CO2 hydrogenation catalysis 9783527824090, 352782409X, 9783527824106, 3527824103, 9783527824113, 3527824111

Citation preview

CO2 Hydrogenation Catalysis

CO2 Hydrogenation Catalysis Edited by Yuichiro Himeda

Editor Dr. Yuichiro Himeda

National Institute of Advanced Industrial Science and Technology AIST Tsukuba West, 16‐1 Onogawa 305‐8569 Tsukuba, Ibaraki Japan Cover

Cover Design: Wiley Cover Image: Courtesy of Yuichiro Himeda

All books published by Wiley-VCH are carefully produced. Nevertheless, authors, editors, and publisher do not warrant the information contained in these books, including this book, to be free of errors. Readers are advised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate. Library of Congress Card No.: applied for British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at . © 2021 WILEY‐VCH GmbH, Boschstr. 12, 69469 Weinheim, Germany All rights reserved (including those of translation into other languages). No part of this book may be reproduced in any form – by photoprinting, microfilm, or any other means – nor transmitted or translated into a machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically marked as such, are not to be considered unprotected by law. Print ISBN: 978‐3‐527‐34663‐9 ePDF ISBN: 978‐3‐527‐82409‐0 ePub ISBN: 978‐3‐527‐82410‐6 oBook ISBN: 978‐3‐527‐82411‐3 Typesetting  SPi Global, Chennai, India Printing and Binding Printed on acid‐free paper 10  9  8  7  6  5  4  3  2  1

v

Contents Preface  xi 1 Introduction  1 Yuichiro Himeda and Matthias Beller 1.1 ­Direct Use of CO2  1 1.2 ­Chemicals from CO2 as a Feedstock  2 1.3 ­Application and Market Studies of CO2 Hydrogenation Products  4 1.3.1 Formic Acid/Formate  4 1.3.2 Methanol  4 Methanation  5 1.3.3 Energy Storage  6 1.3.4 1.4 ­Supply of Materials  6 CO2 Supply  6 1.4.1 1.4.2 Energy and H2 Supply  8 1.5 ­Political Aspect: Tax  9 1.6 ­Conclusion and Perspectives  9 ­References  10 2

2.1 2.2 2.2.1 2.2.1.1 2.2.1.2 2.2.1.3 2.2.1.4 2.2.2 2.2.3 2.3 2.3.1

Homogeneously Catalyzed CO2 Hydrogenation to Formic Acid/Formate by Using Precious Metal Catalysts  13 Wan-Hui Wang, Xiujuan Feng and Ming Bao ­Introduction  13 ­Ir Complexes  14 Ir Complexes with N,N-ligands  14 Tautomerizable N,N-ligands with OH Groups  14 N,N-ligands with NH Group  30 Tautomerizable N,N-ligands with OH and NH Groups  32 Tautomerizable N,N-ligands with Amide Group  33 Ir Complexes with C,N- and C,C-ligands  34 Ir Complexes with Pincer Ligands  35 ­Ru Complexes  37 Ru Complexes with Phosphorous Ligands  38

vi

Contents

2.3.2 2.3.3 2.4 2.5

Ru Complexes with N,N- and N,O-ligands  40 Ru Complexes with Pincer Ligands  41 ­Rh Complexes  46 ­Summary and Conclusions  49 ­References  49

3

Homogeneously Catalyzed CO2 Hydrogenation to Formic Acid/Formate with Non-precious Metal Catalysts  53 Luca Gonsalvi, Antonella Guerriero and Sylwia Kostera ­Introduction  53 ­Iron-Catalyzed CO2 Hydrogenation  55 Non-pincer-Type Iron Complexes  56 Pincer-Type Iron Complexes  63 ­Cobalt-Catalyzed CO2 Hydrogenation  69 ­Nickel-Catalyzed CO2 Hydrogenation  73 ­Copper-Catalyzed CO2 Hydrogenation  77 ­Manganese-Catalyzed CO2 Hydrogenation  78 ­Other Non-precious Metals for CO2 Functionalization  81 ­Conclusions and Perspectives  85 ­References  86

3.1 3.2 3.2.1 3.2.2 3.3 3.4 3.5 3.6 3.7 3.8

Catalytic Homogeneous Hydrogenation of CO2 to Methanol  89 Sayan Kar, Alain Goeppert and G. K. Surya Prakash 4.1 ­Carbon Recycling and Methanol in the Early Twenty-First Century  89 4.2 ­Heterogeneous Catalysis for CO2 to Methanol  91 4.3 ­Homogeneous Catalysis – An Alternative for CO2 to Methanol  92 4.3.1 Benefits of Homogeneous Catalysis  92 4.3.2 CO2 Hydrogenation to Methanol Through Different Routes  92 4.3.3 The First Homogeneous System for CO2 Reduction to Methanol  93 4.3.4 Indirect CO2 Hydrogenation  94 4.3.5 Direct CO2 Hydrogenation  97 4.3.5.1 Through Formate Esters  97 4.3.5.2 Through Oxazolidinone or Formamides  100 4.3.6 CO2 to Methanol via Formic Acid Disproportionation  108 4.4 ­Conclusion  109 ­References  110 4

5

5.1 5.2 5.2.1 5.2.2

Theoretical Studies of Homogeneously Catalytic Hydrogenation of Carbon Dioxide and Bioinspired Computational Design of Base-Metal Catalysts  113 Xiuli Yan and Xinzheng Yang ­Introduction  113 ­H2 Activation and CO2 Insertion Mechanisms  114 Hydrogen Activation  114 Insertion of CO2  115

Contents

5.3 5.3.1 5.3.2 5.4 5.5

­ ydrogenation of CO2 to Formic Acid/Formate  118 H Catalysts with Precious Metals  118 Catalysts with Non-noble Metals  128 ­Hydrogenation of CO2 to Methanol  133 ­Summary and Conclusions  142 ­References  145

6

Heterogenized Catalyst for the Hydrogenation of  CO2 to Formic Acid or Its Derivatives  149 Kwangho Park, Gunniya Hariyanandam Gunasekar and Sungho Yoon ­Introduction  149 ­Molecular Catalysts Heterogenized on the Surface of Grafted Supports  150 ­Molecular Catalysts Heterogenized on Coordination Polymers  157 ­Molecular Catalysts Heterogenized on Porous Organic Polymers  161 ­Concluding Remarks and Future Directions  172 ­References  173

6.1 6.2 6.3 6.4 6.5 7

7.1 7.2 7.3 7.3.1 7.3.2 7.3.3 7.4 7.4.1 7.4.2 7.4.3 7.4.4 7.4.5 7.5 7.6 8 8.1 8.2 8.2.1 8.2.2 8.2.3 8.2.4 8.2.5

Design and Architecture of Nanostructured Heterogeneous Catalysts for CO2 Hydrogenation to Formic Acid/Formate  179 Kohsuke Mori and Hiromi Yamashita ­Introduction  179 ­Unsupported Bulk Metal Catalysts  180 ­Unsupported Metal Nanoparticle Catalysts  181 Metal Nanoparticles Without Stabilizers  181 Metal Nanoparticles Stabilized by Ionic Liquids  182 Metal Nanoparticles Stabilized by Reverse Micelles  183 ­Supported Metal Nanoparticle Catalysts  184 Metal Nanoparticles Supported on Carbon-Based Materials  184 Metal Nanoparticles Supported on Nitrogen-Doped Carbon  185 Metal Nanoparticles Supported on Al2O3  189 Metal Nanoparticles Supported on TiO2  191 Metal Nanoparticles Supported on Surface-Functionalized Materials  194 ­Embedded Single-Atom Catalysts  198 ­Summary and Conclusions  202 ­References  203 Heterogeneously Catalyzed CO2 Hydrogenation to Alcohols  207 Nat Phongprueksathat and Atsushi Urakawa ­Introduction  207 ­CO2 Hydrogenation to Methanol – Past to Present  207 Syngas to Methanol  207 CO2 to Methanol  208 Thermodynamic Consideration – Chemical and Phase Equilibria  212 Catalyst Developments  215 Active Sites and Reaction Mechanisms: The Case of Cu/ZnO Catalysts  217

vii

viii

Contents

8.2.6 8.3 8.3.1 8.3.2 8.3.2.1 8.3.2.2 8.3.2.3 8.3.2.4 8.4

Beyond Industrial Cu/ZnO/Al2O3 Catalysts  223 ­ O2 Hydrogenation to Ethanol and Higher Alcohols – Past to Present  226 C Background  226 Catalysts, Active Sites, and Reaction Mechanisms  227 Modified-Methanol Synthesis Catalyst  227 Modified Fischer–Tropsch Catalysts  230 Rhodium-Based Catalysts  231 Modified Molybdenum-Based Catalysts  232 ­Summary  232 ­References  233

9

Homogeneous Electrocatalytic CO2 Hydrogenation  237 Cody R. Carr and Louise A. Berben ­CO2 Reduction to C─H Bond-Containing Compounds: Formate or Formic Acid  237 Survey of Catalysts  238 Group 9 Metal Complexes  238 Group 8 Metal Complexes  241 Nickel Complexes  244 Iron and Iron/Molybdenum Clusters  246 Hydride Transfer Mechanisms in CO2 Reduction to Formate  247 Terminal Hydrides  247 Bridging Hydrides  248 Kinetic Factors in Catalyst Design  249 Roles of Metal–Ligand Cooperation  249 Roles of Multiple Metal–Metal Bonds  250 Thermochemical Considerations in Catalyst Design  253 Selectivity for Formate over H2 as a Function of Hydricity  254 Solvent Dependence of Hydricity  255 ­Prospects in Electrocatalysis: CO2 Reduction Beyond Formation of One C─H Bond  255 ­References  257

9.1 9.1.1 9.1.1.1 9.1.1.2 9.1.1.3 9.1.1.4 9.1.2 9.1.2.1 9.1.2.2 9.1.3 9.1.3.1 9.1.3.2 9.1.4 9.1.4.1 9.1.4.2 9.2 10

Recent Advances in Homogeneous Catalysts for Hydrogen Production from Formic Acid and Methanol  259 Naoya Onishi and Yuichiro Himeda 10.1 ­Introduction  259 10.2 ­Formic Acid Dehydrogenation  260 10.2.1 Organic Solvent Systems  260 10.2.1.1 Ru  260 10.2.1.2 Ir  266 10.2.1.3 Fe  268

Contents

10.2.2 Aqueous Solution Systems  270 10.2.2.1 Ru  270 10.2.2.2 Ir  272 10.3 ­Aqueous-phase Methanol Dehydrogenation  275 10.3.1.1 Ir  279 10.3.1.2 Non-precious Metals  279 10.4 ­Conclusion  281 ­References  282 Index  285

ix

xi

Preface Carbon dioxide is widely considered to be primarily responsible for global climatic changes. Presently, scientists are facing enormous challenges in mitigating the global CO2 emissions. Significant progress has recently been achieved in the research topic of the catalysis of CO2 hydrogenation, as one of the most important subjects in chemistry. In addition, the paradigm shift from fossil fuels to low‐carbon renewable energy (solar photovoltaics and wind) in recent years will allow for the competition between the CO2 emission by energy consumption and its fixation by CO2 conversion. In future, advancement in the fields of carbon capture and utilization is expected. I would like to thank all the authors, who are all acknowledged as world expert in their area of CO2 hydrogenation, for their enthusiastic efforts to present recent advances in CO2 hydrogenation. Their state‐of‐the‐art research gives exceptionally beneficial information to the researchers, teachers, and students who are interested in the research field of CO2 hydrogenation. I anticipate that their contributions will stimulate further study in CO2 utilization as well as CO2 hydrogenation. I would also like to thank the Wiley‐VCH team for their continuous support. Finally, I deeply appreciate the members of my research group for their valuable assistance, especially Dr. Ryoichi Kanega for the cover design, and Dr. Hide Kambayashi for data survey. In the spring and summer of 2020, the world has been hit by the COVID‐19 pandemic. Despite these difficult times, I am delighted that this book could be completed. July 2020

Yuichiro Himeda National Institute of Advanced Industrial Science and Technology, Global Zero Emission Research Center, Tsukuba, Japan

1

1 Introduction Yuichiro Himeda1 and Matthias Beller2 1 National Institute of Advanced Industrial Science and Technology, Global Zero Emission Research Center, AIST Tsukuba West, 16-1 Onogawa, Tsukuba, Ibaraki, 305-8569, Japan 2 Leibniz-Institut für Katalyse, Applied Homogeneous Catalysis, Albert-Einstein Straße 29a, 18059, Rostock, Germany

Of the final products of the combustion of carbon-based fossil fuels, carbon dioxide (CO2) has the highest oxidation state and is known as the major cause of global warming. Annual CO2 emissions from anthropogenic activity in 2018 were approximately 33.1 Gton, an increase of 1.7% compared with 2017 [1]. Since the Industrial Revolution, two trillion tons of CO2 have accumulated in the atmosphere, and the current atmospheric concentration of CO2 has reached an unprecedented level of over 400 ppm (Figure 1.1) [2]. The anthropogenic emission of CO2 is associated with energy consumption, i.e. the combustion of carbon-based fossil fuels, which currently account for around 85% of the world’s energy. According to the Paris Agreement of the United Nations, an overall limit on total cumulative CO2 emissions is crucial for our future development [3, 4]. According to the 2 °C scenario, further cumulative emissions should be limited to below one trillion ton of CO2. The spread of renewable energy (35%), advances in energy conservation (40%), and carbon capture and sequestration (CCS) technologies (14%) are sure to contribute to addressing the problem (Figure 1.2) [3]. However, it is clear that these methods will not completely solve the issues arising from the vast quantities of emitted CO2. In 2017, the International Energy Agency (IEA) presented the Energy Technology Perspectives (Beyond 2 °C Scenario: B2DS), which placed a much greater emphasis on the role of CO2 utilization for reducing emissions [3]. Indeed, in the next decade, we will still rely on carbon-based products for fuels, polymers, commodity chemicals, cosmetics, detergents, and fabrics in modern life. If these chemicals were to be derived from CO2 instead of fossil oils, a sustainable carbon cycle will be possible.

1.1  ­Direct Use of CO2 Apart from chemical applications, already today, CO2 is used directly in enhanced oil recovery (EOR), beverage carbonation, food processing (e.g. coffee decaffeination and CO2 Hydrogenation Catalysis, First Edition. Edited by Yuichiro Himeda. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.

1 Introduction

CO2 Concentration/ppm

420 400 380 360 340 320 300 1960

1980

2000

2020

Year

Figure 1.1  Atmospheric CO2 concentration at Mauna Loa Observatory. Source: Data from National Oceanic and Atmospheric Administration, Global Monitoring Laboratory [2]. 50 Reference Technology Scenario

40

Renewables

Gton of CO2

2

30

20

Efficiency

2 °C Scenario Fuel switching

10

0 2014 2020

CCS

Nuclear

2030

2040

2050

2060

Year

Figure 1.2  IEA 2 °C Scenario (2DS) in Energy Technology Perspectives 2017. Source: Data from Market-driven future potential of Bio-CC(U)S [3].

drinking water abstraction), welding, as a cleaning agent for textiles, and as a solvent in the electronics industry [5]. These approaches are commercially viable. In particular, 70–80 Mton of CO2 is consumed for EOR in the oil sector. Although such direct utilization of CO2 addresses a significant amount of CO2 emissions, these topics are beyond the scope of this book.

1.2  ­Chemicals from CO2 as a Feedstock CO2 has been recognized as an inexpensive and abundant industrial C1 carbon source. The various chemicals that can be produced by CO2 conversion are shown in Table 1.1 [6]. The largest chemical use of CO2 is in the production of urea from ammonia. However, since a

1.2  ­Chemicals from CO2 as a Feedstock

Table 1.1  Chemicals produced commercially from CO2 . Chemical

Anthropogenic CO2 emissions (2018) Urea [7] Diphenyl carbonate (Asahi Kasei Process) [8]

Scale of production/ton

33 100 000 000 181 000 000 1 070 000

Salicylic acid

90 000

Cyclic carbonate

80 000

Polypropylene carbonate

76 000

Acetylsalicylic acid

16 000

Methanol (CRI process) [9]

4000

Source: From Omae [6]. © 2012 Elsevier.

huge amount of CO2 is emitted during methane steam reforming to supply H2, urea production does not contribute to carbon sequestration at present. The catalytic copolymerization of CO2 with epoxides, which provides a thermodynamic driving force due to the strained three-membered ring, is the most prominent example of the synthesis of CO2-based polymers without formal reduction of the carbon oxidation state. Another example, the manufacture of diphenyl carbonate from ethylene oxide, bisphenol A, and CO2 instead of phosgene was developed beginning in 1977 by Asahi Kasei Chemical Corporation to address environmental and safety issues. The first commercial facility started operation in 2002 [8]. This process produces high-quality polycarbonate and high-purity monoethylene glycol in high yields without waste or wastewater. In addition, the phosgene-free process emits approximately 2.32 ton/tonPC less CO2 than the phosgene process according to life-cycle assessment (LCA). Diphenyl carbonate has a large market (3.6 Mton in 2016) for use in automotive parts and accessories, glazing, and medical devices. The phosgene-free technology has already been licensed to Taiwan, South Korea, Saudi Arabia, China, and Russia. Since 2011, in Iceland, carbon recycling international (CRI) operated the first commercial plant for methanol production from CO2 via syngas by the reverse water-gas shift (rWGS) reaction (George Olah Renewable Methanol Plant) [9]. At present, more than five million liters of methanol per year is produced using low-cost electricity and high-concentration CO2 in the flue gas from an adjacent geothermal power plant. It should be noted that this technology is at present only viable in Iceland; however, if there is a surplus of green electricity in the future from an excess of renewable energy, then this process will be attractive at other places, too. Notably, the amount of CO2 utilized by all these approaches, including urea and carbonate production, is very small compared with the magnitude of anthropogenic emissions. Therefore, CO2 conversion into chemicals is unlikely to significantly reduce emissions. Comparatively, it should be noted that fuels are produced and consumed on a much larger scale than these chemicals.

3

4

1 Introduction

1.3  ­Application and Market Studies of CO2 Hydrogenation Products Hydrogenation of CO2 could be an efficient option for developing more environmentfriendly products as alternatives to fossil-based ones. In terms of practicality, the distribution infrastructure of carbon-based chemicals is well established. However, their manufacturing is currently several times more expensive than their conventionally produced counterparts, mainly due to the costs associated with H2 production. Some of the key features of CO2 hydrogenation products and conventional fuels are given in Table 1.2.

1.3.1  Formic Acid/Formate Formic acid is the first carboxylic acid and is naturally occurring produced by ants, bees, and some plants. In 2016, the global production of formic acid was 1.02 Mton [10]. The general production process of formic acid involves the formal carbonylation of water in a two-step synthesis via methyl formate. Formic acid and its salts (formate) are valuable chemical products used for silage and animal feed (27%), leather and tanning (22%), pharmaceuticals and food chemicals (14%), textile (9%), natural rubber (7%), and drilling fluids (4%) [11]. Recently, formic acid has been recognized as a promising liquid organic H2 carrier (LOHC) because of its low toxicity, low combustibility, stability, environmental friendliness, and 4.4 wt% (53 g l−1) H2 content [12–14]. In addition, compressed hydrogen gas can be supplied only by heating of formic acid using catalysts as a chemical compressor [15]. Therefore, advances in the efficient production of formic acid/formates may eventually lead to their large-scale use as LOHCs (see Chapter 10).

1.3.2  Methanol Methanol, the industrial production of which is mainly from syngas, is in high global demand as a fuel and bulk chemical (Figure 1.3) [17]. One ton of methanol produced by the Table 1.2  Characteristics of various energy vectors.

Compound

Methanol

Energy density (GJ m−3)

Approx. price per energy (US$/GJ)

Boiling point/ Ignition point/ melting point (°C) flash point (°C)

64.55/–97.68

470/15 (open)

Vapor pressure at 25 °C (kPa)

15.8

15

16.9

Formic acid

6.3

100

100.56/8.27

520/59 (open)

43.1

Natural gas (CH4)

8.1 (20 MPa) 2

–161/–183

537/–188

147 (15 °C)

Gasoline

34.5

30

17–220/≤–40

300/≤–43

50–93 (37.8 °C)

Diesel oil

36.3

23

140–400/–29 to –18

250/40–70

≤0.35 (37.8 °C)

Hydrogen

5.1 (70 MPa) 120

–252.87/–259.14

500–571/—

1.65 × 105

1.3  ­Application and Market Studies of CO2 Hydrogenation Products

Others 11% Formaldehyde 26% MTO 23% Acetic Acid 8% Alternative Fuels 20%

MTBE 12%

Figure 1.3  Global methanol demand in 2018. Source: Data from Global methanol demand (Methanol Institute) [16].

established process consumes 37.5 GJ of natural gas and emits 1.49 ton of CO2 [18]. In 2018, the global production of methanol was approximately 91.7 Mton, and since 2015, its production has grown by approximately 16% [16]. Approximately 26% and 8% of the methanol produced worldwide is consumed to produce formaldehyde and acetic acid, respectively, as the conventional demands. Methanol can be used as a fuel for internal combustion engines and fuel cells because it has a comparably high-octane number of 113 and a density approximately half that of gasoline. In addition, methanol can be transformed into gasoline through the methanol-to-gasoline (MTG) process developed by Mobil in the 1970s [19]. Another growing market for methanol is the production of light olefins (i.e. ethylene (152 Mton yr−1) and propylene (103 Mton yr−1) in 2017), [20] which are monomer feedstocks for polyethylene and polypropylene as basic products of the plastics industry [21]. The concept of a so-called methanol economy was independently proposed by Olah and Asinger due to the chemical’s promising characteristics for use as an energy vector and chemical feedstock [22–24]. Therefore, the production of methanol by CO2 conversion is regarded as an attractive and potentially profitable route for CO2 utilization.

1.3.3  Methanation CO2 methanation, also known as the Sabatier process, affords methane by the exothermic reaction of CO2 with H2. The commercial methanation of CO2 is performed at 300–550 °C and above 5 bar. Most CO2 methanation processes are considered to be a linear combination of rWGS and CO methanation. The process is expected to be a power-to-gas concept for converting renewable electrical energy into methane as chemical energy. In other words, the main goal of methanation is the intermediate storage of renewable electricity in methane as an energy carrier. Since fossil-based natural gas is a common fuel, there would be easy access to existing infrastructure. Due to the significant interest in CO2 methanation, the first pilot plant capable of producing 0.5 Nm3 h−1 of synthetic natural gas was built in Japan [25]. In terms of commercial

5

6

1 Introduction

installations, Audi has an operational CO2 methanation facility (max. 325 Nm3 h−1) using renewable H2 (max. 1300 Nm3 h−1) from electrolysis (max. 6.0 MW) in Germany [26, 27].

1.3.4  Energy Storage The two most growing renewable energy sources, solar and wind, are intermittent and thus provide highly fluctuating electrical energy. In addition, the region’s best suited areas for the production of renewable energy are often far from consumption areas, i.e. cities. These cause the two key problems of storage and transport. Certainly, electrical energy is an effective way to transfer energy within 1000 km and can be stored in batteries. However, lowcost solutions for the large-scale storage and long-range transport of electrical energy must be developed to improve energy security and balance energy prices. The transformation of excess renewable energy into chemical energy by converting CO2 is one promising option. CO2-based compounds, such as methane, methanol and formic acid, can store energy as gas or liquids with comparably high-energy densities. Especially liquids can be easily transported and release energy as H2 or electricity through oxidation and fuel cells when there is a greater demand. In other words, CO2 can act as an energy vector between electrical and chemical energy. Recently, the electroreduction of CO2 to chemical fuels has been receiving increasing attention because it allows for the direct use of renewable electricity without conversion to high-cost H2 by water electrolysis (see Chapter 9). Much more CO2 is in demand as a feedstock for fuels than for chemicals and mineralization. In addition, related photo-catalytic processes gain more and more interest.

1.4  ­Supply of Materials The CO2 hydrogenation approach requires H2, CO2, and an energy supply. In particular, how much the energy-intensive hydrogenation process contributes to mitigating CO2 emissions will be dominated by the H2 source. Obviously, H2 must be produced with the help of a renewable electricity source such as water electrolysis and not from fossil fuels.

1.4.1  CO2 Supply Capture, purification, and transport of CO2 are essential for its utilization. Table 1.3 lists several large CO2 sources with their typical amounts and concentrations of CO2 as well as impurities. In present CO2 merchant market (approximately 230 Mton, US$7.7 billion), the fermentation process (i.e. bioethanol production) and ammonia production, which provide close to 100% CO2, are predominantly CO2 sources [5, 30]. The CO2 generated from ethanol fermentation commercially supplies roughly 270 000 ton of CO2 annually for EOR through pipeline from Kansas to Texas [28]. On the other hand, the production of electricity and heat accounts for 41% of global CO2 emissions (Figure 1.4), and the transport and industrial sectors account for an additional 25% and 19%, respectively [32]. However, suitable sources of CO2 for use in chemical transformation are limited. The gases contain various impurities, the separation of which is both energy and cost intensive. To supply CO2 of an

1.4  ­Supply of Material

Table 1.3  Concentration of CO2 and contaminants from various sources. Source

Amount/Mton

Ethanol fermentation [28, 30]

50

CO2 concentration/%

99

Impurities

EtOH, MeOH, H2O, H2S

Anhydrous ammonia

30

>95

Natural deposits

13

90–100

N2, O2, He

Power plants

4287

10–15

N2, H2O, SOx, NOx, CO

Steelmaking

266

18–20

N2, SOx, NOx, O2

220

14–33

SOx, NOx, O2

Cement production [31] Atmosphere

3 200 000

0.04

NH3, CO, H2, H2O

N2, O2, SOx, NOx

Source: Carbon Recycling International; Capturing and Utilizing CO2 from Ethanol: Adding Economic Value and Jobs to Rural Economies and Communities While Reducing Emissions (2017); and Greenhouse Gas Inventory Data [9, 28, 29].

Other 4% Residential 6% Electricity and heat producers 41%

Transport 25%

Industry 19%

Other energy industries 5%

Figure 1.4  CO2 emissions from fuel combustion. Source: Data from IEA, CO2 emissions from fuel combustion, 2020 [32].

appropriate quality for use in chemical conversion processes, capture and separation are required (Table 1.4) [33]. The most effective CO2 capture method as the current industrial standard is chemical absorption in an aqueous solution of an amine-based organic compound. However, the cost (35 US$/ton) and energy consumption (2.5 GJ ton−1) of amine capture must still be reduced to provide economically viable routes from carbon dioxide to fuels [34, 35]. Recently, the direct capture of CO2 from ambient air, called direct air capture (DAC), has received increasing attention [36]. One of the advantages of DAC is that it can be located anywhere, because it is unnecessary for CO2 transport. However, from both engineering and chemistry views, there remains much room for improvements to the sorbents and processes. Additionally, thorough techno-economic analyses of DAC processes are necessary [37].

7

1 Introduction

Table 1.4  CO2 capture technologies. Capture technology

Technical principle

Chemical absorption

Chemical reaction between CO2 and absorbent by a temperature swing.

Physical absorption

Dissolution of CO2 into a liquid, the efficiency of which depends on the solubility of CO2 in the liquid.

Solid absorption

Absorption into solid absorbents, which include porous materials impregnated with amines for low-temperature separation or other solid absorbents for high-temperature separation.

Physical adsorption

Adsorption onto porous solids such as zeolites by a pressure or temperature swing.

Membrane separation

Permeation through a membrane with selective permeability for different gas species.

Source: Based on Styring [33].

2000

1500

TWh

Biofuels

Geothermal

Hydro

Wind

Solar PV

Solar thermal

Tide, wave, ocean 1000

500

K U

ly Ita

SA

az Ca il na da In di a G er m an y R us si a Ja pa n N or wa y

Br

U

hi

na

0

C

8

Figure 1.5  Low-carbon electricity generation by source in 2017. Source: Data from explore energy data by category, indicator, country or region (IEA) [38].

1.4.2  Energy and H2 Supply Another consideration is the energy required to capture and convert CO2, which must certainly be derived from renewable sources (Figure 1.5) [38]. If this energy comes from fossil oils, much more CO2 will be emitted than separated. Fortunately, the renewables now account for over 25% of global power output (hydro: 16%, wind: 5%, PV: 2%), [1] and the costs of PV and wind power become even lower than that of fossil fuels (natural gas and coal) (Figure 1.6) [39]. Thus, electricity from renewable sources can be converted into H2 by water electrolysis, which can be performed on an industrial scale. Nevertheless, H2 produced by electrolysis systems (2.5–6 US$/kgH2) is at present more expensive than that from

1.6  ­Conclusion and Perspective 40

Nuclear Natural Gas Wind

/KWh

30

Coal Solar PV

20

10

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

0

Year

Figure 1.6  Levelized cost of energy comparison: Renewable energy versus conventional generation. Source: Data from Lazard.com, Lazard’s levelized cost of energy analysis [39].

current industrial production based on conventional fossil sources, like natural gas reforming and coal gasification ( 58 > 59 ≈ 60. In particular, complex 61 with an electron-donating Ph3P ligand and anionic imidazolyl ligand gave a high yield of 77% and a high TON of 1530. Complex 58 with OH group showed better activity than 59 and 60, suggesting that OH improves the catalytic activity. In 2018, Huang and coworkers reported a Ru(II)-PN3P pincer complex 62 for CO2 hydrogenation in a THF/H2O biphasic system [81]. A high TOF of up to 13 000 h−1 and TON of up to 33 000 were achieved at 130 °C and 11 MPa H2 after 50 hours. An aqueous solution of N,N,N′,N″,N″′-pentamethyldiethylenetriamine (PMDTA) was used to absorb CO2 from air, and the solution was then hydrogenated with 62. Formate separation and catalyst recovery can be easily achieved in a 2-MeTHF/H2O biphasic system because complex 62 is in the organic phase and formate is in the aqueous phase. A recyclability test suggested that productivity was almost not decreased after five cycles, and only 4.1% catalyst leaching was observed.

Cl N

Ph

PPh2 O

Ph3P N

O Ph

54

P

PPh2

Ru PPh3

Cl

Ru

OH

Cl

HO

P Ph3P

N

N

HO

O

Ru

Ph2P

Cl

Cl N

N Ru N

H Cl H O

55

Figure 2.12  Ru-CNC complexes for CO2 hydrogenation.

56

57

PPh2 Cl OH

45

46

2  Homogeneously Catalyzed CO2 Hydrogenation to Formic Acid/Formate

2.4 ­Rh Complexes In the 1990s, Nicholas and Leitner et al. had extensively studied Rh complexes with various phosphine ligands as strong σ-donors [39, 41, 42, 82]. Considerable progress has been achieved and reviewed [4a, 40]. In this section, we introduce recently developed types of Rh complexes. In 2012, Zou et al. reported a Rh complex 63 (Figure 2.14) with a nitrosyl ligand for CO2 hydrogenation in the presence of DBU at 50 °C and 0.3 MPa H2/CO2 (1/1) in DMSO [83]. A TON of 105 was obtained after 16 hours. In 2013, He and coworkers studied CO2 hydrogenation with in situ complex RhCl3/CyPPh2 in the presence of polyetherimide (PEI, MW = 600 Da) [84]. As a stable, strong, and nonvolatile base, PEI could capture and activate CO2 by forming carbamate or carbonate species to facilitate CO2 hydrogenation. In situ FT-IR study suggested the possible coordination of PEI with RhCl3. A high TON of 852 was achieved in MeOH at 60 °C and 8 MPa H2/CO2 (1/1) for 16 hours. In 2018, Leitner and coworkers developed a base-free system of CO2 hydrogenation to FA in a biphasic n-heptane/DMSO system (Scheme 2.15) by using Rh catalysts 64 and 65 with apolar phosphine ligands tris(4-octylphenyl)phosphine and trioctylphosphine, respectively [85]. The catalysts were dissolved in n-heptane to catalyze the reaction, and the product FA was transferred into the DMSO phase. Therefore, catalyst recycling and FA product separation can be easily performed via phase separation. In the recycling experiments, Rh complex 65 showed higher activity than complex 64 and provided high FA concentration of 1.21–1.37 M in each run at 60 °C and 12 MPa H2/CO2 (8/4). After five runs, a total TON of 675 was obtained with complex 65. By contrast, the TON of complex 64 was only 396 after three runs. Leaching tests suggested that more than 99% of Rh was retained in the n-hexane phase after five runs. However, the activities of both catalysts gradually decreased, and this R Cl H N N N

N Ru N N Cl

OTf

H N

H N

N N

58 R = OH 59 R = H

60 OTf

N N Ph3P

N Ru N

61

N Ru N N OTf

H N

H N N PPh3

Ph2P Ph2P

N

NH CO

Ru

HN N

H N

NH PPh2 PPh2 OC NH Ru Ph2P

PPh2

62

Figure 2.13  Pincer Ru complexes for CO2 hydrogenation.

H N N

2.4 ­Rh Complexe

N

Cy Cy P P Cy Rh Cy NO

Ph2P

63

PPh2

Ph2P

PPh2

PNMeP

dppp

Figure 2.14  Rh complexes and phosphine ligands for CO2 hydrogenation.

phenomenon was attributed to catalyst deactivation. Given that DMSO/FA can form an azeotropic mixture, a co-solvent strategy using AcOH was employed to isolate FA via distillation. Although this approach yielded DMSO-free FA, DMSO was extensively decomposed during distillation, which poses a major limitation to this process. In 2014, Linehan and coworkers reported a Rh(diphoshpine)2 complex 66 (Figure 2.15) that promotes catalytic activity via second and outer coordination sphere effects [44]. CO2 hydrogenation was conducted at 21 °C and 4 MPa H2/CO2 (1/1) in THF in the presence of a strong base 2,8,9-triisobutyl-2,5,8,9-tetraaza-1-phosphabicyclo[3.3.3]undecane (Verkade’s base). Complex 66 shows remarkably higher TOF (920 h−1) than [Rh(dppp)2]+ (dppp: 1,3-bis(diphenylphosphino)propane) (150 h−1) without any amine group. Thus, the NMe moiety plays an important role in catalysis. The author emphasized the effects of amine as an electron donor and an adjacent base that may participate in catalysis. Kubiak and coworkers synthesized a series of complexes [Rh(P2N2)2]+ (67, Figure 2.15) with substituted cyclic diphosphine P2N2 (1,5-diaza-3,7-diphosphacyclooctane) to promote

Cat:

PR3 R3P Rh Cl

64 R = L1 65 R = L2

R3P

C8H17 or C8H17 L2

L1 gas phase

H2/CO2

CO2 + H2 n-heptane phase

CO2 + H2

CO2 + H2

[Rh]

FA•nDMSO

FA•nDMSO

FA•nDMSO

DMSO phase Scheme 2.15  Biphasic system for CO2 hydrogenation developed by Leitner et al. Source: From Jens et al. [85], © 2018 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

47

48

2  Homogeneously Catalyzed CO2 Hydrogenation to Formic Acid/Formate

–

R1

OTf

Ph Ph Ph P P Ph Ru N P Ph P Ph Ph Ph

N

N

R

R

N

O2C

CO2 N

Rh P

P R

66

N

Et2P R1

P Et2

R

–

N

P

P 1

N

R

R1

Rh

P Et2 P Et2

68

67 R = Ph, Cy, R1= Ph, Bn, PhOMe

Figure 2.15  Biometric Rh complexes for CO2 hydrogenation.

hydrogenation by introducing pendant amine into the second coordination sphere [86]. All [Rh(P2N2)2]+ complexes 67 showed lower TOFs (590–720 h−1) than [Rh(depe)2]+ (depe: 1,2-bis(diethylphosphino)ethane) (1070 h−1) without pendant amine under CO2 hydrogenation at 21 °C and 0.2 MPa H2/CO2 (1/1) in THF in the presence of Verkade’s base. In addition, no Rh monohydride was observed in the absence of external base; hence, they deduced that pendant amines were unable to deprotonate the dihydride species. Therefore, the bulky P2N2 ligands hinder the access of external base and decrease the catalytic activity. O’Hagan and coworkers recently incorporated an inactive Rh complex 68 (Figure 2.15) into a protein scaffold to catalyze the CO2 hydrogenation by mimicking an efficient dehydrogenase [87]. The artificial metalloenzyme afforded a TOF of 0.38 h−1 and an average TON of 14 ± 3 at 25 °C and 5.8 MPa H2/CO2 (1/1). A mechanistic study suggested that a protein scaffold improves the catalytic performance by facilitating the interaction between CO2 and a hydride intermediate. The importance of outer-sphere interactions was also recognized. Herrmann and coworkers investigated Rh catalyst 69 (Figure 2.16) with a strong electron-donating bis-NHC ligand for bicarbonate hydrogenation to formate in water [46]. Water solubility was increased by introducing sulfonic acid groups into the side chain of the bis-NHC ligand. A high TON of 3600 was achieved at 100 °C and 0.5 MPa H2 in 2-M KHCO3 aqueous solution for 72 hours. In 2018, Espino and coworkers developed half-sandwich Rh complexes 70 and 71 with N,N-ligands (Figure 2.16) [88]. Complex 70 provided a TON of 5991 at 80 °C and 0.1 MPa H2/CO2 (1/1) in 0.1-M aqueous KOH solution for eight hours. By contrast, no reaction was observed with complex 71. The NH2 group of the 8-aminoquinoline shows a positive effect on the catalytic activity. R N N

Cp*

Cp* Cp* Rh N

Cl N

Cl R

69 R = (CH2)3SO3–

Cl

Rh N N

H

Cl

N N

H2 N

N

H 70

Cl

Rh

71

Figure 2.16  Rh complexes with C,C- and N,N-ligands.

N NH2

 ­Reference

2.5 ­Summary and Conclusions Noble metals show excellent catalytic activity for CO2 hydrogenation to FA/formate. Many ligands do not just act as a “spectator,” they can modulate catalytic activity via electronic effect, pendant-base effect, or outer-sphere metal ligand cooperation. Considerable progress has been achieved in catalytic performance and mechanism, which disclose the catalytic process of different catalysts. Principles for the catalyst design are also provided; hence, fruitful achievements and efficient catalysts are expected to be developed in the near future. Development on other types of tautomerizable ligands and economically viable process for CO2 hydrogenation should be further investigated.

