps. (2) Incorporating UV and optical images available in HST archive, we resolve these fireballs to several point sources seen in UV. (3) With the HST images, we confirm the isolated nature of the previously discovered farthest star, a blue supergiant star (SDSS J122952.66?112227.8) located in the same tail, and speculate the presence of a dozen such individual stars in the fireballs. (4) We find that the FUV SFR of fireballs increases with the distance from the parent galaxy. (5) We propose a model where the fireballs are the vortices that develop due to Kelvin–Helmholtz instability due to velocity shear between colddense stripped gas and hot rare gas/plasma of the intra-cluster medium. (6) The gas clouds in the far away fireballs are actually old compared to that in the fireballs near the galaxy, as they were stripped from the galaxy earlier. Thus, fireballs farther away showing blue optical/UV colour or possibly younger stellar population is counter-intuitive. This can be easily explained in our vortex model where self-gravity takes over turbulence as relative velocity decreases with time or distance from the parent galaxy along the turbulent wake, resulting in higher SFR in the fireballs at the far end of the tail. (7) Such targets can be unique laboratories to understand basics of star formation as well as various high-energy phenomena due to the presence of massive stars well-separated from each other and arranged in chronological order due to ram pressure stripping.

Acknowledgements AH is thankful to the University Grants Commission (India) for the one-time start-up and monthly salary

Page 11 of 12

86

grants, under the Faculty Recharge Programme. This publication uses the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). This publication uses UVIT data processed by the payload operations centre at IIA. The UVIT is built in collaboration between IIA, IUCAA, TIFR, ISRO and CSA. Based on observations made with the NASA/ESA Hubble Space Telescope, and obtained from the Hubble Legacy Archive, which is a collaboration between the Space Telescope Science Institute (STScI/NASA), the Space Telescope European Coordinating Facility (ST-ECF/ESA) and the Canadian Astronomy Data Centre (CADC/NRC/CSA).

References Abadi M. G., Moore B., Bower R. G. 1999, MNRAS 308, 947 Agrawal P. C. 2006, Adv. Space Res. 38, 2989 Cardelli J. A., Clayton G. C., Mathis J. S. 1989, ApJ 345, 245 Cortese L., Marcillac D., Richard J. et al. 2007, MNRAS 376, 157 Croton Darren J., Springel V., White S. D. M. et al. 2006, MNRAS 365, 1 Croft S., van Breugel W., de Vries W. et al. 2006, ApJ 647, 1040 Elmegreen B. G., Elmegreen D. M. 1986, ApJ 311, 554 Etling D. 1989, Meteorology Atmos. Phys. 41, 157 Evans D. A., Fong W.-F., Hardcastle M. J. et al. 2008, ApJ 675, 1057 Fabian A. C. 2012, ARA&A 50, 455 Federrath C., Klessen R. S. 2012, ApJ 761, 156 Fumagalli M., Gavazzi G., Scaramella R., Franzetti P. 2011, A&A 528, A46 George K., Poggianti B. M., Gullieuszik M. et al. 2018, MNRAS 479, 4126 Gullieuszik M., Poggianti B. M., Moretti A. et al. 2017, ApJ 846, 27 Gunn J. E., Gott J. R. I. I. I. 1972, ApJ 176, 1 Hardcastle M. J., Croston J. H. 2020, NewAR, 8801539 Hester J. A. et al. 2010, ApJL 716, L14 Hopkins P. F., Hernquist L., Cox T. J. et al. 2006, ApJS 163, 1 ´ , Bresky W., Daniels J. et al. 2020, J. Geophys. Horva´th A Res.: Atmospheres, 125(4), article id. e32121 Hota A., Rey S.-C., Kang Y. et al. 2012, MNRAS 422L, 38 Hota A., Konar C., Stalin C. S. et al. 2016, J. Astrophys. Astr. 37, 41 Indu G., Subramaniam A. 2011, A&A 535A, 115

86

Page 12 of 12

Ja´chym P., Kenney J. D. P., Rzˇuicˇka A., Sun M., Combes F., Palousˇ J. 2013, A&A 556A, 99 Ja´chym P., Kenney J. D. P., Sun M. et al. 2019, ApJ 883, 145 Jog Chanda J., Solomon P. M. 1992, ApJ 387, 152 Kapferer W., Sluka C., Schindler S., Ferrari C., Ziegler B. 2009, A&A 499, 87 Karachentsev I. D., Kaisina E. I., 2013, AJ, 146, 46 Kenney J. D. P., Geha M., Ja´chym P. et al. 2014, ApJ 780, 119 Kennicutt R. C., Evans N. J. 2012, ARA&A, 50, 531 Krumholz M. R. 2014, Phys. Rev 539, 49 Martin D. C., Seibert M., Neill J. D. et al. 2007, Nature 448, 780 McKee C. F., Ostriker E. C. 2007, ARA&A 45, 565 Mu¨ller A., Poggianti B., Pfrommer C. et al. 2021, Nat. Astron. 5, 159 Oey M. S., Massey P. 1995, ApJ 452, 210 Ohyama Y., Hota A. 2013, ApJ 767L, 29 Oosterloo T., van Gorkom J. 2005, A&A 437L, 19 Poggianti B. M., Moretti A., Gullieuszik M. et al. 2017, ApJ 844, 48

J. Astrophys. Astr.

(2021) 42:86

Salome´ Q., Salome´ P., Combes F., Hamer S. 2016, A&A 595A, 65 Sanders D. B., Mirabel I. F. 1996, ARA&A 34, 749 Santoro F., Oonk J. B. R., Morganti R., Oosterloo T. 2015, A&A 574A, 89 Smith Russell J., Lucey John R., Hammer Derek et al. 2010, MNRAS, 408, 1417 Springel V., Di Matteo T., Hernquist L. 2005, MNRAS 361, 776 Steinhauser D., Schindler S., Springel V. 2016, A&A 591A, 51 Sun M., Donahue M., Voit G. M. 2007, ApJ 671, 190 Tandon S. N et al. 2020, AJ, 159, 158 Tonnesen S., Bryan G. L. 2012, MNRAS 422, 1609 Vollmer B. et al. 2001, ApJ 561, 708 Wu B., Tan J. C. Nakamura F., Loo S. V., Christie D., Collins D. 2017, ApJ, 835, 137 Yagi M., Gu L., Fujita Y. et al. 2013, ApJ 778, 91 Yoshida M., Yagi M., Komiyama Y. et al. 2008, ApJ 688, 918 Zinnecker H., Yorke H. W. 2007, ARAA 45, 481 Zovaro H. R. M., Sharp R., Nesvadba N. P. H. et al. 2020, MNRAS 499, 4940

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:80 https://doi.org/10.1007/s12036-021-09709-3

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

Short-timescale variability of the blazar Mrk 421 from AstroSat and simultaneous multi-wavelength observations RITABAN CHATTERJEE1,* , SUSMITA DAS1, ARCHISHMAN KHASNOVIS1,

RITESH GHOSH2, NEERAJ KUMARI3,4, SACHINDRA NAIK3, V. M. LARIONOV5,6, T. S. GRISHINA5, E. N. KOPATSKAYA5, E. G. LARIONOVA5, A. A. NIKIFOROVA5,6, D. A. MOROZOV1, S. S. SAVCHENKO5, YU. V. TROITSKAYA5, I. S. TROITSKY5 and A. A. VASILYEV5 1

Department of Physics, Presidency University, 86/1 College Street, Kolkata 700 073, India. Visva-Bharati University, Santiniketan, Bolpur 731 235, India. 3 Astronomy and Astrophysics Division, Physical Research Laboratory, Navrangapura, Ahmedabad 380 009, India. 4 Department of Physics, Indian Institute of Technology, Gandhinagar 382 355, India. 5 Astronomical Institute, St. Petersburg State University, Universitetskij Prospekt 28, Petrodvorets, St. Petersburg, Russia 198504. 6 Pulkovo Observatory, St. Petersburg, Russia. *Corresponding Author. E-mail: [email protected] 2

MS received 9 November 2020; accepted 18 January 2021 Abstract. We study the multi-wavelength variability of the blazar Mrk 421 at minutes to days timescales using simultaneous data at c-rays from Fermi, 0.7–20 keV energies from AstroSat, and optical and near infrared (NIR) wavelengths from ground based observatories. We compute the shortest variability timescales at all of the above wave bands and find its value to be  1.1 ks at the hard X-ray energies and increasingly longer at soft X-rays, optical and NIR wavelengths as well as at the GeV energies. We estimate the value of the magnetic field to be 0.5 Gauss and the maximum Lorentz factor of the emitting electrons  1:6  105 assuming that synchrotron radiation cooling drives the shortest variability timescale. Blazars vary at a large range of timescales often from minutes to years. These results, as obtained here from the very short end of the range of variability timescales of blazars, are a confirmation of the leptonic scenario and in particular the synchrotron origin of the X-ray emission from Mrk 421 by relativistic electrons of Lorentz factor as high as 105 . This particular mode of confirmation has been possible using minutes to days timescale variability data obtained from AstroSat and simultaneous multi-wavelength observations. Keywords. Active Galactic Nuclei (AGN)—blazar—variability—Mrk 421—multiwavelength.

1. Introduction Relativistic jets pointed toward the line of sight of the observer is a defining property of blazars (Urry & Padovani 1995), a class of radio loud active galactic nuclei (AGN). Due to relativistic beaming the apparent luminosity of the jet is amplified by a factor of 10– This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

104 in the observer’s frame. Consequently, emission from the other parts of the AGN such as the accretion disk, broad line region, and dusty torus are overwhelmed by that from the jet. Blazars are often bright over a large range of wavelengths from radio to GeV or even TeV c-rays. The spectral energy distribution (SED) of blazars are characteized by two broad peaks: one at the IR-X-ray wavelength range due to synchrotron radiation by relativistic electrons present in the jet (Bregman 1981; Urry & Mushotzky 1982;

80

Page 2 of 7

Impey & Neugenbauer 1988; Marscher 1998) and the other in the GeV range sometimes extending to TeV energies. In the so called ‘‘leptonic scenario,’’ the high energy peak may be due to inverse-Compton (IC) scattering of ‘‘seed’’ photons from the broad line region, dusty torus or the jet itself (Maraschi et al. 1992; Chiang & Bo¨ttcher 2002; Arbeiter et al. 2005; Sikora et al. 1994; Coppi & Aharonian 1999; Bła_zejowski 2000; Dermer et al. 2009; Bo¨ttcher & Dermer 2010; Romero 2012) by the same relativistic electrons that are generating the synchrotron radiation for the lower energy peak. Alternatively, in the so called ‘‘hadronic model,’’ X-rays and c-rays may be produced due to synchrotron radiation by relativistic protons, which are accelerated along with the electrons in the jet, proton-induced particle cascades, or interactions of these high-energy protons with external radiation fields (Mu¨cke & Protheroe 2001; Mu¨cke et al. 2003; Bo¨ttcher 2013). Based on the location of the synchrotron peak (msynch ) in the broad band SED, blazars are divided into three categories: low, intermediate and high synchrotron peaked: LSP (msynch \1014 Hz), ISP (1014 \msynch \1015 Hz), and HSP (msynch [ 1015 Hz), respectively (e.g., Abdo 2010b; Bo¨ttcher 2019). Blazars are characterized by fast and high amplitude variability in emission, which is due to the injection of energy and fluctuations in the density, magnetic field and other physical properties of the jet originally caused by instabilities in the accretion disk (e.g., Malzac 2014). Due to the beaming effect the timescale of variability becomes faster in the obsever’s frame by a factor  10. In the leptonic model, the emission at wavelengths near the two peaks in the SED are caused by the highest energy part of the electron spectrum. Therefore, variability at those wavelengths are expected to be the most pronounced, e.g., by an order of magnitude or more over months to years timescales (e.g., Majumder et al. 2019). Blazar variability is red noise in nature, i.e., amplitude of variability is smaller at shorter timescales (e.g., Chatterjee 2012). But near the SED peak wavelengths, emission may vary by a factor of a few even at subday timescales. Blazar variability at X-ray and other wave bands at longer timescales have been studied extensively (e.g., Chatterjee 2008, 2012; Abdo 2010a; Rajput et al. 2020) but that at sub-day timescales has been less common. It is mainly because even the brighter blazars are not detected with high signal to noise at short exposures. Long and quasi-continuous observations

J. Astrophys. Astr.

(2021) 42:80

with X-ray telescopes are needed to obtain comprehensive information about their short-term fluctuations. The blazar Mrk 421, one of the brightest blazars in the X-ray sky, is an ideal source for such study. It is an HSP blazar with the low energy peak of the SED at the X-ray band (e.g., Abdo 2011; Banerjee 2019). Hence, the X-ray variability is expected to be very pronounced. However, averaged over the last 10 years its brightness is 10 cts/s in Swift-XRT.1 Hence, to probe its short-timescale properties at minutes to days timescale long stares are necessary which are not possible with Swift as it has other priorities. In this paper, we study the minutes to days timescale X-ray variability of Mrk 421 using a quasicontinuous observation by AstroSat (Agrawal 2006; Singh et al. 2014). We compare the same with simultaneous observations at GeV, optical, and near infrared (NIR) wave bands. We analyze the data to test if the multi-wavelength variability properties are consistent with the expectation of the standard model of blazar emission as described above. In particular, we check if the amplitude and timescale of variability are consistent with the leptonic model in which the NIR, optical, and X-ray emission in an HSP blazar such as Mrk 421 are generated by increasingly higher energy part of the electron distribution and the GeV emission is produced by lower energy electrons via IC processes. In Secttions 2, 3 and 4, we describe the data reduction, present the results of the variability analysis, and discuss the implications of the results, respectively.

2. Data 2.1 X-ray and c-ray data Mrk 421 was observed with AstroSat (Obs Id: A05_204T01_9000002856, PI: Ritaban Chatterjee) on April 23–28, 2019. We received the data for individual orbits of the satellite (Level-1 data) from the Indian Space Science Data Center (ISSDC). We first processed the Level-1 Soft X-ray Telescope (SXT; Singh et al. 2016, 2017) data with the sxtpipeline task, which is a part of the SXT software (AS1SXTLevel2, version 1.4b) available at the SXT POC Website.2 Level-2 cleaned event files for the individual orbits were extracted using pipeline calibration of the source events. The extraction process 1

https://www.swift.psu.edu/monitoring/. http://www.tifr.res.in/*astrosat_sxt/sxtpipeline.html.

2

J. Astrophys. Astr.

(2021) 42:80

includes filtering out any contamination by the charged particles due to excursions of the satellite through the South Atlantic Anomaly region and occultation by the Earth. To select only X-rays and avoid counting charged particles, we selected only the events with grade 0–12, i.e., single-quadruple events. Next, we merged the cleaned event files of all the orbits into a single event file using a Julia based software tool developed by G. C. Dewangan in order to avoid any time-overlapping events in consecutive orbits. A circle of radius 15 arcmin was used to extract the source light curve in the energy interval 0.7–8 keV. The Large Area X-ray Proportional Counter (LAXPC) instrument onboard AstroSat has three coaligned proportional counters, LAXPC10, LAXPC20 and LAXPC30 (Yadav et al. 2017; Antia 2017). We did not use the LAXPC30 data due to a gain instability issue caused by gas leakage. LAXPC data were collected in the event analysis (EA) mode. During the observation, the LAXPC10 was operating at a lower gain. Hence, we have used data only from LAXPC20 for our analysis. The light curves and spectra were generated using the LaxpcSoft software package 5. The background was estimated from blank sky observations, where there are no known X-ray sources. We extracted light curves using data from the top layer for the energy interval 4–20 keV. At the GeV energies, we use data from the Large Area Telescope on board Fermi Gamma-Ray Space Telescope (Abdollahi 2020). Fermi has a large fieldof-view (  2.4 Sr) and monitors the whole sky every three hours. We use the light curve at 0.1–300 GeV obtained from the Fermi Science Support Center (FSSC).3

2.2 Optical and near infrared data The optical flux and polarimetric data were obtained using the St. Petersburg University 40-cm LX-200 telescope and the Crimean observatory 70-cm AZT-8 telescope. For the details of data analyses and processing, see Larionov (2008). Near-infrared observations of Mrk 421 in H-band, simultaneous with the AstroSat observations of the source, were carried out with PRL’s 1.2-m telescope at Mount Abu Infrared Observatory (MIRO), India. The photometric observations were performed on the nights of 25, 26 and 27 April 2019 in imaging mode 3

https://fermi.gsfc.nasa.gov/ssc/data/access/lat/msl_lc/.

Page 3 of 7

80

using Near-Infrared Camera and Spectrograph (NICS) instrument. At Mt. Abu, the NICS provides  1 arcsec seeing over 8  8 arcmin2 with 1024  1024 pixels array configuration. A set of five frames of  50 s were taken in H-band (1.635–1.780 lm, central wavelength at 1.49 lm), dithered at five different positions. The data were reduced using IRAF software package as described in Naik et al. (2010). In this procedure, the sky-frame was generated for a particular filter by median combining all the raw and dithered images. After subtracting skyframes from the raw images, clean images were obtained and then corrected for effects like cosmic rays. Aperature photometry was performed using the PHOT task in IRAF. We selected SAO 015832 and SAO 062433 as standard stars.

3. Results The multi-wavelength light curves of Mrk 421 during 2019 April 16–May 06 are shown in Fig. 1. It is evident that the variability in that time interval is most pronounced in the X-ray energies. We compute a variability timescale during both increase and decrease of flux using the following formula (Saito et al. 2013; Roy et al. 2019): sd ¼ Dt

ln 2 ; lnðf2 =f1 Þ

ð1Þ

where f1 and f2 are the flux values at time t1 and t2 , and Dt is the difference between t1 and t2 . We compute sd for all pairs of data points in the light curves and search for the shortest value in each. The shortest timescale of variability that we find at each band along with the corresponding value of Dt are given in Table 1. Value of sd during increase of flux for Hband could not be obtained because the length and sampling of the light curve are not enough for that calculation. In addition, Table 1 exhibits the normalized excess variance (Vaughan 2003) at all the above wave bands. In the GeV band, the variance is dominated by the uncertainties in the data. Therefore, the excess variance is negligible and is not shown on the table. Instead of sd , the exponential growth or decay timescale given by sd =ln 2 may also be used as a characteristic timescale of variability. That will change the result by a constant factor and in most of the cases the value will be consistent with sd within the uncertainties.