R ­ eferences  1 Bredig, G. and Carter, S.R. (1914). Ber. Dtsch. Chem. Ges. 47: 541–545.  2 Inoue, Y., Izumida, H., Sasaki, Y., and Hashimoto, H. (1976). Chem. Lett. 5: 863–864.  3 (a) Klankermayer, J., Wesselbaum, S., Beydoun, K., and Leitner, W. (2016). Angew. Chem. Int. Ed. 55: 7296–7343. (b) Artz, J., Müller, T.E., Thenert, K. et al. (2018). Chem. Rev. 118: 434–504.  4 (a) Leitner, W. (1995). Angew. Chem. Int. Ed. 34: 2207–2221. (b) Jessop, P.G., Joó, F., and Tai, C.C. (2004). Coord. Chem. Rev. 248: 2425–2442. (c) Wang, W., Wang, S., Ma, X., and Gong, J. (2011). Chem. Soc. Rev. 40: 3703–3727. (d) Wang, W.H., Himeda, Y., Muckerman, J.T. et al. (2015). Chem. Rev. 115: 12936–12973. (e) Dong, K., Razzaq, R., Hu, Y., and Ding, K. (2017). Top. Curr. Chem. 375: 23. (f) Onishi, N., Laurenczy, G., Beller, M., and Himeda, Y. (2018). Coord. Chem. Rev. 373: 317–332. (g) Sordakis, K., Tang, C., Vogt, L.K. et al. (2018). Chem. Rev. 118: 372–433.   5 Wang, X., Xia, C., and Wu, L. (2018). Green Chem. 20: 5415–5426.  6 (a) Tanaka, R., Yamashita, M., and Nozaki, K. (2009). J. Am. Chem. Soc. 131: 14168–14169 . (b) Tanaka, R., Yamashita, M., Chung, L.W. et al. (2011). Organometallics 30: 6742–6750.  7 Schmeier, T.J., Dobereiner, G.E., Crabtree, R.H., and Hazari, N. (2011). J. Am. Chem. Soc. 133: 9274–9277.  8 Himeda, Y. (2007). Eur. J. Inorg. Chem. 2007: 3927–3941.  9 Wang, W.-H., Hull, J.F., Muckerman, J.T. et al. (2012). Energy Environ. Sci. 5: 7923–7926. 10 Himeda, Y., Onozawa-Komatsuzaki, N., Sugihara, H., and Kasuga, K. (2007). Organometallics 26: 702–712. 11 Hull, J.F., Himeda, Y., Wang, W.-H. et al. (2012). Nat. Chem. 4: 383–388. 12 Wang, W.-H., Muckerman, J.T., Fujita, E., and Himeda, Y. (2013). ACS Catal. 3: 856–860. 13 Onishi, N., Xu, S., Manaka, Y. et al. (2015). Inorg. Chem. 54: 5114–5123. 14 Azua, A., Sanz, S., and Peris, E. (2011). Chem.-Eur. J. 17: 3963–3967. 15 Maenaka, Y., Suenobu, T., and Fukuzumi, S. (2012). Energy Environ. Sci. 5: 7360–7367. 16 Takaoka, S., Eizawa, A., Kusumoto, S. et al. (2018). Organometallics 37: 3001–3009. 17 Puerta-Oteo, R., Hölscher, M., Jiménez, M.V. et al. (2018). Organometallics 37: 684–696. 18 Gunasekar, G.H., Yoon, Y., Baek, I.-h., and Yoon, S. (2018). RSC Adv. 8: 1346–1350. 19 Suna, Y., Himeda, Y., Fujita, E. et al. (2017). ChemSusChem 10: 4535–4543.

49

50

2  Homogeneously Catalyzed CO2 Hydrogenation to Formic Acid/Formate

20 Kanega, R., Onishi, N., Szalda, D.J. et al. (2017). ACS Catal. 7: 6426–6429. 21 Liu, C., Xie, J.-H., Tian, G.-L. et al. (2015). Chem. Sci. 6: 2928–2931. 22 Jessop, P.G., Ikariya, T., and Noyori, R. (1994). Nature 368: 231–233. 23 Tai, C.C., Pitts, J., Linehan, J.C. et al. (2002). Inorg. Chem. 41: 1606–1614. 24 Muller, K., Sun, Y., and Thiel, W.R. (2013). ChemCatChem 5: 1340–1343. 25 Moret, S., Dyson, P.J., and Laurenczy, G. (2014). Nature Commun. 5: 4017–4023. 26 Lau, C.P. and Chen, Y.Z. (1995). J. Mol. Catal. A-Chem. 101: 33–36. 27 Khan, M.M.T., Halligudi, S.B., and Shukla, S. (1989). J. Mol. Catal. 57: 47–60. 28 Laurenczy, G., Joó, F., and Nadasdi, L. (2000). Inorg. Chem. 39: 5083–5088. 29 Elek, J., Nadasdi, L., Papp, G. et al. (2003). Appl. Catal. A-Gen. 255: 59–67. 30 Federsel, C., Jackstell, R., Boddien, A. et al. (2010). ChemSusChem 3: 1048–1050. 31 Hsu, S.-F., Rommel, S., Eversfield, P. et al. (2014). Angew. Chem. Int. Ed. 53: 7074–7078. 32 Filonenko, G.A., van Putten, R., Schulpen, E.N. et al. (2014). ChemCatChem 6: 1526–1530. 33 Huff, C.A. and Sanford, M.S. (2013). ACS Catal. 3: 2412–2416. 34 Filonenko, G.A., Conley, M.P., Coperet, C. et al. (2013). ACS Catal. 3: 2522–2526. 35 Weilhard, A., Qadir, M.I., Sans, V., and Dupont, J. (2018). ACS Catal. 8: 1628–1634. 36 Scott, M., Blas Molinos, B., Westhues, C. et al. (2017). ChemSusChem 10: 1085–1093. 37 Kar, S., Goeppert, A., Galvan, V. et al. (2018). J. Am. Chem. Soc. 140: 16873–16876. 38 Kothandaraman, J., Goeppert, A., Czaun, M. et al. (2016). Green Chem. 18: 5831–5838. 39 Graf, E. and Leitner, W. (1992). Chem. Commun.: 623–624. 40 Leitner, W., Dinjus, E., and Gassner, F. (1998). CO2 chemistry. In: Aqueous-Phase Organometallic Catalysis, Concepts and Applications (eds. B. Cornils and W.A. Herrmann), 486–498. Weinheim: Wiley-VCH. 41 Gassner, F. and Leitner, W. (1993). J. Chem. Soc., Chem. Commun.: 1465–1466. 42 Ezhova, N.N., Kolesnichenko, N.V., Bulygin, A.V. et al. (2002). Russ. Chem. Bull. 51: 2165–2169. 43 Zhao, G. and Joó, F. (2011). Catal. Commun. 14: 74–76. 44 Bays, J.T., Priyadarshani, N., Jeletic, M.S. et al. (2014). ACS Catal. 4: 3663–3670. 45 Wesselbaum, S., Hintermair, U., and Leitner, W. (2012). Angew. Chem. Int. Ed. 51: 8585–8588. 46 Jantke, D., Pardatscher, L., Drees, M. et al. (2016). ChemSusChem 9: 2849–2854. 47 Himeda, Y., Onozawa-Komatsuzaki, N., Sugihara, H., and Kasuga, K. (2006). J. Photochem. Photobiol. A-Chem. 182: 306–309. 48 (a) Himeda, Y., Onozawa-Komatsuzaki, N., Sugihara, H. et al. (2004). Organometallics 23: 1480–1483. (b) Himeda, Y., Onozawa-Komatsuzaki, N., Sugihara, H., and Kasuga, K. (2005). J. Am. Chem. Soc. 127: 13118–13119. 49 Suna, Y., Ertem, M.Z., Wang, W.-H. et al. (2014). Organometallics 33: 6519–6530. 50 Ohnishi, Y.Y., Matsunaga, T., Nakao, Y. et al. (2005). J. Am. Chem. Soc. 127: 4021–4032. 51 Himeda, Y., Miyazawa, S., and Hirose, T. (2011). ChemSusChem 4: 487–493. 52 Crabtree, R.H. (2011). New J. Chem. 35: 18–23. 53 Onishi, N., Kanega, R., Fujita, E., and Himeda, Y. (2019). Adv. Synth. Catal. 361: 289–296. 54 Ertem, M.Z., Himeda, Y., Fujita, E., and Muckerman, J.T. (2016). ACS Catal. 6: 600–609. 55 (a) Shima, S., Lyon, E.J., Sordel-Klippert, M.S. et al. (2004). Angew. Chem. Int. Ed. 43: 2547–2551. (b) Shima, S., Pilak, O., Vogt, S. et al. (2008). Science 321: 572–575. (c) Shima,

 ­Reference

S. and Ermler, U. (2011). Eur. J. Inorg. Chem.: 963–972. (d) Yang, X. and Hall, M.B. (2009). J. Am. Chem. Soc. 131: 10901–10908. 56 Wang, W.-H., Ertem, M.Z., Xu, S. et al. (2015). ACS Catal. 5: 5496–5504. 57 Manaka, Y., Wang, W.-H., Suna, Y. et al. (2014). Catal. Sci. Technol. 4: 34–37. 58 Xu, S., Onishi, N., Tsurusaki, A. et al. (2015). Eur. J. Inorg. Chem. 2015: 5591–5594. 59 Wang, Z., Lu, S.M., Li, J. et al. (2015). Chem.-Eur. J. 21: 12592–12595. 60 Lu, S.-M., Wang, Z., Li, J. et al. (2016). Green Chem. 18: 4553–4558. 61 Wang, L., Onishi, N., Murata, K. et al. (2017). ChemSusChem 10: 1071–1075. 62 Kanega, R., Onishi, N., Wang, L., and Himeda, Y. (2018). ACS Catal. 8: 11296–11301. 63 Sanz, S., Benitez, M., and Peris, E. (2010). Organometallics 29: 275–277. 64 Siek, S., Burks, D.B., Gerlach, D.L. et al. (2017). Organometallics 36: 1091–1106. 65 (a) Yang, X.Z. (2011). ACS Catal. 1: 849–854. (b) Li, J. and Yoshizawa, K. (2011). Bull. Chem. Soc. Jpn. 84: 1039–1048. (c) Sawatlon, B., Wodrich, M.D., and Corminboeuf, C. (2018). Organometallics 37: 4568–4575. 66 Gunanathan, C. and Milstein, D. (2011). Acc. Chem. Res. 44: 588–602. 67 Aoki, W., Wattanavinin, N., Kusumoto, S., and Nozaki, K. (2016). Bull. Chem. Soc. Jpn. 89: 113–124. 68 (a) Jessop, P.G., Hsiao, Y., Ikariya, T., and Noyori, R. (1996). J. Am. Chem. Soc. 118: 344–355. (b) Munshi, P., Main, A.D., Linehan, J.C. et al. (2002). J. Am. Chem. Soc. 124: 7963–7971. 69 Muller, K., Sun, Y., Heimermann, A. et al. (2013). Chem.-Eur. J. 19: 7825–7834. 70 Gowrisankar, S., Federsel, C., Neumann, H. et al. (2013). ChemSusChem 6: 85–91. 71 (a) Joó, F., Laurenczy, G., Nadasdi, L., and Elek, J. (1999). Chem. Commun.: 971–972. (b) Katho, A., Opre, Z., Laurenczy, G., and Joó, F. (2003). J. Mol. Catal. A-Chem. 204: 143–148. (c) Kovacs, G., Schubert, G., Joó, F., and Papai, I. (2006). Catal. Today 115: 53–60. (d) Horvath, H., Laurenczy, G., and Katho, A. (2004). J. Organometal. Chem. 689: 1036–1045. (e) Erlandsson, M., Landaeta, V.R., Gonsalvi, L. et al. (2008). Eur. J. Inorg. Chem. 2008: 620–627. (f) Laurenczy, G., Jedner, S., Alessio, E., and Dyson, P.J. (2007). Inorg. Chem. Commun. 10: 558–562. 72 Drake, J.L., Manna, C.M., and Byers, J.A. (2013). Organometallics 32: 6891–6894. 73 Thai, T.-T., Therrien, B., and Süss-Fink, G. (2009). J. Organomet. Chem. 694: 3973–3981. 74 (a) Vogt, M., Gargir, M., Iron, M.A. et al. (2012). Chem.-Eur. J. 18: 9194–9197.(b) Huff, C.A., Kampf, J.W., and Sanford, M.S. (2012). Organometallics 31: 4643–4645. 75 Filonenko, G.A., Hensen, E.J.M., and Pidko, E.A. (2014). Catal. Sci. Technol. 4: 3474–3485. 76 Filonenko, G.A., Smykowski, D., Szyja, B.M. et al. (2015). ACS Catal. 5: 1145–1154. 77 Kothandaraman, J., Czaun, M., Goeppert, A. et al. (2015). ChemSusChem 8: 1442–1451. 78 Rohmann, K., Kothe, J., Haenel, M.W. et al. (2016). Angew. Chem. Int. Ed. 55: 8966–8969. 79 Sahoo, A.R., Jiang, F., Bruneau, C. et al. (2017). Catal. Sci. Technol. 7: 3492–3498. 80 Dai, Z., Luo, Q., Cong, H. et al. (2017). New J. Chem. 41: 3055–3060. 81 Guan, C., Pan, Y., Ang, E.P.L. et al. (2018). Green Chem. 20: 4201–4205. 82 (a) Angermund, K., Baumann, W., Dinjus, E. et al. (1997). Chem.-Eur. J. 3: 755–764. (b) Tsai, J.C. and Nicholas, K.M. (1992). J. Am. Chem. Soc. 114: 5117–5124. (c) Hutschka, F., Dedieu, A., Eichberger, M. et al. (1997). J. Am. Chem. Soc. 119: 4432–4443. (d) Leitner, W., Dinjus, E., and Gaßner, F. (1994). J. Organomet. Chem. 475: 257–266. (e) Fornika, R., Görls, H., Seemann, B., and Leitner, W. (1995). J. Chem. Soc., Chem. Commun.: 1479–1481.

51

52

2  Homogeneously Catalyzed CO2 Hydrogenation to Formic Acid/Formate

83 Zou, F., Cole, J.M., Jones, T.G.J., and Jiang, L. (2012). Appl. Organomet. Chem. 26: 546–549. 84 Li, Y.-N., He, L.-N., Liu, A.-H. et al. (2013). Green Chem. 15: 2825–2829. 85 Jens, C.M., Scott, M., Liebergesell, B. et al. (2019). Adv. Synth. Catal. 361: 307–316. 86 Lilio, A.M., Reineke, M.H., Moore, C.E. et al. (2015). J. Am. Chem. Soc. 137: 8251–8260. 87 Laureanti, J.A., Buchko, G.W., Katipamula, S. et al. (2019). ACS Catal. 9: 620–625. 88 Fidalgo, J., Ruiz-Castaneda, M., Garcia-Herbosa, G. et al. (2018). Inorg. Chem. 57: 14186–14198.

53

3 Homogeneously Catalyzed CO2 Hydrogenation to Formic Acid/Formate with Non-precious Metal Catalysts Luca Gonsalvi, Antonella Guerriero, and Sylwia Kostera Consiglio Nazionale delle Ricerche (CNR), Istituto di Chimica dei Composti Organometallici (ICCOM), Via Madonna del Piano 10, Sesto Fiorentino, 50019, Firenze, Italy

3.1 ­Introduction One of the emerging challenges for science and industry in the 21st century is the supply of abundant, safe, and readily available raw materials for the synthesis of bulk chemical and commodities. As an example, their cost represents c. 30% of the gross income of German chemical companies, together with the cost of energy used for the manufacturing processes [1]. To be truly sustainable, the use of renewable feedstocks and more environment-friendly processes must however be combined with the curbing of greenhouse gas (GHG) emissions, among which CO2 is certainly one of the most well-known contributors to climate change and global warming, passing the unwelcome limit of 411 ppm in the atmosphere in March 2019 [2]. A possible contribution, meeting both requirements, is to use CO2 as valuable, renewable, and abundant C1 feedstock for chemical synthesis, provided it can be captured from emission sites in concentrated form, purified and utilized efficiently for selected, cost-intensive chemical production, targeting the highest atom efficiency. Ideal processes for CO2 valorization would imply the use of renewable energy, coming from sources such as hydroelectrical, geothermal, solar, or wind power. One of the major obstacles for the efficient use of CO2 as C1 building block is related to the high thermodynamic stability, with a free enthalpy of formation ΔGf0 = −394.38 kJ mol−1, but this can be counterbalanced by the formation of strong O─H and C─H bonds in catalytic reduction processes. CO2 can be reduced to a number of different chemicals, by 2- to 8-electron processes, from formic acid (HCOOH) to formaldehyde (HCHO), methanol (CH3OH), dimethyl ether (CH3OCH3), methane (CH4), and higher hydrocarbons in the presence of suitable homogeneous and heterogeneous catalysts. A number of excellent review articles and books [3] have summarized the results obtained by world leading groups active in the field during the years. Among the possible reducing agents, hydrogen (H2) is one of the preferred choices, as it conveys high atom efficiency without generating waste byproducts, except water in some cases [4]. The main drawbacks to be considered are the fact that hydrogen is still obtained from fossil feedstocks, and that it has low solubility in CO2 Hydrogenation Catalysis, First Edition. Edited by Yuichiro Himeda. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.

54

3 CO2 Hydrogenation to Formic Acid/Formate with Non-precious Metal Catalysts

most solvents, in turn requiring high pressures causing increased costs and safety concerns. The growth of use of renewable energy sources will decrease the cost of hydrogen production in future, making it more attractive for large-scale utilization from the economic point of view, in turn allowing for a higher viability of its use as reducing agent for CO2. The most atom-efficient (100%) reaction between CO2 and H2 (1  :  1 ratio) gives HCOOH. The reaction is exothermic in the gas phase (ΔH0 = −31.5 kJ mol−1), but has a strongly endergonic character due to the large entropic contribution (ΔS0 = −215 kJ mol−1). The reaction can be made exoergonic using strong bases (i.e. ΔG0 = −9.5 kJ mol−1 in the presence of aqueous NH3), or in adequate solvents such as water. On the other hand, the need to overcome a high kinetic barrier requires the use of suitable, efficient catalysts. As such, HCOOH is a valuable chemical product, finding various large-scale direct applications, such as in agriculture for silage, as preservative in food, as strong acid in wood pulping, leather, and textile industries, and in organic synthesis as acid, formylating or reducing agent, in alcohols esterification, amines formylation, etc. [5]. Formate salts find use in road de-icing, as additives in electroplating and in concretes and as reducing agents for unsaturated functional groups in organic molecules by transfer hydrogenation [4a, 6]. The production of HCOOH reached 770 000 ton yr−1 in 2014 [7] and the demand shows a steady annual increase, especially in the Asian market. It is produced in industry by various methods, the principal one being a two-step, fossil feedstock-based process, involving CH3OH carbonylation in the presence of NaOCH3, followed by hydrolysis of HCOOCH3 so obtained, with high costs of purification of the final product by distillation [5]. Further to its direct use, HCOOH is receiving growing attention as potential hydrogen storage material. In the last two decades, a renewed interest for sustainable hydrogen production and storage methods has stimulated scientists to investigate the use of liquid organic hydrogen-rich molecules (LOHCs), including HCOOH, as suitable candidates for this purpose. The selective decarboxylation versus decarbonylation of HCOOH gives a mixture of CO2 and H2 (1 : 1 ratio) that can be purified from the CO2 component and used in proton exchange membrane (PEM) fuel cells. Thus, the quest for novel, efficient, and selective homogeneous catalysts for HCOOH dehydrogenation (H2 release step) and its reverse reaction (CO2 hydrogenation to HCOOH, H2 storage step) has witnessed a true renaissance in the last decade [8]. Homogeneous CO2 hydrogenation to HCOOH has been achieved using various transition metals and ligand combinations. The most outstanding results, usually measured as turnover numbers (TONs) and turnover frequencies (TOF, h−1) and accounting for the state of the art for this reaction, were obtained using platinum group metals (PGMs) such as Ru, Rh, Ir in combination with organophosphines, bipyridines, cyclic diimines, and pincer-type PNP ligands, as excellently reviewed elsewhere [4a, 9]. The low abundance [10] of such metals in the Earth’s crust (Figure 3.1) and the need for more sustainable, economically viable alternatives, has spurred chemists to find alternatives to PGM-based catalysts, and increasing attention has been given recently to the use of first-row 3d metals such as Fe, Co, Ni, Cu, and Mn. As pointed out by Guan  [11], some features of earth-abundant metals make them in general less suitable than PGM counterparts in homogeneous catalyzed reactions, namely: (i) the general lower thermal stability and (ii) the tendency to generate paramagnetic species. To circumvent these problems, a possible solution is to find ligands that not only can

3.2  Iron-Catalyzed CO2 Hydrogenation VI

VII

VIII

IX

X

XI

Cr

Mn

Fe

Co

Ni

Cu

126

716

43200

24

56

25

Mo

Tc

1.1

Ru

Rh

Pd

Ag

0.0001

6 × 10–5

0.0004

0.07

W

Re

Os

Ir

Pt

Au

1.0

0.0004

5 × 10–5

5 × 10–5

0.0004

0.0025

Figure 3.1  Abundance (ppm) of mid- and late transition metals in Earth’s crust. Source: From Alig et al. [10], © 2018 American Chemical Society. Reprinted with permission of American Chemical Society.

bind strongly to the metals but also promote precious metal-like reactivity. Thus, suggested strategies may involve: (i) to achieve strong ligand chelation to metals, to increase the catalyst stability; (ii) to introduce ancillary co-ligands such as hydride, alkyl/aryl groups, and CO that, providing a strong ligand field, favor low-spin states and diamagnetic species; (iii) to use redox-active or proton-responsive ligands that store and release electrons or protons readily and in close proximity of the reaction center, to avoid in turn the need to change the oxidation states of the metals; (iv) to use ligands that may activate the substrates, such as CO2 and/or H2, via secondary coordination sphere mechanisms [12]. In this chapter, the main results described in the literature for CO2 hydrogenation (limited to H2 as reducing agent) to formic acid and formates using nonprecious transition metals, will be summarized, with particular attention to the class of ligands and their implications in the catalytic mechanisms.

3.2 ­Iron-Catalyzed CO2 Hydrogenation Iron is the most earth-abundant transition metal, and has been widely applied in homogeneous and heterogeneous catalysis for many types of processes. After a first report by Evans and Newell in 1978 [13], describing the synthesis of methyl formate using [HFe3(CO)11]−, in 2003 Jessop and coworkers tested a series of homogeneous catalysts for CO2 hydrogenation to formate using high-pressure combinatorial screening [14]. Metal salts such as FeCl3, MoCl3, and NiCl2 were tested as in situ catalysts formed adding phosphines such as dppe [1,2-bis(diphenylphosphino)ethane], dcpe [1,2-bis(dicyclohexylphosphino)ethane], and PPh3 in the presence of bases such as TMEDA [N,N,N′,N′-tetramethylethylenediamine], bipy (2,2′-bipyridyl), and DBU (1,8-diazabicyclo[5.4.0]undec-7-ene) in DMSO. The best results with Fe were obtained using FeCl3 (15.0 μmol) with dcpe (22.5 μmol), DBU (3.3 mmol), under a total pressure of 100 bar of CO2/H2, at 50 °C for 7.5 hours, reaching TON  =  113 and TOF  =  15.1 h−1. After a few years, during which PGM-based catalysts dominated the field of CO2 hydrogenation, Fe was reconsidered as a promising transition metal due to its abundance and cheapness. Essentially, two classes of catalysts were shown to be active, based, namely on (i) non-pincer type and (ii) pincer-type ligands, and the major contributions will be here described.

55

56

3 CO2 Hydrogenation to Formic Acid/Formate with Non-precious Metal Catalysts

3.2.1  Non-pincer-Type Iron Complexes The first well-defined Fe-phosphine catalyst for the reduction of CO2 and NaHCO3 to formates was described by Beller, Laurenczy, and coworkers in 2010 [15]. The authors studied the effect of the combination of Fe(BF4)2·6H2O with different bi-, tri-, and tetradentate phosphines and aminophosphines in catalysis. The most efficient system was obtained using the tetradentate ligand tris[2-(diphenylphosphino)ethyl]phosphine (PP3). In the presence of 0.14 mol% of catalyst, NaHCO3 was hydrogenated to NaHCO2 (88% yield) in CH3OH at 80 °C, 60 bar H2, 20 hours, reaching TON = 610, close to the value (624) obtained with the precious-metal precursor [{RuCl2(benzene)}2] and diphosphines under the same conditions. The catalyst was also active for the hydrogenation of CO2 to HCO2Me and DMF (dimethylformamide) in the presence of added NEt3 at 100 °C, 90 bar total pressure (CO2/ H2  =  1  :  2), 20 hours, with TONs ranging from 245 to 727. The well-defined complex [FeH(PP3)](BF4) (1) was obtained by reaction of Fe(BF4)2·6H2O and PP3 under 10 bar of H2, in the presence of base in THF, and mechanistic details showed that this species is the active form of the catalyst that activates CO2 in the proposed catalytic cycle. The principal steps involve CO2 coordination to Fe followed by insertion in the Fe─H bond, then coordination of H2 to the metal center to give [Fe(η2-H2)(OCHO)(PP3)](BF4), and finally release of HCOOH to regenerate 1 (Scheme 3.1). Byers and coworkers [16] tested the effect of additives, in particular KHCO3, on the system obtained in situ using either FeCl2 or Fe(BF4)2·6H2O and PP3. It was observed that, using NBu3 as base, under 90 bar total pressure (CO2/H2 = 1 : 2), 100 °C, 21 hours, whereas a modest improvement in TON (from 124 to 187) was reached in the case of Fe(BF4)2·6H2O, a more pronounced effect was noticed for FeCl2 (from 142 to 256). By comparison with other added salts, the authors attributed the role of the additive not to its basicity, but rather to its structure, favoring the formation of a bimetallic resting state that would prevent catalyst decomposition, as shown in Scheme 3.2. NMR studies with M = Ru(η6-p-cymene) substantiated this hypothesis. A significant improvement on the Fe/PP3 catalytic system was later demonstrated by Beller and coworkers, who introduced structural modifications on the tetradentate phosphine ligand  [17]. The ligand tris(2-(diphenylphosphino)phenyl)phosphine (PPhP3) was obtained in a one-pot reaction improving on the previous literature procedure, and used to coordinate Fe(II) from precursor Fe(BF4)2·6H2O, obtaining complex [FeF(PPhP3)](BF4) (2). Complex 2 has a coordinated fluoride ligand (validated in the solid state by X-ray crystal structure determination) derived from abstraction from one BF4− anions of the metal salt, liberating BF3. In the study, different temperature, pressure, and catalyst concentration conditions were tested for NaHCO3 and CO2 hydrogenation. Under optimized conditions (CH3OH, 80 °C, 60 bar H2, 20 hours), the catalyst (0.05 mol%) formed in situ by adding Fe(BF4)2·6H2O and PPhP3 (1 : 1) gave formate yields up to 86% with TON = 1597. By using the well-defined complex 2, similar results were obtained. Further optimization (catalyst loading of 0.01 mol%, 100 °C) gave a higher TON = 7546, albeit with a formate yield of 77%, showing a better thermal stability of the improved catalyst 2 compared to 1. CO2 hydrogenation was also tested in CH3OH in the presence of amines. In the case of NMe3, a c. 1 : 1 mixture of HCOOH and HCOOMe was obtained using 60 bar total pressure (CO2/ H2  =  1  :  1), 100 °C, 20 hours. In the case of dialkylamines as base, formamides were

3.2  Iron-Catalyzed CO2 Hydrogenation

Ph2P

Fe(BF4)2·6H2O

+

H

H

Fe

H

H2, base

THF, PP3

PPh2

Ph2P

– H2

P

+ H2 HCOOH

+

H Ph2P Fe Ph2P

CO2

PPh2

P 1

O H

H

Fe

Ph2P H Ph2P

PPh2

Ph2P

+

CO2

+

O

Fe

H PPh2

P

P O H Ph2P H2

+

O Fe

PPh2

Ph2P P

Scheme 3.1  Proposed catalytic cycle for CO2 hydrogenation in the presence of 1. Source: From Federsel et al. [15], © 2010 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

obtained, with a maximum TON > 5100 for DMF in 74% yield. By NMR measurements, the authors were able to propose a catalytic mechanism, as expected rather similar to the one previously described for 1 (Scheme 3.3). A series of SiP3, BP3, NP3, and CP3 ligands, structurally related to PP3 and PPhP3, were studied by Fong and Peters  [18] to synthesize tris(phosphino) iron species (Figure  3.2), namely [FeH(L)(SiPR3)] (SiPR3 = [Si(o-C6H4PR2)3]−, R = iPr, 3a or Ph, 3b, L = H2, or N2), [Fe(H)3(PMe3)(PhBPiPr3)] (4, PhBPiPr3  =  PhB(CH2PiPr2)3), [FeH(N2)(NPiPr3)](PF6) (5, NPiPr3  =  N(CH2CH2PiPr2)3), [FeH(μ-H)(L)(TPB)] (L  =  N2, 6a or H2, 6b; TPB  =  B(oC6H4PiPr3)3), [FeCl(CPiPr3)] (7, CPiPr3  =  [C(o-C6H4PiPr2)3]−), and [FeCl(CSiPPh3)] (8, CSiPPh3 = [C(Si(CH3)2CH2PPh2)3]−). These ligands feature a different apical atom that may or may not be involved in the coordination to Fe (as the boron-based ones), and impart different geometries and formal oxidation states at iron. The stoichiometric reactivity with H2 and CO2 to give Fe-coordinated

57

58

3 CO2 Hydrogenation to Formic Acid/Formate with Non-precious Metal Catalysts

KO2X

[M]

O

[M]

Cl

O

[M]

O

[M]

X

O

Cl

X

O

X

O

bimetallic resting state X = CO, NO, CCH3, PO2 +

[M]

[NBu3H][HCO2]

O –

O

H2

X

CO2

[M] H

O –

O

X

NBu3H

+

H

+

[M]

O –

H

O

X

NBu3

Scheme 3.2  Proposed mechanism for the role of additives in TON enhancement using [M]Cl2, including [M] = [Fe(PP3)]. Source: From Drake et al. [16], © 2013 American Chemical Society. Reprinted with permission of American Chemical Society.

formate was studied, together with the determination of the Fe─H bond hydricity for 3aH2 that was measured as 54.3 ± 0.9 kcal mol−1 in CH3CN. Under standardized reaction conditions (0.1 mol% catalyst, 29 bar CO2, 29 bar H2, CH3OH, Et3N as base, 100 °C, 20 hours), starting from the chloride analogues as pre-catalysts, complexes 3a, 3b, 6, and 7 showed to be moderately active catalysts giving mixtures of (Et3NH)(HCO2) and HCOOMe, with 3b reaching the highest TON of 200. The authors proposed a catalytic cycle (Scheme  3.4) involving 3a that mirrors Beller’s proposal with 2, albeit in the case of 3a the formation of dihydride complexes was considered less likely to occur due to the system high basicity. A linear tetraphosphine, namely 1,1,4,7,10,10-hexaphenyl-1,4,7,10-tetraphosphadecane (tetraphos-1, P4), was used as rac and meso isomers (Figure 3.3) for the synthesis of both molecularly defined and in situ Fe(II) complexes that were applied for NaHCO3 hydrogenation and HCOOH dehydrogenation [19]. Similar to Beller’s PP3 and PPhP3 systems, starting from precursor Fe(BF4)2·6H2O, complexes [FeF(rac-P4)](BF4) (9a) and [FeF(meso-P4)] (BF4) (9b) were formed in situ upon mixing the reagents in propylene carbonate or THF. Starting from precursor [Fe(MeCN)6](BF4)2, complex cis-[Fe(MeCN)2(rac-P4)](BF4)2 (10a) and a 2 : 1 mixture of trans- and cis-[Fe(MeCN)2(meso-P4)](BF4)2 (11) were obtained from reaction in MeCN. After a full screening of reaction parameters, under optimized conditions, i.e. using catalyst (0.01 mmol); NaHCO3 (10 mmol), CH3OH (20 ml), 80 °C, 30 bar H2, 24 hours, in situ–formed complex 9a gave NaHCO2 in 62% yield with a TON of 620. Comparison between 9a and 9b (under 60 bar H2, 80 °C) showed higher activity for the

3.2  Iron-Catalyzed CO2 Hydrogenation +

F P

P

Fe

P P

+

L P

2

Fe

P

P

2 H2

P

80 °C HF

H HCOOH·base

P

BaseH+, H2

Fe

+L +

H

–L

L = CO2 or THF Base

H P

P

H

BaseH+

P

O H

H P

O P

Fe

H

P = PPh2

H P

P P

P

P

Fe

P

CO2

Scheme 3.3  Proposed catalytic cycle for CO2 hydrogenation in the presence of 2. Source: From Ziebart et al. [17], © 2012 American Chemical Society. Reprinted with permission of American Chemical Society.

former complex (TON = 154 versus 62, respectively). In the case of the well-defined complex 10a, higher TON = 1229 could be obtained using 0.001 mmol of catalyst under 60 bar H2. This behavior was explained by the higher stability of 10a compared to 9a under reaction conditions, and that rac-P4 would favor a cis-geometry for vacant coordination sites, needed to allow coordination and insertion of incoming substrates. The catalytic mechanism (Figure  3.4) involving 9a was later studied by DFT calculations [20]. It was shown that two competing mechanisms are possible. In path A, the in situ formed [FeH(rac-P4)]+ (12) at first coordinates H2 to give [Fe(η2-H2)H(rac-P4)]+ (13) that

59

60

3 CO2 Hydrogenation to Formic Acid/Formate with Non-precious Metal Catalysts L R2P R2P

Fe

PMe3 H H H Fe

H PR2

Si

P P

P

Fe

P

B

PF6

N2

P

N

H P

Ph R = iPr, 3a; Ph, 3b, L = H2, or N2

4, P = PiPr2

L P

Fe

P B

Cl

H

P Fe

H P

L = N2, 6a or H2, 6b; P = PiPr2

5, P = PiPr2

Cl P

P

P P

C

7, P = PiPr2

Fe

P

Si C Si Si

8, P = PPh2

Figure 3.2  Fong and Peters’ iron complexes 3–8 bearing tri- and tetradentate P3-Element (B, C, N, Si) ligands. Source: From Fong and Peters [18], © 2014 American Chemical Society. Reprinted with permission of American Chemical Society.

interacts with HCO3− to give [Fe(H)2(rac-P4)] (14) with CO2 and H2O. The following steps involve insertion of CO2 in the Fe─H bond and elimination of formate to give back 12. This path has a highest barrier of 25.3 kcal mol−1, corresponding to the deprotonation of the H2 ligand in 13 to give 14. In path B, at first two HCO3− molecules disproportionate to (CO2 + CO32− + H2O); next, CO2 coordinates to 12, then insertion of CO2 in the Fe─H bond follows, to generate the Fe formate complex [Fe(O2CH)(rac-P4)]+ (15). The following step, hydrogen coordination to Fe to give complex [Fe(η2-H2)(O2CH)(rac-P4)]+, is the rate-determining step of the mechanism. The free energy barrier is 22.8 kcal mol−1, considering 13 as the resting state. Intramolecular deprotonation of the Fe-coordinated dihydrogen molecule occurs next, resulting in [FeH(HO2CH)(rac-P4)]+, then HCOOH loses a proton to the solution and dissociates the formate to regenerate 12. Path B seems preferred having the lowest overall barrier. In terms of free energies, in both pathways, the overall reaction is exergonic with a reaction-free energy of −5.5 kcal mol−1, in good agreement with the experimental data. Among phosphine-free catalysts, the air- and moisture-stable Knölker’s complex tricarbonyliron[η5-(1,3-bis(trimethylsilyl)-4,5,6,7-tetrahydro-2H-inden-2-one)] (16a) was tested by Yang, Zhou, and coworkers for NaHCO3 and CO2 hydrogenation to formate, together with related complexes (16b–g) derived from arene structural modifications (Figure 3.5) [21]. A maximum TON of 447 was obtained (44.7% formate yield) using NaHCO3 (3 mol), 16a (0.003 mol), EtOH/H2O, 120 °C, 30 bar H2, 24 hours. All related complexes 16b–g showed lower or no activity. Interestingly, CO2 was hydrogenated only after previous addition of

3.2  Iron-Catalyzed CO2 Hydrogenation Cl

(Et3NH)(HCO2)

P

Fe

H2

P

P

(Et3NH)Cl

Si 3a

O H

O P

Fe

(Et3NH)(HCO2)

H2, (Et3NH)Cl

H P

P

P

+

H Fe

P

P Si

(Et3NH)(HCO2)

H2 (Et3NH)Cl

Si

3a-H2

P = PiPr2

H P

H2

H Fe

H2 Et3N H P

P

CO2

Si

(Et3NH)Cl

Scheme 3.4  Proposed catalytic cycle for CO2 hydrogenation in the presence of 3a. Source: From Fong and Peters [18], © 2014 American Chemical Society. Reprinted with permission of American Chemical Society.

Ph

Ph P

P

Ph

Ph P

P Ph

Ph

P

Ph

Ph

P

P

Ph

Ph P Ph

meso-P4

BF4

Fe F

P P

P

(BF4)2

P

NCMe

Fe

NCMe

P

10a

9a

rac-P4

Ph

P P

P

P P Fe P F P 9b

BF4

NCMe P P

Fe

(BF4)2

NCMe

P

P

P

P

Fe P

(BF4)2

P NCMe

NCMe trans-10b

cis-β-10b

Figure 3.3  Fe(II) complexes 9 and 10 bearing the tetradentate phosphine 1,1,4,7,10,10-hexaphenyl-1,4,7,10-tetraphosphadecane. Source: From Bertini et al. [19], © 2015 American Chemical Society. Reprinted with permission of American Chemical Society.

61

Ph TS(6–7) m=1

P

Ph

+

Ph P

Ph2P Fe O

Ph

P

O

O

O TS(7–8) + H2 + H2O + CO32–

H

+

Ph

P

O

H TS(6–7) + H2O + CO32– + H2

Ph

P

PPh2

Ph2P Fe

PPh2

Ph2P Fe

PPh2

+

Ph

P

P

O O

H

H TS(8–9) + H2 + H2O + CO32–

P

TS(9–10) + H2O + CO32–

PPh2

H

O

H TS(9–10) H m=1

O H

H

+

Ph

P

Ph2P Fe

PPh2

Ph2P Fe

O

Ph

+

Ph

P

TS(10–11) m=1

TS(10–11) + H2O + CO32–

12.5 9.2 6.4

4.9

3.9 m=3

2.5

m=1

0.0

6.6

1.8

20.9

1.8

0.6 –2.1

–5.5 –7.9

–8.4

– 1 + 2(HCO3 ) + H2

Ph

Ph

P

P

Ph2P Fe H 1 m=1

PPh2

7 + H2O + CO32– + H2

6 + H 2O + CO32– + H2 +

Ph

Ph

P

P

Ph2P Fe O C

+

Ph

Ph

P

P

PPh2 Ph2P Fe

H 6 m=1

O

+

Ph

P

Ph

7 m=1

O O

+

PPh2

Ph

8 m=1

+

Ph

P

Ph2P Fe O

H

Ph

P

Ph

P

O H 9 m=1

+

Ph

P

PPh2 Ph2P Fe

O H

11 + H2O + CO32–

10 + H2O + CO32–

+ H2

P

PPh2 Ph2P Fe

H

O

2–

9 + H2O + CO3

8 + H2O + CO32– + H2

H

PPh2

Ph

PPh2

OH

11 m=1

P

Ph P

Ph2P Fe

1 + H2O + HCO3– + HCO2– Ph

Ph

P

PPh2

H

H

O 5 m=1

Figure 3.4  NaHCO3 hydrogenation to formate by [FeH(rac-P4)]+ (12), path B. Source: From Marcos et al. [20], © 2018 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

+

P

PPh2 Ph2P Fe

O H

O H H

10 m=1

+

P

Ph2P Fe

H

O

Ph

P

5 + H2O + HCO3–

1 m=1

3.2  Iron-Catalyzed CO2 Hydrogenation

R

SiMe3 O

Ph

Ph

O

O

X

Ph OC

Fe

R CO

CO 16a, R = SiMe3 16b, R = SiEt3 16c, R = SiMe2iPr 16d, R = SiMe2tBu

OC

Fe

SiMe3 CO

CO 16e, X = CH2 16f, X = O

OC

Fe

Ph CO

CO 16g

Figure 3.5  Fe catalysts derived from Knölker’s complex motif. Source: From Zhu et al. [21], © 2018 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

NaOH (3 mmol) to the reaction mixture, via CO2 conversion to HCO3−, which then was reduced to formate (TON = 30.7) using 30 bar H2, 20 bar CO2, EtOH/H2O as solvent mixture, 120 °C, 24 hours. In the same year, Poater, Renaud and coworkers [22] applied complex 16a for NaHCO3 reduction to formate under slightly different conditions, i.e. with a substrate/catalyst ratio of 500, with CH3OH as solvent, 50 bar H2, 100 °C, 1 equiv Me3NO to Fe, 20 hours, obtaining a modest conversion (6.1%, TON = 30). Better results were obtained with a structurally modified arene ligand, namely N,N′-dimethyl-3,4-ethylenediaminosubstituted, bis(diphenyl) cyclopentadienone (17). Without the need of Me3NO as carbonyl scavenger, using a substrate/catalyst ratio of 1000 in DMSO/H2O (1 : 1) solvent mixture, 99.6% conversion was observed (TON = 996) under otherwise identical conditions. TON could be increased up to 1246 using a substrate/catalyst ratio of 10 000, albeit the conversion dropped to 12.5%. Later on, the same complexes were applied for CO2 hydrogenation in the presence of a base by Daturi and Renaud  [23]. Base screening showed that DBU (1,5-diazabicyclo[5.4.0]undec-5-ene) and TMG (1,1,3,3-tetramethylguanidine) efficiently promoted the reaction in the presence of 0.2 mol% of catalyst at 80–100 °C, 20 hours, with the addition of Me3NO as carbonyl scavenger (1 : 1 to Fe). Quantitative formate yield was obtained with DBU using 0.2 mol% of 17 at 80 or 100 °C (TON = 500), while the highest TON was achieved using TMG, 0.02 mol% of 17, 100 °C (yield = 34.5%). With the same base, catalyst loading and other reaction conditions, by addition of a chromium-based metalorganic-framework (MOF) such as MIL-52(Cr) as CO2-sequestration agent (0.05 mol%), TON raised to 3006 with a formate yield of 60.1%. The effect was less remarkable, as expected, using NaHCO3 as substrate. A TON of 1246 was achieved without MOF, and only a slight increase (TON = 1525) was observed upon its addition.