80

Page 4 of 7

J. Astrophys. Astr.

(2021) 42:80

is consistent with the assumption that the X-rays are produced by the highest energy tail of the electron distribution via synchrotron process. A consequence of that assumption is that the variability at the GeV band, being produced by lower energy electrons via IC processes, will be less pronounced, which is also evident. We have found the minimum variability timescale to be 1.1 ks. Assuming the variability is dominated by radiation cooling due to synchrotron only, tcool ’ 7:7  108 ð1 þ zÞd1 B2 c1 eff s;

ð2Þ

where z, d, B and ceff are the redshift of the source, Doppler factor of the jet, magnetic field in the emission region, and effective Lorentz factor of the emitting electrons, respectively (e.g., Zhang 2019; Rybicki & Lightman 1979). The characteristic frequency of the electron distribution responsible for the emission at the synchrotron peak of the SED is given by Rybicki and Lightman (1979) d 2 ð3Þ c B Hz: mch ¼ 4:2  106 1 þ z eff

Figure 1. Variation of the 0.1–300 GeV c-ray flux from Fermi-LAT in daily bins, hard (4–20 keV) and soft (0.7–8 keV) X-ray flux from AstroSat LAXPC and SXT, respectively, H-band flux density from Mount Abu Infrared Observatory NICS instrument, and R-band flux density and optical polarization properties from the St. Petersburg University 40-cm LX-200 telescope and the Crimean Observatory 70-cm AZT-8 telescope of Mrk 421 from April 16–May 06, 2019. The SXT data have been scaled and shifted for clarity. The unscaled light curve is shown in the next figure.

The value of sd is the shortest at the hard X-ray band, slightly higher at the soft X-ray band and much longer at the optical, NIR as well as GeV bands. This

Using z = 0.031, d ¼ 20, and mch ¼ 1018 Hz in the equations (2) and (3), we find B ¼ 0:5 Gauss and ceff ¼ 1:6  105 . These values are consistent within a factor of a few with those obtained from modeling the broad-band SED of Mrk 421 in the standard leptonic scenario (Abdo 2011; Banerjee 2019). The value of d and mch tend to vary from one epoch to another, which may cause appropriate changes in the above estimates of B and ceff by a factor of a few. Alternatively, instead of assuming the value of d and mch , we eliminate ceff from equations (2) and (3) and find B3 d ’ 2:5ð1 þ zÞðmch =1018 Þ1 s2 d ;

ð4Þ

where mch is in Hz and sd is in ks. Hence, using the value of any one of the variables d, B, and mch the other two may be constrained.

Table 1. Variability timescale and excess variance at different wave bands. Wave band GeV Hard X-ray Soft X-ray R-Band H-Band

sd (increase) (ks) 37.2 1.5 2.3 195.3

± ± ± ± –

23.3 0.2 1.4 89.0

Dt (ks) 86.4 2.0 0.6 6.2 –

sd (decrease) (ks) 279.1 1.1 1.5 99.4 366.3

± ± ± ± ±

256.6 0.2 1.1 7.3 34.5

Dt (ks)

Normalized excess variance

432.0 1.0 0.6 7.3 91.6

– 0.29 0.12 0.07 0.07

J. Astrophys. Astr.

(2021) 42:80

Page 5 of 7

From equations (2) and (3), it can be estimated that the ratio of the effective Lorentz factors of electrons contributing to the emission at frequencies mi and mj is proportional to ðmi =mj Þ0:5 . Therefore, the ratio of synchrotron cooling timescales for emission at frequencies mi and mj are proportional to ðmi =mj Þ0:5 , which implies that the ratios of cooling timescales at soft Xrays, optical, and NIR to that in hard X-rays are  3,  70, and  110, respectively. The value of sd from Table 1 are consistent with the above within the uncertainties. Let us suppose that the GeV emission in Mrk 421 is produced by the inverse-Compton scattering of the synchrotron photons generated in the jet itself by the same relativistic electrons (the so-called ‘‘synchrotron self-Compton or SSC process’’). We assume that the up-scattered photons have a mean energy of 1 GeV and most of the seed photons are from the synchrotron peak energy for Mrk 421, which is  1 keV. Then using the simplified formula mf ¼ c2IC mi ;

Figure 2. The cyan open circles with error bars denote the SXT light curve of Mrk 421 during April 23–28, 2019. The red solid line corresponds to the R-band light curve during the same interval. The black dashed line indicates the Rband variability artificially generated based on the SXT light curve and assuming synchrotron origin of the emission at both wave bands. The properties of the actual and artificial optical light curves are consistent with each other.

ð5Þ

where mi and mf are the frequency of photons before and after the IC up-scattering and cIC is the effective Lorentz factor of the energetic electrons, we can estimate that the electrons responsible for the GeV emission have Lorentz factor  103 . Hence, GeV emission variability timescale will be similar to that at the optical-NIR wavelengths, which is consistent with our findings. Alternatively, the energy of electrons responsible for the GeV emission may be obtained using the inverse-Compton cooling timescale given by (e.g., Chiaberge 1999) IC 1 1 ’ 3:0  107 Urad cIC s; tcool

80

ð6Þ

where Urad is the radiation energy density. Using IC ¼ 100 ks and Urad ¼ 102 erg cm3 , we obtain tcool cIC ¼ 3  104 . Here, we estimate Urad to be approximately equal to the magnetic energy density. The obtained value of cIC is uncertain due to the uncertainties in Urad and tcool . The above estimates indicate that the optical variability should be similar but slower than the X-ray variability by a factor of  70 if both are produced by synchrotron radiation by different parts of the electron energy spectrum. To test this we scale the X-ray light curve appropriately to generate an artificial optical light curve, which is correlated with the X-ray variability but is slower by the above factor. We include some additional random fluctuations generated from

the uncertainty of the SXT data. Figure 2 shows the artificial and the actual R-band light curves along with the SXT light curve and it is evident that the nature of the the real and artificial optical light curve segments are consistent with each other.

4. Discussion It can be seen from Table 1 that in the above observations, the variability timescale is shortest at the hard X-ray band and increasingly longer at soft X-ray, optical and NIR bands as well as in the GeV band. This is consistent with the model in which the X-ray to NIR (and longer wavelength) emission in Mrk 421 is generated by synchrotron radiation by the relativistic electrons in the jet. We find that the GeV emission may be produced via SSC process by electrons having energy lower than those producing the X-rays. Hence, the slower variability timescale of GeV emission is also justified. However, the energy density of the seed photons which are up-scattered to produce the GeV emission is not well-constrained in this case. Therefore, the estimate of the electron energy contains the corresponding uncertainty. The obtained value of the magnetic field is consistent with those obtained from studies involving fitting the SED with physical models (e.g., Abdo 2011). The amplitude of variability, as indicated by the normalized excess variance, is also larger at the X-ray

80

Page 6 of 7

than the other wave bands. If the variability is caused by injection of energy or fluctuation in the magnetic field the amplitude of variability may not be different at the above wave bands. However, if the variability is faster in the X-ray band, it is obvious that in a given interval the observed amplitude will be larger. We note that while the amplitude of variability we find here is similar to that found by Zhang (2019) for Mrk 421 from Suzaku data the shortest variability timescale obtained here is an order of magnitude smaller than that found by Zhang (2019). This may be due to the difference in sampling among the X-ray light curves or due to a different brightness state of Mrk 421. It can be seen from Table 1 that sd during increase and decrease at all wavebands are not systematically different. In blazar variability it is usually assumed that the timescale of acceleration of electrons to the highest energies (e.g., ceff ¼ 105 in this case) is faster than the cooling timescale such that the acceleration may be considered instantaneous. However, that will indicate sd during increase will be systematically shorter than that during decrease, which is not the case here. This implies that the acceleration timescale may not be negligible compared to the cooling timescales and may be constrained using short-timescale variability as analyzed here. We note that the variability of the polarization fraction and electric vector polarization angle (EVPA) is more pronounced than that of the optical flux. This may be due the effect of turbulence (Marscher 2014; Liodakis 2020). More data are needed to draw detailed inference about the polarization variability. Blazars vary at a large range of timescales often including minutes to years. The above quantitative results, obtained from the shortest end of the large range of timescales at which blazars vary, are consistent with the standard leptonic scenario, in which the lower energy peak of the spectral energy distribution of Mrk 421 is due to synchrotron radiation by the relativistic electrons in the jet and is located at the X-ray frequencies. Similar observation of a sample of blazars by AstroSat along with a more dedicated simultaneous multi-wavelength coverage will provide more specific constraints on the physical properties of the blazar class.