3.2.2  Pincer-Type Iron Complexes The use of pincer-type transition metal complexes in catalysis is well established for a large family of processes, ranging from C-element unsaturated bond hydrogenation, to C–C coupling reactions, etc. [24]. This class of ligands features high versatility, easier tunability of electronic and donor properties than multidentate phosphines, exclusive tridentate

63

64

3 CO2 Hydrogenation to Formic Acid/Formate with Non-precious Metal Catalysts

meridional coordination to transition metals, and the possibility for the ligand backbone to act as non-innocent redox or proton-responsive site that can be exploited in metal-ligand cooperation (MLC) mechanisms of substrate activation in catalysis [25]. Examples of successful application of pincer-type complexes for CO2 hydrogenation to formate include the state-of-the-art iridium(III) trihydride PNP-pincer complex developed by Nozaki and coworkers  [26], reaching the outstanding TON  =  3 500 000 and TOF  =  150 000 h−1. These results inspired other research groups to apply pincer-type complexes to CO2 hydrogenation, including those based on earth-abundant metals. In 2011, Milstein and coworkers reported on the first example of Fe-PNP complexes able to catalyze CO2 hydrogenation under low pressure, exhibiting noble metal activities [27]. The authors focused their attention on the dihydridocarbonyl complex trans-[Fe(H)2(CO) (tBuPNP)] (18, tBuPNP  =  2,6-bis(di-tert-butylphosphinomethyl)pyridine), at first for the hydrogenation of NaHCO3 to NaHCO2. Using 5 mmol of substrate, 0.005 mmol of 18 and a low H2 pressure (8.3 bar), the highest TOF = 320 was obtained at 100 °C for 16 hours. Next, CO2 hydrogenation under basic conditions (NaOH) was studied, and under optimized conditions (80 °C, five hours, NaOH 2 M) a mixture of CO2/H2 (1 : 2, total pressure c. 10 bar) was successfully reduced to NaHCO2 in H2O/THF (10 : 1) using 0.1 mol% of 18, reaching a maximum TON of 788 and TOF of 156 h−1, rivalling with noble metal performances known at the time of publication. Reaction parameters screening showed that the maximum formate yield reached 53.2% under slightly different conditions (80 °C, 10 hours, NaOH 1 M). The catalytic mechanism (Scheme 3.5) was studied by NMR spectroscopy method. The first step involves insertion of CO2 into the Fe─H bond, then formate ligand displacement by H2O to give [FeH(H2O)(CO)(tBuPNP)]+. The water molecule is then replaced by incoming hydrogen to give [FeH(η2-H2)(CO)(tBuPNP)]+. The dihydrogen ligand is then deprotonated H PR2

H2O Fe

N

CO2

CO

PR2 H 18

R = tBu H

OH H

H

H

H

PR2 Fe

N

O

PR2 CO

or

N

PR2

CO + H2O

Fe

N

PR2 H

–

N

OH2

+

H

Fe

PR2 H

CO

H

+

H2O

PR2

PR2 OH

Fe

PR2 H

H

O PR2

N

CO H2O

H2

Fe

CO –

HCO2

PR2 H

Scheme 3.5  Proposed catalytic cycle for CO2 hydrogenation in the presence of 18. Source: From Langer et al. [27], © 2011 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

3.2  Iron-Catalyzed CO2 Hydrogenation

by OH− giving H2O and 18. As an alternative for the latter step, the authors propose a MLCtype pyridine ligand dearomatization and deprotonation on one methylene arm of the PNP ligand by OH−, giving the formally neutral complex [FeH(η2-H2)(CO)(tBuPNP*)] (tBuPNP* = dearomatized pincer ligand). In this case, 18 is regenerated by an intramolecular, ligand-assisted dihydrogen deprotonation step. The same research group explored ligand variations a few years later, replacing the pyridine in the tBuPNP ligand with pyrazole [28]. Complex [FeCl(H)(CO)(tBuPNzP)] (19, tBuPNzP  =  2,6-bis(di-tert-butylphosphinomethyl)pyrazole), with H trans to Cl, was used as pre-catalyst for NaHCO3 and CO2 hydrogenation to NaHCO2 in H2O/THF (10 : 1). The best TONs (149) were obtained with NaHCO3 using 0.1 mol% of 19, 6.5 bar of H2 and 45–65 °C for 16 hours, whereas TONs up to 388 were measured using CO2 under the conditions described above for 18, but using higher concentrations of NaOH (3 M) at 55 °C. Mechanistic studies showed that treatment of 19 with a strong base such as tBuOK led to pyrazole dearomatization. In THF, aggregation leads to the polynuclear species [Fe(H)(CO)(tBuPNzP*)]n (20, tBuPNzP*  =  dearomatized pincer ligand), where the free N atom of pyrazole binds intermolecularly to the vacant coordination site of the Fe center of a second complex molecule. Complex 20 was characterized in the solid state by X-ray single crystal data diffraction, showing a hexanuclear cyclic arrangement (n = 6). On the other hand, repeating this reaction under 1 bar of H2 gave the dihydride analogue of 19, namely [Fe(H)2(CO)(tBuPNzP)] (21). Interestingly, the reaction of 21 with CO2 gave as expected complex [Fe(H) (OCHO)(tBuPNzP)] (22), whereas in the case of 20, complex trans-[Fe(H)(CO)(tBuPNzPt Bu-COO)] (23) formed, derived from [1,3]-cycloaddition of CO2 to the exocyclic methine carbon of the deprotonated arm of the pincer ligand and the iron metal center. The authors describe a proposed mechanism (Scheme 3.6), highlighting the participation of the PNP ligand in a MLC mechanism of activation of CO2.

O

H

H PR2

N

N

Fe

PR2 Cl 19

CO

O PR2

tBuOK

N

–KCl

Fe

N

PR2

CO2 CO

N

– CO2

PR2

HCO2Na, H2O

20

R = tBu

CO

PR2

n

N

Fe

N

– H2

Cl

H2

23

H

NaOH O

O PR2

N

Fe

PR2

N

CO

PR2 22

H2

H

N

Fe

CO

CO2

PR2 H

H – CO2

21

CO2

Scheme 3.6  Proposed catalytic cycle for CO2 hydrogenation showing possible MLC contribution, starting from 19 as pre-catalyst. Source: From Rivada-Wheelagan et al. [28], © 2015 American Chemical Society. Reprinted with permission of American Chemical Society.

65

66

3 CO2 Hydrogenation to Formic Acid/Formate with Non-precious Metal Catalysts

A variation on the pyridine-based PNP pincer motif was obtained using the 2,6-(diaminopyridyl) scaffold, yielding P(N)N(N)P ligands. Here, the PR2 binding moieties are connected to the pyridine ring via N–H, N-alkyl, or N-aryl spacers. Gonsalvi, Kirchner and coworkers applied the iron hydridocarbonyl complex [FeBr(PNPH-iPr)H(CO)] (24) and the corresponding N–Me derivative, [FeBr(PNPMe-iPr)H(CO)] (25) as catalysts for CO2 and NaHCO3 hydrogenation [29]. In the case of NaHCO3 hydrogenation, the highest formate yield (98%) was obtained using 24 (0.05 mol%) in H2O/THF (4  :  1), at 80 °C, 90 bar H2, 24 hours, with a TON of 1964. The highest TON = 4560 was obtained using 0.005 mol% of catalyst, with a drop in formate yield to 23%. In the case of CO2 hydrogenation, solvent and base screening helped to determine the best conditions for efficient catalysis. Complex 24 gave the best activity (TON = 1220, yield = 98%) using 0.08 mol% of catalyst, H2O/THF (4  :  1), at 80 °C, 80 bar CO2/H2 (1  :  1), 21 hours. Under the same conditions, 25 showed comparable activity to 24 (TON = 1153, yield = 92%) but by changing the solvent to EtOH and using DBU as base. Further optimization showed that, with this combination, 25 could give CO2 hydrogenation either at low pressure (8.5 bar CO2/H2) and 80 °C, with TON = 480, formate yield = 48%, or at room temperature under 80 bar CO2/H2 (1 : 1), with TON = 856, formate yield  =  86%. By decreasing the catalyst loading to 0.01 mol%, excellent performances were still achieved (TON = 9840, formate yield = 98%), and TON could be increased to 10 275 by lowering the catalyst amount further to 0.005 mol%, albeit yield dropped to 21%. A catalytic mechanism was proposed, involving the formation of cis- and trans[Fe(H)2(PNPMe-iPr)(CO)] (26), CO2 insertion to give the hydridoformate complex [Fe(OCHO)(PNPMe-iPr)H(CO)] (27) followed by hydrogenolysis and base-assisted formate elimination. The role of the solvent (EtOH) was established by NMR studies, in promoting formate decoordination in 27 and stabilizing the highly reactive pentacoordinate cationic Fe(II) hydridocarbonyl species (Scheme 3.7). The highest barrier of the whole pathway was calculated by DFT methods as 14 kcal mol−1, corresponding to the step involving hydrogen coordination to the Fe center.

Fe

Me

N PR2

CO

CO2

Fe

N

–

HCOO

Me

–

Br

PR2

N

PR2

N N

H

Me

H

Me

HCOO

25

CO

N PR2 O 27

H O

H

Me

EtOH

Fe

N PR2 CO

N

CO

N PR2 H

PR2

N

Me

Fe

N PR2

–

EtOH

cis-26

H2/DBU

PR2

N

Me (DBUH)(HCO2)

H

Me N

H

EtOH

HCOO

CO

trans-26

PR2

N N Me

Me

H

Me

Fe

N

H

Me PR2

N

DBU, H2

Fe

CO

N PR2 O

H

Scheme 3.7  Proposed catalytic cycle for CO2 hydrogenation starting from 25. Source: From Bertini et al. [29], © 2016 American Chemical Society. Reprinted with permission of American Chemical Society.

3.2  Iron-Catalyzed CO2 Hydrogenation

A dramatic improvement in Fe-pincer catalyzed CO2 hydrogenation was disclosed in 2015 by Hazari, Bernskoetter and coworkers, introducing PNP ligands containing either a secondary or a tertiary amine in the backbone, and adding Lewis acid (LA) co-catalysts to the reaction mixture [30]. This class of complexes was independently developed in 2014 by Beller’s group for the selective hydrogenation of aromatic and aliphatic (di)nitriles [31] and esters [32], and by Schneider, Hazari, Bernskoetter and coworkers for HCOOH dehydrogenation  [33]. The five-coordinate iron(II) formate carbonyl hydride species [FeH(RPNP) (CO)] (R = iPr, 28a; Cy, 28b), derived from the deprotonation of the corresponding amine PNP ligand bis(2-(diisopropylphosphino)ethyl)amine, to give a formal amido moiety able to strongly bind to Fe, gave interesting results in CO2 hydrogenation. Using 69 bar of CO2/ H2 (1 : 1), 0.78 mmol of catalyst, THF, DBU (1500 equiv), LiBF4 (1 : 2 to DBU) at 80 °C, 24 hours, TONs > 1000 and yields in the range 67–82% were achieved. By switching to LiOTf (7.5  :  1 ratio to DBU; OTf  =  CF3SO3−) and increasing the quantity of base to 5000 equiv, TON raised to >3000 with 28b. The ligand was modified by replacing the secondary amine-bridging group to the corresponding N–Me tertiary amine, obtaining ligand 2-(dialkylphosphino)-N-(2-(dialkylphosphino)ethyl)-N-methylethanamine (RPNMeP; R  =  iPr, Cy) that were used to obtain the precatalysts [FeH(RPNMeP)(CO)(κ1-H-BH4)] (R = iPr, 29a; Cy, 29b). Complexes 29 readily afford the corresponding dihydrido counterparts [Fe(H)2(RPNMeP)(CO)] (R =  iPr, 30a; Cy, 30b) by reaction with NEt3. Complexes 29 were shown to be competent catalysts for CO2 hydrogenation under the conditions described above, reaching quantitative formate yields and TONs of 7660 (29a) and 6900 (29b). Further optimization (0.3 μmol of catalyst, DBU/Fe  =  79 600, DBU/LiOTf  =  5) allowed a significant increase of TON to c. 60 000 with 29a, setting the state-of-the-art for Fe-catalyzed CO2 hydrogenation to formate. Mechanistic studies demonstrated that, for the secondary amine-based complexes, formate extrusion is rate limiting, and the Li+ co-catalyst facilitates this step by disrupting an intramolecular hydrogen bond between the Fe-coordinated formate and the amide ligand (Scheme 3.8). On the other hand, for the RPNMeP-based catalysts, the authors proposed that Li+ facilitates the displacement of formate by incoming dihydrogen to generate the cationic complex [FeH(RPNMeP)(η2-H2)(CO)]+, which is then deprotonated by DBU to give back the active dihydride species (Scheme  3.9). Complexes 28a and 29a were later shown to be highly active for the Fe-catalyzed synthesis of a small library of formamides from CO2, H2, and amines, with the intermediate formation of (protonated amine) formate, reaching H

O H

E N

Li

Fe

CO E

Li

E

+

N

+

H

O

H

O

H

+

O

H CO

Fe H

E

DBU

H

O

(DBU)Li

E N

O Fe H

E

C O

E = PiPr2, PCy2

Scheme 3.8  Proposed mechanism for Li co-catalyst promoting effect starting from 28. Source: From Zhang et al. [30a], Licensed under CC by 3.0 unported. Public Domain.

67

68

3 CO2 Hydrogenation to Formic Acid/Formate with Non-precious Metal Catalysts H

O Me

E N

CO2

O Fe H

H Me

E N

B

H Fe H

29a

CO

HCO2Li

E

H Me

H CO

E

Li+, H2

DBU

E N

H Fe

– (DBU)(BH3) H

H H E

Me CO

N

Fe

E H

+

CO E

30a

E = PiPr2 DBUH+

DBU

Scheme 3.9  Proposed catalytic mechanism for Li-assisted CO2 hydrogenation with 29a. Source: From Zhang et al. [30], Licensed under CC by 3.0 unported.Public Domain.

TONs > 4500 using c. 69 bar total pressure, 0.02 mol% of catalyst, THF, 120 °C, and four hours reaction time [34]. In an attempt to improve even further the performance of this class of catalysts for CO2 hydrogenation, Bernskoetter, Hazari and coworkers introduced ancillary ligand modifications on 28 and 29, namely by replacing the CO ligand with isonitriles  [35]. Complexes [FeH(iPrPNP)(C≡NR)] (R = 2,6-dimethylphenyl, 31a; 4-methoxyphenyl, 31b) were obtained and tested under optimized conditions (0.3 μmol of catalyst, DBU/Fe  =  79 600, DBU/ LiOTf = 7.5, THF, 80 °C, 24 hours); however, modest TONs were observed (613 and 333, respectively) when compared to 28a (TON  =  6030). Next, complexes [FeH(iPrPNMeP) (C≡NR)(κ1-H-BH4)] (iPrPNMeP = MeN(CH2CH2PiPr2)2, R = 2,6-dimethylphenyl, 32a; tertbutyl, 32b; adamantyl, 32c) have been prepared and crystallographically characterized, and their activity was compared to that of 29a and 31a [36]. Under the conditions described above for 31, these catalysts gave better performances than the iPrPNP counterparts, i.e. reaching TONs of 5300 (32a), 1300 (32b), and 710 (32c) that were however much lower than the value observed with 29a under these conditions (42 000). The authors attributed this behavior to the lower stability of the isonitrile catalysts compared to the carbonyl analogues under reaction conditions. One of the challenges of sustainable, green syntheses starting from CO2 is to use this feedstock once recycled, purified, and captured from flue gases. Olah and Prakash described the screening of a series of amines to capture CO2 in water solutions, followed by Ru or Fe-PNP catalyzed hydrogenation of ammonium carbamate and/or carbonate directly to the corresponding formates  [37]. The best system, combining the highest CO2 absorption capacity and catalyst activity, was obtained with TMG as base, catalyst [RuH(PhPNHP) (CO)(κ1-H-BH4)] (also known as Ru-MACHO-BH), 50 bar of H2, 55 °C, 20 hours, giving a formate yield of 95% and TON of 7375. The bromide analogue of 28a, namely [FeBr(H)

3.3  Cobalt-Catalyzed CO2 Hydrogenation

(iPrPNHP)(CO)] (33), was also tested, using PEHA (pentaethylenehexamine), 80 bar of H2, 50 °C, 10 hours, showing however lower efficiency (yield = 53%, TON = 255), compared to the above-mentioned Ru catalyst (yield = 68%, TON = 325) under the same conditions. Catalyst recycling was also addressed, using a biphasic system H2O/2-MeTHF, where the products are confined in the water phase and the catalyst in the organic one (Figure 3.6). In the case of 33, using DABCO (diazabicyclo[2.2.2]octane), 50 bar H2 at 55 °C for 10 hours, over five consecutive hydrogenation cycles were obtained, with a slight loss in rate for each cycle (TOFs decreasing from 141 h−1 for the first run to 115 h−1 for the fifth run). Also in this case, the Ru catalyst proved superior by 4–5 times. The CO2 capture-hydrogenation-catalyst recycling system was further optimized by the same research group in 2018 [38]. Key to the improvement was the use of MOH (M = Na, K, Li, Cs aqueous solutions) to enhance capture in up to 25 mmol of CO2 per gram of base (M = Na). The best results were obtained with the Ru-PNP complex [RuCl(H)(tBuPNHP) (CO)] that was able to convert 13.8 mmol of captured CO2 in KOH/H2O/2-MeTHF in less than 30 minutes, with TON of 2710, TOF > 5420 h−1, and formate yield of 98%. Catalyst recyclability was very high (yield > 90% over five consecutive runs). Furthermore, the aqueous formate solution could be used to feed a direct formate fuel cell (DFFC) without purification and, at 10 mA constant current, a relatively steady voltage of about 0.7 V was observed for more than 13 hours. The Fe-PNP complex 33 was also tested with NaOH as base; however, the reaction time was raised to 379 minutes to obtain a 98% yield in HCO2Na, with TON = 2706 and TOF = 428 h−1.

3.3 ­Cobalt-Catalyzed CO2 Hydrogenation Although less investigated than iron counterparts, cobalt complexes have received some attention as homogeneous catalysts for bicarbonate and CO2 hydrogenation to formate.

96%

700 600

Ru-MACHO-BH 33

94% 97%

94%

94%

CO2

Catalyst

(Ru- or FePNP complex)

HCO3– +

HNR3 NR3

Me-THF 55 °C H2

Biphasic H2O

Catalyst

(Ru- or FePNP complex)

HCOOH

TOF (h–1)

500 400 300 200

HCO2–

96%

95%

100

+

HNR3

0

NR3

1

2

95%

3

96%

4

95%

5

Cycle

(a)

(b)

Figure 3.6  Catalyst recycling scheme (a) and observed TOFs (b) using Ru-MACHO-BH and 33. Source: From Kothandaraman et al. [37], © 2016 Royal Society of Chemistry. Reprinted with permission of Royal Society of Chemistry.

69

70

3 CO2 Hydrogenation to Formic Acid/Formate with Non-precious Metal Catalysts

Based on the previous work with Fe described above [15], Beller and coworkers reported the efficient combination of Co(II) precursors with PP3 to generate in situ an active catalyst for these reactions [39]. In the case of NaHCO3, using Co(BF4)2·6H2O, PP3, CH3OH, 80 °C, 60 bar H2, 20 hours, NaHCO2 was obtained in 94% yield and TON = 645. A temperature increase to 120 °C gave higher TON of 3877, with a yield of 71%. CO2 hydrogenation in CH3OH in the presence of NEt3 gave HCO2Me in up to 83% yield with a TON of 659 under a total pressure of 90 bar of CO2/H2 (1 : 2) at 100 °C. Formamides such as DMF (dimethylformamide) and formylpiperidene were also obtained in the presence of the corresponding amines, with high productivities (TON = 1308 and 1254, respectively). The well-defined Co(I) precatalyst [Co(η2-H2)(PP3)](BPh4) (34) was also synthesized and used instead of the in situ–formed catalyst, showing same activity for CO2 hydrogenation to methyl formate. The authors propose the cationic Co-dihydride [Co(H)2(PP3)]+ (35) as the active species in catalysis and that the key step of the mechanism is the insertion of CO2 in the Co─H bond. Although the description of catalysts able to reduce CO2 to products other than formates is out of the scope of this chapter, it is worth mentioning that subtle changes to the Co(BF4)2·6H2O/multidentate phosphine system gave very interesting results in the transformation of CO2 to CH3OH [40] and mixtures of different ratios of CH3OH, methyl formate, and dialkoxymethane ethers [41], as reported independently and almost at the same time by the groups of Beller and Klankermayer. Key to these active catalytic systems is the combination of Co(II) precursors such as Co(BF4)2·6H2O or Co(III) precursors such as Co(acac)3 (acac = acetylacetonate) with the tridentate ligand 1,1,1-tri(diphenylphosphino methyl)ethane (triphos) and its xylyl and tolyl analogues [41], in the presence of a Brønsted acid such as HBF4·Et2O or HNTf (Tf = trifluoromethanesulfonyl). The effect of aryl group modulation was particularly evident in the increase of TON in the synthesis of dimethoxymethane, reaching TON = 157 with the most electron-rich tolyl-substituted triphos derivative (TON = 92 for phenyl, 120 for xylyl). Muckerman, Himeda, Fujita and coworkers described the synthesis of water-soluble η5pentamethylcyclopentadienyl (Cp*) complexes bearing proton-responsive ligands such as 4,4′- and 6,6′-dihydroxy-2,2′-bypiridine (4DHBP and 6DHBP, respectively) [42]. This class of complexes can easily respond to pH variations, i.e. by deprotonating the hydroxyl functions on the bipyridine in neutral and basic solutions, and were extensively studied with noble metals such as Rh, Ir, and Ru for CO2 and NaHCO3 hydrogenation in water phase [43]. In the case of cobalt, complexes [Cp*Co(L)Cl](PF6) (L = 4DHBP, 36-Cl; 6DHBP, 37-Cl), [Cp*Co(L)(H2O)](PF6)2 (L = 4DHBP, 36-H2O; 6DHBP, 37-H2O), and their unsubstituted bipyridine (bipy) analogues, [Cp*Co(bipy)Cl](PF6) (38-Cl) and [Cp*Co(bipy)(H2O)](PF6)2 (L = 38-H2O), were synthesized and applied as catalysts for the hydrogenation of CO2 in aqueous solution of NaHCO3. The production of formate in concentrations ranging from 15 to 24 mM was reached with complexes 36-Cl and 36-H2O, with a maximum TOF of 39 h−1 at 80–100 °C, 40 bar CO2/H2 (1 : 1) within one to two hours, in NaHCO3 1-M water solutions. Lower activity was observed in the case of the 6DHBP complexes 37, due to lower thermal stability under catalytic conditions, while the bipy complexes 38 were inactive, as expected for systems lacking the proton-responsive moieties. In the same year, Linehan and coworkers disclosed results on highly active Co catalysts able to carry out CO2 hydrogenation under ambient conditions [44]. The concept developed by the authors was to match the catalyst’s hydride donor ability, i.e. the free energy

3.3  Cobalt-Catalyzed CO2 Hydrogenation

(ΔGH−) required to cleave a hydride from a M–H moiety, with the acidity (pKa) of the base needed for deprotonating the dihydride formed upon oxidative addition of H2 to the metal center. Starting from complex [CoH(dmpe)2] (39, dmpe  =  1,2-bis(dimethylphosphino) ethane) and using Verkade’s base (2,8,9-triisopropyl-2,5,8,9-tetraaza-1-phosphabicyclo[3.3.3]undecane, Vkd), a TOF as high as 74 000 was reached, corresponding to a TON of 9400, under very mild reaction conditions (0.040 mM catalyst concentration; 20 bar CO2/H2 1 : 1, 21 °C, THF), albeit a rather high amount of this exotic and expensive super-base (up to 740 mM) was needed. Importantly, 39 was shown to be a competent catalyst even at 1 bar of CO2/H2 (1  :  1). The simplified mechanism proposed by the authors is shown in Scheme 3.10. The key steps for the process were established as (i) CO2 insertion into the Co─H bond; (ii) elimination of formate to give a coordinatively unsaturated [Co(dmpe)2]+ intermediate; (iii) H2 oxidative addition to give cis-[Co(H)2(dmpe)2]+ (40); and iv) the deprotonation of 40 by Verkade’s base to regenerate 39. More in-depth analysis of the system, including reaction conditions optimization, catalyst deactivation pathways [45], and the study of the mechanism of hydride transfer to CO2 by DFT calculations [46], was later obtained. Interestingly, high CO2/H2 ratios (up to 5 : 1) caused the formation of undesired, inactive Co species such as [(μ-dmpe){Co(dmpe)}2]2+ and [Co(dmpe)2(CO)]+, the latter likely derived from competing RWGS (reverse water-gas shift) reaction that may be triggered under these conditions. DFT calculations showed that two pathways for CO2 hydrogenation are possible using 39, namely an associative one, where CO2 binds to the metal first, and a direct hydride transfer one, where the hydride is directly transferred from the metal center to CO2. The former mechanism is slightly favored by only 1.4 kcal mol−1, and the calculations results were also supported by kinetic isotope effects (KIE) experiments. H

VkdH+

P Co

P Me2

Me2 P

CO2 HCO2–

P Me2 39

Vkd

+

+ Me2 H P H Co P PMe2 Me2 P Me2

Me2 P Co P Me2

Me2 P P Me2

40

H2

Scheme 3.10  Proposed catalytic mechanism for CO2 hydrogenation with 39. Source: From Jeletic et al. [44], © 2013 American Chemical Society. Reprinted with permission of American Chemical Society.

71

72

3 CO2 Hydrogenation to Formic Acid/Formate with Non-precious Metal Catalysts

In an effort to avoid the need of large amounts of costly strong bases, the same group developed the Co(I) tetraphosphine complex [Co(L)(CH3CN)](BF4), (41, L = 1,5-diphenyl3,7-bis(diphenylphosphino)-propyl-1,5-diaza-3,7-diphosphacyclooctane) that proved to be an efficient pre-catalyst for CO2 hydrogenation to formate under ambient conditions [47]. In the presence of 0.6 M 2-tert-butyl-1,1,3,3-tetramethylguanidine (TMG) and 0.031 mM solution of 41 in CD3CN, 25 °C, 1.7 bar of CO2/H2 (1  :  1), an initial TOF of 150 h−1 was reached. Detailed studies revealed that the pendant amine arms do not act as proton relays as they are not basic enough to deprotonate the formed Co(III) dihydride trans-[Co(L) (H)2]+ (42), but rather help in disfavoring the formation of trigonal bipyramidal metal complexes. A possible catalytic cycle (Scheme 3.11), involving also the monohydrido complex [Co(L)H](BF4), (43) was proposed. Another efficient Co-based system was developed by Bernskoetter and coworkers [48]. Using ligand iPrPNP (iPrPNP = MeN[CH2CH2(PiPr2)]2) previously applied for Fe-catalyzed CO2 hydrogenation [30] with the appropriate Co precursor and synthetic pathways, complexes [CoCl(iPrPNP)] (44), [Co(iPrPNP)(CO)2]X (X  =  Cl, 45-Cl; BArF4, 45-BArF4, BArF4  =  B[2,5-(CF3)2C6H3]4), [CoH(iPrPNP)(CO)] (46), and [CoH(κ2-iPrPNP)(CO)2] (47) were obtained. Catalyst screening using 0.3 μmol catalyst, 24 mmol DBU, 4.8 mmol LiOTf in THF at 80 °C for 16 hours with a total pressure of 69 bar of CO2/H2 (1 : 1), showed that the highest TON of c. 10 000 was reached with 45-Cl, with a modest yield in formate (13%). Further optimization showed that changing solvent from THF to CH3CN and lowering the amount of LiOTf to 3.2 mmol gave TON enhancement to 18 000 (with TOF = 12 000 h−1), which was further increased to 29 000 running the tests at 45 °C (TOF = 5700 h−1). A proposed mechanism starting from 45 involves de-coordination of the N-donor of the iPrPNP Ph N H P

TMGH+ N

Ph

CO2

PPh 2 PPh 2

–

HCO2

–MeCN 43

Ph N H P N

Co

P H

TMG

Ph

H

+

H Co

P H

+

Ph N H P

PPh 2 PPh 2

Ph

N

P H

H 42

Co

PPh 2 PPh 2

NCMe MeCN

41

H2

Scheme 3.11  Proposed catalytic cycle for CO2 hydrogenation starting from 41. Source: From Burgess et al. [47], © 2017 American Chemical Society. Reprinted with permission of American Chemical Society.

3.4 ­Nickel-Catalyzed CO2 Hydrogenation

ligand upon coordination of H2, followed by dihydrogen ligand deprotonation by DBU to yield 47, which, in turn, undergoes CO2 insertion into the M─H bond to give [Co(OCHO) (κ2-iPrPNP)(CO)2] (48). Formate elimination from 48 regenerates 45 closing the catalytic cycle (Scheme 3.12).

3.4 ­Nickel-Catalyzed CO2 Hydrogenation Historically, the Ni(0) complex [Ni(dppe)2] (49, dppe  =  1,2-bis(diphenylphosphino) ethane)) was the first earth-abundant, first-row metal-based homogeneous catalyst tested for CO2 hydrogenation to formate in the presence of base [49]. It was shown that at 25 °C, under a total pressure of 50 bar of CO2/H2 (1 : 1), with 0.1 mmol of 49, 50 mol of NEt3, H2O, 20 hours, formate could be obtained in 7% yield. Almost thirty years later, Jessop and coworkers re-assessed the properties of nickel for this reaction, in particular using Ni(II) complexes such as [NiCl2(dppe)] (50) and [NiCl2(dcpe)] (51, dcpe = 1,2-bis(dicyclohexylphosphino)ethane)) [14]. The catalysts were at first screened as in situ–formed complexes. Using NiCl2 (15.0 μmol), ligand (22.5 μmol), DBU (3.3 mmol) under a total pressure of 100 bar of CO2/H2, at 50 °C for 7.5 hours in DMSO, in the case of dppe a TON of 45 and TOF of 6.0 h−1 were reached, rising to a TON of 117 and TOF of 15.6 h−1 with dcpe. Next, the well-defined complex 51 was used under slightly different conditions, namely 5.0 μmol of catalyst, 60 mmol DBU, a total pressure of

H E Me

+

H Co

N

CO

DBU

E

H2

DBUH+

CO 47 +

Me

E N

Co

Me

CO

N

E 45

Me N

E Co

CO E

CO

H

O

E HCO2–

H

H E

CO

46

O CO2 Co

E

Co N

CO

Me

E = PiPr2

CO E

CO 48

Scheme 3.12  Proposed catalytic cycle for CO2 hydrogenation starting from 45, involving κ3- to κ2-ligand switch. Source: From Spentzos et al. [48], © 2016 American Chemical Society. Reprinted with permission of American Chemical Society.

73

74

3 CO2 Hydrogenation to Formic Acid/Formate with Non-precious Metal Catalysts

200 bar, 216 hours, and a TON of 4400 was obtained. Shorter reaction time (21 hours) and lower amount of DBU (6 mmol), with a higher catalyst loading (15.0 μmol), gave a TON of 330. Pincer-type Ni complexes were studied by Hazari, Kemp and coworkers to assess the effect of the nature of the ligand on the thermodynamics of CO2 insertion into the Ni─H bond [50]. By a combination of experimental methods and DFT calculations, it was found that the nature of the ligand trans to the hydride is crucial to determine the favorability of such an insertion; thus, the reaction is favored in the presence of PCP and PSiP-supported species versus PNP counterparts, to yield stable Ni-formate complexes. Interestingly, the mechanism of insertion follows a single-step, four-membered transition state rather than the usual two-step, outer-sphere mechanism. Later on, the study was extended to other LnNi–E complexes (Ln = PCP-type ligands; E = H, OH, NH2) by stopped-flow kinetic measurements  [51]. The kinetics of CO2 insertion into [Ni(OH)(tBuPCP)] (52-OH; t BuPCP = 2,6-C6H3(CH2PtBu2)2) to yield the κ1-bicarbonate complex [Ni(tBuPCP){OC(O) OH}] (52-O3CH) were compared to those for CO2 insertion in M─H bond of complexes [NiH(RPCP)] (R = tBu, 53a; Cy, 53b; iPr, 53c), together with an assessment of the effect of solvents on the rate of insertion in 52-OH. It was observed that the rate of insertion increases in the order E = H  160) were obtained using Cu(OAc)2·H2O (0.02 mmol), DBU (10 mmol), 60 bar of CO2/H2 (1 : 1), 100 °C, albeit reaching only 34% of formate yield. It was possible to increase the yield to 60% by increasing the quantity of Cu salt (0.05 mmol) and decreasing the amount of DBU (5 mmol) under the same reaction conditions, but in this case TON dropped to 60. The authors studied preliminary reaction mechanism. Upon reaction of CuI with DBU in CH3CN, complex [CuI(DBU)2] was isolated, and once tested in catalysis, showed comparable activity to the in situ system CuI/DBU. Thus, it was postulated that the active species [CuH(DBU)n] (60), derived from hydrogen activation, would then bring about CO2 insertion and base-assisted formate release. This study was extended to the assessment of ligand effects in promoting catalytic performance under the conditions described above. Bidentate N,N-ligands such as 1,10-phenanthroline (phen), 3,4,7,8-tetramethyl-1,10-phenanthroline (Me4phen), 4,7-diphenyl-1,10-phenanthroline (Ph2phen), 2,9-dimethyl-1,10-phenanthroline (Me2phen), 2,2′-bis(2-oxazoline), 2,2′-isopropylidenebis[(4R)-4-benzyl-2-oxazoline], and 2,2′-isopropylidenebis(2-oxazoline) were added to the reaction mixture in 1 : 1 ratio to the metal. Interestingly, whereas the bipyridine-type ligands proved to be detrimental to catalysis, a TON enhancement was observed in the presence of the oxazoline ones, reaching a maximum TON of 207 (41% formate yield) with 2,2′-isopropylidenebis(2-oxazoline) [59]. Higher activities and thermal stability were obtained by Appel and coworkers [60] using the well-defined complex [Cu(triphos)(MeCN)](PF6) (61-MeCN, triphos  =  1,1,1-tris(diphenylphosphinomethyl)ethane) as catalyst precursor, in combination with DBU under 40 bar of CO2/H2 (1 : 1) in CH3CN. The highest TON of 500 was obtained with 0.05 mM solutions of 61-MeCN, 50 mM of DBU, at 140 °C for 20 hours. Interestingly, base screening showed that no correlation between pKa and catalytic rate was present, instead reaction rates are higher with the base (DBU) that has the higher ability to bind to the metal center. Preliminary studies showed that indeed the catalyst resting state is complex [Cu(triphos) (DBU)]+ (61-DBU), which, in turn, activates H2 to give neutral [CuH(triphos)] (62). Complex 62 gives CO2 insertion into the Cu─H bond and formate release, as shown in the proposed mechanism (Scheme  3.14). Further detailed studies on the nature of catalytic intermediates  [61] showed that, upon reaction of 61 with H2, the H-bridged bimetallic complex [(triphos)Cu(μ-H)Cu(triphos)]+ (63) is formed. Complex 63 reacts stoichiometrically with CO2 to give the formato complex [Cu(OCHO)(triphos)] (64) and 61. When 63 was tested as well-defined pre-catalyst under the conditions described above, comparable activity to 61 was observed, confirming its possible role as catalytic intermediate.

77

78

3 CO2 Hydrogenation to Formic Acid/Formate with Non-precious Metal Catalysts + P Co P

P DBU

NCMe 61-MeCN

–MeCN +

–

P

HCO2

H2 Co

P DBU

DBU

HDBU+

P 61-DBU

P

P Co

P

O

Co O

P H

P

H

P 62

64

CO2

Scheme 3.14  Simplified catalytic cycle for CO2 hydrogenation starting from the cationic Cu(I) complex 61-DBU upon reaction of 61-MeCN with DBU. Source: From Zall et al. [60], © 2015 American Chemical Society. Reprinted with permission of American Chemical Society.