Acknowledgements We thank the anonymous referee for comments and suggestions that made the manuscript more

J. Astrophys. Astr.

(2021) 42:80

comprehensive. RC thanks Presidency University for support under the Faculty Research and Professional Development (FRPDF) Grant, ISRO for support under the AstroSat archival data utilization program, and IUCAA for their hospitality and usage of their facilities during his stay at different times as part of the university associateship program. RC received support from the UGC start-up grant. RC acknowledges financial support from BRNS through a project grant (sanction no: 57/14/10/2019-BRNS) and thanks the project coordinator Pratik Majumdar for support regarding the BRNS project. This work has made use of data from the AstroSat mission of the ISRO, archived at the Indian Space Science Data Centre (ISSDC). This work has been performed utilizing the calibration databases and auxiliary analysis tools developed, maintained and distributed by AstroSatSXT and AstroSat-LAXPC teams with members from various institutions in India and abroad and the SXT and LAXPC Payload Operation Center (POC) at the TIFR, Mumbai for the pipeline reduction. We are also thankful to the AstroSat Science Support Cell hosted by IUCAA and TIFR for providing the necessary data analysis software. The work has made use of software, and/or web tools obtained from NASA’s High Energy Astrophysics Science Archive Research Center (HEASARC), a service of the Goddard Space Flight Center and the Smithsonian Astrophysical Observatory.

References Abdo A. et al. 2010a, ApJ, 722, 520 Abdo A. et al. 2010b, ApJ, 716, 30 Abdo A. et al. 2011, ApJ, 736, 131 Abdollahi S. et al. 2020, ApJS, 247, 1 Agrawal P. C. 2006, Adv. Space Res., 38, 2989 Antia H. M. et al., 2017, ApJS, 231, 10 Arbeiter C., Pohl M., Schlickeiser R. 2005, ApJ, 627, 62 Banerjee B. et al. 2019, MNRAS, 487, 845 Bła_zejowski et al. 2000, ApJ, 545, 107 Bo¨ttcher M., Dermer C. D. 2010, ApJ, 711, 445 Bo¨ttcher M. et al. 2013, ApJ, 768, 54 Bo¨ttcher M. 2019, Galaxies, 7, 20 Bregman J. N. et al. 1981, Nature, 293, 714 Chatterjee R. et al. 2008, ApJ, 689, 79 Chatterjee R. et al. 2012, ApJ, 756, 13 Chiaberge M., Ghisellini G. 1999, 306, 551 Chiang J., Bo¨ttcher M. 2002, ApJ, 564, 92 Coppi P. S., Aharonian F. A. 1999, ApJ, 521, L33 Dermer C. D., Finke J. D., Krug H., Bo¨ttcher M. 2009, ApJ, 692, 32 Impey C., Neugenbauer G. 1988, Astron. J., 95, 307

J. Astrophys. Astr.

(2021) 42:80

Liodakis I. et al. 2020, ApJ, 902, 61 Larionov V. M. et al. 2008, A&A, 492, 389 Majumder A., Mitra K., Chatterjee R., Urry C. M., Bailyn C. D., Nandi P. 2019, MNRAS, 490, 124 Malzac J. 2014, MNRAS, 443, 299 Maraschi L., Ghisellini G., Celotti A. 1992, ApJ, 397, L5 Marscher A. P. 1998, in Zensus J. A., Wrobel J. M., Taylor G. B., eds, Radio Emission from Galactic and Extragalactic Compact Sources, Astronomical Soc. Pacific Conf. Ser., 144, 25 Marscher A. P. 2014, ApJ, 780, 87 Mu¨cke A., Protheroe R. J. 2001, Astroparticle Phys., 15, 121 Mu¨cke A., Protheroe R. J., Engel R., Rachen J. P., Stanev T. 2003, Astroparticle Phys., 18, 596 Naik S., Banerjee D. P. K., Ashok N. M., Das R. K. 2010, MNRAS, 404, 367 Rajput B., Stalin C. S., Rakshit S. 2020, A&A, 634, 80 Romero G. E. et al. 2012, SSRv, 207, 5 Roy N., Chatterjee R., Joshi M., Ghosh A. 2019, MNRAS, 482, 743

Page 7 of 7

80

Rybicki G. B., Lightman A. P. 1979, New York, WileyInterscience, 393 p. Saito S., Stawarz Ł., Tanaka Y. T., Takahashi T., Madejski G., D’Ammando F. 2013, ApJ, 766, 11 Sikora M., Begelman M. C., Rees M. J. 1994, ApJ, 421, 153 Singh K. P., Tandon S. N., Agrawal P. C. et al. 2014, Proc. SPIE, Space Telescopes and Instrumentation 2014: Ultraviolet to Gamma Ray, 9144, 91441 Singh K. P., Stewart G. C., Chandra S. et al. 2016, Proc. SPIE, in Space Telescopes and Instrumentation 2016: Ultraviolet to Gamma Ray, 9905, 99051 Singh K. P., Stewart G. C., Westergaard N. J. et al. 2017, J. Astrophys. Astr., 38, 29 Urry C. M., Mushotzky R. F. 1982, ApJ, 253, 38 Urry C. M., Padovani P. 1995, PASP, 107, 803 Vaughan S. et al. 2003, MNRAS, 345, 1271 Yadav J. S., Agrawal P. C., Antia H. M., Manchanda R. K., Paul B., Misra R. 2017, arXiv e-prints arXiv:1705.06440 Zhang Z. et al. 2019, ApJ, 884, 25

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:88 https://doi.org/10.1007/s12036-021-09705-7

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

SCIENCE RESULTS

Potential UV standards for UVIT filter system S. G. BHARGAVI RRI Quarters, II Main Road, Vyalikaval, Bangalore 560 003, India. E-mail: [email protected] MS received 30 November 2020; accepted 27 December 2020 Abstract. We propose a set of eleven faint UV white dwarf stars as potential UV standards for the UVIT filter system on-board AstroSat mission. Synthetic photometry of these stars has been carried out using the in-orbit filter curves of UVIT, GALEX, UVOT and model spectra in the literature. The synthetic photometry of GALEX and UVOT is well correlated with their respective observations. The synthetic UVIT photometry is found to be strongly correlated with that of GALEX. The UV extinction Ak for these stars has been computed using CCM-89 extinction model and its value for the kmean of each UVIT filter band has been listed. These results motivate us to recommend this set of stars as additional UVIT flux calibrators. Keywords. UVIT—synthetic photometry—UV colors—potential standards.

1. Introduction AstroSat, the first Indian astronomy satellite designed for simultaneous multi-wavelength astronomy was launched in september 2015 (Tandon 2017a). The Ultra-Voilet Imaging Telescope (UVIT), one of the five instruments on board AstroSat is equipped with multiple broad and narrow band filters in far-UV, near-UV and visual regions of the electro-magnetic spectrum to carry out imaging observations. In-orbit calibrations of the UVIT have been carried out by Tandon (2017b) and Rahna (2017). The science goals of UVIT include studies of a variety of astrophysical objects both stellar and non-stellar. The successful operation and performance of the mission and in particular UVIT has resulted in a large number of scientific journal papers (e.g. Subramaniam 2016, 2017; Rao 2018a, b, 2020), and several other investigators). Additional in-flight calibrations have also been presented recently (Tandon 2020). While typical astronomical observations span a range in spectral and luminosity classes, the available flux standards in UV are limited in number, sky This article is part of the Special Issue on ‘‘AstroSat: Five Years in Orbit’’.

coverage, brightness range and spectral type. The contemporary UV missions HST, GALEX, Swift/ UVOT use 1 to 4 primary standards as flux calibrators and these are brighter than V  14. One of the aims of any space based observatory should be to expand the list of flux calibrators. This can be achieved by identifying the suitable candidate stars, carrying out their repeated observations and calibration until arriving at reliable photometry to establish their candidature. This process should continue until the space-craft remains in the orbit. Bhargavi and Pati (2009) had commenced a compilation of an all-sky catalog of stellar sources which would be suitable as standards for UVIT. As a first step, the magnitude scales and UV colours for UVIT filters were examined for standard stars with known Spectral Energy Distributions (SED) and later being extended to suitable stars (particularly those with UV observations) from the existing catalogs in order to span a range of spectral type and colour. As an extension and expansion to all-sky secondary standards for UVIT program here, I present a detailed analysis for a set of faint UV stars which could be potential UVIT standards. In the following section, we describe the methodology followed in computing the synthetic photometry, details of the filter sets and stellar libraries used. In Section 3, we have examined

88

Page 2 of 7

the UV magnitudes and colours of a catalog of faint UV White dwarfs that could be potential standards for UVIT and also computed UV extinctions for these set of stars. Discussion and conclusions are presented in Section 4.