3.6 ­Manganese-Catalyzed CO2 Hydrogenation In spite of the well-established use of Mn complexes for photo- and electrochemical catalytic reactions, including CO2 conversion to CO, the application of this transition metal (the third most abundant one in the Earth’s crust after Fe and Ti, see Figure 3.1) is still in its infancy, but getting increased attention in the last few years  [12, 62]. Mn-catalyzed homogeneous CO2 hydrogenation was at first explored by predictive computational methods using DFT calculations by Pathak and coworkers, focusing on Mn(I)-bipy cationic complexes and assessing the role of σ-donors and π-acceptor ligands on the individual steps of the reaction pathway [63]. It was found that σ-donor ligands such as PMe3 or PH3 favor the hydride transfer step, whereas the π-acceptor ligands (CO/PF3) favor heterolytic hydrogen cleavage. With Mn-complexes containing σ-donor and π-acceptor ligands, hydride transfer is rate limiting, so a proper combination of such ancillary ligands should favor catalytic rate. In 2017, the first report on Mn(I)-catalyzed efficient CO2 hydrogenation to formate was disclosed by Gonsalvi, Kirchner and coworkers  [64]. Two well-defined Mn(I)

3.6 ­Copper-Catalyzed CO2 Hydrogenation

hydridocarbonyl complexes, stabilized by PNP-type ligands based on the 2,6-(diaminopyridyl) scaffold analogues of 24 and 25, namely [MnH(PNPH-iPr)(CO)2] (65) and [MnH(PNPMe-iPr)(CO)2] (66), differing only by the group linked to amine N atom in the backbone (H and Me, respectively), were synthesized and tested in the presence of DBU as added base. The best results were obtained using 65. Under standard test conditions (80 °C in THF/H2O, 80 bar H2/CO2  =  1  :  1, 24 hours) using a 65/DBU ratio of 1  :  1000, formate was obtained in quantitative yields with a TON = 1000). At lower catalyst loadings (65/DBU = 1 : 10 000; [65] = 0.18 μmol/ml), a drop in yield was observed (55%) after 24 hours (TON  =  5520) and at catalyst concentration of 0.036 μmol/ml (65/ DBU = 1 : 50 000) even lower yields were obtained (16%) but with a significantly increased TON (9100). Longer reaction times (48 hours) at 65/DBU ratio of 1  :  10 000 gave high yields of formate and TON (86%, TON = 8600). Next, the effect of Lewis acid co-catalysts was tested, and by adding LiOTf (0.5 mmol; 65/DBU = 1 : 10 000; 65/LiOTf = 1 : 1000), formate was obtained in quantitative yields within the first 24 hours of reaction (TON = 10 000). Under optimized conditions, i.e. 100 °C, 24 hours, 65/DBU = 1 : 50 000; 65/LiOTf = 1 : 5000), a TON of 31 600 (63% yield) was obtained. The reaction mechanism involving 65 was studied by a combination of NMR spectroscopy and DFT calculations. It was shown that two mechanisms are possible (Scheme 3.15). In the first one (left), a purely metal-centered mechanism without participation of the N─H bond of the PNP ligand is active, where H2 coordination to Mn has the highest energy transition state of the entire mechanism (ΔG  = 17.9 kcal mol−1). This pathway has an overall barrier of 27.6 kcal mol−1 and the entire reaction is thermodynamically favorable with ΔG = –11.2 kcal mol−1. In the O

H

H

H

DBUH

PR2

N

O

Mn

N

i

R = Pr CO

N PR2 + DBU

H

CO

(7.4)

H

H

Mn

CO

N PR2 H

H

N PR2

(17.9) O

N

CO

O

N

CO

DBUH

H

O

PR2 Mn

N PR2 H

O

H

H N

(–9.7)

(–4.2)

(5.9)

H

(13.2)

N PR2 H

O

CO

CO

O

H

PR2 Mn

PR2 Mn

N

H

H

N

H

N (9.6)

O

(7.0)

CO

H

H

O

PR2

N N

CO2

+

H

(20.0)

65 (0)

CO

CO

DBU +

H H

(5.2)

Scheme 3.15  Proposed (simplified) competing catalytic cycles for CO2 hydrogenation in the presence of 65. Free energy values in kcal mol−1, transition state energies in italics. Source: Bertini et al. [64], © 2017 Royal Society of Chemistry.

79

80

3 CO2 Hydrogenation to Formic Acid/Formate with Non-precious Metal Catalysts

second pathway (right), a bifunctional mechanism with participation of the N─H bond of the PNP ligand is competing. After an initial step leading to Mn-bound H-coordinated formate, common to both pathways, the base deprotonates the N–H group of the PNP ligand and releases the formate. Next, the five-coordinated Mn neutral complex, bearing a negatively charged dearomatized PNP*-iPr ligand, coordinates H2 giving the intermediate [Mn(PNP*-iPr)(η2-H2)(CO)2]. Solvent-assisted H─H bond splitting then regenerates 65 forming a Mn─H hydride bond and reprotonating the PNP ligand. This step has the highest barrier of 20.0 kcal mol−1. The overall barrier for this pathway is 29.7 kcal mol−1, only 2 kcal mol−1 higher than the previous one; hence, both mechanisms can be active for 65 in catalytic conditions. As in the case of Hazari and Bernskoetter’s Fe(PNP) complexes 28, the LiOTf Lewis acid helps in stabilizing the Mn-bound H-coordinated formate over the Mn-bound O-coordinated formate (resting state) by decreasing the free energy difference between the two isomers by c. 3.4 kcal mol−1. At the same time, Nervi, Khusnutdinova and coworkers published results on the use of Mn(I)-bipy type complexes as catalysts for CO2 hydrogenation to formate and formamide  [65]. The bio-inspired Mn-bromotricarbonyl complexes 67–70 contain modified bipy-ligands bearing o- and p-substituents such as OH, NH2, OMe (Figure 3.8) that were introduced to test the effect in catalysis of proton-shuttling groups either in proximity or at distance from the metal center. The best results were obtained with the o-OH substituted complexes 67a and 69a, where these substituents enhance the ability of adjacent oxo-groups to participate in H2 cleavage. In 1,4-dioxane in the presence of DBU as a base under 60 bar of H2/CO2 (1 : 1), 80 °C, 24 hours, 97% yield of formate salt was obtained with HO

OH

Br

Br N

N Mn

R OC

Mn

CO

OC

CO

68

OTf

NCMe

Mn OC

R CO

CO 69a, R = OH 69b, R = OMe

HO

NCMe

OH

OTf

N

N

N

N

CO CO

67a, R = OH 67b, R = H 67c, R = OMe 67d, R = NH2

R

N

N R

Mn OC

CO CO 70

Figure 3.8  Substituted-bipyridine Mn(I) bromotricarbonyl complexes 67–70 used for CO2 hydrogenation to formate and formamide. Source: From Dubey et al. [65], © 2017 American Chemical Society. Reprinted with permission of American Chemical Society.

3.7 ­Other Non-precious Metals for CO2 Functionalization

67a, reaching a TON of 1224. By changing solvent to CH3CN, comparable results were obtained at 65 °C using 5 μmol of 67a (TON  =  1313, 96% yield) and 69a (TON  =  1299, yield  =  99%). Even better results (TON  =  6250, 98% yield) were obtained at 65 °C in CH3CN with lower catalyst loading (1 instead of 5 μmol) in the presence of 67a. Furthermore, in the presence of diethylamine and 70 bar of CO2/H2 (2 : 5), catalyst 67a gave diethylformate with a TON of c. 588 after 24 hours at 80 °C. The reaction mechanism was elucidated by Das and Pati by DFT calculations [66]. It was shown that the main steps are heterolytic dihydrogen cleavage and hydride transfer to CO2, and that first one is assisted by the external base. Interestingly, calculations suggest that the role of the pendant oxygen atoms is not in lowering the barriers of these two steps, but rather as an indirect effect in forcing the correct geometry making the final hydride transfer step more feasible. Indeed, as the pendant oxygen and the H2 molecule occupy equatorial and axial sites in the active catalyst, this causes significant geometric distortions during the basefree H2 activation step. Thus, forcing these two ligands in geometrical proximity (same equatorial plane) contributes to reduce the structural distortion and make H2 activation more facile. Only a few months after publication of these two reports, Prakash and coworkers disclosed their results on Mn(I)-PNP pincer catalyzed sequential one-pot homogeneous CO2 hydrogenation to CH3OH by molecular H2 in the presence of amines [67]. Although the process does not yield formate as product, the first step of the sequential reaction was proposed to be the 2-electron reduction of CO2 to formate, which then reacts with the amine to give an intermediate formamide that is finally reduced to CH3OH, giving back the initial amine. The authors used as pre-catalysts the Mn(I)-bromodicarbonyl complexes [MnBr(RPNP)(CO)2] bearing the PNP ligands bis(2-(dialkylphosphino)ethyl)amine (R = alkyl = iPr, 71a; Cy, 71b). The amine formylation step occurs by a MLC-type mechanism, with prior amine-amido ligand deprotonation that facilitates H2 heterolytic splitting, giving the corresponding [MnH(RPNP)(CO)2] complex (72). Next, CO2 inserts into the Mn─H bond and releases formate that react smoothly with the amine to give the corresponding formamide.

3.7 ­Other Non-precious Metals for CO2 Functionalization Second- and third-row metals not belonging to the platinum group such as Re, Mo, and W were also tested in catalytic CO2 hydrogenation to formate giving, in general, either less active catalysts than first-row or precious metals counterparts or stable formate complexes. MoCl3 was tested in the presence of PPh, dppe, and dcpe by Jessop and coworkers [14]. Under a total pressure of 100 bar of CO2/H2, at 50 °C for 7.5 hours in DMSO, the highest TON of 63 and TOF of 8.4 h−1 were obtained with dcpe. An interesting approach was later used by Berke and coworkers [68], by adding the rhenium complex [ReBr(H)(NO)(PR3)2] (R =  iPr, 72a; Cy, 72b) to a solution of B(C6F5)3 and exposing the resulting mixture to CO2 (1 bar), observing the instantaneous formation of the frustrated Lewis pair (FLP)-type species [ReBr(H)(NO)(PR3)2(η2-O=C=O−B(C6F5)3)] (R =  iPr, 73a; Cy, 73b), where the metal hydride side acts as the FLP base. The Re(I) η1formato dimer [{Re(μ-Br)(NO)(η1-OC(H)=O−B(C6F5)3)(PiPr3)2}2] (74a) was then

81

82

3 CO2 Hydrogenation to Formic Acid/Formate with Non-precious Metal Catalysts

generated from 73a and, by following exposure to H2 gas, the stable formato dihydrogen complex [ReBr(η2-H2)(NO)(η1-OC(H)=O−B(C6F5)3)(PiPr3)2] (75a) was obtained. The activity of the in situ catalytic system 72a/B(C6F5)3 was tested in the presence of 2,2,6,6-tetramethylpiperidine (TMP). The best results were obtained under 20 bar of CO2 and 40 bar of H2 in THF, 1 mmol of TMP, 72a/B(C6F5)3 (0.5 mol%, 1 : 2) at 80 °C for 15 hours, reaching formate salt (HTMP)+(HCO2)− yield of 52%, corresponding to a TON of 104. The proposed catalytic cycle is shown in Scheme 3.16. Another rhenium-based system tested for CO2 activation was complex cis-[Re(tBuPNP*) (CO)2] (76), studied by Milstein and coworkers [69]. Upon dearomatization of the pyridinebased PNP pincer ligand, [1,3]-addition of CO2 is triggered in a MLC mechanism yielding cis-[Re(tBuPNP–COO)(CO)2] (77) by Re─O and C─C bond formation. Reaction of 77 with H2 gave free CO2 and the re-aromatized hydride complex cis-[ReH(tBuPNP)(CO)2] (78). The reaction was proved to be reversible under a pressure of CO2 at high temperature (>75 °C), whereas at lower temperature (160 °C) to form appreciable amounts of methanol and the highest methanol formation was observed at 240 °C. The CO2 hydrogenation proceeded through the formation of CO by reverse water gas-shift reaction. CO was also observed as a side product, along with CH4, whose formation increased drastically beyond 240 °C due to further reduction of the formed methanol. Besides Ru3(CO)12, other metal carbonyls such as Rh4(CO)12, Ir4(CO)12, W(CO)6, Mo(CO)6, Fe2(CO)9, and Co2(CO)8 did not show appreciable catalytic activity for either methanol, CO, or methane formation. The presence of KI in the system was crucial to stabilize the metal carbonyl cluster complexes formed during the reaction (KCl and KBr also worked to some extent). Without any KI additive, Ru3(CO)12 decomposed to ruthenium metal, leading to the extensive formation of CH4 through methanation. Although the system described a homogeneous way to hydrogenate CO2 to methanol, temperatures, and pressures as high as those encountered for heterogeneous systems were necessary.

93

4  Catalytic Homogeneous Hydrogenation of CO2 to Methanol

4.3.4  Indirect CO2 Hydrogenation For more than a decade after the initial study by Tominaga et al., no significant progress was reported toward homogeneous CO2 hydrogenation to methanol using molecular complexes. In 2011, in a step forward, Milstein and coworkers described ruthenium-based pincer complexes for the catalytic hydrogenation of dimethyl carbonate, methyl carbamates, and methyl formates to methanol (Figure 4.5) [21]. These carbonates, carbamates, and formate esters can be conveniently synthesized from CO2 and thus provide an indirect stepwise approach for obtaining methanol from CO2. Under the reaction conditions (110 °C, 10 atm H2, tetrahydrofuran [THF]), hydrogenation of dimethyl carbonate and methyl formate proceeded faster compared to methyl carbamate and a highest turnover number (TON) of 4400, 98, and 4700 was obtained for carbonate, carbamate, and formate, respectively, using catalyst C-2. Since this study, many other catalysts were reported for the efficient hydrogenation of formate esters and carbonates to methanol. Also in 2011, Milstein and coworkers reported another indirect CO2 to methanol route through catalytic hydrogenation of urea derivatives using complex C-2 (Figure 4.6) [22]. The hydrogenation rate was expectedly slower due to the electron-rich nature of the urea carbonyl center. With dimethyl urea as the substrate, the hydrogenation proceeded in the presence of C-2 (2 mol%) at 110 °C under 13.6 atm H2 pressure in THF solvent to produce

Milstein and co-workers O

C-1/C-2

+ 3H2

O

3O

H R CO2

–H

3 CH3OH

1,4-Dioxane/THF/neat 145 °C

H

O

2C

94

TON C-1 = 2500 TON C-2 = 4400

O

2

NH2

–H2O H 2, CH 3O –H H 2O

O N H

R

C-2

+ 3H2

O

O H

O

2 CH3OH

THF 110 °C

TON C-1 = 1155 TON C-2 = 4700 H

H P Ru Et

N

NH2

2 CH3OH

1,4-Dioxane/THF/neat 110 °C/145 °C

N

R

98%

C-1/C-2

+ 3H2

+

t t

Bu

N

Bu

CO

P Ru

N

t

Bu

t

Bu

CO

Et C-1

C-2

Figure 4.5  Step-wise hydrogenation of CO2 to methanol through carbonate, carbamate, or formate esters.

4.3  ­Homogeneous Catalysis – An Alternative for CO2 to Methanol

Milstein and co-workers [22] O R

CO2 +2 RNH2

–H2O

N H

N H

R + 3H2

C-2 1,4-Dioxane/THF/neat 145 °C

Urea derivative

CH3OH + 2RNH2 93% yield TON C-2 = 47 R = CH3

Figure 4.6  Indirect CO2 hydrogenation to methanol through urea derivatives.

CH3OH with 93% yield after 72 hours, corresponding to a TON of 47. Since then, other instances of urea hydrogenation to methanol have been reported [23]. Complexes C-1 and C-2 can catalyze the hydrogenation of different carbonyl derivatives due to their unique ability to split molecular hydrogen using metal-ligand cooperation (Figure 4.7). Complex C-1, in the presence of molecular H2, forms the corresponding dihydride species (C-1A). The pyridine moiety of the ligand is aromatized during the process, which facilitates this H2 splitting reaction. The dihydride species in turn can transfer one hydride and proton through an outer sphere mechanism to a suitable carbonyl group. This process forms back complex C-1, thus completing the catalytic cycle. In 2012, Ding and coworkers demonstrated the hydrogenation of cyclic carbonates in a similar fashion to Milstein and coworkers using several ruthenium PNP pincer complexes with very high catalytic activities (Figure  4.8) [24]. The authors used low-cost ethylene carbonate as substrate to obtain ethylene glycol and methanol as hydrogenation products. Ethylene carbonate can be easily synthesized from CO2 using ethylene oxide (ethylene oxide in turn can be obtained from ethylene and O2 as in the first step of the only monoethylene glycol advantage [OMEGA] process). Importantly, ethylene carbonate is already synthesized on an industrial scale from ethylene oxide and CO2, which is then hydrolyzed to obtain ethylene glycol (and carbonic acid, H2CO3). Thus, the process reported by Ding and coworkers provides an alternative method to obtain ethylene glycol from ethylene carbonate alongside an important value-added product, methanol, as opposed to H2CO3. The authors screened several ruthenium PNP pincer complexes (along with one iridium PNP pincer, which was not very active). Among these, Ru-Macho (C-3) was found the most H N N

H

H H H2

N

P

Ru

N

CO

Ru

P CO

H

C-1 P = PtBu2 N = NEt2

H

O

OH R

H

R′

R

R′

C-1A

Figure 4.7  Molecular mechanism of hydrogenation by complex C-1. Source: Based on Kar et al. [7].

95

96

4  Catalytic Homogeneous Hydrogenation of CO2 to Methanol

efficient to catalyze this reaction. With Ru-Macho, they observed a TON as high as 84 000 for methanol formation through ethylene carbonate hydrogenation. Importantly, beside cyclic carbonates, linear polycarbonates were also catalytically hydrogenated using this method. Recently, the research groups of Leitner, Milstein, and Rueping have independently reported the hydrogenation of ethylene carbonate employing inexpensive manganesebased catalysts C-4, C-5, and C-6, respectively (Figure 4.8) [25]. As expected, the catalytic activity of all of these manganese complexes was inferior when compared to the ruthenium pincer complexes. Nonetheless, the manganese-based catalysts all provided high methanol yields (>95%) at catalyst loadings of 0.5–2 mol% and a temperature range of 110–140 °C. Similar to catalysts C-1 and C-2, catalyst C-3 can also split H2 at a molecular level and subsequently transfer it to suitable carbonyl substrates (Figure 4.9). This process involves the formation of an amido complex (C-3A) formed in situ from complex C-3 in presence of a base. The amido complex splits H2 to form the dihydride species C-3B. The dihydride

O O

O

+ CO2

N

P Ru

P

Ph

H

Ph Ph

CO

Ph

Base, solvent, 100–140 °C 16–72 h

iPr

N

P iPr

HO

OH

Br

H

H H

H2, Cat.

O

iPr

P

iPr

Mn

N

CO

t

Bu

CO

P t

Mn

+ CH3OH

+ –

CO H

H N t Bu

N

CO

P

N Mn

Ph Ph

Br

CO

C-4

Bu CO C-5

CO

C-3 Ding and co-workers TON max = 84000

Leitner and co-workers TON max = 400

Milstein and co-workers TON max = 50

Rueping and co-workers TON max = 200

Cl

C-6

* Highest methanol TONs for the catalytic hydrogenation of ethylene carbonate

Figure 4.8  Indirect CO2 to methanol route through cyclic carbonates.

H2 H H

N

P

H

H P

Ru CO Cl C-3

KOtBu –KCl –tBuOH

N P

H P

Ru

N

P

CO

P Ru CO H

C-3A

C-3B XH

P = PPh2 Y

H

Z

X Y Z X = O, N

Figure 4.9  Hydrogenation mechanism with complex C-3. Source: From Kar et al. [7].

4.3  ­Homogeneous Catalysis – An Alternative for CO2 to Methanol

species subsequently provides the hydride and proton to the carbonyl substrate and form back the amido complex, thus closing the reaction cycle.

4.3.5  Direct CO2 Hydrogenation Apart from the studies on catalytic hydrogenation of CO2 derivatives to methanol, direct CO2 hydrogenation to methanol has also been investigated. The reported processes are generally carried out in the presence of alcohols, amino alcohols, or amine additives and proceed through the formation of formate esters, oxazolidinone, or formamides, respectively. 4.3.5.1  Through Formate Esters

In 2011, Huff and Sanford demonstrated an elegant sequential CO2 hydrogenation process that can produce methanol from a CO2/3H2 gas mixture (40 bar) at 135 °C [26]. The whole sequential process involved three different steps catalyzed by three different catalysts – (i) hydrogenation of CO2 to formic acid catalyzed by C-7, (ii) esterification to generate a formate ester catalyzed by Sc(OTf)3, and finally (iii) hydrogenation of the formate ester to 2 equivalents of methanol, catalyzed by complex C-2 (Figure 4.10). When all three steps were performed in the same pot with a CO2/3H2 mixture and CD3OH as a solvent, a TON of 2.5 was observed for the methanol formation. The low TON was attributed to the deactivation of the formate ester hydrogenation catalyst (C-2) by Sc(OTf)3. The authors were able to increase the methanol turnover to 21 by physically separating the cross-reactive catalysts of steps b and c using a vial in the center of the vessel. The first two steps were carried out inside the vial at 75 °C, and the produced methyl formate was transferred to the outer vessel simply by increasing the reaction temperature to 135 °C. Nonetheless, the requirement of three different catalysts was still a drawback of the system. In 2012, Leitner and coworkers employed a single ruthenium phosphine complex to achieve homogeneously catalyzed hydrogenation of CO2 to methanol (Figure 4.11) [27]. Complex C-8, [(Triphos)Ru-(TMM)] (TMM = trimethylenemethane) was used as a catalyst precursor along with an acid additive (typically methanesulfonic acid (MSA) or triflimide Huff and Sanford [26]

a. H2

C-7

OH

2H2 –ROH

C-2 b.

O H

Me3P

c.

H

ROH –H2O

CO2

+

3H2

PMe3

N

Ru Me3P

O

Sc(OTf) 3

H

PMe3

CH3OH + H2O

3H2 + CO2

P

C-7, C-2

Sc(OTf) 3 135 °C, 16 h CD3OH

N

Figure 4.10  Cascade catalysis of CO2 to methanol.

C-7

Bu

CO C-2

CH3OH + HCOOCD3 2.5

Bu

t

Ru

OAc Cl

OR

t

34

= TON

97

98

4  Catalytic Homogeneous Hydrogenation of CO2 to Methanol Leitner and co-workers [27, 28] Cat. HNTf2

CO2 + 3H2

CH3OH + H2O

140 °C, 24 h, THF 10 mmol EtOH

20 bar 60 bar

TON = 221

NTf2 Ph2P

PPh2 Ru

PPh2

Ph2P

PPh2

Ru PPh2

H2 H

S C-8

C-8A Proposed active species S= solvent or substrate

Ph2P

PPh2 O

Ru PPh2

H O

S C-8B Catalyst resting state S= solvent or free coordination site

Ph2P

NTf2

PPh2 O Ru

PPh2

O

S C-9

Figure 4.11  Ruthenium-catalyzed CO2 to methanol system developed by Leitner and coworkers.

(HNTf2)). The authors surmised the formation of complex C-8A as the active species under the reaction conditions responsible for catalysis. As in the case of the previously discussed cascade system by Huff and Sanford, the present system likely also proceeded through the formation of formate esters, and methanol was obtained in the presence of ethanol as an alcohol additive. Under the reaction conditions of CO2/3H2 (80 bar), 140 °C, the authors obtained a TON of 221 after 24 hours. The same authors improved upon the system and, in a study published in 2015, they carried out the hydrogenation without the need of any acid or alcohol additives [28]. From mechanistic studies, the presence of ruthenium formate species (C-8B), with a bidentate formate ligand, was observed. In solution and in the presence of H2, this formate complex can produce methanol. However, the isolation of complex C-8B in a pure solid powder form was challenging. The analogous acetate complex (C-9) was isolable, and the authors obtained a methanol TON of 165 with C-9 as the catalyst in the absence of acid and alcohol additives at a pressure of 80 bar of CO2/3H2 and a temperature of 140 °C. A mechanistic cycle based on the formation of ruthenium hydride (C-8A), formate (C-8B), hydroxymethanolate (C-8C), and methoxy (C-8D) species was proposed based on density functional theory calculations (Figure  4.12). Finally, the authors also demonstrated the successful recycling of the catalyst by using a biphasic 2-methyltetrahydrofuran (2-MTHF) and water solvent system and obtained a total TON of 769 over four cycles (TON of 247, 222, 191, and 110 in consecutive cycles). All the above-mentioned studies of direct CO2 hydrogenation to methanol utilized ruthenium-based catalysts. Recently, however, first-row transition metal–based catalysts (cobalt) have also been reported by Beller and coworkers for this process [29]. Similar to the ruthenium-based systems described by Leitner and coworkers, the cobaltbased system also utilized triphos ligands coordinated to the cobalt center (Figure 4.13). The active catalyst was prepared in situ using cobalt acetylacetonate (or cobalt acetate, which also worked well, as the cobalt precursor), triphos (L1) as the ligand, and triflimide as an acid additive. In the absence of acid additive, no appreciable amount of methanol formation was observed. Again, the presence of ethanol as an alcohol additive was required to stabilize the formic acid intermediate through the formation of ethyl

4.3  ­Homogeneous Catalysis – An Alternative for CO2 to Methanol

CH3OH

P3Ru

CO2

H

C-8A

H2

CO2 + 3H2 O

P3Ru

OCH3 C-8D

P3Ru CH3OH

H2 O

H O C-8B H2

H H2

P3Ru

O H

O H C-8C

Figure 4.12  Mechanistic cycle of CO2 to methanol with C-8 in the absence of any alcohol additive.

Beller and co-workers [29, 30]

CO2 + 3H2 20 bar 70 bar PPh2 Ph2P

PPh2 Triphos L-1

Co(acac)3, Ligand L Cat. HNTf2 100 °C, 24 h THF/EtOH

CH3OH + HCOOEt TON = 50

P(o-Tol)2 (o-Tol)2P

P(o-Tol)2 L-2

TON = 2

P(p-Tol)2 (p-Tol)2P

P(p-Tol)2 L-3

P(m-Xyl)2 (m-Xyl)2P

P(m-Xyl)2 L-4

Figure 4.13  Cobalt-catalyzed CO2 to methanol system.

formate ester. Using the Co(acac)3/Triphos/HNTf2 catalytic system, the authors achieved a highest methanol TON of 78 at 100 °C after 96 hours with 20 and 70 bar of CO2 and H2 pressure, respectively. In 2019, Beller and coworkers reported an improvement over the previously described system using slightly modified triphos ligands [30]. When the phenyl substituents of the triphos ligand were changed to p-tolyl (L-3) or m-xylyl (L-4) groups, the methanol TON increased from 60 (for triphos) to 125 and 115, respectively, under similar reaction conditions. On the other hand, substitution in the ortho position (L-2) completely inhibited methanol formation. According to the authors, the electron-donating groups in meta and para positions increase the electron density in the P atoms and the Co center, which enables a faster H2 activation. The substitutions might prevent the formation of dihydrido bridged dimers as well. The authors also screened different first-row transition metal precursors for the hydrogenation, but no significant methanol formation was observed with Mn(acac)2, Fe(acac)3, Ni(acac)2, or Cu(acac)2. Finally, the presence of CO, water, and methanol was found to be detrimental to methanol formation.

99

100

4  Catalytic Homogeneous Hydrogenation of CO2 to Methanol

4.3.5.2  Through Oxazolidinone or Formamides

All the aforementioned systems for direct CO2 hydrogenation to methanol were used in slightly acidic or neutral media. However, most of the existing CO2 capture techniques rely on basic aqueous amine or amino alcohol solutions. Thus, if we desire to synthesize methanol as a means to recycle captured CO2, these systems would first require desorption and compression steps, where pure pressurized CO2 is produced. These desorption and compression steps are energy intensive and add substantial cost to the overall process of CO2 recycling to methanol. Thus, from an economic point of view, it would be beneficial to devise systems where CO2 can be hydrogenated directly in a basic medium. Moreover, the high reaction temperature and closed reactor can also theoretically produce compressed CO2 gas inside the reactor, making intermediate desorption and compression unnecessary. Such a system would be ideal to integrate CO2 capture with its subsequent conversion to methanol. Milstein and coworkers reported in 2015 such an integrated system for CO2 capture and conversion in two steps [31]. In the first step, CO2 at low pressure (1–3 bar) was captured using amino alcohols at 150 °C in the presence of a cesium carbonate catalyst (10 mol%) in dimethylsulfoxide solvent to form oxazolidinones (Figure  4.14). It should be noted here that this mode of capture is different from the industrially practiced way of capturing CO2 in aqueous amino alcohol solutions, which proceeds at room temperature or slightly higher and forms primarily carbamate and bicarbonate salts. In the second step, the formed oxazolidinone was hydrogenated without any intermediate purification under 60 bar H2 pressure in the presence of C-2 as the hydrogenation catalyst. A large excess of tBuOK (25 mol%) was required to neutralize the acidic byproducts of CO2 capture and obtain high methanol yield. Using this procedure, the authors obtained a highest methanol yield of ~50%. This study described for the first time an integrated approach for CO2 capture and conversion to methanol, although the capture of CO2 in the form of oxazolidinone required high reaction temperatures and dimethyl sulfoxide (DMSO) as a solvent. An important breakthrough came in 2015 when Sanford and coworkers described the hydrogenation of carbamates to methanol under homogeneous conditions using the Milstein and co-workers [31] Cat. Cs2O3 CO2

+

RHN

H2O

OH R′

1–3 bar

O N H PtBu2 Ru N CO C-2

RHN MeOH

OH

+ R′

RN

O

R′

Oxazolidinone

Ru-pincer cat 3H2

Figure 4.14  Combined CO2 capture and conversion to methanol through oxazolidinones.

4.3  ­Homogeneous Catalysis – An Alternative for CO2 to Methanol

Ru-Macho-BH catalyst (Figure 4.15) [32]. The reaction proceeded through the formation of free gaseous CO2 and amine, generated at the reaction temperatures (as opposed to the direct hydrogenation of carbamate to formamides). The free CO2 was then hydrogenated to formamide products in the presence of amine through ammonium formate salts. The produced formamide salts were further hydrogenated by the catalyst to methanol and the amine was formed back. Using the same catalyst, the authors demonstrated that the hydrogenation of CO2 gas to methanol is also possible in the presence of dimethylamine as an additive. Using a temperature-ramping strategy (95 °C for 18 hours, to form the formamide product; then 155 °C for 36 hours for the formamide hydrogenation), the authors synthesized methanol from a gas mixture of CO2 (2.5 bar) and excess H2 (50 bar) with Ru-Macho-BH (C-10) as the catalyst, with a highest TON of 550. The intermediate products, such as the formamide and formate salts, were also observed upon termination of the reaction, indicating an incomplete formamide hydrogenation after 36 hours. Ding and coworkers also reported in 2015 a sequential route to obtain methanol from CO2 through formamide intermediates (Figure 4.16) [33]. In a first step, an amine (typically morpholine) was formylated using a CO2/H2 gas mixture (70 bar) at 120 °C in THF solvent.

Sanford and co-workers [32] O C O +

Conversion to methanol

H H

2 NHR2 O

Cat. [Ru] H2

CO2 Capture

R2N

O C

H

– HNR2 –H2O

N R

R

Cat. [Ru] 2H2 – HNR2

CH3OH

Ph

TONmax = 550

N

P Ru

P Ph

Ph Ph

CO H

Formamide

BH3

Ru-Macho-BH (C-10)

O

H2NR2

Figure 4.15  Amine-assisted CO2 to methanol process. Ding and co-workers O CO2 + 35 atm

H2 + 35 atm

N H

C-3 (0.01 mol%)

H2 (50 atm)

1 : 1 CO2:H2 (70 atm)

KO tBu (0.12 mol%) 160 °C, 1 h

THF, 120 °C 40 h

H

Me N Ph

P Ph

Ru

Ph P Ph CO

Cl

C-11

H

O H CH3OH + (36%)

N O

H N

P R R

P Ru

R R

CO Cl

R = Ph (C-3) = iPr (C-12) = Cy (C-13) = tBu (C-14) = Ad (C-15)

Figure 4.16  Morpholine-mediated CO2 to methanol through formamide.

(63%)

O + N H

101

102

4  Catalytic Homogeneous Hydrogenation of CO2 to Methanol

The authors screened several ruthenium pincer complexes among which complexes C-11 to C-15 were found to be highly active. In the second step, the formamide product was hydrogenated in the same pot using pure H2 gas (50 atm) at 160 °C, and a methanol yield of 36% (TON = 3600) was obtained with catalyst C-3 after one hour. Following the report of Sanford and coworkers on the amine-assisted CO2 to methanol process, Prakash and coworkers devised a system where CO2 from air was successfully captured by aqueous amine solutions and converted to methanol (Figure 4.17) [34]. The use of aqueous solutions for the capture is desirable due to its benign nature, along with the use of high boiling amines to avoid atmospheric amine contamination. The authors proposed a carbon capture and utilization cycle where CO2 is captured in a first step using the high boiling point polyamine pentaethylenehexamine (PEHA). The resulting carbamate product is then hydrogenated to methanol through formamide. Importantly, the amine is regenerated after the hydrogenation and can thus be theoretically employed again for CO2 capture. In the first part of the study, the authors focused on the hydrogenation of gaseous CO2 to methanol in the presence of PEHA as the amine, and screened a series of ruthenium and iron pincer complexes, among which Ru-Macho-BH (C-10) was found to be most efficient (Figure  4.18). The metal-ligand cooperation of the PNP ligand was essential to produce methanol as demonstrated by the fact that complex C-11 with a N-Me functionality did not produce any methanol but only the formamide intermediate. Complex C-3, Ru-Macho, was

Prakash and co-workers [34] H N

CH3OH

N

Cat.

Step 1 Step 4

2H2

CO2

n

H

CO2 capture

H

O N

H2 N H

H2O

n

N

Carbon capture and utilization

N COO

H2

Step 2

Step 3

Cat. CO2 utilization Step 2–4

H2 N

N H – HCOO

–

n

Figure 4.17  Carbon capture and utilization cycle to form methanol from CO2.

n

4.3  ­Homogeneous Catalysis – An Alternative for CO2 to Methanol Prakash and co-workers [34] C-10 PEHA

Overall equation: CO2 + 3H2

H H

Ph

N

P Ph

Ru

H H

Ph P Ph CO

Ph

H

N Ru

P Ph

H H

Ph P Ph CO

Cl

BH3 C-10

CH3OH + H2O

Triglyme 125–155 °C

19 bar 56 bar

C-3

iPr

N

P iPr

P Fe

iPr iPr

CO

H

H H

N

S

Me S

Ru CO

Ph

N P Ph

Ph Ph

P Ru

CO

Br

Cl

Cl

C-16

C-17

C-11

Figure 4.18  PEHA-assisted CO2 to methanol.

also similarly active in methanol formation as C-10. On the other hand, the iron-based complex C-16 didn’t produce any methanol. The use of K3PO4 in the system increased the methanol TON to some extent. Under the reaction conditions (CO2/3H2 75 bar, 155 °C, 40 hours, 10 ml triglyme), a TON of 1060 was observed with catalyst C-10. The catalyst was also reused in the study over multiple cycles. Between the cycles, methanol was conveniently distilled out from the reaction solution (trigyme being a high boiling solvent), and after five cycles, a total TON of 1850 was obtained. The second part of the study focused on integrated CO2 capture from air and conversion to methanol (Figure 4.19). In the first step, CO2 was captured by passing synthetic air through 15 ml of an aqueous solution of PEHA (3.4 mmol) with a flow rate of 200 ml min−1. After 64 hours, 5.4 mmol of CO2 had been captured in the form of bicarbonate and carbamate salts. Subsequently, this aqueous solution was hydrogenated using Ru-Macho-BH (C-10) as the hydrogenation catalyst, triglyme as additional co-solvent, at 155 °C under 50 bar of H2. After 55 hours, methanol was produced in 79% yield, corresponding to a TON of 215. Thus, it was demonstrated that CO2 from air can be conveniently captured through aqueous polyamine solutions and converted to methanol through an amine-assisted CO2 to methanol route. The authors improved upon the system and reported in 2018 the successful recycling of the catalyst and regeneration of the amine using a biphasic 2-MTHF/water solvent system Prakash and co-workers [34] 5.4 mmol CO2 captured as carbamate and bicarbonate/carbonate salts Air (400 ppm CO2)

H2N

+

CO2 capture

H N

rt, 64 h H 2O

4

PEHA (3.4 mmol)

NH2

In situ hydrogenation Cat. C-10 50 bar H2 + Triglyme 55 h, 155 °C

Figure 4.19  Conversion of CO2 from air to methanol.

CH3OH + H2O 79% + H N 2

H N

4

NH2

103

104

4  Catalytic Homogeneous Hydrogenation of CO2 to Methanol

for the hydrogenation step (Figure 4.20) [35]. Following the CO2 capture with polyamines in aqueous solution, the capture products were hydrogenated after the addition of 2-MTHF as a co-solvent (rather than water-miscible triglyme or THF). After hydrogenation, the catalyst remained in the top organic phase, whereas the amine was in the bottom aqueous phase, allowing for easy separation. Among different polyamines the authors screened, PEHA was found to be the most suitable for CO2 capture and conversion to methanol, along with C-10 as the most efficient catalyst. Using the amine-catalyst pairing of PEHA and Ru-Macho-BH, the authors recycled both the amine and catalyst three times with a total turnover of 582. In 2019, Prakash and coworkers reported an alternate integrated CO2 capture and conversion process that utilizes amines immobilized onto solid supports for CO2 capture in the absence of any solvent (Figure  4.21) [36]. After capture, the CO2-loaded amines were placed with 80 bar of H2 at 150 °C into a pressure vessel to obtain methanol and regenerate the amine. The immobilization of the amine enabled the authors to easily filter out the immobilized amine from the hydrogenation solution and reuse it in the subsequent capturing cycle. Polyamines covalently bonded to a silica support were more stable than the physically impregnated immobilized amines under the hydrogenation conditions involving high temperatures. Using covalently bonded Prakash and co-workers [35]

2-MTHF

Cat. Amine

Cat.

H2

CO2 Carbamate/ bicarbonate

H2O

Cat. CH3OH

Cat. CH3OH

CH3OH Amine

CH3OH Amine

CH3OH

Figure 4.20  Biphasic system for catalyst and amine recycling. Prakash and co-workers [36] CO2/air CO2 THF, Cat.

CO2

H2

Filtration

Heat CH3OH

= Cat. + THF = Solid-supported amine (SSA)

Figure 4.21  Catalyst and amine recycling based on amine immobilization. Source: From Kar et al. [36], © 2019 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

4.3  ­Homogeneous Catalysis – An Alternative for CO2 to Methanol

polyethylenimine (PEI) and catalyst C-10, CO2 was captured and converted to methanol over multiple cycles. Also in 2019, Prakash and coworkers investigated the catalytic mechanism of the amine assisted CO2 to methanol process involving Ru-PNP catalysts C-3, C-10, and C-12 to C-14 [37]. The authors observed high methanol yields only with Ru-PNPPh (C-3, C-10), although all C-3, C-10, and C-12 to C-14 were active in catalyzing the individual processes of CO2 to formamide and formamide hydrogenation to methanol (Figure 4.22). This selectivity was ascribed to the formation of deactivating biscarbonyl complexes, generated in the presence of in situ–formed CO gas. According to the authors, in case of C-3, the electron-withdrawing Ph group on the P atoms renders the ruthenium center less electron rich, which results in high lability of the axial carbonyl ligand. Consequently, the C-3 biscarbonyl complex reverts to the active catalytic dihydride complex responsible for formamide reduction to methanol. Based on the experimental observations, the authors proposed an inner-sphere mechanism for the first hydrogenation step of CO2 to formate/formamide without metalligand cooperation. On the other hand, the formamide hydrogenation was surmised to proceed through an outer-sphere mechanism involving the PNP ligand (Figure 4.23). Finally, through further optimization of the reaction conditions (catalyst amount, solvent amount), the authors obtained a highest methanol TON of 9900 with Ru-Macho-BH (C-10) at 145 °C and CO2/3H2 pressure of 75 bar after continuous reaction over 10 days. In 2017, Everett and Wass also reported an amine-assisted system to obtain methanol from CO2 via formamide intermediates (Figure 4.24) [38]. The authors screened several ruthenium complexes among which complex C-20 provided the best catalytic activity toward methanol. The authors found that the metal ligand cooperation was essential for methanol production, and thus surmising a formamide hydrogenation pathway proceeding through an outersphere mechanism. The structure of the amine also influenced the methanol formation rate. When a bulky amine was used, the corresponding sterically hindered formamide was more reactive toward hydrogenation. As a result, methanol TON increased almost 10-fold (from 240 to 2300) when the amine was switched from dimethylamine to diisopropylamine. Using complex C-20 and diisopropylamine, the authors obtained a highest methanol turnover of 8900 at 50 nmol catalyst loading at 180 °C, CO2/3H2 of 40 bar after two hours of reaction time. Prakash and co-workers [37] PEHA Cat.; K3PO4

CO2 + 3H2

H H

Ph

N Ru

P

H BH 3

Ru-Macho-BH C-10 TON MeOH

1050

H

Ph Ph

CO

Ph

H

H P

Ph

N

P Ru

P

Cl

Ru-Macho C-3 1040

H

Ph Ph

CO

Ph

formamide + CH3OH + H2O

Triglyme 145 °C, 40 h

iPr

N

P iPr

H P

Ru

iPr iPr

CO Cl

RuHClPNPiPr(CO) C-12 320

H

N

P Cy Cy

H P

Ru

Cy Cy

CO Cl

RuHClPNPCy(CO) C-13 50

H

N

tBu P tBu

P Ru

tBu tBu

CO Cl

RuHClPNPtBu(CO) C-14 0

Figure 4.22  Change in methanol TON with different substitutions on the P atoms.

105

106

4  Catalytic Homogeneous Hydrogenation of CO2 to Methanol H H

N

P Ru

P

Prakash and co-workers [37]

CO

P = PR2

Cl C-X Base – –Cl H2 H

H N

R2NH

P Ru

P H2

R

CO

H

H

NH2 R

C-XE H

H

H

H

N

Step I formate synthesis

P Ru

P

C-XF

N

P

CO

P Ru CO H C-XA

R

NH2

H H

R –

HCOO

R

R

P

NH2

Formate to formamide

CO 2

N

P Ru CO H

O

CH3OH

O C-XB

C-XC

H2

H2O

C-XA O H

N

H

R

R

O

OH

R

+ N

R

H

H

R

R H

CO + H

H N

P

H P

Ru

NH

Step II formamide reduction

CO

N P

H

R2NH3 P

Ru CO C-XC

H R2NH

N

P

P Ru CO CO C-XD

Resting state

C-XA H2

H2

Figure 4.23  Proposed mechanism of amine-assisted CO2 to methanol process.

In 2019, Kayaki and coworkers reported a CO2 to methanol system assisted by linear and branched polyethylenimines using Ru-Macho-BH catalyst (C-10) (Figure 4.25) [39]. The other catalysts screened by the authors did not produce significant amounts of methanol.

4.3  ­Homogeneous Catalysis – An Alternative for CO2 to Methanol Everett and Wass [38] Cat. R2NH

CO2 + nH2

Cl

P

P

P

Cl P

Cl

N N H HH Cl H

C-18

C-19

P

P

Maximum TON MeOH = 8900 Cl

P

Ru

Ru

HC(O)NR2 + CH3OH + H2O

NaOEt, toluene 180 °C

P

P

Ru

N H

N Cl

N

H

C-20

Cl

P

P Ru

N

N

Cl P Ru N

Cl

Cl

C-21

C-22

P = PPh2

Figure 4.24  Amine-assisted CO2 to methanol system reported by Everett and Wass. Source: Based on Everett and Wass [38].

Kayaki and co-workers [39] NH2

NH2

n

BPEI or H N

H2N

N

N H

N

N H

+ CO2 + 3H2 NH2 n

Cat.

CH3OH + H2O +

THF 130–150 °C

H2N

H

Ph

N

P Ph

Ru

H

H Ph P Ph CO

H BH3 C-10

NH2 n

LPEI

LPEI

H

n

BPEI or H N

H

Ph

N

P Ph

Ru

Ph P Ph CO

Cl C-3

H

N

P Cy Cy

Ru

H H

Cy P Cy CO

Cl C-13

Ph

N

P Ph

Ph P Ph

Ru Cl

N N

C-23

Figure 4.25  PEI-mediated CO2 to methanol through formamides.

Methanol turnover increased steadily as the temperature increased from 130 °C (TON: 33) to 160 °C (TON: 238), while the amount of formamide intermediate steadily decreased, indicating that a high temperature is required for formamide hydrogenation. When branched polyethylenimines (BPEIs) were used, the methanol TON increased with decreasing molecular weights (102 for BPEI10k compared to 362 for BPEI600). In the case of linear PEIs, no clear correlation with molecular weights could be determined. The authors obtained a highest TON of 599 with Ru-Macho-BH (C-10) as catalyst and linear PEI (Mn 25 000) as the amine, at a CO2/3H2 pressure of 80 bar and a reaction temperature of 150 °C after 100 hours. All the studies described so far for CO2 hydrogenation to methanol through formamide intermediates utilized ruthenium-based complexes. In 2017, Prakash and coworkers reported manganese-based pincer complexes for this transformation (Figure  4.26) [40].

107

108

4  Catalytic Homogeneous Hydrogenation of CO2 to Methanol Prakash and co-workers [40]

CO2

H

[Mn] t-BuOK RR′NH, H2 –H2O

H

P

P Mn CO Br C-4

R′

N

2H2

R TON max = 840

CO N

[Mn] t-BuOK

O

H

CO N

P

R

H N

R′

Yield up to 84% TON max = 36

H

P

CO N

Mn CO O

P = PiPr 2

CH3OH +

C-4A O

H

P

P Mn CO

H C-4B

Figure 4.26  Manganese-catalyzed CO2 hydrogenation to methanol.

The authors devised a two-step process, where in the first step amine (morpholine or benzylamine) was formylated with a CO2/H2 (70 bar) gas mixture at 110 °C, catalyzed by complex C-4. In the second step, the in situ–generated formamide was hydrogenated by the same complex (without any further addition of catalyst between the steps) under pure H2 pressure (80 bar) at 150 °C. The authors obtained a maximum TON of 840 for formamide synthesis and 36 for methanol formation using C-4 and morpholine. The formation of a manganese formate complex (C-4A) was observed in the solution after the formylation reaction in the first step. The formate complex was active for the hydrogenation of formamides to methanol. The authors surmised that at high H2 pressures, the formate complex gets decarboxylated to produce the hydride species (C-4B), which is the active catalyst for the hydrogenation of formamides. One-step direct hydrogenation of CO2 to methanol in a CO2/3H2 gas mixture with C-4 was not possible due to the sluggishness of the second step in the presence of CO2. Also in 2017, Pombiero and coworkers reported an iron scorpionate complex for CO2 to methanol in presence of PEHA with a maximum reported TON of 2260 [41].

4.3.6  CO2 to Methanol via Formic Acid Disproportionation Following the report by Goldberg and coworkers in 2013, catalytic formic acid disproportionation using molecular complexes has gained increased attention in recent years as a means to obtain methanol from formic acid in aqueous solutions at low temperatures [42]. In 2016, the research groups of Himeda and Laurenczy proposed a novel reaction scheme to obtain methanol from CO2 based on formic acid disproportionation (Figure 4.27) [43]. In the first step, CO2 is hydrogenated to formic acid (Eq. 1). In the second step, the formic acid disproportionation occurs to produce methanol and CO2 in a 1 : 2 ratio (Eq. 2). The authors successfully used an aqueous solution of iridium-based complex C-24 to catalyze both the CO2 hydrogenation to formic acid and formic acid disproportionation

4.4 ­Conclusio

steps at low temperatures (~70 °C) in a CO2/3H2 mixture (80 bar). Although methanol formation from a CO2/3H2 gas mixture was observed even without H2SO4 additive at room temperature after six days’ reaction, a higher reaction temperature and the presence of acid increased the rate of methanol formation through formic acid disproportionation. Using complex C-24 (15.9 μmol) in 2.5 m H2SO4, the authors obtained a TON of around 8 from a CO2/3H2 gas mixture at 70 °C after 50 hours. Whereas the catalytic activity clearly needs to be improved, this is the only report to date where CO2 was converted to methanol in a completely aqueous solution, without requiring any organic co-solvent. Also, the reaction proceeds at an overall lower reaction temperature compared to the other CO2 to methanol conversion processes.

4.4  ­Conclusion In recent years, significant advancements have been achieved in CO2 recycling to methanol through homogeneous catalysis. While methanol synthesis from CO2 via CO required high temperatures (>200 °C) in the first report by Tominaga and coworkers, the introduction of new hydrogenation catalysts for formate ester, organic carbonate, carbamate, urea, and formamide hydrogenation has opened up new doors for CO2 hydrogenation to methanol at low temperatures. CO2 can be hydrogenated to methanol in the presence of an alcohol additive, through formate ester, as demonstrated by the groups of Leitner (with ruthenium triphos complexes) and Beller (with cobalt complexes). Alternatively, in the presence of an amine, CO2 to methanol via formamide intermediates has been reported by the groups of Sanford, Prakash, and others. The amine-assisted CO2 to methanol process also allows for integration of CO2 capture with conversion. Following this strategy, CO2 from air, which is the ultimate CO2 sink, was captured and converted to methanol over multiple cycles. Furthermore, a novel CO2 to methanol route based on formic acid disproportionation has also been recently reported by the groups of Himeda and Laurenczy. From the viewpoint of large-scale application, the homogeneous CO2 to methanol process is still in need of improvement. Research efforts should be geared toward the development of better catalysts with improved TON and turnover frequencies (TOFs) for methanol formation. The number of catalysts that give high TOFs for this transformation over long periods of time is limited (Ru-Macho-BH being one of the most active and robust catalysts).

Himeda and Laurenczy and co-workers [43] OH 2+ CO2 + H2

HCOOH

(1)

3 HCOOH

CH3OH + H2O + 2CO2

(2)

CO2 + 3H2

CH3OH + H2O

Cp

*

N Ir

H2O

3x(1)+(2)

Figure 4.27  CO2 to methanol via formic acid disproportionation.

N C-24

OH

109

110

4  Catalytic Homogeneous Hydrogenation of CO2 to Methanol

Furthermore, the major problem of catalytic deactivation through CO formation also needs to be addressed. The homogeneous studies reported so far utilize batch reaction conditions. A flow reaction system based on homogeneous catalyst deposited onto solid support can perhaps enable production of methanol in a similar infrastructural setup as the ones for heterogeneous catalysis but at lower temperatures (120–150 °C). Based on the increasing interest displayed by the scientific community, along with the significant progress achieved over the last decade, the homogeneous CO2 to methanol process could play a vital role for the realization of the methanol economy concept, involving renewable methanol.

­References  1 CO2-earth. https://www.co2.earth/ (accessed 14 April 2019).  2 IPCC (2014): Climate Change 2014: Synthesis Report. Summary for Policymakers. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Core Writing Team, R.K. Pachauri and L.A. Meyer (eds.)]. IPCC, Geneva, Switzerland.  3 (a) Chang, J., Ciais, P., Viovy, N. et al. (2016). Global Change Biol. 22: 338–350.(b) Pan, S., Tian, H., Dangal, S.R.S. et al. (2015). Earth’s Future 3: 15–35.(c) Hao, Y., Chen, H., Wei, Y.-M., and Li, Y.-M. (2016). J. Cleaner Prod. 131: 667–677.  4 (a) Sanz-Pérez, E.S., Murdock, C.R., Didas, S.A., and Jones, C.W. (2016). Chem. Rev. 116: 11840–11876.(b) Goepper, A., Czaun, M., Prakash, G.K.S., and Olah, G.A. (2012). Energy Environ. Sci. 5: 7833–7853.(c) Boot-Handford, M.E., Abanades, J.C., Anthony, E.J. et al. (2014). Energy Environ. Sci. 7: 130–189.   5 (a) Owusu, P.A. and Asumadu-Sarkodie, S. (2016). Cogent Eng. 3: 1167990.(b) Obama, B. (2017). Science 355: 126–129.(c) Olah, G.A., Goeppert, A., and Prakash, G.K.S. (2009). J. Org. Chem. 74: 487.   6 Methanol Institute. https://www.methanol.org/ (accessed 23 April 2019).   7 Kar, S., Kothandaraman, J., Goeppert, A., and Prakash, G.K.S. (2018). J. CO2 Util. 23: 212–218.  8 (a) Olah, G.A., Goeppert, A., and Prakash, G.K.S. (2018). Beyond Oil and Gas: The Methanol Economy, Third, Updated and Enlarged Edition. Wiley-VCH Verlag GmbH & Co. KGaA.(b) Olah, G.A. (2005). Angew. Chem. Int. Ed. 44: 2636.(c) Olah, G.A., Goeppert, A., and Prakash, G.K.S. (2009). J. Org. Chem. 74: 487–498.(d) Olah, G.A., Prakash, G.K.S., and Goeppert, A. (2011). J. Am. Chem. Soc. 133: 12881–12898.  9 Carbon Recycling International (CRI). http://www.carbonrecycling.is/george-olah/ (accessed on 23 April 2019). 10 (a) Grabow, L.C. and Mavrikakis, M. (2011). ACS Catal. 1: 365–384.(b) Fujita, S., Usui, M., Ito, H., and Takezawa, N. (1995). J. Catal. 157: 403–413.(c) Yang, Y., Mims, C.A., Mei, D.H. et al. (2013). J. Catal. 298: 10–17. 11 (a) Dang, S., Yang, H., Gao, P. et al. (2019). Catal. Today 330: 61–75.(b) Behrens, M. (2014). Angew. Chem. Int. Ed. 53: 12022–12024.(c) Behrens, M., Studt, F., Kasatkin, I. et al. (2012). Science 336: 893. 12 (a) Studt, F., Sharafutdinov, I., Abild-Pedersen, F. et al. (2014). Nat. Chem. 6: 320.(b) Nguyen, H.K.D., Dang, T.H., Nguyen, N.L.T. et al. (2018). Can. J. Chem. Eng. 96:

 ­Reference

832–837.(c) Tang, Q., Shen, Z., Huang, L. et al. (2017). Phys. Chem. Chem. Phys. 19: 18539–18555. 13 (a) Martin, O., Martín, A.J., Mondelli, C. et al. (2016). Angew. Chem. Int. Ed. 55: 6261– 6265.(b) Frei, M.S., Capdevila-Cortada, M., García-Muelas, R. et al. (2018). J. Catal. 361: 313–321. 14 (a) Li, Y.-N., Ma, R., He, L.-N., and Diao, Z.-F. (2014). Catal. Sci. Technol. 4: 1498–1512.(b) Hertrich, M.F. and Beller, M. (2019). Metal-catalysed hydrogenation of CO2 into methanol. In: Organometallics for Green Catalysis (eds. P.H. Dixneuf and J.-F. Soulé), 1–16. Cham: Springer International Publishing. 15 (a) Courtemanche, M.-A., Légaré, M.-A., Maron, L., and Fontaine, F.-G. (2013). J. Am. Chem. Soc. 135: 9326–9329.(b) Courtemanche, M.-A., Légaré, M.-A., Maron, L., and Fontaine, F.-G. (2014). J. Am. Chem. Soc. 136: 10708–10717.(c) Huang, F., Zhang, C., Jiang, J. et al. (2011). Inorg. Chem. 50: 3816–3825. 16 (a) Riduan, S.N., Zhang, Y., and Ying, J.Y. (2009). Angew. Chem. Int. Ed. 48: 3322–3325.(b) Riduan, S.N., Ying, J.Y., and Zhang, Y. (2013). ChemCatChem 5: 1490–1496. 17 (a) Amao, Y. and Watanabe, T. (2004). Chem. Lett. 33: 1544–1545.(b) Boston, D.J., Xu, C., Armstrong, D.W., and MacDonnell, F.M. (2013). J. Am. Chem. Soc. 135: 16252–16255.(c) Kuwabata, S., Nishida, K., Tsuda, R. et al. (1994). J. Electrochem. Soc. 141: 1498–1503.(d) Wu, J., Huang, Y., Ye, W., and Li, Y. (2017). Adv. Sci. 4: 1700194.(e) Barton, E.E., Rampulla, D.M., and Bocarsly, A.B. (2008). J. Am. Chem. Soc. 130: 6342–6344. 18 (a) Obert, R. and Dave, B.C. (1999). J. Am. Chem. Soc. 121: 12192–12193.(b) El-Zahab, B., Donnelly, D., and Wang, P. (2008). Biotechnol. Bioeng. 99: 508–514.(c) Dibenedetto, A., Stufano, P., Macyk, W. et al. (2012). ChemSusChem 5: 373–378. 19 (a) Ohya, S., Kaneco, S., Katsumata, H. et al. (2009). Catal. Today 148: 329–334.(b) Kobayashi, T. and Takahashi, H. (2004). Energy Fuels 18: 285–286.(c) Giesbrecht, P.K. and Herbert, D.E. (2017). ACS Energy Lett. 2: 549–555. 20 (a) Tominaga, K.-i., Sasaki, Y., Kawai, M. et al. (1993). J. Chem. Soc., Chem. Commun.: 629–631.(b) Tominaga, K., Sasaki, Y., Watanabe, T., and Saito, M. (1995). Bull. Chem. Soc. Jpn. 68: 2837–2842. 21 Balaraman, E., Gunanathan, C., Zhang, J. et al. (2011). Nat. Chem. 3: 609–614. 22 Balaraman, E., Ben-David, Y., and Milstein, D. (2011). Angew. Chem. Int. Ed. 50: 11702–11705. 23 (a) Kothandaraman, J., Kar, S., Sen, R. et al. (2017). J. Am. Chem. Soc. 139: 2549–2552.(b) Xie, Y., Hu, P., Ben-David, Y., and Milstein, D. (2019). Angew. Chem. Int. Ed. 58: 5105–5109. 24 Han, Z., Rong, L., Wu, J. et al. (2012). Angew. Chem. Int. Ed. 51: 13041–13045. 25 (a) Kumar, A., Janes, T., Espinosa-Jalapa, N.A., and Milstein, D. (2018). Angew. Chem. Int. Ed. 57: 12076–12080.(b) Zubar, V., Lebedev, Y., Azofra, L.M. et al. (2018). Angew. Chem. Int. Ed. 57: 13439–13443.(c) Kaithal, A., Hölscher, M., and Leitner, W. (2018). Angew. Chem. Int. Ed. 57: 13449–13453. 26 Huff, C.A. and Sanford, M.S. (2011). J. Am. Chem. Soc. 133: 18122–18125. 27 Wesselbaum, S., Vom Stein, T., Klankermayer, J., and Leitner, W. (2012). Angew. Chem. Int. Ed. 51: 7499–7502. 28 Wesselbaum, S., Moha, V., Meuresch, M. et al. (2015). Chem. Sci. 6: 693–704. 29 Schneidewind, J., Adam, R., Baumann, W. et al. (2017). Angew. Chem. Int. Ed. 56: 1890–1893.

111

112

4  Catalytic Homogeneous Hydrogenation of CO2 to Methanol

30 Scharnagl, F.K., Hertrich, M.F., Neitzel, G. et al. (2019). Adv. Synth. Catal. 361: 374–379. 31 Khusnutdinova, J.R., Garg, J.A., and Milstein, D. (2015). ACS Catal. 5: 2416–2422. 32 Rezayee, N.M., Huff, C.A., and Sanford, M.S. (2015). J. Am. Chem. Soc. 137: 1028–1031. 33 Zhang, L., Han, Z., Zhao, X. et al. (2015). Angew. Chem. Int. Ed. 54: 6186–6189. 34 Kothandaraman, J., Goeppert, A., Czaun, M. et al. (2016). J. Am. Chem. Soc. 138: 778–781. 35 Kar, S., Sen, R., Goeppert, A., and Prakash, G.K.S. (2018). J. Am. Chem. Soc. 140: 1580–1583. 36 Kar, S., Goeppert, A., and Prakash, G.K.S. (2019). ChemSusChem 12: 3172–3177. 37 Kar, S., Sen, R., Kothandaraman, J. et al. (2019). J. Am. Chem. Soc. 141: 3160–3170. 38 Everett, M. and Wass, D.F. (2017). Chem. Commun. 53: 9502–9504. 39 Yoshimura, A., Watari, R., Kuwata, S., and Kayaki, Y. (2019). Eur. J. Inorg. Chem. 2019 (18): 2375–2380. 40 Kar, S., Goeppert, A., Kothandaraman, J., and Prakash, G.K.S. (2017). ACS Catal. 7: 6347–6351. 41 Ribeiro, A.P.C., Martins, L.M.D.R.S., and Pombeiro, A.J.L. (2017). Green Chem. 19: 4811–4815. 42 (a) Miller, A.J.M., Heinekey, D.M., Mayer, J.M., and Goldberg, K.I. (2013). Angew. Chem. Int. Ed. 52: 3981–3984.(b) Savourey, S., Lefèvre, G., Berthet, J.-C. et al. (2014). Angew. Chem. Int. Ed. 53: 10466–10470.(c) Chauvier, C., Thuéry, P., and Cantat, T. (2016). Chem. Sci. 7: 5680–5685.(d) Sordakis, K., Tsurusaki, A., Iguchi, M. et al. (2017). Green Chem. 19: 2371–2378. 43 Sordakis, K., Tsurusaki, A., Iguchi, M. et al. (2016). Chem. Eur. J. 22: 15605–15608.

113

5 Theoretical Studies of Homogeneously Catalytic Hydrogenation of Carbon Dioxide and Bioinspired Computational Design of Base-Metal Catalysts Xiuli Yan and Xinzheng Yang State Key Laboratory for Structural Chemistry of Unstable and Stable Species, Institute of Chemistry Chinese Academy of Sciences, Zhongguancun North First Street No. 2, Beijing, 100190, China Beijing National Laboratory for Molecular Sciences, Zhongguancun North First Street No. 2, Beijing, 100190, China

5.1 ­Introduction Carbon dioxide is one of the primary greenhouse gases closely related to global climate change. The rapid increase of atmospheric CO2 caused by the consumption of fossil fuels has prompted people’s strong desire for the conversion and utilization of CO2 as a cheap, nontoxic, and abundant C1 building block for the synthesis of valuable chemicals and fuels [1, 2]. An ideal way for CO2 recycling is the hydrogenation of CO2 to methanol (CO2 + 3H2 → CH3OH + H2O), which is not only widely used for the production of formaldehyde, acetic acid, olefins, dimethyl ether, and so forth, but also acts as a high-density hydrogen carrier (12.6 wt%) for different types of fuel cells. Although significant progresses have been achieved in developing homogeneous catalysts for hydrogenation of CO2 to formic acid/formate and methanol [2–5], most of the reported catalysts contain scarce noble metals and have rather low efficiency even in rigorous reaction conditions. The effective design of high-efficiency base-metal catalysts requires a deep understanding of related catalytic mechanisms, especially the metal–ligand cooperations (MLCs) between metal centers and redox non-innocent ligands. With the advent of modern quantum chemistry methods, especially the density functional theory (DFT), and dramatic increases of computer power in recent decades, theoretical and computational chemistry can now contribute much more to our knowledge – from understanding electronic structures and reaction mechanisms to predicting new structures and reaction pathways for rational catalyst design – and has led to numerous discoveries in organometallic chemistry and catalysis [6, 7]. In this chapter, we review recent advances in theoretical studies of homogeneously catalytic CO2 hydrogenation reactions, and discuss the key factors prompting H2 activation and CO2 reduction. Furthermore, we introduce some bioinspired computational design of basemetal catalysts as a mimic of the active sites of hydrogenases for hydrogenation of CO2 to formic acid and methanol. We hope that this chapter would provide some general guidelines and design principles for the development of base-metal catalysts for CO2 reduction reactions. CO2 Hydrogenation Catalysis, First Edition. Edited by Yuichiro Himeda. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.

114

5  Theoretical Studies of Homogeneously Catalytic Hydrogenation of Carbon Dioxide

5.2 ­H2 Activation and CO2 Insertion Mechanisms Because of the inertia, lack of polarity, and weak acidity of H2 and CO2, H2 activation and CO2 insertion are usually the turnover-limiting steps in catalytic CO2 hydrogenation reactions. The cleavage of strong covalent bond of H2 requires a rather high energy of 104.2 kcal mol−1 [8]. Fortunately, people have learned from nature that hydrogenases can efficiently catalyze the reversible oxidation of H2 with only base metal iron and nickel in their active sites [9]. The structural characters and catalytic mechanisms of the active sites of hydrogenases enlighten the design of efficient base-metal catalysts for H2 activation and CO2 reduction [10, 11]. Here we briefly introduce the active site structures of hydrogenases and common ways of H2 activation.

5.2.1  Hydrogen Activation Figure 5.1 shows the active site structures of three types of hydrogenases, [Fe]- [13, 14], [FeFe]- [15–17], and [NiFe]-hydrogenase [18, 19] (Figure 5.1a), and two common ways for heterolytic H2 cleavage, the in-sphere formation of a metal dihydrogen bond (Figure 5.1b) and the outer-sphere proton transfer to an external base (Figure 5.1c). The H2 activation catalyzed by [Fe]-hydrogenase is typically a heterolytic cleavage of H2 with the formation of an Fe─Hδ− … Hδ+─O dihydrogen bond. Based on Shima et al. crystal structure of wild-type [Fe]-hydrogenase [20], Yang and Hall proposed a “trigger” mechanism for the catalytic activation of H2, where the arrival of substrate

OC OH2

(a)

S

Guanyl nucleotide

Cys

O

C N HO CO

Fe

NC OC

L

M

Fe

C O

H2

H

H

M

L

M

S Cys S CO Fe CN X CN

S

[Fe4S4]

S

Ni

CN CO

S

Cys X = O or OH

[FeFe]-hydrogenase

H2 L

Cys

S S

Fe

[Fe]-hydrogenase

Cys

Cys

HN

[NiFe]-hydrogenase

‡

H

H

L

M

(b)

H2

H2 L (c)

M

L

M

H H

Base L

M

Base

TS

H

L

HBase

M

H

L

M

BaseH+

Figure 5.1  (a) The structural diagrams of the active sites of three types of hydrogenases. Source: From Xu et al. [12], © 2015 Elsevier. Reprinted with permission of Elsevier. (b) The inner-sphere formation of metal dihydrogen bond. (c) The outer-sphere proton transfer to an external base.

5.2  ­H2 Activation and CO2 Insertion Mechanisms

N5,N10-methenyl-tetrahydromethanopterin (MPT+) triggered the electron release from the pyridone ligand and allowed heterolytic cleavage of H2 between Fe and MPT+ in an outersphere way [21]. Later on, based on the crystal structure of a mutated [Fe]-hydrogenase reported by Hiromoto et al. [22], Yang and Hall proposed a new H2 activation mechanism with the formation of a strong Fe─Hδ− … Hδ+─O dihydrogen bond in the resting state of the catalytic reaction (Scheme  5.1) [23]. Different with the “trigger” mechanism proposed based on the crystal structure of wild-type [Fe]-hydrogenase with two open sites, the essential role of the iron center in this new mechanism is H2 capture and hydride formation without MPT+, while the pyridone’s special role involves the protection of the hydride by the formation metal dihydrogen bond. Meanwhile, the new mechanism features an unexpected dual pathway for inner-sphere H2 cleavage with proton transfer to Cys176-sulfur or 2-pyridinol’s oxygen for the formation and regeneration of the resting state with a strong Fe─Hδ− … Hδ+─O dihydrogen bond. Therefore, the pyridone ligand in the active site of [Fe]hydrogenase played an essential role for the heterolytic cleavage of H2. Inspired by the above H2 activation mechanisms and the active site structures of hydrogenases, as well as the characters of some recently reported redox non-innocent ligands, Yang and coworkers proposed a series of Mn, Fe, and Co complexes and computationally examined their catalytic activities for hydrogenation of CO2 to formic acid [24, 25] and methanol [26–28]. The predicted structures and reaction mechanisms of those newly designed base-metal complexes will be further introduced in Sections 5.3 and 5.4. With the participation of an external base, H2 can be split into an outer-sphere way with direct proton transfer. In a theoretical study of CO2 hydrogenation catalyzed by an aromatic PNP (PNP = 2,6-bis(di-isopropylphosphinomethyl)pyridine) iridium(III) trihydride complex with the presence of OH−, Yang proposed a hydroxyl-participating direct proton transfer mechanism for the cleavage of H2 and predicted two base-metal complexes, trans-(PNP) Fe(H)2CO and (PNP)CoH3, as potential catalysts for hydrogenation of CO2 [29]. Chen and coworkers [30] reported a tetraphos-ligated cobalt complex for catalytic hydrogenation of CO2 with the aid of NH(CH3)2. Pati and coworker [31] studied the mechanism of the Mn-catalyzed hydrogenation of CO2 to formate with the involvement of an external base, DBU (1,8-diazabicyclo[5.4.0]undec-7-ene). The critical roles of these external bases for H2 activation and detailed mechanistic insights for hydrogenation of CO2 will be introduced in Section 5.3.

5.2.2  Insertion of CO2 As summarized by Yoshizawa and coworker [32], there are usually two ways for the insertion of CO2 into a metal-hydride bond, the pre-coordination of CO2 to the metal center followed by hydride migration to CO2 (path 1 in Figure 5.2) and the direct hydride transfer from metal to the carbon atom in CO2 followed by the rotation of formate (path 2 in Figure 5.2). A representative example of above two pathways is the dehydrogenation of formic acid catalyzed by an Fe complex, [(PP3)FeH]+ (PP3  =  tris[2-(diphenylphosphino)ethyl]phosphine), reported by Beller and coworkers [33]. Yang’s computational study of this reaction revealed two possible pathways: the cationic pathway with β-hydride elimination (Cycle 1 in Scheme 5.2a) and direct hydride transfer from formate to Fe (Cycle 2 in Scheme 5.2b) [34].

115

+

H3C

S O C H OC Fe N +

H

MHPT

OH

H O CO

H-transfer +

+ MPT

H3C

SO C H OC Fe N H

H3C

SO C H OC Fe N H

0

+

H3C

SO C H OC Fe N H H O MPT CO 0

OH

TS

SO C H OC Fe N

OH

HMPT

SO C H OC Fe N H2

OH

MPT + =

H2 cleavage

O

H O CO

H2N

H3C

SO C H OC Fe N H

H O CO

0

OH

H3C OC H2

H

+ N

N

HN

SO C H OC Fe N

N

N H

SO C Fe N

CH3

H2

H

H

H CH3

+ H2 +

H3C

TS OH

– HMPT – H+

O

CO

OH

H O CO

0

H3C

H O CO

+

H3C

CO

OH

O H

0

OH

– H+

H O CO

TS Scheme 5.1  Yang and Hall’s mechanism for H2 activation catalyzed by [Fe]-hydrogenase with the formation of a strong Fe─Hδ− … Hδ+─O dihydrogen bond. Source: From Yang and Hall [23], © 2009 American Chemical Society. Reprinted with permission of American Chemical Society.

5.2  ­H2 Activation and CO2 Insertion Mechanisms O

Path 1 M

O

C

M

H

O

H + CO2

M O C O

O

Path 2

M

H

M

C

H

H C O

O

Figure 5.2  Two typical ways for CO2 insertion into a metal–hydride bond. Source: Redrawn from Li and Yoshizawa [32]. +

H +

P3PFe

HCOOH

H

TS

P3PFe H

H+ transfer

O

P3PFe H2

TS

H2

PPh2

CO2 PP3Fe = P

+

O

PPh2

O

P3PFe

+

P3PFe β-H elimination

+

O

+

O C O

TS

H

O

Cycle 1

Fe PPh2

C

(a)

O

H

O

+

H

O

O

TS

H

H

P3PFe O

H

TS

P3PFe

O O

H

P3PFe H –

+ HCOO H2

–

+

P3PFe

HCOO

H

TS PPh2 +

H2

P3PFe

PP3Fe = P

H

PPh2 PPh2

H

–

O

O

Fe

P3PFe H

Cycle 2 TS – H transfer +

H

P3PFe (b)

O C O

H

H

H

CO2

P3PFe H

Scheme 5.2  Two plausible cycles for the dehydrogenation of formic acid catalyzed by [(PP3)FeH]+ with β-hydride elimination (a), and direct hydride transfer (b). Source: From Yang [34],© 2013 Royal Society of Chemistry. Reprinted with permission of Royal Society of Chemistry.

117

118

5  Theoretical Studies of Homogeneously Catalytic Hydrogenation of Carbon Dioxide

DFT calculation results confirmed that the rate-determining step in the formic acid dehydrogenation catalyzed by [(PP3)FeH]+ is the β-hydride elimination process for hydride transfer from formate to Fe (Scheme 5.2a). A pathway with direct hydride transfer from HCOO− to Fe is possible, but 2.3 kcal mol−1 is less favorable than Cycle 1.

5.3 ­Hydrogenation of CO2 to Formic Acid/Formate 5.3.1  Catalysts with Precious Metals In 1976, Inoue et al. [35] reported the first catalytic hydrogenation of CO2 to formic acid with precious metals, such as Ru, Rh, and Ir, as the catalysts. Following their pioneering work, significant progresses have been achieved in understanding the mechanistic insights of transition metal–catalyzed CO2 hydrogenation to formic acid/formate, as well as computational design of non-noble catalysts for such reactions. In 2000, Sakaki and coworker [36] reported a theoretical study of ruthenium-catalyzed hydrogenation of CO2 to formic acid using a simplified model structure of cis-RuH2(PH3)4, and proposed a new type of σ-bond metathesis mechanism (Scheme 5.3). Their DFT calculations indicate that the insertion CO2 into the Ru─H bond is difficult in cis-RuH2(PH3)4 but much easier in cis-RuH2(PH3)3, the activation energy of six-membered σ-bond metathesis of RuH(η1-OCOH)(PH3)3(H2) is much lower than the activation energies of four-membered σ-bond metathesis and five-membered H─OCOH reductive elimination, and the rate-determining step is CO2 insertion irrespective of whether PH3 is dissociated from cisRuH2(PH3)4. Later on, Sakaki and coworkers [37] computationally investigated the same reaction using the real catalyst, cis-Ru(H)2(PMe3)3, and found that the reaction takes place more easily because of the stronger donating ability and trans-influence of PMe3 than PH3. The rate-determining step in the whole catalytic reaction is still the insertion of CO2 into

L3Ru HCOOH

L3Ru

CO2 insertion CO2

H

Six-centered σ−bond metathesis H

H

H

O C

H

L3Ru H

O

C

O

H

O

L = PMe3

H

H2

Scheme 5.3  Sakaki’s mechanism for the hydrogenation of CO2 to formic acid catalyzed by cis-RuH2(PMe3)3 with CO2 insertion and six-centered σ-bond metathesis. Source: From Ohnishi et al. [37], © 2005 American Chemical Society. Reprinted with permission of American Chemical Society.

5.3  ­Hydrogenation of CO2 to Formic Acid/Formate

the Ru(II)─H bond. Such results indicate that the use of real catalyst structures without simplification is important to obtain correct reaction barriers in computational studies. In 2009, Nozaki and coworkers [38] reported a hydrogenation of CO2 reaction in aqueous potassium hydroxide with a PNP-Ir(III) trihydride complex, [Ir(H)3PiPrNPyPiPr], as the catalyst, which achieved an unprecedented high activity for the formation of potassium formate with turnover numbers (TONs) up to 3 500 000 at 120 °C and turnover frequencies (TOFs) up to 150 000 h−1 at 200 °C and 50 bar pressure. They also proposed a plausible mechanism based on experimental observations (Scheme  5.4). Ahlquist [39] reported a theoretical study of this reaction using a very simplified structure model of the Ir(III) complex, in which the isopropyl groups in the PNP pincer ligand were replaced by hydrogen atoms. Ahlquist proposed a two-step mechanism for the insertion of CO2 into an Ir─H bond, the formation of an H-bound formate complex followed by the coordination of a H2 molecule to iridium. The generation of the iridium(III) trihydride complex is calculated to be the rate-limiting step with a total free energy barrier of 26.1 kcal mol−1. Following their works, Yoshizawa and coworker [32] examined two routes for the insertion of CO2 into the Ir─H bond of the [Ir(H)3PiPrNPyPiPr] complex, hydride migration to CO2 with the precoordination of CO2 to the metal center and the direct addition of hydride to CO2 via nucleophilic attack. Their computational results indicate that the direct addition of hydride to CO2 is more favorable. The rate-limiting step in the whole catalytic reaction is the regeneration of the active complex with a barrier of 15.6 kcal mol−1. Based on molecular orbital and natural bond orbital analyses, they found that this trihydride complex consists of two kinds of hydrides with distinct Ir─H bond properties, and the type of metal–hydride bond may affect the route of CO2 insertion. Later on, Nozaki and coworkers computationally studied this Ir(III) catalytic system and proposed a detailed mechanism with two competitive reaction pathways for the hydrogenation of CO2 (Scheme 5.5) [40]. The rate-determining steps of those two pathways are deprotonative dearomatization (TSc) and hydrogenolysis (TSe). The calculated free energy PR2 H N Ir H H PR2

H2

PR2 H N Ir H

CO2

PR2 H N Ir H O O PR2 H

PR2

H2O – HCO2

OH–

Scheme 5.4  Nozaki and coworkers’ plausible mechanism for hydrogenation of CO2 catalyzed by (PNP)IrH3. Source: From Tanaka et al. [38], © 2009 American Chemical Society. Reprinted with permission of American Chemical Society.

119

120

5  Theoretical Studies of Homogeneously Catalytic Hydrogenation of Carbon Dioxide

TSd

iPr P 2

H

H H i P Pr2 Ir H N H

TSa

O

H

H iPr P PiPr2 Ir 2 H N H

A

+ CO2

iPr P 2

H H

Ir N

H

C O PiPr2

+ H2

iPr

H 2P

Ir N

TSb

A′*H2O

H PiPr2

TSe iPr P 2

– H2O

OH–

H PiPr2 Ir H N H

+

H OCOH PiPr2 Ir H N

– OCOH–

H OH2 iPr P PiPr2 Ir 2 H N

TSc

H

iPr P 2

iPr P 2

iPr

H 2P

H

Ir N

OH PiPr2 H H

H + Ir H N

PiPr2

+ OH–

Scheme 5.5  Mechanism for the hydrogenation of CO2 catalyzed by (PNP)IrH3 proposed by Nozaki and coworkers based on DFT calculations. Source: From Tanaka et al. [40], © 2011 American Chemical Society. Reprinted with permission of American Chemical Society.

profiles of those cycles are dependent on the strength of the base, hydrogen pressure, and solvent. Almost at the same time, Yang computationally studied the mechanism of this Ir(III) catalytic system and proposed two base-metal complexes, trans-(PNP)Fe(H)2CO and (PNP) CoH3, as promising catalysts for the hydrogenation of CO2 to formate with the presence of OH− [29]. As shown in Scheme 5.6, Yang’s mechanism reveals the essential role of hydroxyl in the catalytic CO2 reduction cycle and suggests that the incorporation of strong bases and unsaturated ligands may be critical for the design of new catalysts for efficient hydrogen activation. His calculation results indicate that the direct out-sphere H2 cleavage by OH− without the participation of the PNP ligand is about 20 kcal mol−1 more favorable than the ligand-participating H2 cleavage by dearomatization and rearomatization of the pyridine ring in the PNP ligand. The newly proposed iron and cobalt pincer complexes have total enthalpy barriers of 21.9 and 22.6 kcal mol−1, respectively. Such low barriers indicate those base-metal complexes are promising low-cost and high-efficiency catalysts for CO2 reduction under mild conditions. It is worth to note that the computationally predicted iron complex, trans-(arPNP)Fe(H)2CO, was independently synthesized by Milstein and coworkers [41] in their experimental study of hydrogenation of ketones catalyzed by iron pincer

N

M

H– transfer

PiPr2

H

N

TS

L

P H iPr 2

–

CO2

PiPr2

H

N

M

L P iPr H 2 C O O

HCOO–

M

–

HCOO

H 2O

N

TS H2 cleavage

OH

PiPr2

H

+

–

N

M

TS L P H H2 cleavage iPr 2 H O H

M

H M

P H2 iPr 2

PiPr2 L

H2O

N

N

M O

L P C H iPr 2 O

HCOO–

OH

TS H transfer

PiPr2

H M

P OH 2 iPr 2

H

M

P H2 iPr 2

PiPr2 L

H2

H2O

–

HCOO

H2

H

H

H2O N

L

–

+

H

L

TS H+ transfer

PiPr2

H

-

N

OH 2

H

H2

OH

H2O

H

L

P H2 iPr 2

TS H+ t rans f er

N

PiPr2

H

M

P

iPr

PiPr2

H

N

L

P iPr 2

H2

OH

PiPr2

H

H

PiPr2

M L P O C H iPr 2 O

N P iPr 2

H M

PiPr2 L

Scheme 5.6  Yang’s mechanism for the hydrogenation of CO2 to HCOO− catalyzed by (PNP)IrH3, (PNP)Fe(H)2CO, and (PNP) CoH3. Source: From Yang [29], © 2011 American Chemical Society. Reprinted with permission of American Chemical Society.