J. Astrophys. Astr. (2021)42:88

F

eff

R

kSk Fk dk ¼ R : kSk dk

ð1Þ

˚ Effective fluxes Here, Fk is in units of erg=s=cm2 =A. thus obtained were then converted into magnitudes in Vega system as follows:

2. Synthetic photometry

mk ¼ ðZPÞVega  2:5  log10 ðF eff Þ:

In Bhargavi and Pati (2009), calculations of instrumental magnitudes for a set of HST’s primary standard stars were carried out through pre-flight filter curves of UVIT, assuming unit collecting area for all telescopes, unit efficiency in the reflectivity of mirrors and unit detector response in all filter bands since actual detector and other telescope characteristics were not available at that time. The instrumental UV magnitudes and colours computed for the filter systems of UVIT, Tauvex, GALEX as well as optical were inter-compared. The Color–Magnitude Diagrams (CMD) and Color–Color (C–C) plots showed that the proposed UVIT filters provide a large spread of colours and thus will be able to offer better discrimination with temperature. In the subsequent run we took into account the actual measured parameters such as mirror reflectivity, detector response etc., and studied the UV colour distribution for a larger sample of stars spanning a range in luminosity and spectral classes across the H-R diagram.

The Zero Point (ZP) of the magnitude scale in Vega system, ðZPÞVega ¼ 21:175. It is the magnitude corresponding to the monochromatic flux F0;k ¼ 3:39  ˚ of Vega star outside the Earth’s 109 erg=s=cm2=A atmosphere at all wavelengths. The AB magnitudes can be obtained using Equation (3) below:

2.1 Methodology The methodology to compute the synthetic photometry of stellar sources with known absolute Spectral Energy Distribution (SED) through the bandpasses of UVIT is detailed below. In the discussion following, the words ‘filter or bandpass’ refers to the over all sensitivity of the photometric system, which includes the effects of filters, detector as well as telescope optics and mirror, expressed in units of Effective Area (EA) (cm2 ). The details of filter sets and SEDs used in this work are detailed in the next section. To compute the photometric magnitudes, we obtained effective flux F eff by integrating the stellar flux (Fk ) of each star through the filter sensitivity (Sk ) normalized to unit filter area as given in Equation (1) below. In order to achieve uniform wavelength grids for both SED and bandpass we interpolated the functions using a cubic spline function. Typically, grid ˚ was used to achieve a good fit. width of dk  35 A The ‘SPLINE’ and ‘SPLINT’ tasks of Press (1992) were used for interpolation and integration.

mAB ¼ ðZPÞAB  2:5  log10 ðFm Þ:

ð2Þ

ð3Þ

Here, Fm is computed using the relation: Fm ¼ F eff 

ðkeff Þ2 : c

ð4Þ

Here, c is velocity of light and keff is effective wavelength given by R kSk Fk dk : ð5Þ keff ¼ R Sk Fk dk The zero point of the magnitude scale in AB system is ðZPÞAB ¼ 48:6 and it corresponds to the flux F0;m ¼ 3:5  1020 erg=s=cm2 =Hz in all bands. While the traditional Vega system is preferred, it was decided to compute AB as well, due to its popularity among modern surveys.

2.2 Filter sets We have used the following filter sets to compute the synthetic photometry. Table 1 contains filter abbreviations adopted by Tandon (2017a) followed by the filter names (in brackets) in column 2. The mean wavelengths kmean in column 3 are computed using Equation (6). We have used the latest in-orbit filter curves available at the UVIT page1. The filter curves Sk (i.e., plots of wavelength vs. EA) of UVIT are not shown here instead, the reader is referred to UVIT page or Tandon (2017b). Effective areas of GALEX bandpasses are obtained from GALEX home page2 and those of 1

http://uvit.iiap.res.in/Instruments/Filters. See http://www.GALEX.caltech.edu.

2

J. Astrophys. Astr. (2021)42:88

Page 3 of 7

Table 1. Filter sets.

UVIT-FUV

UVIT-NUV

GALEX UVOT

88

Table 2. UV faint stars.

Filter Abbr. (Name)

˚) kmean (A

Sl. No.

F148W (Caf2-1) F148Wa (Caf2-2) F154W (Baf2) F169M (Sapphire) F172M (Silica) N219M (NUV-B15) N245M (NUV-B13) N263M (NUV-B4) N279N (NUV-N2) N242W (NUV-Silica) G-FUV G-NUV NUV-W2 NUV-M2 NUV-W1

1520.12 1524.56 1555.81 1602.36 1718.03 2211.20 2448.38 2625.39 2792.37 2421.47 1538.62 2315.66 2139.53 2271.85 2677.44

1 2 3 4 5 6 7 8 9 10 11

UVOT from Swift home page3. The effective wavelengths computed using Equation (5) for UVIT filters are presented in Table A1 in Appendix.

2.3 Spectral Energy Distributions (SEDs) of stars In the following, we list the various sources of SEDs used in this work. The SEDs we obtained from published atlasses and catalogs are calibrated, absolute fluxes. Where ever necessary FITS files were converted into ascii using the tool MRDFITS of IDL package or Fortran programs we wrote.

Star Id J002806.49?010112.2 J083421.23?533615.6 J092404.84?593128.8 J103906.00?654555.5 J134430.11?032423.2 J140641.95?031940.5 J144108.43?011020.0 J150050.71?040430.0 J173020.12?613937.5 J231731.36-001604.9 J235825.80-103413.4

Av (mag) 0.067 0.122 0.079 0.058 0.082 0.108 0.127 0.014 0.126 0.126 0.101

(ii) Faint UV white dwarfs: Model spectra presented by Siegel (2010) of eleven faint UV white dwarf stars previously identified as standards for GALEX and Swift/UVOT. 3. Examining the suitability of faint UV stars as potential standards for UVIT filter system About a dozen faint (U  17) DA White dwarf stars observed by GALEX, Swift/UVOT and SDSS have been identified as faint UV secondary standards (see Table 2). The basic parameters of these stars along with Swift/UVOT and GALEX photometry are available in Tables 1 and 2 in Siegel (2010). The

(i) SEDs of HST fundamental standard stars: The reference spectra used in the calibration of the HST instruments are available to the public for the absolute calibration of ground-based or satellite data. This database known as CALSPEC4 contains composite ultraviolet and optical absolute calibrated stellar spectra of the HST standard stars. We obtained SEDs of 29 standards to calibrate the UVIT filter system. The spectra span a wavelength range of 1050– ˚ compiled from various sources Calspec 10000 A stores the spectra in multiple extension .FITs table format. Tasks to convert to ASCII table and also to plot the spectra are available in IRAF v2.11.2 onwards as well as in IDL graphic package5. 3

See https://swift.gsfc.nasa.gov/. 4 See www.stsci.edu and MAST 5 idlastro.gsfc.nasa.gov

Figure 1. Top panel: Observed vs. synthetic AB magnitudes of UV faint standards. Points marked (*) are for GALEX FUV filters and (?) for GALEX NUV filters. Bottom panel: Residuals of the fit.

88

Page 4 of 7

J. Astrophys. Astr. (2021)42:88

Table 3. GALEX magnitudes of UV faint stars. Star# 1 1 2 3 4 5 6 7 8 9 10 11

FUV

NUV

FUV

NUV

FUV

NUV

2 13.938 13.008 14.256 14.344 13.025 14.623 13.109 14.335 14.592 13.052 13.930

3 14.480 13.585 14.717 14.857 13.564 15.092 13.694 14.820 15.046 13.566 14.398

4 14.116 13.332 14.466 14.498 13.243 14.910 13.447 14.372 14.926 13.387 14.198

5 14.658 13.908 14.926 15.011 13.782 15.378 14.031 14.857 15.379 13.899 14.665

6 16.446 15.429 16.716 16.859 15.434 16.994 15.487 16.779 17.069 15.499 16.416

7 16.801 15.893 17.041 17.206 15.880 17.433 16.015 17.162 17.408 15.893 16.722

Column 1: Star IDs as in Table 2; Column 2, 3 synthetic mags; Columns 4, 5: synthetic, reddened mags; Columns 6, 7: GALEX observations (Siegel 2010).

model fluxes are in Table 8 of online version. Here we use their photometric data and model spectra to examine their suitability to UVIT as secondary standards. The model spectra are convolved through GALEX, UVOT and in-flight filter curves of UVIT to obtain synthetic magnitudes. In the top panel of Fig. 1, the observed GALEX magnitudes are plotted against those we synthesized for GALEX filters. The FUV and NUV points are marked by ‘*’ and ‘?’ respectively. A simple linear fit of the form Y ¼ mX þ c gives slope m ¼ 0:972  0:02 and c ¼ 1:96  0:35 for FUV and for NUV the slope is m ¼ 0:98  0:005 and c ¼ 1:99  0:08. The residuals of the fit are shown in the bottom panel of Fig. 1 and are within 1% error (x ¼ 0:58%; 0:91% for FUV and NUV respectively). We repeated these computations for swift/UVOT filters and also obtained less than 1% errors of residuals.