122

5  Theoretical Studies of Homogeneously Catalytic Hydrogenation of Carbon Dioxide

complexes. Later on, Milstein and coworkers [42] reported a very similar trans-(PNP) Fe(H)2CO complex, in which the isopropyl groups were replaced by tert-butyl, for lowpressure hydrogenation of CO2 and achieved TONs up to 788 and TOFs up to 156 h−1. In 2011, Hazari and coworkers [43] reported an air-stable and highly active water-soluble Ir(III) complex with aliphatic PNP pincer ligand, IrH3[(iPr2PC2H4)2NH], as a catalyst for CO2 reduction and achieved a maximum TON of 348 000 and a high TOF of 18 780 h−1. They also computationally predicted the thermodynamic favorability of CO2 insertion into Ir(III) hydrides, and proposed an outer-sphere mechanism with the displacement of formate by H2 as the rate-determining step for hydrogenation of CO2 (Scheme 5.7). The catalytic cycle contains three key steps, the formation of formate by CO2 insertion, the displacement of coordinated formate by H2, and the deprotonation of H2 for the regeneration of the trihydride complex. The DFT calculation results indicate that a hydrogen bond donor in the secondary coordination sphere is likely to facilitate CO2 insertion and formate production. Later on, Hazari and coworkers [44] investigated a series of Ni(II) and Pd(II) hydride complexes with PCP and PSiP ligands for catalytic CO2 insertion reactions. DFT calculations indicate that the strong trans influence of the anionic carbon donor makes CO2 insertion thermodynamically favorable by destabilizing the metal–hydride bond in the complexes containing PSiP- and PCP ligands. In contrast, the CO2 insertion in the PNP pincer complexes is thermodynamically unfavorable. The insertion of CO2 into the metal– hydride bond goes through a four-centered transition state. The activation energy of this transition state decreases with the increase of the trans influence of the anionic donor in the pincer ligand. In addition to the catalysts with aromatic and aliphatic PNP pincer ligands, metal complexes with N,N–chelated ligands have also been well studied in recent years. Thiel and coworkers [45] reported a series of pyridinylazolato Ru(II) complexes, [(N–N′)RuCl(PMe3)3], and proposed a hydride migration mechanism for catalytic hydrogenation of CO2 based on H N

H 2O

E

H Ir H

–

OH

CO2

H E

A H

E = PiPr2 O

H N

H E

H +

Ir H

C H N

H E

E

O Ir H

H E

B

C HCOO–

H2

Scheme 5.7  Hazari and coworkers’ mechanism for the hydrogenation of CO2 catalyzed by IrH3[(iPr2PC2H4)2NH]. Source: From Schmeier et al. [43], © 2011 American Chemical Society. Reprinted with permission of American Chemical Society.

5.3  ­Hydrogenation of CO2 to Methanol

DFT calculations (Scheme 5.8). The rate-determining step was found to be the dissociation of the formate hydrogen atom from the ruthenium center for the formation of a 16-electron intermediate. The calculation results indicate that the activation energy barrier rises with the increase of electron-withdrawing ability of pyrazolylpyridine series. Himeda and coworkers [46] recently reported a proton-switchable iridium catalyst, which is the first reversible and recyclable hydrogen storage system under mild conditions using CO2, formate, and formic acid. To understand the essential role of THBPM (THBPM  =  4,4′,6,6′-tetrahydroxy-2,2′-bipyrimidine) in catalytic hydrogenation of CO2 reactions, they compared the activities of [Ir(Cp*)(4DHBP)(OH2)]2+ (nDHBP  =  n,n′dihydroxy-2,2′-bipyridine), [Ir{(Cp*)(OH2)}2(THBPM)]4+, and [Ir{(Cp*)(OH2)}2(BPM)]4+ (BPM = 2,2′-bipyrimidine). It is worth noting that the THBPM ligand can be fully protonated (pH  5) with a midpoint at pH 3.8. They computationally investigated the mechanism of CO2 hydrogenation to formic acid catalyzed by the Ir complex [Ir{(Cp*)(OH2)}2(THBPM)]4+. As shown in Scheme  5.9, the catalytic cycle involves three steps: heterolytic cleavage of H2 for the formation of a metal hydride, rate-determining CO2 insertion into the Ir─H bond, and the dissociation of HCOOH. [Ir{(Cp*) (OH2)}2(THBPM)]4+ can achieve a rather high TOF of 53 800 h−1 and TON of 153 000 N

N N

R

HCOOH

N

R

N

Ru

CO2

R

HH

N

O O

N

H Ru

C

R

R H

N

Ru

H N

R N

R N

Ru

H

H

R

C O

R = PMe3 ligand N

O

O C

N

O

R N

H

Ru O

N

H C

O

R N

R N

Ru

H

N H

H

O C

N R

R N

Ru O

N

O

H C

O

R

H2 Scheme 5.8  Thiel and coworkers’ mechanism for the hydrogenation of CO2 catalyzed by (pyridinylazolato)ruthenium complexes. Source: From Muller et al. [45], © 2013 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

123

124

5  Theoretical Studies of Homogeneously Catalytic Hydrogenation of Carbon Dioxide OH

HO Cp* H2O

N

N Ir

Ir N HO

N OH

Cp* OH2

4H+ + 4H+

Cp* H2O

N

N Ir

Ir N O

N

O

O

O

O

Cp*

2H2O

OH2 + 2H2O

O

[{Ir(Cp*)(OH2)}2(THBPM)] [{Ir(Cp*)(OH2)}2(THBPM)]4+ pH > 5 pH < 2 pKa = 3.8

N

N

N

N

(Cp*)Ir

Ir( )(Cp*)

O

O

H2 + H2 OH

O N

N

N

N

(Cp*)Ir

Ir(H)(Cp*)

O

a. HCO2–, H+ b. + H2

O

+ CO2 OH

O N

N

N

N

(Cp*)Ir O

Ir(OCHO)(Cp*) O

Scheme 5.9  Himeda and coworkers’ proton-switchable iridium catalyst for reversible and recyclable hydrogen storage using CO2. The box symbols in the artworks of those schemes represent vacant positions at the metals. Source: From Hull et al. [46], © 2012 Springer Nature. Reprinted with permission of Springer Nature.

(5 MPa pressure and 80 °C) for the transformation of 1 : 1 mixture of H2 and CO2 to formate. The high catalytic activity of this Ir complex is attributed to the pendant hydroxy groups, which act as proton reservoirs and strong π-donors for such reversible conversion of H2 and CO2 to formate in aqueous solvent under mild temperatures. Furthermore, Himeda and coworkers [47] reported a series of water-soluble Ir(III) complexes, [Cp*Ir(6DHBP)(OH2)]2+ and their analogues [Cp*Ir(6,6′-R2-bpy)(OH2)]SO4 (bpy = bipyridine; R = OMe, Me, H) for hydrogenation of CO2 at ambient temperatures and pressures. The nuclear magnetic resonance (NMR) experiments and DFT calculations indicate that [Cp*Ir(6DHBP)(OH2)]2+ is easier to form Ir─H species than [Cp*Ir(4DHBP) (OH2)]2+. The calculation results also confirmed that the heterolysis of H2 is the rate-determining step under weak basic conditions (pH 8.3), and suggested that the CO2 insertion process is stabilized by a weak hydrogen-bonding interaction with the deprotonated pendant base (structure E in Scheme 5.10). Moreover, the reaction rates were enhanced with the movement of substituents from 4,4′ to 6,6′ positions in the 2,2′-bipyridine ligand. Theoretical studies suggested that the rate enhancement was also caused by the presence of a pendant base in the second-coordination sphere, which facilitates the heterolytic cleavage of H2. The above experimental and computational studies clearly show the participation a solvent water molecule in the heterolytic H2 cleavage catalyzed by [Cp*Ir(6DHBP)(OH2)]2+ and [Cp*Ir(N2)(OH2)]2+ bearing a pendant base [48], and stabilizing the transition state as a proton transfer shuttle (Scheme 5.11). Using [Cp*Ir(6DHBP-2H+)] as the catalyst, Himeda and coworkers found that the direct heterolysis of H2 is 3.4 kcal mol−1 higher than waterassisted H2 cleavage in free energy. Further computational investigations for the

5.3  ­Hydrogenation of CO2 to Formic Acid/Formate HCO2 O N

Ir

O C

N

Ir

Ir

H2O

N

Ir

O

B O

Ir

N

Cp*

O

A*

H

N

E

N

OH2

O

N

CO2

Ir

H2

Cp*

N

N

O

pH = 8.3 1atm 1 Minsolvent

O

+ H2O Cp*

A

O C

O

N

O

F

H

N O

Cp*

N

O

O

H

H

Cp*

O

O H H

N

+ H+

Cp*

Ir

+

H

N

O

D

H Cp*

O

C

Scheme 5.10  Himeda and coworkers’ mechanism for the hydrogenation of CO2 catalyzed by [Cp*Ir (6DHBP)(OH2)]2+. The box symbols in the artworks of those schemes represent vacant positions at the metals. Source: From Wang et al. [47], © 2012 Royal Society of Chemistry. Reprinted with permission of Royal Society of Chemistry.

N

Ir Cp*

N

PT

0

O

O H

H2

N

H2O

N

O

(a)

Ir

O

H H Cp*

H

0

PT

N

Ir

N

H

H O

H

Cp* O

O

(b)

0

O H

(c)

Scheme 5.11  Mechanism for H2 heterolysis assisted by the pendant base and water-assisted proton transfer (PT). The box symbols in the artworks of those schemes represent vacant positions at the metals. Source: From Wang et al. [48], © 2013 American Chemical Society. Reprinted with permission of American Chemical Society.

mechanisms of the interconversion between CO2 and formic acid catalyzed by similar half sandwich Ir and Co complexes have been well established in a series of studies by Ertem et al. [49]. In 2014, Himeda and coworkers [50] reported another combined experimental and computational study of the positional effects of hydroxy groups on the activity of CO2 hydrogenation catalyzed by similar proton-responsive half-sandwich Cp*Ir(III) complexes with two hydroxy groups at the 3,3′-, 4,4′-, 5,5′-, or 6,6′-positions (3DHBP, 4DHBP, 5DHBP, or

125

126

5  Theoretical Studies of Homogeneously Catalytic Hydrogenation of Carbon Dioxide

6DHBP) under acidic and basic conditions. Among those complexes, 6DHBP exhibits enhanced catalytic activity with a TOF of 1400 h−1 compared to 4DHBP (TOF = 650 h−1) for hydrogenation of CO2. DFT calculations indicate that the H2 heterolysis catalyzed by 4DHBP is 2.9 kcal mol−1 less favorable than the reaction catalyzed by 6DHBP with a waterassisted proton relay. The ortho-oxyanion in 6DHBP is the key factor for faster formation of formate by facilitating the rate-determining H2 cleavage process as a pendant base. In addition to the above studies of half-sandwich Ir complexes, Pidko and coworkers [51] performed a combined in situ NMR spectroscopy and DFT investigation for the impact of MLC in the hydrogenation of CO2 catalyzed by aromatic (PNP)Ru pincer complexes. They observed a TOF of 14 500 h−1 with complex C in Scheme 5.12 as the catalyst under 70 °C and 40 bar H2/CO2, while a higher TOF of 21 500 h−1 for complex D with the presence of DBU. They also found that the pathways involving ligand participation do not contribute to the catalytic cycle. The extra water molecule could convert C–F and re-activate the reaction. Their DFT calculation results indicate that the Ru-formate complex E is the most stable compound and also the resting state in the catalytic cycle. Moreover, the formation of formic acid goes through an outer-sphere mechanism over complex D, in which a noncoordinated HCOO− anion assists heterolytic cleavage of H2 in a fashion of frustrated Lewis pairs.

N (PtBu)2

Ru H

C O

P(tBu)2

R = tBu

Cl

A

KOtBu O N R2P H

Ru

PR2

H2

N R2P H

H C O D

CO2

H

Ru C O

C O

CO2

PR2

B H2

N

PR2 H O

E*

R2P O

H

Ru C O

N R2P H

HCOOH

N R2P

Ru

PR2 O H

E

C O

O

PR2

C

H2O

H2, DBU

N R2P

O

Ru

H

Ru C O

PR2 O

O

HO

F

Scheme 5.12  The transformations of H2 and CO2 catalyzed by (PNP)Ru complexes with the presence of DBU observed by Pidko and coworkers. Source: From Filonenko et al. [51], ©2013 American Chemical Society. Reprinted with permission of American Chemical Society.

5.3  ­Hydrogenation of CO2 to Formic Acid/Formate

Subsequently, Pidko and coworkers [52] performed a further DFT study for CO2 hydrogenation catalyzed by Ru-PNP pincer complexes with the presence of DBU. As shown in Scheme  5.13, they found three alternative reaction channels (Cycles I, II, and III). The C-CO2 P

CO2

H N H Ru P P H C H O

C

D*

N Ru P H C H O O C O

P

N Ru H C H O

Cycle I

C

O

P

N Ru P H C H C-FA O H O O P

B°-FA

H N Ru P P HO H C O O D-H2 H

N Ru

P H H C H O

HCO2

–

HDBU+ HCO2

P

A-H2

H2

N Ru H C O

A

O C O

P

CO2

P

DBU

O C

OH N Ru P P H C H O C

F

A-FA

DBU

O C N O O Ru P P P H – O HDBU+ HCO 2 H C H2 O H O B O C

N Ru P H C O O O H

H Cycle II

N Ru P P H H C H O

N Ru P H C

O

H H

H2

HDBU+ HCO2–

DBU

D

H Cycle Ia H P P

Cycle III

P

N Ru H C O

N Ru H C O

P

A-CO2

O C

O

P

B°-H2

O O C

N OH N Ru P P Ru P P H C H C O E O H O CO2 C O E-CO2

O C

OH

Scheme 5.13  Pidko and coworkers’ mechanism for CO2 hydrogenation to formates catalyzed by (PNP) Ru complexes (tBu groups are omitted for clarity) [52]. Source: Redrawn from Filonenko et al. [52].

127

128

5  Theoretical Studies of Homogeneously Catalytic Hydrogenation of Carbon Dioxide

barrier of direct hydride transfer from bis-hydrido Ru-PNP complex (complex C in Cycle I) to CO2 is only about 5 kcal mol−1. However, the formation of a stable formato-complex and the difficulty of H2 insertion into the strong Ru─OCHO bond led the total barrier of the catalytic reaction to near 15.5 kcal mol−1. They also suggested that the reaction rate can be promoted under H2-rich conditions. In 2016, Dang and coworkers [53] computationally explored the steric and electronic effects of bidentate phosphine ligands in three different [Ru(diphosphine)2H2] complexes for catalytic hydrogenation of CO2 to formic acid. As shown in Scheme 5.14, the hydrogenation reaction consists of three major steps: cis–trans isomerization of Ru dihydride complexes, CO2 insertion into the Ru─H bond, and H2 insertion into the ruthenium and formate ions. For all three ruthenium complexes, CO2 insertions always have the lowest barriers in those three steps. The calculation results also indicate that the bidentate phosphine ligands have rather weak steric effects, but significant electronic effects in cis–trans isomerization and H2 insertion steps. Kühn and coworkers [54] synthesized water-soluble rhodium and iridium complexes with bis-N-heterocyclic carbene (bis-NHC) for hydrogenation of bicarbonate to formate and dehydrogenation of formic acid. The rhodium complex has higher activity and reached a TOF of 39 000 h−1 and a TON of 449 000 for H2 production. They also proposed an overall mechanism with three steps, the replacement of the chloride ligand by bicarbonate, the reduction of bicarbonate to formate, and the replacement of formate by bicarbonate, based on DFT calculations (Scheme 5.15), and found that a second catalyst molecule is involved in the reaction by providing an external hydride as a reducing agent for the reduction of bicarbonate. Ke and coworkers [55] systematically studied the mechanisms of CO2 hydrogenation catalyzed by half-sandwich transition metal complexes [Cp*M(6,6′-R2-bpy)(OH2)]2+ (M = Co, Rh, Ir, bpy = 2,2′-bipyridine, R = OH) and analyzed the metal effect in reaction through DFT calculations (Scheme  5.16). They found all three metal complexes process a similar mechanism with the heterolytic cleavage of H2 as the rate-determining step. The Ir complex has the lowest activation free energy of 13.4 kcal mol−1. The d orbital back-donating abilities of metal atoms were attributed as the primary influence for the H2 cleavage barriers. Their computational study also reveal the special role of the bipyridinol ligand, which significantly facilitates the catalytic cycle by acting as a base (deprotonated) for intramolecular heterolytic cleavage of H2, and a Brønsted acid (protonated) to lower the lowest unoccupied molecular orbital (LUMO) of CO2 for hydride transfer.

5.3.2  Catalysts with Non-noble Metals With the concerns of cost and environmental impact, the development of efficient basemetal catalysts for hydrogenation of CO2 is highly desired and attracting increasing attention in recent years. In 2013, Badiei et al. [56] reported a series of water-soluble half-sandwich Cp*Co(III) complexes with proton-responsive bipyridine ligands (4,4′- and 6,6′-dihydroxy-2,2′-bipyridine, 4DHBP and 6DHBP, respectively) as cobalt analogues of previously reported half-sandwich iridium complexes for hydrogenation of CO2 in aqueous conditions [46–48]. However, the cobalt complexes containing 4DHBP ligands have rather

H

CO2

H Ru P

P

P

A: PCH2CH2P B: PCH2CH2CH2P C: PCH2NMeCH2P

Ru P

O

O

C H

P

P

P

Ru

P

H

O

P P

H

H H

P

Ru

P

H H

P

O

P

A: ∆G≠ = 23.9 kcal mol−1 B: ∆G≠ = 23.4 kcal mol−1 C: ∆G≠ =18.7 kcal mol−1

C

P

P

P

P Ru P

H C O O

P

H

Activation (cis–trans)

H

HCOOH

H P

P Ru P

H2 H2 insertion A: ∆G≠= 24.7 kcal mol−1 B: ∆G≠ = 24.8 kcal mol−1 C: ∆G≠ = 29.4 kcal mol−1

P

H

Scheme 5.14  The overall process of CO2 hydrogenation catalyzed by Ru complexes with bidentate phosphine ligands. Source: From Zhang et al. [53], © 2016 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

130

5  Theoretical Studies of Homogeneously Catalytic Hydrogenation of Carbon Dioxide R

Cp*

N

Rh

N

Cl

R

N N –

HCO3

R

Cp*

–

N

Rh

R = CH2CH2SO3

Cl

–

HCO2

R

Cp*

N

Rh

N

O O

R

H N

N

N

OH

N N

+

H

R

+L

R

Cp*

N

Rh

–

HCO3

R N N

H2

N O Cp* Rh

O N

N

R

H –

+ OH

O

R

Cp*

N

Rh

N

O

R

L

R

N N

‡ H OH

N N

Scheme 5.15  Proposed overall mechanism for the reduction of bicarbonate to formate catalyzed by water-soluble bis-NHC rhodium complex. Source: From Jantke et al. [54], © 2016 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

low catalytic activities (TOF = 49 h−1) for the formation of formate at moderate temperature (80–100 °C) under a 4–5 MPa pressure, and the complexes with 6DHBP ligands have even lower and negligible activities. The thermal instabilities of those cobalt complexes were believed to be the major causes of their low activities. The calculation results confirmed that water-assisted H2 heterolysis is the rate-determining step in the reaction catalyzed by the 6DHBP complex. In addition to the (PP3)Fe catalysts, novel cobalt dihydrogen complexes with similar tetraphos ligands were reported by Beller and coworkers [57] for catalytic hydrogenation of CO2 with improved performance. Chen and coworkers [30] computationally investigated the mechanism of this cobalt-catalyzed CO2 hydrogenation reaction and proposed monohydride (Path I) and dihydride (Path II) pathways (Scheme 5.17). Their calculation results indicated that Path II is 3.7 kcal mol−1 more favorable than path I in free energy, and confirmed that the active catalyst is the dihydride complex. In 2016, Bertini et al. [58] reported a combined experimental and computational study of Fe(II) hyrido carbonyl complexes with 2,6-(diaminopyridyl)diphosphine pincer ligands for the catalytic hydrogenation of CO2 and NaHCO3 to formate under mild conditions (Scheme 5.18). Their DFT calculations reveal an outer-sphere hydride transfer mechanism for the reduction of CO2 with a hydrido formate complex as the resting state. The driving force of the catalytic cycle is the reaction between the product (formic acid) and extra base in the solvent.

5.3  ­Hydrogenation of CO2 to Formic Acid/Formate H2O

O

N

H2O

Cp*

N

Cp*

N

‡

O

M H

N

H

M H H O H

O

O

H

TS

H2 O Cp*

N

M N

‡

O

HCOOH

Cp*

N

H2O O

CO2

M

M

H2O

N

H2O

O

Cp*

N

H

N

H O C OH O

OH

TS M = Co, Rh, Ir

–

HCOO

H2O

O

‡

O

M H

N O

C O

Cp*

N

Cp*

N

+

H

M H

N

O

O

CO2

TS

Scheme 5.16  Ke and coworkers’ mechanism for the hydrogenation of CO2 catalyzed by halfsandwich Cp* complexes with bipyridinol ligands. Source: From Hou et al. [55], © 2014 American Chemical Society. Reprinted with permission of American Chemical Society.

Inspired by the H2 activation mechanism and the active site structure of [Fe]-hydrogenase (Figure 5.3a), synthesized model complex (Figure 5.3b), and experimentally reported aliphatic PNP iron complexes (Figure 5.3c), Yang and coworkers proposed a series of Fe and Co complexes with experimentally readymade acylmethylpyridinol and aliphatic PNP pincer ligands, and computationally predicted their catalytic activities for hydrogenation of CO2 to formic acid (Figure 5.3d,e) [24, 25]. They found that the acylmethylpyridinol ligand plays an essential role in inner-sphere heterolytic H2 cleavage and CO2 hydrogenation by forming strong M─Hδ− … Hδ+─O dihydrogen bonds. The calculated total free energy barriers for the formation of formic acid catalyzed by the most active iron (with R1 = R2 = Me) and cobalt (with R1  =  Me, R2  =  OH) complexes are 23.9 kcal mol−1 in tetrahydrofuran (THF) and 23.1 kcal mol−1 in water, respectively. In 2018, Ke and coworkers [66] proposed a formate anion–assisted deprotonation mechanism for the hydrogenation of CO2 catalyzed by a Co(dmpe)2H (dmpe: 1,2-bis(dimethylphosphino)ethane) complex, which was experimentally reported by Linehan and coworkers [67] as an efficient base-metal catalyst for the production of formate from CO2 and H2 with a TOF of 3400 h−1 under mild conditions (room temperature and 1 atm, TOF  =  74 000 h−1 at 20 atm). The mechanism involves three steps: oxidative

131

132

5  Theoretical Studies of Homogeneously Catalytic Hydrogenation of Carbon Dioxide H Ph2P

Path I

H PPh2

Co

Ph2P

NH(CH3)2

NHH(CH3)2

P

H

TS

Ph2P O H C

H Ph2P

PPh2

Co

P

O Ph2P

H

tion

riza

P

H2 HCOO

TS H Ph2P

Ph2P

O

C

O

H Ph2P

Co

Ph2P H

O PPh2 H

P

PPh2

P

Path II

Ph2P

O

H2

HCOO

CO2

Co

C

Co

Ph2P

PPh2

P

Ph2P

O

H

H Co

Ph2P

CO2

me

P

PPh2

Co

Ph2P

iso

Ph2P

PPh2

Co

Ph2P

H

P

C

Ph2P

O Co

Ph2P

PPh2 H

isomerization PPh2 H

H2

P

HCOO

H2 H

HCOO Ph2P Ph2P

H Co

H

PPh2

P

Scheme 5.17  Predicted catalytic cycles with the monohydride (Path I) and dihydride (Path II) pathways for the hydrogenation of CO2 catalyzed by PP3 cobalt complexes. Source: From Gao et al. [30], © 2015 Royal Society of Chemistry. Reprinted with permission of Royal Society of Chemistry.

addition of H2 to cobalt, formation of cobalt–monohydride complex by the deprotonation/ heterolysis of H2, and the reductive elimination of CO2, which is the TOF-determining step in the whole reaction with a total free energy barrier of 20.9 kcal mol−1, which is 2.4 kcal mol−1 more favorable than direct hydride transfer from cobalt to CO2. In addition,

5.4  ­Hydrogenation of CO2 to Methanol Me

H N N

Fe

N PR2 Me

–

CO2

PR2

HCOO

–

HCOO

Br Me

DBU, H2

H N N

EtOH Me

Me

PR2 Fe

N PR2

H N N

CO

H

Me

N

N

Fe

PR2 Fe

H

–

+

HCOO HDBU

H2/DBU

CO

N PR2 O Me

CO

H

–

HCOO EtOH Me

H N N

N PR2 CO Me

PR2 Fe

O

N PR2

H N

N

PR2

EtOH Me

H

Me

CO

PR2 Fe

N PR2 Me O

CO

H

Scheme 5.18  Predicted mechanism for CO2 hydrogenation catalyzed by PNP–Fe complexes. Source: From Bertini et al. [58], © 2016 American Chemical Society. Reprinted with permission of American Chemical Society.

the oxidative addition of H2 is more feasible than the direct H2 heterolysis assisted by Verkade’s base or formate anion. Very recently, we computationally investigated the hydrogenation of CO2 to formate and formamide reactions catalyzed by Khusnutdinova and coworkers’ Mn bipyridinol complex [68], and proposed an unexpected three-site two-proton transfer MLC mechanism with the participation of both ortho-O atoms in the bipyridinol ligand for the formation of formamide from formic acid and diethylamine (Scheme 5.19) [69]. Figure 5.4 shows the optimized transition state structures of simultaneous two-proton transfer for the activation of diethylamine and formic acid (TSd), as well as the formation of formamide and water (TSe). Our findings not only reveal the essential role of the bipyridinol ligand in the catalytic reaction, but also enlighten the way to designing new base-metal catalysts for CO2 reduction.

5.4 ­Hydrogenation of CO2 to Methanol Although the catalytic hydrogenation of CO2 to formic acid/formate has been well developed and achieved rather high efficiency under mild conditions, the conversion of H2 and CO2 to methanol is still a big challenge. To date, although a few homogeneous transition metal catalysts for the direct and indirect transformation of CO2 to methanol were reported, most of those catalysts contain noble metals and require rigid reaction conditions. Therefore, the development of high-efficiency base-metal catalysts for the hydrogenation of CO2 to methanol is still highly desirable. The direct conversion of H2 and CO2 to methanol usually contains three cascade catalytic cycles, hydrogenation of CO2 to formic acid, hydrogenation of formic acid to formaldehyde and water, and hydrogenation of formaldehyde to methanol. The conversion of formic acid to formaldehyde is usually the most turnover-limiting step in the whole reaction because of the difficulties in the reduction of formic acid and the breaking of strong C─O bond. In addition to [Fe]-hydrogenase, the active sites of [FeFe]- and [NiFe]-hydrogenases are also enlightening for the design of base-metal catalysts for CO2 hydrogenation.

133

(c)

(b)

(a) Cys 176 S OC ?

O

Fe CO

R

C N

O

OC

HO

PiPr2

C

Fe CO

Shima

Fe

H N

N

H

HO

PiPr2

H CO and H N

PiPr2 Fe

H CO

PiPr2

Beller, Guan, Jones, Hazari

Chen and Hu

(d)

(e) EO C

H N

O S

Fe E

EO C

R1 N O

R2 R1

H N

Co E

R1 N

HO

R2 R1

E = PiPr2; R1 = H, Me, etc. R2 = H, Me, OH, OMe, etc.

Figure 5.3  The active site structure of [Fe]-hydrogenase reported by Shima and coworkers (a) [22, 59], synthesized model active site structure of [Fe]-hydrogenase by Hu and coworkers. The symbol “?” means the ligand coordinates to Fe at this position is unknown [20] (b) [60, 61], aliphatic (PNP)Fe complexes reported by Beller, Guan, and Hazari groups (c) [62–65], and Fe and Co complexes proposed by Yang and coworkers (d, e) [24, 25]. Source: From Ge et al. [24], © 2016 American Chemical Society. Reprinted with permission of American Chemical Society.

O H C H N HO O Mn OC CO CO O

O

HO

N OC

C

O

TSc

H N OH Mn CO CO

HO

N

C

H

TSf

N O N HO O Mn OC CO CO

HCOOH

O

Et2NH

C

H

N O N HO OH Mn OC CO CO

H Et Et C N H O N N HO O Mn OC CO CO 11b H O

CO2

TSb

Et 2NH Br N N N HO OH HO O Mn Mn OC CO OC CO Et 2NH2B r CO CO H2O N

Pre

HO

N OC

H N Mn CO

HCONEt 2 two-proton transfer H Et H O TSe Et

H2 OH

CO

TSd

C N

TSa

HO

H2 N O Mn OC CO CO N

CO2 + H2 + Et2NH

TSh H2O HCONEt2

Cat.

O

N OH N OH Mn CO CO

OC

HCONEt2 + H2O

two-proton transfer

H Et Et CN H N HO N HO O Mn OC CO CO HO

TSg

O

HO

N OC

C

H

O N Mn CO

CO

Et2NH

Scheme 5.19  Predicted mechanism for the hydrogenation of CO2 to formamide catalyzed by the Mn(CO)3 bipyridinol complex. Source: From Yan et al. [69], © 2018 Royal Society of Chemistry. Reprinted with permission of Royal Society of Chemistry.

O

5  Theoretical Studies of Homogeneously Catalytic Hydrogenation of Carbon Dioxide

03

1 Et

0.9

91

1.835

1

1.316

Et

Et

(a)

1.8 2

2

99 1.2

1.35

8 1.8

1.469

2.091

1.0 1.144

1.01 71 .57 2

Mn

Mn

1.3 79

136

TSd

(b)

TSe

Et

Figure 5.4  Optimized transition state structures of two-proton transfer for (a) the activation of diethylamine and formic acid (TSd), and (b) the formation of formamide and water (TSe). Source: From Yan et al. [69], © 2018 Royal Society of Chemistry. Reprinted with permission of Royal Society of Chemistry.

Inspired by the active site structures of [FeFe]- and [NiFe]-hydrogenases and Bullock’s catalysts with pendant amines (Figure  5.5) [70–77], Yang and coworkers proposed a series of Mn, Fe, and Co complexes with diphosphine pendant amine ligands as promising catalysts for hydrogenation of CO2 to methanol (Figure  5.6) [26–28]. Scheme  5.20 shows the computationally predicted mechanism, including three cycles, hydrogenation of CO2 to formic acid, hydrogenation of formic acid to formaldehyde and water, and hydrogenation of formaldehyde to methanol, for the reaction catalyzed by newly

+ Bn

N

Ph P

BnN

P P

CO

(a) Ph

tBu

PhN

Co

ArF

P

ArF

tBu

(b)

NCMe

2+

tBu

N PPh2

tBu

NCMe Co

P

tBu

(d)

H

P

tBuN

PPh2

P

C5F4N

Co

H

tBu

(c)

Ph N

2+

P

H

tBu P

ArF

H

N

H

N ArF

H Mn

P Ph

+

Ph

H

NCMe NCMe

Figure 5.5  Liu, Dubois, and Bullock’s Mn (a) and Co (b, c, and d) complexes with pendant amines [70–77]. Source: Redrawn from Refs. [70–77].

5.4  ­Hydrogenation of CO2 to Methanol tBu

tBu

N1 tBuN

P

tBu

CN

Fe

P tBu

tBu H P tBuN

CN

CO

(a)

N1

Fe P tBu

CO CO

R

(b)

R = H, Cl, NO2, CN, C5H4N, CH3, NH2, OCH3, OH, CHO, COOH, COCH3, COOCH3, CONH2, CONHCOOH tBu

tBu

N1

N1 tBuN

P P tBu

(c)

tBu

CO

Mn R3

tBuN

P

tBu

P tBu

CO (d)

CN

Co

CN

R3

R3 = H, CN, NO2, CH3, NH2, OH, CHO, COOH, COCH3, COOCH3

Figure 5.6  Yang and co-workers’ computationally designed Fe (a, b), Mn (c), and Co (d) complexes with diphosphine pendant amine ligands [27, 28]. Source: From Chen et al. [27], © 2017 Royal Society of Chemistry. Reprinted with permission of Royal Society of Chemistry.

proposed Co complexes. The Mn and Fe complexes have similar catalytic cycles, but different relative energies in the overall catalytic reaction. Although the breaking of the C─O bond in formic acid for the formation of formaldehyde has a relatively high barrier, the formation of unstable hydroxymethanolate anion is the rate-determining step in the whole catalytic reaction. However, the rate-determining step might be different in different metal complexes with various electron donor groups at that R3 position. The lowest free energy barriers of hydrogenation of CO2 to methanol catalyzed by those newly designed Fe, Co, and Mn complexes are only 23.7, 24.9, and 26.6 kcal mol−1, respectively. In addition, the ligands with stronger electron donor groups at the R3 position usually have higher catalytic activities. Although the above computationally designed base-metal complexes are promising catalysts for hydrogenation of CO2 to methanol, the difficulties in the synthesis of air- and moisture-sensitive phosphine ligands in them limited their application. Therefore, Knölker’s iron cyclopentadienone (IC) complex, [2,5-(SiMe3)2-3,4-(CH2)4(η5C4COH)]Fe(CO)2H, has attracted people’s attention because of its phosphine-free structure and high catalytic activities for hydrogenation reactions [78–80]. Yang and coworkers computationally predicted the catalytic activities of Knölker’s IC complex for hydrogenation of CO2 to methanol and obtained a total free energy barrier of 26.0 kcal mol−1 [81]. Instead of direct H2 heterolysis by the MLC of Fe and the cyclopentadienone ligand, their mechanism features methanol-assisted proton transfer for the activation of H2 in three interrelated

137

-

–

tBu

TS

CH2O

tBuN

tBuN

CH3OH H2O

tBuN

CH2O

TS

CO2

–

HCOO

R3 = COOH OH H H C N1 H tBu O P CN Co P CN tBu R3 TS

tBu

TS tBuN

tBu

H2

N1

tBuN

H HO H C N1 tBu OH P CN Co P CN tBu R3

tBu

P P

tBu

Co

CN CN

tBu

tBuN

R3

TS

tBu

HOCH2OH tBuN

N1 tBu H2 P CN Co P CN tBu R3

OH HO C H N1 tBu H P CN Co P CN tBu R3

tBu

TS

H N1 tBu H P CN tBuN Co P CN tBu R3

TS

– H O H C H tBu N1t Bu OH P CN tBuN Co P CN tBu R3

N1 tBu OH2 P CN Co P CN tBu R3

TS

N1 TS tBu OH P CN Co P CN tBu R3

tBuN

O

H C O N1 tBu H P CN Co P CN tBu R3

tBu

tBu tBuN

O H C H N1 tBu O P CN Co P CN tBu R3

–

HCOO

TS

tBuN

H

O H H C H N1 H tBu P CN Co P CN tBu R3

TS

tBu tBu

tBu

tBu

HCOOH tBuN

+ H N1 tBu H2 P CN Co P CN tBu R3

tBu

tBuN

+

H N1 tBu P P

Co

tBu

CN CN

R3

H2

Scheme 5.20  Yang and coworkers’ mechanism of the hydrogenation of CO2 to methanol catalyzed by a computationally designed Co complex with pendant amines. Source: From Chen et al. [27], © 2017 Royal Society of Chemistry. Reprinted with permission of Royal Society of Chemistry.

5.4  ­Hydrogenation of CO2 to Methanol

catalytic cycles for the hydrogenation of CO2 to formic acid (Scheme 5.21), the hydrogenation of formic acid to formaldehyde and water (Scheme 5.22), and the hydrogenation of formaldehyde to methanol (Scheme 5.23). They also built a series of Fe, Co, and Mn complexes with cyclopentadienone ligands (Figure 5.7), and computationally predicted their catalytic activities. Their calculation results indicate that the new IC complex, [2,5-(SiMe3)2-3,4-CH3CHSCH2(η5-C4COH)]Fe(CO)2H (1Fe-Casey-S-CH3), is the most active one with a total free energy barrier of 25.1 kcal mol−1. Inspired by the structures and catalytic mechanisms of the active site of [Fe]-hydrogenase [21, 23], Meyer and coworkers’ iron monoamidate complexes [82], Hitomi et al.’s pentadentate manganese complexes [83], as well as Zhao and coworkers’ cobalt complexes with pentadentate polypyridyl–amine ligands [84–86], Yang and coworkers [87] built another series of phosphine-free and pentadentate amidate–ligated Fe and Co complexes with experimentally available N-heterocyclic pyridinol groups (Figure 5.8) and computationally evaluated their catalytic activities for hydrogenation of CO2 to methanol. After initial screening and detailed reaction mechanism analysis, they found that the Co complex with a 2-[bis(pyridine-2-ylmethyl)]amino-N-3,9-purin-2-one ligand has a rather low total free energy barrier of 23.3 kcal mol−1 for catalytic hydrogenation of CO2 to methanol (Scheme 5.24). The calculation results indicate that the transition state for the formation of formic acid is the rate-determining step, while the special role of N-heterocyclic pyridinol groups in pentadentate ligands is that of a proton reservoir for fast heterolytic H2 cleavage by MLC.

R OH

R = TMS

R O OC

H2

Fe CO

R Fe R CO

R

CH3OH

R H2

R OH Fe R H OC CO

TS O

OC

CO2 Fe R OC H TS TS CO

CH3OH

Cat.

H2 + CO2

OH Fe R O H C OC CO O –

HCOO H O

OH

HCOOH

OC

Cycle 1 HCOOH

R O

Fe R CO –

R

Fe R H O O OC C CO

+

R

CH3

HCOO OH

TS

O Fe R O C OC CO H

Scheme 5.21  Yang and coworkers’ mechanism for the hydrogenation of CO2 to formic acid catalyzed by Knölker’s IC complex. Source: From Ge et al. [81], © 2017 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

139

R

HCOOH R

TS OH

Fe R OC H CO

HOCH2OH

O H O Fe R O C OC H CO H H

TS CH2O

CH3OH H2 + HCOOH

Cat.

R

CH2O + H2O

R O

OC

Cycle 2

R

OH

Fe R

Fe R O OC CO H

CO

OH OC

R

O OH Fe R H C OC OH CO H

Fe R H H O CO CH3

TS

CH3OH OC

R O Fe R H2 CO

R

H2 H2O

O OC

Fe R O H CO H

TS R =TMS

Scheme 5.22  Yang and coworkers’ mechanism for the hydrogenation of formic acid to formaldehyde and water catalyzed by Knölker’s IC complex. Source: From Ge et al. [81], © 2017 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

5.4  ­Hydrogenation of CO2 to Methanol

R OH

CH3OH

Fe R OC H CO

CH2O

TS R

R

O

OH OC

Fe R H H O CO CH3

TS CH3OH

Cy c l e3

H2 + CH2O Cat.

OH Fe R H C OC H CO H

CH3OH CH3OH

R

R

O

O Fe R H2 OC CO

OC

Fe R CO

H2

R = TMS

Scheme 5.23  Yang and coworkers’ mechanism for the hydrogenation of formaldehyde to methanol catalyzed by Knölker’s IC complex [81]. The box symbols in the artworks of those schemes represent vacant positions at the metals. Source: From Ge et al. [81], © 2017 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

In addition to direct hydrogenation of formic acid, disproportionation of formic acid is an attractive alternative for the production of methanol from CO2. We recently investigated the mechanistic insights into the disproportionation of formic acid to methanol catalyzed by a half-sandwich iridium complex [Cp*Ir(bpy)OH2]2+ [88], which achieved a TON of 1314 with methanol selectivity up to 47.1% under mild reaction conditions (50–60 °C and 5.2 MPa pressure of H2) in the experimental work reported by Himeda and coworkers [89]. Two competitive pathways, methanol-assisted proton transfer and direct deprotonation of hydroxyl in methanediol, for the formation of formaldehyde were found. The formation of formaldehyde from methanediol through direct cleavage of a C─O bond after hydroxyl deprotonation has a free energy barrier of 25.9 kcal mol−1, which is 1.9 kcal mol−1 more favorable than methanol-assisted proton transfer and matches well with the observed reaction rate at 50–60 °C. In 2018, Rueping and coworkers [90] reported the first base metal–catalyzed hydrogenation of CO2-derived carbonates to alcohols using a new bench stable PNN–Mn(CO)3 complex as the catalyst under mild experimental conditions. Their computational study reveals a mechanism with heterolytic cleavage of three dihydrogen molecules through MLC in three catalytic cycles, including the hydrogenation of ethylene carbonate to 2-hydroxyethyl formate, the hydrogenation of 2-hydroxyethyl formate to ethylene glycol and formaldehyde, and the hydrogenation of formaldehyde to methanol, for hydrogenation of ethylene carbonate to methanol. The rate-determining states were found to be the hydride transfer from Mn to the unsaturated carbon in ethylene carbonate (D–E) and H2 cleavage assisted by methoxide with a total activation barrier of 25.0 kcal mol−1 (Scheme 5.25).

141

142

5  Theoretical Studies of Homogeneously Catalytic Hydrogenation of Carbon Dioxide

TMS

TMS OH L

M CO

TMS H

OH

O L

1M

OH

Ph L

M CO

Ptol

MeO2C

OH

MeO2C

Ph H

L

1M-Williams-1

R1

CO

M = Co, L = CN; M = Mn, L = NO.

TMS H

1M-Casey

Ph

Ph

M

M CO

Ptol H

M = Co, L = CN; M = Mn, L = NO.

1M-Williams-2

Ph OH

O

M R2 H L CO

1M-Wills-1 : R1 = H, R2 = Ph; 1M-Wills-2 : R1 = Me, R2 = Ph; 1M-Wills-3 : R1 = Me, R2 = TMS; 1M-Wills-4 : R1 = H, R2 = TMS; 1M-Wills-5 : R1 = Me, R2 = TBDMS.

M = Co, L = CN; M = Mn, L = NO.

From 1M-Wills-1 to 1M-Wills-5 TMS OH

S L

M CO

1M-Casey-S

TMS H

TMS OH

H N L

M CO

TMS H

M = Fe, L = CO; M = Co, L = CN M = Mn, L = NO.

1M-Casey-N

Figure 5.7  Newly constructed iron, cobalt, and manganese complexes [81]. Source: From Ge et al. [81], © 2017 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

5.5 ­Summary and Conclusions In this chapter, we briefly introduced theoretical studies of the mechanistic insights into homogeneously catalytic hydrogenation of CO2 to formic acid/formate and methanol, as well as some computational designs of base-metal catalysts for those reactions. Those progresses clearly show that the dramatic development of quantum chemistry methods and

5.5  ­Summary and Conclusion OH2 N N

(a)

Fe

2+

N

N

N

N

(b)

H3C H

H

(d)

N

Fe

OH

R1

N

(e)

N

Zhao

N

R1 =

1 R2 = HN

R2

N

Co

H N

1e

R2 O

O

N

N

N

N

O

N

N N

1c

1b

N

N N

1d

N

1Co

1Fe

N

N

O

1a N

R1

R2

O

N

R1 =

(c)

+

Yang and Hall

O

N

O

R2

O R2 =

N

Co N

H O CO

N

N

Hitomi

S O C Fe

OC

N

N

N

O

Meyer

N

Mn

N

3+

OH2

+

1f

N

N

N

N

1g

1h

Figure 5.8  Meyer’s monomeric amidate-ligated Fe complex (a) [82], Hitomi’s monomeric amidateligated Mn complex (b) [83], Zhao’s cobalt complex with pentadentate polypyridyl-amine ligands (c) [84–86], Yang and Hall’s structure model of the active site of [Fe]-hydrogenase (d) [21, 23], and the newly proposed iron and cobalt complexes (e). Source: From Wang et al. [87], © 2019 Royal Society of Chemistry. Reprinted with permission of Royal Society of Chemistry.

tremendous upswing of computer power in recent years have pushed our understanding of the chemical and physical insights of chemical reactions to the electronic structure level and significantly benefitted the rational catalyst design for the conversion and utilization of carbon dioxide. Although the mechanisms of transition metal–catalyzed CO2 hydrogenation reactions have been well evaluated, the design of low-cost, high-efficiency, and environmentally benign catalytic systems for practical application is still a big challenge. The bioinspired design of base-metal complexes with redox non-innocent ligands for rapid two-electron transfer through metal-ligand co-operations has been demonstrated as an effective way to developing new catalysts for the activation and reformation of inert chemical bonds. It is worth noting that although the state-of-the-art DFT methods have reached the rather high accuracy in predicting the molecular structures and relative energies, the finding of suitable functional for a specific transition metal system is still a big challenge and controversial work in computational reaction mechanism studies. For the situations without experimental references, a sound evaluation of the performance of different levels of density functionals with various percentages of Hartree–Fock exchanges would be strongly recommended to ensure reliable computational predictions.

143

OH2 N Co N N N N O N N N

+

O CO H N N O Co N N N N O N N

Hydrogenation of CH2O

CH3OH H2O

CO2

TS O N

N

N N

N

OH Co N

N N O

HO H C HO H N N O Co N N N N O N N

+

N

O N

N N

N

Co N

N N

+

H2 N

O

N

O

N N

N

H2 Co N

N N O

O TS2,3a–Co

N

N N

=

N

N

N N

N

H Co N

O

N

N N

TS Hydrogenation of HCOOH

N

N N

N

Co

N N

N

2+

O HCOO–

H CO O N N O Co N N N N O N N H

B

+ CH2(OH)2

+

N N

H O N

C

B

A

H

+

+

Hydrogenation of CO2

H C O O

TS

TS CH2O

+

TS

H

HO H C H O H N N O Co N N N N O N N

HCOO–

H

CH2O

TS

TS

O

HCOOH

TS

H O N

N

N N

N

H2 Co N

N N

+

H2

O

Scheme 5.24  Yang and coworker’s mechanism for the formation of methanol from CO2 and H2 catalyzed by a computationally designed pentadentate Co complex. Source: From Wang et al. [87],© 2019 Royal Society of Chemistry. Reprinted with permission of Royal Society of Chemistry.

+

  ­Reference HH

H

H

G–H 22.1

HH

O H H H P N Mn CO N CO

O H H P N Mn CO CO N H

G 17.1

MeOH 4.0

H

H

C

0.0 P N Mn CO CO N O

H 8.9

O

O

Cycle 1

H2 H H N

F –1.0

HH

O

O P Mn CO CO N

H O H HO

e O M OH

O D H P 5.5 N Mn CO CO N H

E –3.9 O

N

O H

O

O

P Mn CO CO N

D–E 16.4

Scheme 5.25  Rueping and coworkers’ mechanism for the hydrogenation of ethylene carbonate into 2-hydroxyethyl formate catalyzed by Mn–PNN complex. Source: From Zubar et al. [90], © 2018 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

­References  1 Maeda, C., Miyazaki, Y., and Ema, T. (2014). Catal. Sci. Technol. 4: 1482–1497.  2 Wang, W., Wang, S., Ma, X., and Gong, J. (2011). Chem. Soc. Rev. 40: 3703–3727.  3 Wang, W.-H., Feng, X., and Bao, M. (2018). Transformation of CO2 to methanol with homogeneous catalysts. In: Transformation of Carbon Dioxide to Formic Acid and Methanol (ed. S.K. Sharma), 53–87. Singapore: Springer https://doi. org/10.1007/978-981-10-3250-9.  4 Kar, S., Kothandaraman, J., Goeppert, A., and Prakash, G.K.S. (2018). J. CO2 Util. 23: 212–218.  5 Appel, A.M., Bercaw, J.E., Bocarsly, A.B. et al. (2013). Chem. Rev. 113: 6621–6658.   6 Hammes-Schiffer, S. (2017). Acc. Chem. Res. 50: 561–566.  7 Cheng, G.-J., Zhang, X., Chung, L.W. et al. (2015). J. Am. Chem. Soc. 137: 1706–1725.  8 Kubas, G.J. (1988). Acc. Chem. Res. 21: 120–128.  9 Lubitz, W., Ogata, H., Rüdiger, O., and Reijerse, E. (2014). Chem. Rev. 114: 4081–4148. 10 Simmons, T.R., Berggren, G., Bacchi, M. et al. (2014). Coord. Chem. Rev. 270–271: 127–150. 11 Chen, X. and Yang, X. (2016). Chem. Rec. 16: 2364–2378.

145

146

5  Theoretical Studies of Homogeneously Catalytic Hydrogenation of Carbon Dioxide

12 Xu, T., Chen, D., and Hu, X. (2015). Coord. Chem. Rev. 303: 32–41. 13 Hu, B., Chen, D., and Hu, X. (2014). Chem. Eur. J. 20: 1677–1682. 14 Song, L.-C., Hu, F.-Q., Zhao, G.-Y. et al. (2014). Organometallics 33: 6614–6622. 15 Nicolet, Y., Lemon, B.J., Fontecilla-Camps, J.C., and Peters, J.W. (2000). Trends Biochem. Sci. 25: 138–143. 16 Frey, M. (2002). ChemBioChem 3: 153–160. 17 Evans, D.J. and Pickett, C.J. (2003). Chem. Soc. Rev. 32: 268–275. 18 Ogata, H., Kramer, T., Wang, H. et al. (2015). Nat. Commun. 6: 7890. 19 Ogata, H., Nishikawa, K., and Lubitz, W. (2015). Nature 520: 571–574. 20 Shima, S., Pilak, O., Vogt, S. et al. (2008). Science 321: 572–575. 21 Yang, X. and Hall, M.B. (2008). J. Am. Chem. Soc. 130: 14036–14037. 22 Hiromoto, T., Ataka, K., Pilak, O. et al. (2009). FEBS Lett. 583: 585–590. 23 Yang, X. and Hall, M.B. (2009). J. Am. Chem. Soc. 131: 10901–10908. 24 Ge, H., Jing, Y., and Yang, X. (2016). Inorg. Chem. 55: 12179–12184. 25 Yang, X. (2015). Chem. Commun. 51: 13098–13101. 26 Chen, X., Jing, Y., and Yang, X. (2016). Chem. Eur. J. 22: 8897–8902. 27 Chen, X., Ge, H., and Yang, X. (2017). Catal. Sci. Technol. 7: 348–355. 28 Chen, X. and Yang, X. (2016). J. Phys. Chem. Lett. 7: 1035–1041. 29 Yang, X. (2011). ACS Catal. 1: 849–854. 30 Gao, H., Chen, L., Chen, J. et al. (2015). Catal. Sci. Technol. 5: 1006–1013. 31 Das, S. and Pati, S.K. (2018). Catal. Sci. Technol. 8: 3034–3043. 32 Li, J. and Yoshizawa, K. (2011). Bull. Chem. Soc. Jpn. 84: 1039–1048. 33 Boddien, A., Mellmann, D., Gartner, F. et al. (2011). Science 333: 1733–1736. 34 Yang, X. (2013). Dalton Trans. 42: 11987–11991. 35 Inoue, Y., Izumida, H., Sasaki, Y., and Hashimoto, H. (1976). Chem. Lett. 5: 863–864. 36 Musashi, Y. and Sakaki, S. (2000). J. Am. Chem. Soc. 122: 3867–3877. 37 Ohnishi, Y.Y., Matsunaga, T., Nakao, Y. et al. (2005). J. Am. Chem. Soc. 127: 4021–4032. 38 Tanaka, R., Yamashita, M., and Nozaki, K. (2009). J. Am. Chem. Soc. 131: 14168–14169. 39 Ahlquist, M.S.G. (2010). J. Mol. Catal. A: Chem. 324: 3–8. 40 Tanaka, R., Yamashita, M., Chung, L.W. et al. (2011). Organometallics 30: 6742–6750. 41 Langer, R., Leitus, G., Ben-David, Y., and Milstein, D. (2011). Angew. Chem. Int. Ed. 50: 2120–2124. 42 Langer, R., Diskin-Posner, Y., Leitus, G. et al. (2011). Angew. Chem. Int. Ed. 50: 9948–9952. 43 Schmeier, T.J., Dobereiner, G.E., Crabtree, R.H., and Hazari, N. (2011). J. Am. Chem. Soc. 133: 9274–9277. 44 Suh, H.-W., Schmeier, T.J., Hazari, N. et al. (2012). Organometallics 31: 8225–8236. 45 Muller, K., Sun, Y., Heimermann, A. et al. (2013). Chem. Eur. J. 19: 7825–7834. 46 Hull, J.F., Himeda, Y., Wang, W.H. et al. (2012). Nat. Chem. 4: 383–388. 47 Wang, W.-H., Hull, J.F., Muckerman, J.T. et al. (2012). Energy Environ. Sci. 5: 7923–7926. 48 Wang, W.-H., Muckerman, J.T., Fujita, E., and Himeda, Y. (2013). ACS Catal. 3: 856–860. 49 Ertem, M.Z., Himeda, Y., Fujita, E., and Muckerman, J.T. (2016). ACS Catal. 6: 600–609. 50 Suna, Y., Ertem, M.Z., Wang, W.-H. et al. (2014). Organometallics 33: 6519–6530. 51 Filonenko, G.A., Conley, M.P., Copéret, C. et al. (2013). ACS Catal. 3: 2522–2526. 52 Filonenko, G.A., Hensen, E.J.M., and Pidko, E.A. (2014). Catal. Sci. Technol. 4: 3474–3485. 53 Zhang, P., Ni, S.F., and Dang, L. (2016). Chem. Asian J. 11: 2528–2536.

  ­Reference

54 Jantke, D., Pardatscher, L., Drees, M. et al. (2016). ChemSusChem 9: 2849–2854. 55 Hou, C., Jiang, J., Zhang, S. et al. (2014). ACS Catal. 4: 2990–2997. 56 Badiei, Y.M., Wang, W.-H., Hull, J.F. et al. (2013). Inorg. Chem. 52: 12576–12586. 57 Federsel, C., Ziebart, C., Jackstell, R. et al. (2012). Chem. Eur. J. 18: 72–75. 58 Bertini, F., Gorgas, N., Stöger, B. et al. (2016). ACS Catal. 6: 2889–2893. 59 Hiromoto, T., Warkentin, E., Moll, J. et al. (2009). Angew. Chem. Int. Ed. 48: 6457–6460. 60 Chen, D., Scopelliti, R., and Hu, X. (2011). Angew. Chem. Int. Ed. 50: 5671–5673. 61 Chen, D., Scopelliti, R., and Hu, X. (2012). Angew. Chem. Int. Ed. 51: 1919–1921. 62 Koehne, I., Schmeier, T.J., Bielinski, E.A. et al. (2014). Inorg. Chem. 53: 2133–2143. 63 Werkmeister, S., Junge, K., Wendt, B. et al. (2014). Angew. Chem. Int. Ed. 53: 8722–8726. 64 Alberico, E., Sponholz, P., Cordes, C. et al. (2013). Angew. Chem. Int. Ed. 52: 14162–14416. 65 Chakraborty, S., Dai, H., Bhattacharya, P. et al. (2014). J. Am. Chem. Soc. 136: 7869–7872. 66 Zhang, Z., Li, Y., Hou, C. et al. (2018). Catal. Sci. Technol. 8: 656–666. 67 Jeletic, M.S., Mock, M.T., Appel, A.M., and Linehan, J.C. (2013). J. Am. Chem. Soc. 135: 11533–11536. 68 Dubey, A., Nencini, L., Fayzullin, R.R. et al. (2017). ACS Catal. 7: 3864–3868. 69 Yan, X., Ge, H., and Yang, X. (2018). Catal. Sci. Technol. 8: 5735–5739. 70 Liu, T., Dubois, D.L., and Bullock, R.M. (2013). Nat. Chem. 5: 228–233. 71 Liu, T., Wang, X., Hoffmann, C. et al. (2014). Angew. Chem. Int. Ed. 53: 5300–5304. 72 Bullock, R.M. and Helm, M.L. (2015). Acc. Chem. Res. 48: 2017–2026. 73 Hulley, E.B., Kumar, N., Raugei, S., and Bullock, R.M. (2015). ACS Catal. 5: 6838–6847. 74 DuBois, D.L. and Bullock, R.M. (2011). Eur. J. Inorg. Chem. 2011: 1017–1027. 75 Wiedner, E.S., Appel, A.M., DuBois, D.L., and Bullock, R.M. (2013). Inorg. Chem. 52: 14391–14403. 76 Wiedner, E.S., Roberts, J.A., Dougherty, W.G. et al. (2013). Inorg. Chem. 52: 9975–9988. 77 Fang, M., Wiedner, E.S., Dougherty, W.G. et al. (2014). Organometallics 33: 5820–5833. 78 Knölker, H.-J., Heber, J., and Mahler, C.H. (1992). Synlett 1992: 1002–1004. 79 Knölker, H.-J. and Heber, J. (1993). Synlett 1993: 924–926. 80 Knölker, H.-J., Baum, E., Goesmann, H., and Klauss, R. (1999). Angew. Chem. Int. Ed. 38: 2064–2066. 81 Ge, H., Chen, X., and Yang, X. (2017). Chem. Eur. J. 23: 8850–8856. 82 Coggins, M.K., Zhang, M.T., Vannucci, A.K. et al. (2014). J. Am. Chem. Soc. 136: 5531–5534. 83 Hitomi, Y., Iwamoto, Y., and Kodera, M. (2014). Dalton Trans. 43: 2161–2167. 84 Wang, P., Liang, G., Reddy, M.R. et al. (2018). J. Am. Chem. Soc. 140: 9219–9229. 85 Rowland, J.M., Olmstead, M.M., and Mascharak, P.K. (2002). Inorg. Chem. 41: 2754–2760. 86 Danopoulos, A.A., Wright, J.A., Motherwell, W.B., and Ellwood, S. (2004). Organometallics 23: 4807–4810. 87 Wang, W., Qiu, B., and Yang, X. (2019). Dalton Trans. 48: 8034–8038. 88 Yan, X. and Yang, X. (2018). Organometallics 37: 1519–1525. 89 Tsurusaki, A., Murata, K., Onishi, N. et al. (2017). ACS Catal. 7: 1123–1131. 90 Zubar, V., Lebedev, Y., Azofra, L.M. et al. (2018). Angew. Chem. Int. Ed. 57: 13439–13443.

147

149

6 Heterogenized Catalyst for the Hydrogenation of CO2 to Formic Acid or Its Derivatives Kwangho Park, Gunniya Hariyanandam Gunasekar and Sungho Yoon Chung-Ang University, Department of Chemistry, Seoul, 06974, Republic of Korea

6.1 ­Introduction The social requirements of environmental friendliness and sustainable energy or chemical production have considerably increased due to the sudden climate change resulting from global warming as well as the fear associated with the depletion of fossil-based fuels [1–6]. The hydrogenation of CO2 to formic acid and its derivatives has been considered to satisfy the societal requirements with pronounced versatility for (i) the utilization of emitted CO2 [7–12] as well as (ii) the strategy of efficient hydrogen storage [13–20]. Recently, remarkable attempts have successfully focused on the use of formic acid as a fuel-cell energy source for public transportation, which brightens the future of the hydrogen economy [21, 22]. However, affordable supply of CO2-based formic acid on a large scale is essential to realize the commercialization of this technology. Multiple attempts have been focused on the hydrogenation of CO2 to formic acid or formate using a homogeneous catalytic system [23–28]. A number of reported studies revealed that homogeneous catalytic systems can be implemented on the scale of mini plants for continuous production of formic acid or formate, exhibiting superior catalytic efficiency [29–35]. However, the complexity of the separation and recyclability of the catalyst, which requires more sophisticated operation, especially for noble-metal-derived catalysts, limits its use for actual commercialization. In contrast, conventionally, heterogeneous catalysts are more preferred for implementation in industrial areas due to the facile handling and solid characteristics, which can be separated from the reaction medium. However, compared to homogeneous catalytic systems, heterogeneous catalytic systems are still in their infancy in terms of the catalytic efficiency and stability [25, 36–39]. For decades, bridging the advantageous characteristics of homogeneous and heterogeneous catalytic systems exhibits a significant challenge in several research and industrial fields. In recent years, among considerable attempts, the heterogenization of molecular complexes on catalytic supports has attracted considerable attention [40]. In heterogenized catalysis, support materials provide abundant fixed ligand sites, which stabilize the molecular metal species during metalation to be highly dispersed on the support surface. CO2 Hydrogenation Catalysis, First Edition. Edited by Yuichiro Himeda. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.

150

6  Heterogenized Catalyst for the Hydrogenation of CO2 to Formic Acid or Its Derivatives

In addition, owing to the analogous structure with a homogeneous counterpart, the solid ligands induce pronounced electronic and geometrical effects on the metal center, enhancing the catalytic reaction rate and selectivity. Furthermore, the multifunctional characteristics of the catalytic supports, including inherent physicochemical stability, high porosity, and an acidic or a basic surface, can render additional durability on the immobilized molecular-level metal centers and facilitate interactions between the catalyst and substrate, thereby improving the catalytic efficiency and long-term stability of the catalyst under the reaction conditions. In this chapter, we will review the recently reported studies on heterogenized catalytic systems for the hydrogenation of CO2 to formic acid or formate in the presence of the base. Until now, several reviews have worthily presented studies about heterogenized catalysis systems [36–39]. However, most reviews mainly discuss on general aspects of heterogenized catalysts and hydrogenation conditions and reactivity. Herein, contributions on the synthetic method, investigation of the coordination environment, and consequent catalytic performance and stability induced by the interaction of molecular metal sites are predominantly highlighted. By summarizing the results obtained from studies reported previously and recent studies, we hope to provide insights for developing a novel design of heterogenized catalysts, ultimately realizing the sustainable hydrogenation of CO2 in the future.

6.2 ­Molecular Catalysts Heterogenized on the Surface of Grafted Supports As a general approach for preparing a heterogeneous catalyst, the immobilization of molecular metal complexes on conventional solid supports such as silica, alumina, and polymers has been reported [41]. Owing to the thermal or pressure resistance and various textural properties with high porosities of the supports, catalytic systems have been used for a wide range of applications. Besides, abundant heteroatoms placed on a support surface, which was derived from an inherent structure, such as a hydroxide on silica and an alumina support, or grafted onto a support by post-modification, can provide stable binding conditions for the metal complexes. Hence, by limiting the agglomeration of metal complexes into nanoparticles, which, in turn, decreases the surface-to-volume ratio versus isolated single atoms, the increased number of catalytically active sites eventually leads to the improvement in the catalytic activity. In particular, the relatively facile synthetic process enables the exploitation of a wide range of functional groups for grafting the support, affording high versatility of heterogeneous catalysis systems for the hydrogenation of CO2 to formate. With such advantages, there has been notable progress toward the development of novel catalytic systems, viz. representative molecular Ru- and Ir-based catalysts tethered on grafted supports for the hydrogenation of CO2 to formic acid or formate, exhibiting improved catalytic efficiency and stability. Following the remarkable catalytic efficiency of homogeneous Ru–phosphine complexes for the synthesis of formate in supercritical CO2 solutions and the inspiring results obtained from sol–gel-derived silica-supported RuCl2[P(CH3)3]4 for the synthesis of dimethylformamide (DMF), the Zheng group has reported the use of functionalized silica materials

6.2  ­Molecular Catalysts Heterogenized on the Surface of Grafted Support

(including mesoporous MCM-41) containing various coordinating ligands (e.g. NR2, CN, and SH) on the support surface for the immobilization of Ru species (Figure 6.1a) [42–45]. The catalytic performance of the prepared catalysts was predominantly examined in supercritical CO2 solutions (Table  6.1, entries 1–4). Almost all of the prepared catalysts have been reported to be inactive for hydrogenation under the investigated reaction conditions; however, the activity can be enhanced if hydrogenation is performed in the presence of external additives or ligands. These results indicated the formation of an active catalyst in situ upon hydrogenation, containing external ligands as one of the ligands in the coordination sphere of the Ru species. Among the examined external ligands or additives (e.g. PPh3, AsPh3, NPh3, and Ph2P(CH2)2PPh2 (dppe)), the bidentate ligand dppe exhibits a higher catalytic activity than that of the monodentate ligand. The smaller bite angle of dppe (313.082°, 313.418°) compared to PPh3 (314.76°), which reduces the steric hindrance of the dppe ligand with the Ru ion, favors the facile coordination with the Ru ion, and leads to the improved activity for hydrogenation. Studies, which reported on the effect of supports, demonstrated that compared to the silicasupported catalyst, the MCM-41-supported catalyst exhibits higher activity. Some of the factors possibly responsible for this result are the high surface area (850 m2 g−1) and uniform pore structure (3.5 nm) of MCM-41. In addition, the authors have utilized a polystyrene (PS) support to immobilize Ru species (Figure  6.1b) [46]. However, the efficiency of the PS-supported Ru catalysts is less than that of the MCM-41-supported catalysts (Table  6.1, entry 5). The exact reason(s) for the diminished activity of PS-supported catalysts is not known. The alteration of surface functional groups has been reported to strongly affect the outcome of the catalysts. The amine-functionalized catalyst surface exhibits improved catalytic efficiency compared to nitrile- and thiol-functionalized catalysts, related to the stronger coordinating and electron-donating abilities of amine groups. This rationale was further evaluated while simultaneously comparing the efficiencies of primary, secondary, and tertiary amine catalysts; compared to primary- and tertiary amine catalysts, the secondary amine catalyst exhibits higher activity. Despite the fact that amine-functionalized catalysts R R RuIIICl3

R

EtO Si OEt OEt

OHOHOH MCM-41

Si O O O

Si O O O

Silica or MCM-41

Silica or MCM-41

Ligands*

R = NH2, NH(CH2)3CH3), N(CH2CH3)2), CN, SH

(a) PS

Cl

PS

N H

R

PS

N H

R

RuIIICl3

Ligands*

R = NH2, CSCH3, PPh2

(b)

*Ligands : PPh3, AsPh2, NPh3, Ph2P(CH2)2PPh2

Figure 6.1  Schematic representation for synthesis of silica (a) and PS (b) supported Ru complexes.

151

6.2  ­Molecular Catalysts Heterogenized on the Surface of Grafted Support

exhibit higher activity, considerable deactivation is observed upon recycling the catalyst. Conversely, a thiol-functionalized catalyst exhibits improved stability compared to the other catalysts, possibly related to a compact clutch between the Ru and S atoms originating from the better back-donation ability of thiol groups. Following a similar strategy, Han and coworkers have reportedly immobilized a Ru complex (“Si”-(CH2)3NH(CSCH3)-RuCl3-PPh3) for the hydrogenation of CO2 to formate and its subsequent isolation to formic acid using an ionic liquid (IL) as the reusable base (Figure 6.2). Monoamine ([mammim][TfO])- and diamine ([DAMI][TfO])-functionalized ILs are used to obtain 1 : 1 and 2 : 1 molar ratios of formic acid to the IL, respectively. The maximum TOF value of 920 h−1 is obtained by using [DAMI][TfO] IL at 80 °C under a pressure of 18 MPa (Table 6.1, entry 6). Notably, free formic acid is isolated from the IL and catalyst, which is then reused for several consecutive runs [47, 48]. In 2013, Hicks coworkers were the first to report on the use of an immobilized Ir catalyst for hydrogenation [49]. The silica-tethered iminophosphine (P–N)-functionalized hybrid material was used as the support to immobilize IrCl3 via a P─N coordination bond. Bidentate P–N-functionalized silica (PN/SBA-15) was prepared by a Schiff base reaction between o-(diphenylphosphino)benzaldehyde and the amine groups of mesoporous silica (SBA-15). The post-synthetic metalation with IrCl3 affords the P–N-coordinated Ir catalyst (Ir-PN/SBA-15) (Figure 6.3). For comparison, monodentate phosphine and amine counterparts were prepared, in addition to the homogeneous Ir-PN complex. X-ray photoelectron spectroscopy (XPS) analysis was employed to confirm the bidentate coordination of Ir with the P–N sites in Ir-PN/SBA-15. The catalytic performance of the as-prepared Ir catalysts was examined in an aqueous triethylamine (Et3N) solution with an Ir concentration of 100 μM. Under the investigated reaction conditions, only phosphine-containing catalysts are active for hydrogenation. Among the phosphine-containing catalysts, Ir-PN/SBA-15 exhibits the highest activity, with a maximum turnover number (TON) of 2700 at 60 °C in 20 hours (Table 6.1, entry 7). The activity of the unsupported catalyst is almost 20 times less than that of Ir-PN/SBA-15, demonstrating the potential of catalyst immobilization for the improvement in activity and stability. The filtrate of Ir-PN/SBA-15, which catalyzes the reaction solution, does not suggest any detectable quantity of Ir by inductively coupled plasma (ICP)-optical emission spectrometry (OES) analysis, whereas a high metal content is observed in the case of monodentate catalysts. These results indicated that the bidentate catalyst is more stable than the monodentate catalyst under the investigated reaction conditions. The reusability of the Ir-PN/SBA-15 catalyst was investigated at a short reaction

S

RuIIICl3

S

Cl

NH

Si

Si

Si

O O O

O O O

O O O

Silica

Silica

Silica

NH

RuIII(PPhx)yClz

CO2 + H2 [DAMI] +– [TfO]

Cl

EtO Si OEt OEt

2HCOOH [DAMI] + – [TfO] [DAMI] +– [TfO] 2HCOOH

Figure 6.2  Schematic representation for synthesis of “Si”-(CH2)3NH(CSCH3)-RuCl3-PPh3 and IL-based CO2 hydrogenation process.

153

154

6  Heterogenized Catalyst for the Hydrogenation of CO2 to Formic Acid or Its Derivatives

Ph2P

Ph2P

O

N OHMe

MeO Si OMe

NH2

MeO

Si

OMe OMe

SBA-15

Ir-PN/SBA-15

PHPh2 N IrIIICl3

Si(OMe) O O SBA-15

Figure 6.3  Schematic representation for synthesis of Ir-PN/SBA-15. Source: From Xu et al. [49], © 2013 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

time of 0.5 hours at 60 °C under a total pressure of 4 MPa (H2/CO2  =  1) The catalyst is active, albeit with reduced efficiency, even after 10 runs with an average TON of 70. XPS analysis of the reused catalyst revealed that the binding energy of Ir is identical to that of fresh Ir-PN/SBA-15. The slight decrease in activity upon consecutive runs is related to the catalyst loss during workup. Considering the excellent CO2 capture capacity of polyethyleneimine (PEI) polymers, Hicks et  al. have developed tunable, multifunctional PEI-based Ir materials for the combined capture and conversion of CO2 to formate (Scheme 6.1) [50]. These multifunctional materials, which can serve as CO2 capture materials, a formate stabilizer, and a catalyst O H N

H N

N

Ph2P

H N

H N

N

N

NH2 Ph2P

H N

H N

N

Cl3Ir

N

Ph2P

PEI-PN/Ir

Scheme 6.1  Synthesis of PEI-PN/Ir.

6.2  ­Molecular Catalysts Heterogenized on the Surface of Grafted Support

support, are synthesized by the modification of a branched PEI backbone with an iminophosphine ligand by the Schiff base reaction with o-(diphenylphosphino)benzaldehyde, followed by post-metalation with IrCl3 via a P─N coordination bond (PEI-PN/Ir). The structure–property relationship was elucidated by the investigation of the effects of the material structures and reaction conditions on the CO2 capture and catalytic activity (Table 6.1, entry 8). Similar to their previous study on Ir-PN/SBA-15, the performance of the bidentate Ir catalyst is considerably greater than that of the monodentate catalyst. Studies on the catalyst loading, which can be altered by the variation in the percentage of the total primary amines that converted to PN/Ir sites in the PEI backbone, revealed that the ability of Ir catalysts is extremely dependent on the ratio of the available amines to the active Ir(III) centers: the material with a catalyst loading of 65% exhibits high activity. As has been reported by the authors, one of the reasons for the low activity of materials with catalyst loadings of 25% and 95% is related to the generation of Ir nanoparticles, which is confirmed by transmission electron microscopy (TEM) and XPS analyses. Studies on the effect of the PEI molecular weight (MW) revealed that low-MW materials are typically more active for CO2 conversion and at low temperatures, whereas high-MW materials are relatively more stable during the recycling reactions. The inferior stability of the low-MW materials during recycling experiments is related to the increased solubility as well as the formation of catalytically inactive Ir(0) species during hydrogenation. By performing studies on the ratio between water and the external base, the authors have proposed that the amine functionality on the PEI backbone serves as the base for formate stabilization and H2 heterolysis. However, the extremely low yield of formate in the absence of an external base suggested that the amines in the PEI backbone do not significantly stabilize the product under these conditions. Yamashita et  al. have exploited titanate nanotubes (TNTs) as the catalyst support to immobilize the tethered Ir molecular complex on the iminophosphine groups on PEI (Figure 6.4) [51]. In their study, synthesis was carried out in two steps. First, the incorporation of the imino-phospine ligand groups on PEI was conducted by the Schiff base reaction to the PEI chain end. As a result, the strong electron-donating ability of the P and N sites induced on PEI facilitates the interaction with Ir complexes, rendering it to be a prominent metal scaffold. Besides, the presence of abundant amines on the PEI structure, i.e. primary, secondary, and tertiary amines, again facilitates the proximity of the catalytically active sites for the condensation of CO2 and its subsequent conversion to formic acid. Second,

TNT

H N

H N

N N Ph2P

H N

TNT IrCl3

H N

N

Cl3Ir

N

Ph2P

Ir-PN-PEI@TNT(Na+) Figure 6.4  Schematic representation for synthesis of Ir-PN-PEI@TNT(Na+). Source: From Kuwahara et al. [51], © 2017 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

155

156

6  Heterogenized Catalyst for the Hydrogenation of CO2 to Formic Acid or Its Derivatives

considering the enhancement of the catalyst longevity, the Ir-PN-PEI catalyst was introduced onto the TNT supports. TNTs are selected as the solid support due to their high surface area, large pore volume (6 nm), and high tolerance to alkaline conditions, making them suitable for the hydrogenation of CO2. In particular, abundant Ti metals on the TNT pore surface induce rich acidic sites, playing a key role for the strong binding with the Ir-PN-PEI catalyst. To investigate the effect of the support, different types of TNT were prepared, i.e. TNT(Na+) and TNT(H+), by controlling the pH during washing. The catalytic activity was screened by the variation of time, temperature, and NaOH concentration, reaching the maximum TON of 1012 under rather mild conditions at 140 °C at a pressure of 2.0 MPa (1.0 MPa H2/1.0 MPa CO2) in 20 hours with Ir-PN-PEI@TNT(Na+) (Table 6.1, entry 9). Compared to Ir-PN-PEI, the heterogenized catalyst exhibits noticeably enhanced activity, related to the assistance derived from the high CO2 adsorption capacity of the TNT supports. In addition, the maximum loading amount of the Ir-PN-PEI catalyst on the TNTs is reported to reach up to 44 wt%. However, with the incorporation of 40 wt% of Ir-PN-PEI on TNT(Na+), the catalytic activity decreases due to the pore blockage. Meanwhile, Ir-PNPEI@TNT(Na+) exhibits more improved stability in the reusability test (~73% after the 3rd cycle) compared to Ir-PN-PEI@TNT(H+) due to the more basic characteristics of the pore surface, which hinder the formation of the Ir nanoparticles. In the same year, the group again reported a Ru molecular catalyst atomically located on the surface of a layered double hydroxide (LDH) structure, i.e. hydrotalcite (Figure  6.5) [52]. The support material with a formula of Mg10Al2(OH)24](CO3)·mH2O was prepared by co-precipitation by using Mg and Al precursors in an aqueous solution, and the morphology of the material was determined by the pH and temperature control. Thereafter, the Ru species was immobilized on the LDH surface, where the Ru precursor was pretreated with the NaOH aqueous solution to convert it into a more electron-rich form, i.e. Ru(OH)63−, preventing the dissolution of the basic sites on LDH during metalation. Upon meticulous analyses, the Ru species were placed onto triads of a hydroxide on the brucite layer surface of LDH with the coordination environment of Ru(IV)(OHLDH)3(OH)(H2O)2, referred to as Ru/LDH. Owing to the strong electron-donating ability of the hydroxyl groups on the supports, the isolation of Ru complexes on a single atomic level is stabilized without further agglomeration. The hydrogenation of CO2 to formate was performed by using Ru/LDH, as

H2O

OH H2O Ru

O O O

LDH Ru/LDH

Mg2+

Al3+ –

OH

H2O CO32–

Figure 6.5  Schematic representation of Ru/LDH. Source: From Mori et al. [52], © 2017 American Chemical Society. Reprinted with permission of American Chemical Society.

6.3  ­Molecular Catalysts Heterogenized on Coordination Polymer

well as other Ru catalysts, for comparison. The catalytic ability of Ru/LDH is more effective than the others, affording a TON of 461 under 2.0 MPa (H2/CO2 = 1) at 100 °C for 24 hours in a 1.0 M NaHCO3 aqueous solution. By the investigation of the correlation between the XPS-binding energy for Ru3p and TON, the strong electron-donating ability of the LDH support, which affords the lowered binding energy for the Ru metal center, leads to the increase in the catalytic activity. Furthermore, the catalytic performance increases by using NaOH and KOH as the bases, affording TONs of 698 and 489, respectively (Table 6.1, entry 10). In particular, Ru/LDH catalyzed the hydrogenation of CO2 to formate even under neutral conditions, possibly related to the key role of abundant hydroxide on the LDH support surface as an efficient base. To elucidate the effects of surface basicity on the catalytic performance, a series of LDH supports were prepared with various M2+/M3+ components and ratios. Among these supports, a Mg2+/Al3+ ratio of 5 was selected to be optimum due to the fact that the basicity of the LDH surface was determined by the arrangements of cations, which is related to the concomitant proximity of CO2 to the metal sites by chemical adsorption. The catalyst was directly recovered by centrifugation and found to be recyclable at least three times, affording a hydrogenation yield of 90% from its original catalytic performance without the structural change or agglomeration of metal complexes. Recently, Copéret et al. have developed an immobilized Ru molecular catalyst on a lutidinederived bis-N-heterocyclic-carbene (NHC) CNC-pincer-ligand-tethered silica-based support (Figure 6.6) [53]. The synthesis of the hybridized catalyst was carried out in three steps. First, azidopropyl-functionalized SBA silica was prepared as an inorganic support by the sol–gel method using tetraethoxysilane, 3-(azidopropyl)trimethoxysilane, and a triblock copolymer surfactant, i.e. P123, affording a high surface area (706 m2  g−1) and mesopores (7.2 nm). Second, functional SBA silica was grafted with the pre-synthesized lutidine-derived pincer ligand by the click reaction between the substituted alkyne groups on the ligand and the azido group dangling on the SBA support surface. Then, the support was passivated with Me3SiCl to prevent the possible side coordination of Ru with the OH groups remaining on the silica surface. Finally, metalation was performed using RuHCl(CO)(PPh3)3 as the precursor by the treatment of LiHMDS and LiBr for the deprotonation of C2-H on imidazolium for the Ru-NHC coordination. The intact structure of the Ru complexes heterogenized on the support, attaining the RuHCl(CO)–CNC coordination, was investigated by meticulous analyses. The hydrogenation of CO2 to formate was performed under a pressure of 4 MPa (0.5 MPa of CO2 with 3.5 MPa of H2) at 85 °C by using 1,8-diazabicyclo(5.4.0)undec-7-ene (DBU) as the base in a DMF solution (Table 6.1, entry 11). The catalyst exhibits pronounced activity with a maximum TON of 18 000 for 24 hours; this value exceeds the TONs of Ru precursors and homogeneous analogs. However, the catalyst is deactivated within the extended reaction time, the reason for which is not elucidated thus far.