Table 4. Synthetic UVIT (AB) magnitudes. Star#

F148W

F148Wa

F154W

F169M

F172M

N219M

N245M

N263M

N279N

N242W

1 2 3 4 5 6 7 8 9 10 11

13.927 12.997 14.250 14.333 13.014 14.617 13.098 14.323 14.587 13.044 13.924

13.930 13.000 14.253 14.336 13.017 14.619 13.102 14.327 14.590 13.047 13.927

13.952 13.022 14.269 14.358 13.039 14.635 13.123 14.349 14.604 13.065 13.943

13.986 13.055 14.296 14.391 13.073 14.661 13.157 14.382 14.627 13.094 13.969

14.075 13.146 14.372 14.478 13.161 14.736 13.247 14.467 14.698 13.176 14.044

14.457 13.575 14.699 14.826 13.541 15.086 13.684 14.773 15.048 13.564 14.389

14.544 13.651 14.766 14.919 13.627 15.138 13.762 14.884 15.086 13.617 14.445

14.606 13.707 14.814 14.984 13.689 15.177 13.819 14.961 15.116 13.657 14.486

14.671 13.771 14.866 15.050 13.753 15.224 13.886 15.032 15.156 13.707 14.534

14.533 13.641 14.760 14.908 13.617 15.135 13.752 14.871 15.085 13.612 14.440

See Table 2 for star IDs.

Table 5. Computed UV extinction. Filter

˚ kmean A

Star 1

Star 2

Star 3

Star 4

Star 5

Star 6

Star 7

Star 8

Star 9

Star 10

Star 11

F148W G-FUV F154W F169M F172M N219M G-NUV N242W N245M N263M N279N

1520.12 1538.62 1555.81 1602.36 1718.03 2211.20 2315.50 2421.47 2448.38 2625.39 2792.37

0.176 0.175 0.174 0.171 0.168 0.209 0.188 0.166 0.162 0.142 0.130

0.322 0.319 0.317 0.312 0.306 0.381 0.342 0.304 0.295 0.258 0.238

0.208 0.207 0.205 0.202 0.198 0.247 0.222 0.197 0.191 0.167 0.154

0.153 0.152 0.151 0.148 0.145 0.181 0.163 0.144 0.141 0.123 0.113

0.216 0.214 0.213 0.210 0.205 0.256 0.230 0.204 0.199 0.173 0.160

0.285 0.282 0.281 0.276 0.271 0.338 0.303 0.269 0.262 0.228 0.211

0.335 0.332 0.330 0.325 0.318 0.397 0.356 0.316 0.308 0.269 0.248

0.0369 0.0366 0.0363 0.0358 0.0351 0.0438 0.0393 0.0348 0.0339 0.0297 0.0273

0.332 0.329 0.327 0.322 0.316 0.394 0.354 0.314 0.305 0.267 0.246

0.332 0.329 0.327 0.322 0.316 0.394 0.354 0.314 0.305 0.267 0.246

0.266 0.264 0.262 0.258 0.253 0.316 0.284 0.251 0.245 0.214 0.197

See Table 2 for star ID.

J. Astrophys. Astr. (2021)42:88

Page 5 of 7

88

Figure 2. Synthetic AB magnitudes: GALEX-FUV vs. UVIT far-UV (column 1); corresponding residuals (column 2); GALEX-NUV vs. UVIT near-UV (column 3); corresponding residuals (column 4).

Figure 3. Synthetic flux points for two GALEX filters and ten UVIT filters of stars 1 to 4 (left panel) and of stars 5 to 8 in (right panel) superimposed on respective model spectra.

As an additional test, we reddened the model spectra of each star and repeated the run to compute the magnitudes through GALEX filters. To do this, UV extinctions (Ak ) are computed adopting Av

values from Siegel (2010) and using the CCM-89 model as detailed in the next section. The right panel of Fig. 4 shows UV extinction curves for eleven faint stars.

88

Page 6 of 7

J. Astrophys. Astr. (2021)42:88

residuals are within 0:06. Note that the residuals are higher for narrow band filters (F172M, N279N and N219M) than the broader bands. Such GALEX-UVIT correlation have been used for the in-flight photometric calibration of UVIT by Rahna (2017). In conclusion, given this detailed analysis including the UV extinctions, and the resulting strong correlations between (i) (synthetic vs. observed) quantities of GALEX as well as UVOT and (ii) synthetic photometry of (UVIT vs. GALEX) filter systems, it is strongly recommended that these faint stars be observed and included as UVIT flux standards. 3.1 UV extinction Figure 4. Left panel: Synthetic flux points for two GALEX filters and ten UVIT filters of stars 9 to 11 superimposed on respective model spectra. Right panel: UV extinction curves for eleven UV faint stars using CCM-89 model.

In Table 3, the columns 2 (FUV) and 3 (NUV) give unreddened synthetic magnitudes, the columns 4 (FUV) and 5 (NUV) give reddened synthetic magnitudes computed for GALEX FUV and NUV filters respectivly. The columns 6 (FUV) and 7 (NUV) are actual GALEX observations from Siegel (2010). The reddened magnitudes when corrected by the fit coefficients match the observed GALEX magnitudes within the errors. Next, we performed the synthetic photometry of UV faint standards for UVIT filters. Table 4 gives the synthetic UV magnitues in columns 1 to 10. We have listed the AB magnitudes although magnitudes have been computed for both Vega and AB systems. In Fig. 3 and left panel of Fig. 4 the synthetic fluxes are superimposed on the corresponding model spectra. To obtain the expected magnitudes from UVIT observations we need to apply the fit coefficients and extinction (see next section). Figure 2 demonstarates strong GALEX vs. UVIT correlation of synthetic magnitudes. The column 1 is plots of GALEX-FUV vs. UVIT far-UV and column 3 is plots of GALEX NUV vs. UVIT near-UV. Filter names can be seen within the panels. The corresponding residuals are plotted on the right side panels (columns 2 and 4). The slopes of linear fit for various UVIT filters are as follows: F148W: 1:0018  0:001, F148Wa: 1:0016 0:001, F154W: 0:999  0:0003, F169M: 0:996 0:002, F172M: 0:938  0:016, N219M: 0:995 0:007, N245M: 0:992  0:004, N263M: 0:989  0:011, N279N: 0:983  0:016, N242W: 0:993  0:003. The

Study of UV extinction is interesting due to its variation as a function of wavelength as well as along different sight lines. Several authors in literature have modelled UV extinction of which we prefer the model of (Cardelli et al. 1989; CCM-89). This model provides Ak =Av , the extinction normalized by visual extinction Av . The values of visual extinction for each star were adopted from the Siegel’s model parameters. The UV faint stars studied here are located at high Galactic lattitudes and extinctions are small. However it is necessary to account for the extinction in precise photometry and particularly for candidate standards. The extinction curves we computed for the faint stars are presented in th right panel of Fig. 4. In Table 5, we have tabulated the extinction (Ak ) at the mean wavelength kmean of each filter. The mean wavelengths kmean given in column 2 are computed using the equation below: R kSk dk : ð6Þ kmean ¼ R Sk dk 4. Discussion and conclusions We propose a set of 11 faint UV white dwarf stars as potential UV standards for UVIT. A detailed analysis of these stars using synthetic photometry has been carried out. The set of 11 UV faint stars analysed here satisfy following selection criteria for all-sky secondary standards for UVIT: (i) Stars known to be photometrically stable and single. (ii) Location to be  30 degree above or below the galactic lattitude.

J. Astrophys. Astr. (2021)42:88

Page 7 of 7

88

Table A1. Effective wavelengths. Star#

F148W

F148Wa

F154W

F169M

F172M

N219M

N245M

N263M

N279N

N242W

1 2 3 4 5 6 7 8 9 10 11

1488.76 1488.88 1491.03 1488.91 1488.84 1491.28 1488.94 1488.83 1492.02 1490.24 1491.14

1493.33 1493.45 1495.60 1493.49 1493.42 1495.84 1493.51 1493.41 1496.58 1494.81 1495.71

1530.48 1530.49 1532.11 1530.66 1530.53 1532.26 1530.54 1530.68 1532.79 1531.51 1532.19

1587.21 1587.19 1587.96 1587.34 1587.23 1588.02 1587.24 1587.39 1588.29 1587.69 1588.01

1714.71 1714.65 1714.89 1714.77 1714.72 1714.88 1714.65 1714.82 1714.92 1714.78 1714.88

2206.81 2206.97 2207.05 2206.72 2206.82 2207.25 2206.96 2206.47 2207.43 2207.25 2207.20

2438.91 2439.39 2439.55 2438.64 2438.94 2440.12 2439.35 2437.89 2440.63 2440.15 2439.98

2617.24 2617.33 2617.76 2617.20 2617.27 2618.01 2617.25 2616.91 2618.29 2617.92 2617.95

2791.37 2791.36 2791.43 2791.38 2791.37 2791.45 2791.35 2791.36 2791.48 2791.43 2791.45

2344.90 2345.53 2349.64 2344.70 2345.12 2351.66 2344.95 2342.53 2354.13 2350.63 2351.15

See Table 2 for star IDs.

(iii) Absence of any other bright source in the FoV for the safety of the on-orbit detectors. (iv) The FoV to be uncrowded. (v) Extinctions to be low in the line of sight. (vi) Availability of UV observations or SEDs. DA white dwarfs are ideal candidates for flux calibration. They have pure hydrogen atmospheres and hence lack spectral features in UV regime making it simpler to compute the model spectra. Siegel (2010) constructed the model spectra for 11 UV faint stars using the spectroscopic data of Sloan survey as well as photometry from GALEX and UVOT. Their models could reproduce the photometric measurements from GALEX and UVOT. We used these model spectra to compute synthetic photometry for UVIT, GALEX and UVOT filter systems and compared the results. The fact that (i) these stars were identified as standards for GALEX and UVOT, (ii) the correlation between synthetic vs. observed photometry with GALEX as well as UVOT filter system and (iii) the strong correlation between synthetic UVIT vs. GALEX filter systems makes them potential UV standards for UVIT. Finally, having a common network of standard stars across many UV missions allows photometric transformation and intercomparison easier and reliable.

Acknowledgements This work has made use of MAST, SIMBAD and vizier data bases. IDL graphic package and Fortran 77 are used to write own codes. The work presented here is carried out at home without any institutional support. The software programs used here are advanced versions of preliminary programs developed during the tenure at

Indian Institute of Astrophysics (IIA), Bangalore. I wish to acknowledge that the investigations presented here basically began from the initial discussions with Professor A. K. Pati (IIA) to compile ‘An All-Sky catalog of secondary standards for UVIT’ and then were developed further. The UVIT is built in collaboration between IIA, IUCAA, TIFR, ISRO and CSA. This publication uses the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). Appendix

The effective wavelengths are given in Table A1.

References Bertelli G. et al. 1994, Astron. Astrophys. Suppl. Ser., 106, 275 Bhargavi S. G., Pati A. K. 2009, Poster presented at the ASI meeting held in Indian Institute of Astrophysics, Bangalore, March 2009 Cardelli J. A., Clayton G. C., Mathis J. S. 1989, ApJ, 345, 245. Pickles A. 1998, PASP, 110, 863 Press W. H. et al. 1992, in Numerical Recipes in Fortran, second edition, Cambridge University Press, Cambridge Rahna P. T. 2017, MNRAS 471, 3028 Rao N. K. et al. 2018a, A&A, 620, 138 Rao N. K. et al. 2018b, A&A, 609L, 1 Rao N. K. et al. 2020, PASP, 132, 4201 Siegel M. H. et al. 2010, Astronom. J., 725, 1215 Subramaniam A. et al. 2016, ApJ, 833, 27 Subramaniam A. et al. 2017, AJ, 154, 233 Tandon S. N. et al. 2017a, J. Astrophys. Astr., 38, 28 Tandon S. N. et al. 2017b, Astron. J., 154, 128 Tandon S. N. et al. 2020, AJ, 159, 158

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021) 42:78 https://doi.org/10.1007/s12036-021-09744-0

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

BEYOND ASTROSAT

Beyond AstroSat: Astronomy missions under review P. SREEKUMAR1,2,*

and V. KOTESWARA RAO3

1

Indian Space Research Organisation, Antariksh Bhavan, New BEL Road, Bengaluru 560 094, India. Indian Institute of Astrophysics, II Block, Koramangala, Bengaluru 560 034, India. 3 U. R. Rao Satellite Centre, Bengaluru 560 017, India. *Corresponding Author. E-mail: [email protected] 2

MS received 20 November 2020; accepted 30 March 2021 Abstract. India has an expanding program in using space as a platform for research. Astrophysics research from satellites increasingly complement ground-based observations with unique wavelength coverage, more frequent temporal coverage and diffraction-limited observations. India’s first dedicated space astronomy mission, AstroSat has completed five years in orbit and continues to generate important results. Most onboard systems are healthy and the mission is expected to continue to operate for many more years. Plans for space astronomy missions beyond AstroSat, are under discussion for some time. These are based on responses from the Indian research community to an announcement of Opportunity call in early 2018. Here we discuss, an outline of the science focus of future space astronomy missions, under consideration. Keywords. Space astronomy—AstroSat—Indian space missions.

1. Introduction With the advent of India’s space program in the 60’s, the country has sustained a modest but expanding program in space astronomy. As the country’s launch capability rose, piggyback experiments evolved into full-scale dedicated satellites. AstroSat, India’s first multiwavelength astronomy satellite (Rao et al. 2009), was a culmination of this effort with its launch in 2015 from the Satish Dhawan Space Center, Sriharikota using the highly dependable, PSLV rocket. AstroSat took two decades to emerge from initial discussions to realisation and launch. It had an important early phase involving numerous discussions and meetings among the research community in the country and ISRO teams, in evolving the thematic multiwavelength focus of the mission and to arrive at specific wavelength coverage desired of key experiments. It blended mature instrumentation techniques in the country with exploratory ones (e.g., UV detectors, X-ray optics and use of compound semiconductor X-ray This article is part of the Special issue on ‘‘AstroSat: Five Years in Orbit’’.

detectors). AstroSat’s unique proposal-driven observational program was a new experience for ISRO. It was also designed to respond quickly to Target-of-Opportunities when unexpected events/states occurring in cosmic sources, may force repointing of the spacecraft to new targets in the sky. This demands quick decisionmaking and disrupts ongoing observations, to maximise the science yield from the Observatory. The creation of Payload Operations Centers for each payload at associated lead institutes, has created an infrastructure for data transfer, data processing to higher level products and training of user scientists. This can be a continuing asset that can serve future space science programs. Finally, the integrated data repository for AstroSat, the Indian Space Science Data Center at Bylalu, near Bengaluru, has an experienced team in providing proposal processing support, data storage, archive and dissemination in accordance with ISRO release guidelines, to users around the world. AstroSat created a new collaborative research environment in the country across many academic institutions and ISRO centers. The yearly cycle inviting scientific proposals to use AstroSat, has resulted in the creation of an active community of

78

Page 2 of 4

Indian astronomers, with many of the young researchers having matured adequately to contribute significantly to future programs. The AstroSat mission supplemented national efforts through the inclusion of international contributions (UK and Canada), thus building new bridges for global collaboration. This legacy built around the success of AstroSat, clearly should be leveraged to build and strengthen future space astronomy research in India.

2. Upcoming near-term astronomy missions There are two astronomy missions already approved by the government and expected to be launched by ISRO in 2021, Xposat and Aditya-L1.

2.1 Xposat Xposat will be the first of a new wave of X-ray polarimeters, emerging nearly 45 years after the first and only dedicated X-ray polarimeters in space which discovered polarisation from the Crab Nebula (Novick et al. 1972; Weisskopf et al. 1976). Xposat carries two X-ray payloads, POLIX and XSPECT. The X-ray polarimeter, POLIX (Rishin et al. 2010) measures intensity of polarised photons in the 8–30 keV energy range using the principle of Thompson scattering wherein the incident photon polarisation state defines a preferred scattering plane in aizmuth. It is made up of four orthogonally arranged position-sensing proportional counters into which incident X-ray photons are scattered from a Berilliyum disk scatterer. Xposat also carries a non-imaging, soft X-ray spectrometer (XSPECT), operating in the 0.8–15 keV range using an array of 16 large area swept charge detectors (each of 2 cm2 geometric area) and good spectral resolution (*160 keV @ 6 keV). This is a near identical copy of the CLASS X-ray spectrometer, currently onboard Chandrayaan-2, being used to detect X-rays from the lunar surface (Radhakrishna et al. 2020). The observational program of POLIX involves long exposures (*a few weeks) at cosmic sources to reach a Minimum Detectable Polarisation (MDP) of 3% for bright sources ([40 milliCrab). A slow rotation (*0.2 rpm) about the view axis is planned to reduce impact of any unmodeled response of any of the four detectors, on the measured polarisation. This makes Xposat mission interesting for spacecraft operators, tasked with the delicate optimisation of maximising observation time with a rotating spacecraft during the night-side orbit and a non-rotating

J. Astrophys. Astr.

(2021) 42:78

spacecraft during the sunlit part for charging of onboard batteries. The near-continuous, month-long spectral observations of the same source by the co-aligned XSPECT payload, adds unique data on the science of state transitions in binaries and could link polarisation and spectral state of source.

2.2 Aditya-L1 The question of what heats the solar corona is an important focus for the first dedicated solar mission from India. Utilising a slew of 7 payloads, Aditya-L1 observes the Sun with a high observational duty cycle from the unique vantage point of L1 (Seetha & Megala 2017). A Visible Emission Line Coronograph (VELC) (Prasad et al. 2017) observes specific lines of Fe XIV (530.3 nm), Fe XI (789.2 nm) and Fe XIII (1074.4 nm) with a spectral resolution of *30 mAng/pixel in the visible and *200 mAng/pixel in IR. The ability of the coronograph to observe the corona as close as 1.08Rsun is an important capability of this instrument. The second major instrument is a UV imager with 11 filters (includes few narrow bands) covering the 200–400 nm range and used to monitor the total irradiance and to trace chromospheric activities (Tripathi et al. 2017). Two X-ray spectrometers provide the broad-band X-ray spectrum from solar flare events (Sankarasubramanian et al. 2017). These are well complemented by in-situ observations of particles and magnetic field measurements (Janardhan et al. 2017). This makes Aditya-L1 a unique solar observational facility in space. Together with the Parker probe (NASA) (Neugebauer et al. 2020) and the Solar Orbiter (ESA) (Muller et al. 2020), Aditya-L1 forms a trio of contemporaneous advanced solar missions for the global solar research community.

3. Future astronomy missions AstroSat has created a large pool of user scientists and students in the country. For future programs to build on this success, it must further enhance the number and scope of new users and attract newer institutions into the critical areas of design and development of space payloads, advanced instrument calibration, analysis techniques and data modeling. Further, it must provide a platform for increased collaborations across institutes and academia and train a larger number of post-graduate students in the country in space science research.

J. Astrophys. Astr.

(2021) 42:78

In early 2018, ISRO invited proposals for new missions/experiments under an Announcement of Opportunity call, towards evolving a robust plan for future space astronomy missions following AstroSat. More than twenty proposals addressing missions to study the microwave background, UV, IR and exoplanet spectroscopy, X-ray polarimeters, solar missions and observational electromagnetic counterpart of gravitational merger events, were received and reviewed. The review committee recommended the need to merge independent experiment proposals addressing both solar physics and those more aligned to space weather research, into a comprehensive solar mission as a follow-on mission to Aditya-L1. Motivated by the success of AstroSat, and a desire to build on the expected outcome from the soon-to-be launched Xposat mission, the high-energy community in India has expressed its keen interest for a broad-band (soft to hard X-rays) X-ray polarimetry mission. Multiple proposals addressing this research area were also requested to combine the requirements into a more optimised X-ray mission, retaining the polarimetric focus. This program requires initiating steps to strengthen key technologies that include multi-layer X-ray optics development, essential to increase the numerical aperture at hard X-rays. This exercise towards evolving a comprehensive, globally-competitive X-ray mission is still underway. A major mission concept addressing the next advancement in the cosmic microwave background studies, was also proposed. The associated development of many new technologies and the overall high cost of such a mission, could be partially justified through capacity building in the country in specific areas. Developmental initiatives in select high technology areas at academic/research institutes could enable such opportunities. The review process short listed a few missions in these specific areas of research. These are discussed in brief:

3.1 Exoplanet research The last 25 years of research on exoplanets has resulted in the emergence of an exciting field of study with current focus in two broad areas: (1) discovery of more planets and (2) improved characterisation of known ones. The astronomy community in India and a small group from UK, led by the Indian Institute of Space

Page 3 of 4

78

Science and Technology (IIST), has embarked on a focused program to address the second area, specifically to study atmospheric composition of exoplanets using moderate to low-resolution spectroscopic observations across IR, optical and UV bands. To achieve adequate signal-to-noise ratio and photometric precision, this mission demands a large collecting optics of *2 meters and a parking orbit that shields the experiment from the thermal effects of Sun and Earth. Currently, a study has been undertaken to explore the feasibility of accommodating a large aperture telescope into ISRO’s rocket with the largest faring, for a mission that places this experiment at about 1.5 million kilometers away from Earth, at the Sun–Earth Lagrange point 2.

3.2 Gravitational wave followup The exciting discovery of gravitational waves by LIGO (Abbott et al. 2016) has paved the way for a long anticipated pursuit of gravitational wave astronomy to study astrophysical sources. The discovery of exciting signal from the coalescence of a binary neutron star system in 2017 (Abbott et al. 2017) and the subsequent discovery of associated emission at other wavelengths, triggered the hunt for detection of prompt emission from such events. A team led by the Indian Institute of Technology Bombay has proposed to build an all-sky X-ray detector that can see the prompt emission from such merger events and provide significantly improved localisation for other observatories to pursue deep, sensitive search for counterparts. The proposed twin satellite system which is required to ensure all-sky coverage, free of Earth occultation, has to be prototyped, designed and built in a short time to exploit the current observational gap.

3.3 Search for epoch of re-ionisation signal The discovery of a redshifted 21-cm line feature at *70 MHz (Bowman et al. 2018) from the period of cosmic dawn has triggered a new interest globally to search for clear signatures from the Epoch of reionisation. It is believed that this is the period when the first stars and galaxies were formed. This is a very faint signal buried in the bright radio synchrotron emission of our galaxy from electrons interacting with the galactic magnetic field. The proposed experiment attempts to improve upon the ground-based developments (Girish et al. 2020) in this effort and reach

78

Page 4 of 4

sensitivities that are nearly a million times weaker than the foreground galactic emission using the unique vantage point of a far side lunar orbit. The blocking of the Radio Frequency Interference (RFI) noise from Earth by more than 70 dB, enables such a high sensitivity observation. This effort is being led by scientists at the Raman Research Institute (RRI).

3.4 Advanced UV imaging and spectroscopy UV astronomy research has grown with the availability of high-resolution images from UVIT on AstroSat. This has triggered a strong desire for a UV spectroscopy mission. A new mission concept, led by the Indian Institute of Astrophysics (IIA) in partnership with Canada, is under review. It proposes a UV mission with moderate spectral resolution and an order of magnitude improvement in angular resolution over UVIT. It incorporates multi-object spectroscopic observation mode for a subset of the field-of-view (FoV) and slit-less spectroscopy over the full FOV. In addition to the usual astrophysical targets for focussed studies, this mission also attempts to carry out limited surveys of star-forming galaxies, tracing quasar outflows and UV studies of interacting galaxies. Such high angular resolution imaging and wide-field spectroscopy require fine-guidance systems, large advanced detector arrays, capability to select specific objects in the FoV for spectroscopic studies, etc. These are new and interesting challenges for the research community to pursue and are being partly addressed under a pre-project funding program.

4. Summary The successful AstroSat mission has brought together a larger community of astronomers who are keen to work across wavelength regimes to address specific astrophysical problems. Major ground-based facilities accessible to Indian astronomers like the upgraded GMRT, the upgraded Ooty radio telescope, the upcoming large TeV Imaging telescope (MACE) and the new 3.6-m Devasthal Optical Telescope, are now

J. Astrophys. Astr.

(2021) 42:78

being expanded in a major way with the Indian participation in Mega-projects like the Square Kilometer Array (SKA), the Thirty Meter Telescope (TMT) and the LIGO-India program. The new Indian space astronomy missions discussed here, build on existing strengths, both in research areas and experimental capabilities, but is also designed to push the community in India into new, challenging areas of technology, research pursuits and wider collaborations. An early success with the ongoing technology prototyping phase of short-listed proposals, will be key towards seeking formal government approvals for some of these proposed space astronomy missions.

Acknowledgement Important contributions made by the Payload Shortlisting Committee and the Apex Science Board towards identification of priority proposals for future astronomy missions, are gratefully acknowledged.

References Abbott B. P. et al. 2016, PRL, 116, 061102 Abbott B. P. et al. 2017, PRL, 119, 161101 Bowman J. D. et al. 2018, Nature, 555, 67 Girish B. S. et al. 2020, J. Astron. Instrum., 09, 2050006 Janardhan P. et al. 2017, Curr. Sci., 113, 620 Koteswara Rao V., Agrawal P. C., Sreekumar P. et al. 2009, Acta Astronautica, 65, 6 Muller D. et al. 2020, A&A, 642, A1 Neugebauer M. et al. 2020, ApJS, 246, 19 Novick R., Weisskopf M. C., Berthelsdorf R., Linke R., Wolff R. S. 1972, Astrophys. J., 174, L1 Prasad R. et al. 2017, Curr. Sci., 113, 613 Radhakrishna V. et al. 2020, Curr. Sci., 118, 219 Rishin P.V. et al. 2010, in Bellazzini R., Costa E., Matt G., Tagliaferri G., eds, X-ray Polarimetry: A New Window in Astrophysics, Cambridge University Press, p. 83 Sankarasubramanian K. et al. 2017, Curr. Sci., 113, 625 Seetha S., Megala S. 2017, Curr. Sci., 113, 610 Tripathi D. et al. 2017, Curr. Sci., 113, 616 Weisskopf M. C., Cohen G. G., Kestenbaum H. L., Long K. S., Novick R., Wolff R. S. 1976, Astrophys., J. 208, 125

Ó Indian Academy of Sciences

J. Astrophys. Astr. (2021)42:97 https://doi.org/10.1007/s12036-021-09770-y

Sadhana(0123456789().,-volV)FT3](012345 6789().,-volV)

ASI outreach The outreach component of the AstroSat mission started even before its launch and continued well into 2019. Prior to launch, a booklet was written that explained the rationale behind AstroSat, its various payloads, and its science case, aimed at college students. Multi-lingual FAQs accompanied this effort. In the week of the launch, press conferences were organised in three different cities. In the following months, a set of well-designed posters were created and printed that described the key features of a Space Observatory. The longest running outreach program was AstroSat Picture of the Month (APOM)

(https://astron-soc.in/outreach/apom/), which featured a monthly image from AstroSat accompanied by explanatory text and links for further reading. Many of these APOMs were also recast as articles that were published in many newspapers. All of these resources reached the student and teacher communities in India through social media, some of which were dedicated to AstroSat. These activities were organised by the Public Outreach and Education Committee of the Astronomical Society of India in collaboration with the AstroSat Training and Outreach Team.