6.3 ­Molecular Catalysts Heterogenized on Coordination Polymers Metal–organic frameworks (MOFs), which are a subclass of coordination polymers, are well-known materials in several research fields, including gas storage and separation [68], sensors [69], and catalysts or catalytic supports [70–72], due to their versatile structural and

157

OEt EtO Si OEt OEt + OEt EtO Si OEt +

N3

Pluronic P-123

N

N3

N

N

+

N N N

O

N

OH + Si Si O O O O O

O

N

+

N

OH Si Si O O O O O

N

+ –

–

2Cl

N

N 2Cl

N

N N N

N

O

+

N

N

OTMS + Si Si O O O O O

N

+

–

2Cl

N

N N N

O

+

N

N

N

N

Cl RuCO H N

OTMS + Si Si O O O O O

Figure 6.6  Schematic representation for synthesis of lutidine-derived bis-NHC CNC-pincer-ligand-tethered silica-based Ru molecular catalyst. Source: From Lo and Copéret [53], © 2018 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

6.3  ­Molecular Catalysts Heterogenized on Coordination Polymer

chemical properties. MOFs inherently exhibit high crystallinity and ultrahigh porosity based on their host structure. In particular, recently, MOFs are emerging as relevant catalytic supports for the heterogenization of molecular metal catalysts due to the facile tunability of functional groups embedded into the framework, which could provide a diverse coordination scaffold for the catalytic-active metal centers, affording superior catalytic activity and stability for various catalytic reactions. However, the use of MOFs as a catalytic support for the hydrogenation of CO2 to formic acid is limited, because wet and basic conditions, typically required for hydrogenation, can lead to the structure destruction. This section reports examples of the successful use of MOF as the catalyst support that overcomes past limitations. Lin coworkers have investigated molecular iridium complexes heterogenized on MOFs for the hydrogenation of CO2 to formic acid by using a Soxhlet-type reflux system (Figure 6.7) [54]. MOF-based supports were prepared using Zr, biphenyl-4,4′-dicarboxylate, and a 2,2′-bipyridine-5,5′-dicarboxylate ligand with or without 6-OH, referred to as mbpyOH-UiO or mbpy-UiO, respectively. The bipyridyl sites in MOF-based supports play a key role for the isolation of an Ir complex, i.e. IrCl3, via the stable coordination of the framework in the bidentate mode, preventing the polymerization of the dimer or oligomer without steric hindrance by bulky ligands such as Cp*. In particular, the incorporated hydroxyl groups exhibit strong electron donation to the bipyridine sites and simultaneously provide hydrogen-bonding sites, which facilitate the hydrogenation of CO2. However, MOFs are inherently vulnerable to basic conditions, rendering them unsuitable for practical applications for the hydrogenation of CO2. Hence, in their study, the catalyst plate is separated from the reaction mixture by suitable placement under the condenser of a Soxhlet extractor. Hydrogenation was conducted at the interface of a dynamic triphasic system of H2/CO2, hot water, and a solid catalyst under ambient pressure and a temperature of 85 °C. The as-formed formic acid penetrates through the plate and collects at the bottom of the flask containing NaHCO3, finally producing formate. The novel catalytic system for the hydrogenation of CO2 exhibits a maximum TON of 6149 ± 50 and TOF of 410 ± 3 in 15 hours (Table 6.1, entry 12). Furthermore, mbpyOH-[IrIII]-UiO maintains catalytic activity without any clear loss for three recycling runs. As another type of an MOF, MIL-101(Cr)-NH2 has been recently utilized as a catalyst scaffold to develop a versatile CO2 hydrogenation catalyst. Ma coworkers have functionalized MIL-101(Cr)-NH2, which was well known for its high porosity and thermal stability, especially its chemical stability, by the post-synthetic modification with salicylic aldehyde and 2-diphenyl phosphinobenzaldehyde (DPPBde), referred to as MIL-101(Cr)-Sal and MIL-101(Cr)-DPPB (Figure 6.8a) [55]. During the immobilization, each of the MOF-based supports exhibits different binding states for Ru in their respective coordination mode, i.e. Ru-N by amine groups of MIL-101(Cr)-NH2, Ru-NO by the Schiff base moiety, phenolic hydroxide in salicylic aldehyde, and Ru-NP by DPPBde. The hydrogenation of CO2 to formate was initially investigated by using a pressure of 60 bar (CO2/H2 = 1 : 1) at 80 °C for two hours in an Et3N aqueous solution: Compared to the other catalysts, RuCl3@ MIL-101(Cr)-DPPB exhibits a higher catalytic activity with a TON of 105, possibly related to the stronger electron-donating ability by the PN bidentate ligands derived from DPPBde. The catalytic performance was further optimized by the increase in the temperature to 120 °C, achieving a TON of 242. Thereafter, the group examined the effect of the reaction

159

mbpyOH-[IrIII]-UiO COOH

COOH R

R N +

N

ZrCl4

N

N

COOH

CO2 + H2

COOH

R N N

R

R N N

Cl

Cl

N

OH

H2O

IrIII

IrIII

Cl

H2O H2O

Solv.

N

HCOO– Hbase

+

Figure 6.7  Schematic representation for synthesis of mbpyOH-[IrIII]-UIO and CO2 hydrogenation via Soxhlet extractor. Source: From An et al. [54], © 2017 American Chemical Society. Reprinted with permission of American Chemical Society.

6.4  ­Molecular Catalysts Heterogenized on Porous Organic Polymer

solvent for the hydrogenation of CO2 to formate, indicating that the high polarity of a solvent positively affects the catalytic activity following the polarity order of dimethyl sulfoxide (DMSO) > methanol > ethanol > isopropanol. In particular, significant performance enhancement is observed by using a mixture of water and DMSO due to the key role as a hydrogen-bond donor for stabilizing formate as well as for rendering high mutual solubility of water to DMSO and triethylamine. In addition, with the additional electronic assistance of the external PPh3 ligand to the complex, the catalytic activity of RuCl3@ MIL-101(Cr)-DPPB reaches up to a TON of 831 under optimized conditions (Table  6.1, entry 13). However, the investigation of the catalyst state after the hydrogenation and recycling was not addressed in the studies. Recently, the same group has reported a series of Ru molecular catalysts heterogenized on azolium-based MOFs for the catalytic hydrogenation of CO2 to formate (Figure 6.8b) [56]. The azolium-containing MOF support prepared by the coordination of a zinc ion with an azolium carboxylate linker, which was the precursor for the N-heterocyclic carbene moiety, exhibits thermal stability up to 250 °C and a Brunauer–Emmett–Teller (BET) surface area of 52 m2 g−1. During post-synthetic metalation, Ru species were immobilized onto the MOF support by using RuCl3, [RuCp*Cl2]2, and [Ru(C6Me6)Cl2]2 as the precursors via the Ru-NHC coordination environment, which was examined to be highly stable due to the strong σ-electron donation by the NHC ligand. The catalytic hydrogenation ability of CO2 to formate by using a Ru-NHC-MOF catalyst was screened under various temperatures and pressures, as well as using different additives to obtain the optimum conditions. The catalytic performance is enhanced in the presence of the C6Me6 ligand compared to the others, related to the higher electron-donating ability from C6Me6. Thereafter, a maximum TON value of 3803 is obtained under a pressure of 80 MPa at 120 °C in a solution of triethylamine and DMF for two hours with K2CO3 as the additive, possibly related to the assistance of the additive as a ligand as well as the high polarity of DMF, facilitating the insertion of CO2 into the Ru─H bond (Table 6.1, entry 14). Ru3-NHC-MOF was recycled five runs without the critical loss of activity, again indicative of the intact presence of Ru complexes on the NHC support after hydrogenation.

6.4 ­Molecular Catalysts Heterogenized on Porous Organic Polymers Porous organic polymers (POPs) are emerging catalytic supports for various catalytic applications, in particular for the hydrogenation of CO2 to formic acid [36–39, 73–75]. POPs are typically amorphous materials that exhibit abundant microporosity, high surface area, and thermal or chemical stability even under aggressive conditions, inducing long-term stability to catalytic systems. Besides, based on the rational selection of building blocks and linking strategy, the tuning of various factors on the network structures renders POPs relevant candidates as catalytic supports for developing a novel heterogenized catalytic system for CO2 hydrogenation. In particular, the incorporation of well-defined functional groups or ligand sites on building blocks of POPs offers a coordination environment for the active metal center similar to that offered to homogeneous analogs, enriching the chemical and geometrical functions of the heterogeneous catalytic systems. In this section, we will

161

162

6  Heterogenized Catalyst for the Hydrogenation of CO2 to Formic Acid or Its Derivatives R

R:

R RuCl3

: CrIII

NH2,

, N

N HO

RuCl 3@MIL-101(Cr)-R PPh2

(a)

: ZnII N

Ru3-NHC-MOF

N N

Cl

RuCl2

N

Ligand

(b)

: *Ligand: Cl, C5Me5 (Cp*), C6Me6

Figure 6.8  Schematic representation of RuCl3@MIL-101(Cr)-R (a) and Ru3-NHC-MOF (b). Source: Wang et al. [55], © 2019 Elsevier. Reprinted with permission of Elsevier, and from Wu et al. [56], © 2019 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

discuss decades of prominent progress in molecular catalysts heterogenized on the POPbased supports. In 2015, Liu coworkers had reported the first study utilizing a POP as a support for hydrogenation [57]. Similar to the rationale reported by McNamara and Hicks [50] for PEI-based Ir catalysts, these authors have designed a microporous organic polymer derived from Troger’s base (TB-MOP) as a CO2 adsorbent and a support for Ru(III) immobilization (Figure 6.9). TB-MOP was obtained as a brown powder by the one-pot reaction between tris(4-aminophenyl)amine and dimethoxymethane in trifluoroacetic acid at ambient temperatures. The prepared TB-MOP exhibited a BET surface area of 802 m2 g−1 and a pore volume of 0.50 cm3 g−1, with a stability up to 300 °C. By the immobilization of RuCl3 with 4.87 wt% of TB-MOP-Ru, the BET surface area and pore volume decrease to 540 m2/g and 0.33 cm3 g−1, respectively. Compared with that of TB-MOP, the CO2 adsorption capacity of TB-MOP-Ru

NH2

3

H2N

O

O

N RuCl3 N

NH

2

Figure 6.9  Schematic representation for synthesis of TB-MOP-Ru. Source: From Yang et al. [57], © 2015 Royal Society of Chemistry. Reprinted with permission of Royal Society of Chemistry.

6.4  ­Molecular Catalysts Heterogenized on Porous Organic Polymer

TB-MOP-Ru was slightly lower at a pressure of 1 bar, corresponding to its lower surface area caused by the complexation with the Ru(III) ion. The catalytic ability of TB-MOP-Ru was investigated in a neat Et3N solution at 40 °C under a total pressure of 12 MPa (H2/ CO2 = 1) for 24 hours. The catalytic efficiency of TB-MOP-Ru is extremely low (TON = 25) under the investigated reaction conditions; however, the activity substantially improves when hydrogenation is performed in the presence of PPh3 (TON = 2254) (Table 6.1, entry 15). On the other hand, while the catalyst is reused, the catalytic performance of TB-MOP-Ru decreases during the second run, corresponding to the leaching of the Ru species into the solution. The main cause of leaching was assumed to be related to the weaker complexing ability of the Troger’s base compared to PPh3 with the Ru species. Considering the high thermal and chemical stability, excellent acid–base resistivity, and rigid and tunable pore structures with high surface areas, as well as the potential to immobilize various transition metal complexes onto the pore walls of covalent triazine frameworks (CTFs), Yoon coworkers have employed CTFs as support materials for developing practically viable heterogeneous catalytic entities for the successful production of formic acid via CO2 hydrogenation in the industry [58]. CTFs is a class of porous organic frameworks (POFs) comprising triazine building blocks, which are typically synthesized by the ionothermal trimerization of aromatic dinitriles at elevated temperature using ZnCl2. Compared with those of other POFs, the physicochemical properties of CTFs such as surface area and pore volume, which are highly essential for the smooth diffusion of the reaction species, including substrates and products, can be easily tuned by the simple variation in the reaction parameters, such as temperature and ZnCl2 loading. In their first study, 2,2′-bipyridine (bpy)-incorporated CTF (bpy-CTF) was used as the support material for the immobilization of a half-sandwich Ir complex, i.e. [IrCp*(bpy)Cl]Cl, corresponding to one of the derivatives of a highly active homogeneous catalyst (Figure 6.10a, 1). The [IrCp*(bpyCTF)Cl]Cl catalyst was prepared by the treatment of the Ir precursor [(IrCp*Cl2)2] with bpy-CTF at 400 °C using five equivalents of ZnCl2, in methanol/chloroform at a reflux temperature for 24 hours. As hypothesized, the coordination environment of [IrCp*(bpy-CTF) Cl]Cl is confirmed to be similar to its homogeneous counterparts. Considering the superiority of the aqueous Et3N solution toward the yield of formate, which was obtained by screening various bases using homogeneous Ir catalysts, the catalytic activity of [IrCp*(bpyCTF)Cl]Cl was investigated in an aqueous solution using Et3N as the base. Maximum TON and TOF values of 5000 and 5300 h−1, respectively, are obtained at 120 °C under a total pressure of 8 MPa (Table 6.1, entry 16). However, the catalytic efficiency is slightly reduced while reusing the catalysts in consecutive runs. In addition, the authors have investigated the use of a heptazine-based organic framework (HBF-2) as the support for the immobilization of the homogeneous catalyst [IrCp*(bpy)Cl]Cl (Figure 6.11) [59]. The XPS analysis of the [IrCp*(HBF-2)Cl]Cl catalyst demonstrated that the electron density on Ir is slightly less than that on the homogeneous catalyst [IrCp*(bpy)Cl]Cl, possibly related to the inferior electron-donating ability of the heptazine ligands. The catalytic activity of [IrCp*(HBF-2)Cl]Cl was investigated again in an aqueous medium by using Et3N as the base. The catalyst exhibits a maximum TON of 6400 and a TOF of 1500 h−1 at 120 °C under a total pressure of 8 MPa (Table 6.1, entry 17). Similar to their previous study, the catalytic activity of [IrCp*(HBF-2)Cl]Cl is reduced with the reuse of the catalyst.

163

CN

N

3 N

[MLx(bpy-CTF)X]X

N

N CN

N

N N

N

MLxXy – – 1 : M = IrIII, L = C5Me5 , –X = 2Cl –

N

N

N

–

–

2 : M = RhIII, L = C5Me5 , X = 2Cl

N

N

–

4 : M = RuIII, L = 2acac , X = Cl – 5 : M = RuIII, X = 3Cl

N

N N

(a) CN

Cl N

3

X

Ir N

N X

N N

N

N

N

Ir-NHC-CTF

N N

N

X N

X

N

CN

N N

N

Cl

*Cp

N

N

N

–

3 : M = RuIII, L = C6Me6 , X = 2Cl

N

N

N

(b)

Figure 6.10  Schematic representation for synthesis of MLx(bpy-CTF)X]X (a) and Ir-NHC-CTF (b). Source: From Park et al. [58], © 2015. Reprinted with permission of John Wiley and Sons, and from Gunasekar et al. [61], © 2017 American Chemical Society. Reprinted with permission of American Chemical Society.

N

[IrCp*(HBF-2)Cl]Cl N N

N N

N Ir

3

*Cp

Cl

N

NH2

N

N N

+

N

N N

Cl

N

N H2N

N N

N

Figure 6.11  Schematic representation for synthesis of [IrCp*(HBF-2)Cl]Cl. Source: From Hariyanandam et al. [59], © 2016 ELSEVIER. Reprinted with permission of Elsevier.

166

6  Heterogenized Catalyst for the Hydrogenation of CO2 to Formic Acid or Its Derivatives

Bavykina et al. have prepared porous, mechanically rigid CTF-based spheres to overcome the practical difficulties of using CTFs as powder materials (Figure 6.12) [60]. The CTF, which was prepared using 4,4′-biphenyldicarbonitrile and 2,6-dicyanopyridine building blocks, was utilized. The CTF sphere was synthesized using the Matrimid• polyimide as the binder by the phase inversion method. Stability tests in water at different temperatures revealed that the CTF sphere is highly stable in aqueous media; however, its stability in organic solvents (such as acetonitrile and hexane) is not reported. During the immersion of the CTF powder within the spheres, the original surface area of CTF decreases to 465 m2 g−1, possibly related to the partial penetration of the host polymer chain. Then, the fabricated CTF sphere (diameter of a few millimeters) underwent metalation with (IrCp*Cl2)2 in the DMF solvent to coordinate IrCp* with the pyridine moieties of CTF, denoted as Ir@sphere. SEM/EDS analysis revealed that a majority of Ir is located in the outer shell of the sphere. In contrast to powder CTF, the metalated CTF sphere exhibits a slightly greater N2 uptake than that of the non-metalated sphere, possibly related to the partial itching of some binder by the DMF solvent. The catalytic activity of the Ir@sphere was investigated using 50 mg of the catalyst under a total pressure of 2 MPa in an aqueous solution using K2CO3 as the base (Table 6.1, entry 18). Compared to the powder catalyst, the Ir@CTF sphere exhibits diminished efficiency (40–90%); however, the deviation in recycling experiments was significantly lower (5%), indicative of the recycling of the facile catalyst and the improved reproducibility of the Ir@CTF sphere. To improve the efficiency of CO2 hydrogenation catalysts, Yoon coworkers have used an NHC-functionalized CTF (NHC-CTF) as the support for the immobilization of the [IrCp*(N-C)Cl]Cl complex (Figure 6.10b) [61]. The authors hypothesized that the strong electron-donation ability of the carbene ligand (via σ-donating and weak π-accepting characteristics of CTFs) can localize the high electron density on the Ir metal center, and consequently enhance the catalytic activity. The NHC-CTF was synthesized by the ionothermal trimerization of 1,3-bis(5-cyanopyridyl)-imidazolium bromide with ZnCl2 at 400 °C. The heterogenized carbene catalyst Ir-NHC-CTF was prepared by the post-synthetic metalation of NHC-CTF with the Ir precursor [(IrCp*Cl2)2] in the presence of Et3N. By meticulous analyses, the generation of N–C coordination with the IrCp* unit was confirmed. As hypothesized, compared to [IrCp*(bpy-CTF)Cl]Cl, the prepared Ir-NHC-CTF catalyst exhibits improved initial TOFs (16 000 h−1) and TONs (24 300) under similar reaction conditions (Table 6.1, entry 19). Nevertheless, the leaching of Ir is observed, and the efficiency of Ir-NHC-CTF is gradually reduced upon consecutive runs. In their follow-up study on the immobilization of the half-sandwich complexes onto bpyCTF, Yoon and coworkers have reported a detailed study on the effect of various parameters such as the central metal atom, CTF architecture, and metal-leaching pathway of half-sandwich heterogenized catalysts [62]. By following the above procedures, the authors have prepared Rh ([RhCp*(bpy-CTF)Cl]Cl) counterparts to compare their activities with that of an Ir-based catalyst and investigate the effect of the central metal atom on the efficiency of heterogenized catalysts (Figure 6.10a, 2 and 3). Similar to [IrCp*(bpy-CTF)Cl]Cl, the coordination identity of the prepared Rh catalyst is confirmed to be similar to their respective homogeneous counterparts. However, hydrogenation performed using the heterogenized Rh catalyst affords relatively lower TOFs (i.e. 960/h, respectively) compared with that of the Ir counterpart (5300/h), suggesting that the half-sandwich Ir catalyst is

Covalent triazine framework

Syringe pump N N

N H2O

CTF suspension Matrimid solution

Ir

N N

N

Water

Ir@CTF sphere (a)

(*OTf)2 N

N

500 μm (b)

Figure 6.12  Scheme of CTF sphere preparation (a) and SEM image of CTF sphere (b). Source: Reproduced with permission from Bavykina et al. [60], Copyright 2016 Wiley-VCH Verlag GmbH & Co. KGaA.

more relevant for the effective hydrogenation of CO2 to formate compared to the Rh catalysts (Table 6.1, entry 20). Then, the effect of the synthesis temperature of the CTF architecture (two- versus three-dimensional CTF) on the activity of the heterogenized catalyst was investigated. Owing to its high activity, Ir was selected as a metal center in the halfsandwich complexes for further studies. Although the surface area of bpy-CTF synthesized at 500 °C is greater than that of bpy-CTF synthesized at 400 °C, the activity of the Ir catalyst immobilized on as-prepared bpy-CTF at 500 °C is significantly lower (TOF = 1360 h−1). The exact reason for this inferior activity is not known; however, the authors attribute it to the hydrophilic properties. In addition, the author varied the metal loading on [IrCp*(bpyCTF)Cl]Cl and found that an Ir loading of 4.7 wt% is optimum for efficient hydrogenation. Notably, by the treatment of an excess amount of the Ir precursor with bpy-CTF, Ir nanoparticles are generated during the synthesis. Thus, the authors concluded that there is a critical balance between the metal loading and CTF porosity to obtain a desired structure for these complexes. Finally, the authors have provided insights on the diminished recyclability of the heterogenized half-sandwich complexes. By the ICP-atomic emission spectroscopy (AES) analysis of the filtrate and XPS analysis of the recovered catalyst, the inferior recyclability of the half-sandwich heterogenized catalysts is mainly related to the leaching of the metal species. As all of the heterogenized half-sandwich catalysts exhibit this inferior behavior, albeit in varied ranges, the authors have envisaged that the main pathway of leaching is possibly the same for all of the half-sandwich catalysts. To investigate the leaching pathway, density functional theory (DFT) calculations were carried out for [IrCp*(bpy-CTF) Cl]Cl to identify the general deactivation route. Based on the calculated energy barriers and energy input, the authors have suggested that the undesirable interaction of H2 with the N-sites of bpy during H2 heterolysis is the major pathway for the leaching of metals into the solution. To inhibit this deleterious interaction and develop recyclable, efficient heterogeneous catalytic systems, the authors have proposed twofold design strategies that either employ a tridentate pincer framework as the support material or replace Cp* in the metal center by oxyanionic ligands (e.g. acetylacetonate, or acac, carboxylate, and carbonate). In their follow-up study, the authors have replaced Cp* with acac ligands and reported excellent recyclability upon consecutive runs. In their continuous efforts toward the development of practically viable heterogeneous catalysts, Yoon coworkers focused on the improvement of the recyclability and efficiency of the catalyst [63]. Considering the leaching pathway of half-sandwich-based catalysts, which was investigated by DFT studies, a new design strategy that introduces oxyanionic ligand(s) in the coordination sphere has been proposed. The authors have postulated that the oxyanionic ligand(s) in the coordination sphere can simultaneously deter the undesirable interactions between H2 and the bpy N-site, which was the plausible route of leaching, and activate the heterolysis of H2. Acetylacetonate was selected as the oxyanionic ligand to demonstrate the viability of this design strategy. An acac-based Ru catalyst, [bpy-CTFRu(acac)2]Cl, was prepared by the treatment of bpy-CTF with RuCl3 in a methanol solution and subsequently with an acac ligand, exhibiting a surface area and a pore volume of 502 m2 g−1 and 0.29 cm3 g−1, respectively (Figure  6.10a, 4) The catalytic activity of [bpyCTF-Ru(acac)2]Cl was investigated under various reaction conditions in an aqueous solution by using Et3N as the base. The catalyst exhibits efficient hydrogenation of CO2 to

6.4  ­Molecular Catalysts Heterogenized on Porous Organic Polymer

formate with a TON of 21120, affording the highest formate concentration of 1.8 M in five hours (Table  6.1, entry 21). Upon consecutive runs, the catalytic activity of [bpy-CTFRu(acac)2]Cl is considerably maintained in 1 M and 3 M Et3N solutions. Approximately 20.5 g of the formate adduct is generated in four cycles. The characterization of the recovered catalyst demonstrated that [bpy-CTF-Ru(acac)2]Cl serves as a pre-catalyst, and the actual active catalyst is generated in situ during the hydrogenation. By XPS analysis, the authors have suggested that the species generated in situ maintains a 3+ oxidation state for Ru similar to that in [bpy-CTF-Ru(acac)2]Cl; however, the intensity of peaks corresponding to the acac species is reduced, and new peaks corresponding to the H─O─C═O species are generated. Nevertheless, the actual active species have not been completely identified. In addition, the authors have performed DFT calculations to examine the mechanism for the hydrogenation catalyzed by [bpy-CTF-Ru(acac)2]Cl, which will be discussed later in this chapter. Regina Palkovits coworkers first introduced phosphine-based cross-linked polymers as catalyst supports for the hydrogenation of CO2 to formate [64]. In their study, a series of phosphine-based supports, i.e. pTPP, pDPPE, pTPPB, and pTPrP, were prepared by subsequent reactions, i.e. polylithiation of bromo-linker derivatives and crosslinking with PCl3 and 1,2-bis(dichlorophosphino)ethane (Scheme 6.2). The supports exhibit a high thermal stability up to 340 °C and abundance of phosphine atoms homogeneously distributed on the support surface, albeit with variable surface areas of 290 m2 g−1 for pTPPB, 209 m2 g−1 for pTPP, and 33 m2 g−1 for pDPPE, and no measurable porosity for pTPrP. However, all supports could provide an accessible immobilization site for the Ru complex on the support, which is facilitated by the effect of the strong electron donation by abundant P atoms. Catalytic processes were successively conducted to evaluate the effects of the base-salt concentration, catalyst-to-substrate ratio, temperature, pressure, and Ru loading for the hydrogenation of CO2 to formate. High salt concentration induces precipitation in the solution, leading to the dramatic decrease of the catalytic activity. With the further optimization of the reaction conditions, Ru@pDPPE exhibits excellent catalytic performance, reaching a

P

P

P

P

pTPP

P

pTPrP

pTPPB

pDPPE

(a)

Cl P

Ru P

Cl

Ru Cl Cl Cl Ru

P

P

Ru@pDPPE

(b)

Scheme 6.2  Representative structure of porous phosphine polymers (a) and the (b) synthesis of Ru@pDPPE .

169

170

6  Heterogenized Catalyst for the Hydrogenation of CO2 to Formic Acid or Its Derivatives

maximum TON of 13 170 at 120 °C and 100 bar (CO2/H2 = 1) in the presence of a 2.59 M K2CO3 aqueous solution for four hours (Table 6.1, entry 22). Particularly, the increase in the Ru-loading amount on the support polymer leads to the lower reactivity due to the limited internal diffusion of H2 to the active metal centers. In recycling studies, the catalyst exhibits a significant drop in the catalytic performance from the second step. Several meticulous analyses were performed by the authors to elucidate the state of the catalyst after hydrogenation. As a result of the stable binding of Ru atoms and the phosphine-based polymer support, Ru atoms remain the same as those on the fresh catalyst, affording a leaching of only 4% of Ru atoms up to the last run and maintaining a 2+ oxidation state after hydrogenation. However, the structure of the Ru complex was confirmed to significantly change in the coordination environment. The cymene and chlorine of the fresh catalyst were substituted by the carbonyl ligand and hydride generated under hydrogenation conditions. Hence, the poisonous effect of the carbonyl ligand is thought to lead to catalyst deactivation. Tao Zhang and coworkers have envisaged that the incorporation of a pincer-type ligand into solid supports for the heterogenized catalyst leads to the remarkable enhancement in the catalytic performance (Figure 6.13) [65]. In this respect, the aminopyridine-based POP was synthesized using 1,3,5-benzenetricarbonyl chloride and 2,6-diaminopyridine via amide formation as a linking method, referred to as AP-POP. The polymer support exhibits mesoporosity (~7.6 nm) with a BET surface area of 43 m2/g and a reasonable thermal stability of up to 300 °C, which is relevant to hydrogenation. In particular, the presence of welldefined C═O and N─H groups in the framework, which is regarded to be strong metal-chelating groups, plays a key role for the stabilization of metal complexes. Ir/AP-POP was prepared by the deposition of an Ir metal precursor, i.e. H2IrCl6, with NaBH4, followed by the synthesis of Ir/AC and Ir/C3N4 for comparison. The Ir species immobilized on AP-POP does not undergo reduction into nanoparticles. In contrast, 2-nm-sized particles are observed on the supports of Ir/AC and Ir/C3N4. This enhanced stability of the isolated single metal atoms corresponds to the strong interactions between abundant heteroatoms and metal complexes, thereby hindering the metal from undergoing reduction to nanoparticles. By accountable analyses, the group has elucidated that the Ir complex is immobilized on the ONO-based pincer-type coordination environment induced by the carbonyl and pyridine groups on the framework. The hydrogenation of CO2 to formate was investigated under the conditions of a pressure of 60 bar (CO2/H2 = 1 : 1) at 120 °C in a 1 M TEA aqueous solution for 24, affording a TON of 6784 with an AAR of ~0.7, which is the thermodynamic limit. Particularly, the catalytic performance of Ir/AP-POP reached the maximum TON of 25 135, with a lower loading of 0.66% (Table  6.1, entry 23). By recycling experiments, the catalyst was easily recovered by centrifugation and recycled four times, indicating that the catalytic activity significantly decreases and the active Ir species do not change. Recently, Yoon coworkers have reported a simple, efficient, and recyclable catalyst for hydrogenation [66]. The [bpy-CTF-RuCl3] catalyst was prepared by the simple treatment of commercially available RuCl3·xH2O on bpy-CTF in a methanol solution (Figure 6.10a, 5). The coordination environment of Ru was confirmed by the comparison of the XPS results of its homogeneous counterpart. The surface area and pore volume of the prepared catalyst are 477 m2 g−1 and 0.17 cm3 g−1, respectively. The catalyst efficiently converts CO2 to

O

H2N

N

H2O H O

Cl

O Ir H N HN

NH2 Cl

O

NH Ir/AP-POP

Cl

O

O HN N O H N O

N

NH

H N

H N O

O

N

H N O

Figure 6.13  Schematic representation for synthesis of Ir/AP-POP. Source: From Shao et al. [65], © 2019 Elsevier. Reprinted with permission of Elsevier.

formate with a TON of 20 000 and an unprecedented initial TOF of 38 800 h−1 (Table 6.1, entry 24). In addition, the catalyst is extremely selective toward the production of formate, i.e. it does not generate any side products such as CO or methanol. The catalyst converts ~12% of the applied CO2 to formate, which is similar to the conversion obtained for the commercial synthesis of methanol via CO2 hydrogenation. The maximum formate concentration of 2.1 M is obtained in 2.5 hours at 120 °C and a total pressure of 8 MPa. The performance of [bpy-CTF-RuCl3] was well maintained upon recycling the catalytic system. Similar to [bpy-CTF-Ru(acac)2]Cl, the actual active species is generated in situ, and the new peaks corresponding to C─O, C═O, and H─O species are observed in the recovered catalyst. In their continued efforts on preparing simple and efficient catalysts, Yoon and coworkers very recently reported the preparation of phenanthroline-functionalized porous organic polymer (phen-POP) and its subsequent metalation with IrCl3 for the hydrogenation [67]. The phen-POP was synthesized by knitting commercially available bathophenanthroline (4,7-diphenyl-1,10-phenanthroline) with CH2Cl2 through AlCl3-catalyzed Friedel–Crafts reaction (Figure 6.14). The phen-POP was constructed without other reactive/coordinating functional groups in the polymer skeleton (like imine and alkyne bonds) with a BET surface area of 560 m2 g–1. The presence of non-reactive/non-coordinative functional groups in the skeleton with the well-defined and isolated phen sites of phen-POP offered the preparation of single-site metal catalyst for reported reaction. By following their general procedure (i.e. hydrogenation in a solution of aqueous Et3N), the catalytic ability of the catalyst has been demonstrated. Although the generation of real active species for the hydrogenation was not demonstrated, it is expected to follow the pathway proposed for the [bpy-CTFRuCl3] catalyst. The IrCl3-coordinated phen-POP demonstrated superior catalytic performance over other Ir-based catalysts; a maximum initial TOF of 40 000 h−1 and TON of 14 400 has been achieved using this catalyst (Table 6.1, entry 25).

6.5 ­Concluding Remarks and Future Directions In the last decades, the dramatic development of homogeneous catalysts capable of efficiently producing formic acid and formic acid derivatives, which are energy-transferring substances through carbon dioxide hydrogenation, has been reported. The molecular structure of the catalytic sites required for efficient hydrogenation was more fully understood.

N

+ Cl

N

Cl

N N

IrCl3

Phen-POP-Ir

Figure 6.14  Schematic representation for synthesis of Phen-POP-Ir. Source: From Gunasekar and Yoon [67], © 2019 Royal Society of Chemistry. Reprinted with permission of Royal Society of Chemistry.

 ­Reference

Inspired by this academic achievement, a number of intensive studies have been conducted to realize the catalyst having the advantages of the homogeneous catalyst in its activity and the heterogeneous catalyst in its physical properties by heterogenizing the active site of the homogeneous catalyst over past few years. Through these, it was possible to prepare a catalyst immobilized on a porous support capable of producing a 2.1 M formate reaching a maximum initial TOF of 40 000 h−1 and TON of 14 400, which is comparable to the catalytic activity of reported homogeneous catalysts [66]. It is notable that such a catalyst is very easy to separate from the reaction product and maintains the catalytic activity upon reuse. While developed heterogenized catalysts have both high catalytic activity and reusability, the main issues in the commercial possibility of catalytic hydrogenation of CO2 into formic acid relate to the use of expensive base and the development of continuous processes in large scale. In many reported cases, the cost of the base used is higher than the cost of the formic acid derivative produced, making it difficult for these catalytic conversion reactions to be economically viable. To overcome this problem, industrial researches have been carried out to reproduce bases by decomposing formate chemicals and separating formic acid. Also, most of the reported results have been carried out in a batch reactor, which has limitations at the current level for application to continuous processes. Indeed, many challenges remain with regards to creating economically viable continuous catalytic CO2 hydrogenation process into formic acid while recycling bases without loss. It is believed that the recently patented catalytic system with the CTF-based RuCl3 catalyst by using trickle-bed reactor [76], which is one of the representative fixed-stage catalytic reactors, shows a gradual movement of research in this direction.

R ­ eferences 1 Chen, Q., Lv, M., Tang, Z. et al. (2016). Opportunities of integrated systems with CO2 utilization technologies for green fuel & chemicals production in a carbon-constrained society. J. CO2 Util. 14: 1–9. 2 Bennett, S.J., Schroeder, D.J., and McCoy, S.T. (2014). Towards a framework for discussing and assessing CO2 utilisation in a climate context. Energy Procedia 63: 7976–7992. 3 Abanades, J.C., Rubin, E.S., Mazzotti, M., and Herzog, H.J. (2017). On the climate change mitigation potential of CO2 conversion to fuels. Energy Environ. Sci. 10: 2491–2499. 4 Bruhn, T., Naims, H., and Olfe-Kräutlein, B. (2016). Separating the debate on CO2 utilisation from carbon capture and storage. Environ. Sci. Pol. 60: 38–43. 5 Schakel, W., Fernández-Dacosta, C., Van Der Spek, M., and Ramírez, A. (2017). New indicator for comparing the energy performance of CO2 utilization concepts. J. CO2 Util. 22: 278–288. 6 Yang, Y. and Lee, J. (2019). Toward ideal carbon dioxide functionalization. Chem. Sci. 10: 3905. 7 Yuan, Z., Eden, M.R., and Gani, R. (2015). Toward the development and deployment of largescale carbon dioxide capture and conversion processes. Ind. Eng. Chem. Res. 55: 3383–3419. 8 Dimitriou, I., García-Gutiérrez, P., Elder, R.H. et al. (2015). Carbon dioxide utilisation for production of transport fuels: process and economic analysis. Energy Environ. Sci. 8: 1775–1789.

173

174

6  Heterogenized Catalyst for the Hydrogenation of CO2 to Formic Acid or Its Derivatives

  9 Otto, A., Grube, T., Schiebahn, S., and Stolten, D. (2015). Closing the loop: captured CO2 as a feedstock in the chemical industry. Energy Environ. Sci. 8: 3283–3297. 10 Naims, H. (2016). Economics of carbon dioxide capture and utilization—a supply and demand perspective. Environ. Sci. Pollut. Res. 23: 22226–22241. 11 Liu, Q., Wu, L., Jackstell, R., and Beller, M. (2015). Using carbon dioxide as a building block in organic synthesis. Nat. Commun. 6: 5933. 12 Aresta, M., Dibenedetto, A., and Angelini, A. (2013). Catalysis for the valorization of exhaust carbon: from CO2 to chemicals, materials, and fuels. Technological use of CO2. Chem. Rev. 114: 1709–1742. 13 Boddien, A., Federsel, C., Sponholz, P. et al. (2012). Towards the development of a hydrogen battery. Energy Environ. Sci. 5: 8907–8911. 14 Onishi, N., Laurenczy, G., Beller, M., and Himeda, Y. (2018). Recent progress for reversible homogeneous catalytic hydrogen storage in formic acid and in methanol. Coord. Chem. Rev. 373: 317–332. 15 Sordakis, K., Tang, C., Vogt, L.K. et al. (2017). Homogeneous catalysis for sustainable hydrogen storage in formic acid and alcohols. Chem. Rev. 118: 372–433. 16 Johnson, T.C., Morris, D.J., and Wills, M. (2010). Hydrogen generation from formic acid and alcohols using homogeneous catalysts. Chem. Soc. Rev. 39: 81–88. 17 Enthaler, S., von Langermann, J., and Schmidt, T. (2010). Carbon dioxide and formic acid—the couple for environmental-friendly hydrogen storage? Energy Environ. Sci. 3: 1207–1217. 18 Grasemann, M. and Laurenczy, G. (2012). Formic acid as a hydrogen source – recent developments and future trends. Energy Environ. Sci. 5: 8171–8181. 19 Mellmann, D., Sponholz, P., Junge, H., and Beller, M. (2016). Formic acid as a hydrogen storage material – development of homogeneous catalysts for selective hydrogen release. Chem. Soc. Rev. 45: 3954–3988. 20 Loges, B., Boddien, A., Gärtner, F. et al. (2010). Catalytic generation of hydrogen from formic acid and its derivatives: useful hydrogen storage materials. Top. Catal. 53: 902–914. 21 van Putten, R., Wissink, T., Swinkels, T., and Pidko, E.A. (2019). Fuelling the hydrogen economy: Scale-up of an integrated formic acid-to-power system. Int. J. Hydrog. Energy 44: 28533–28541. 22 GRT Group (2019), Hyform-PEMFC. https://grtgroup.swiss (accessed 10 May 2019). 23 Leitner, W. (1995). Carbon dioxide as a raw material: the synthesis of formic acid and its derivatives from CO2. Angew. Chem. Int. Ed. 34: 2207–2221. 24 Jessop, P.G., Ikariya, T., and Noyori, R. (1995). Homogeneous hydrogenation of carbon dioxide. Chem. Rev. 95: 259–272. 25 Wang, W., Wang, S., Ma, X., and Gong, J. (2011). Recent advances in catalytic hydrogenation of carbon dioxide. Chem. Soc. Rev. 40: 3703–3727. 26 Wu, L., Liu, Q., Jackstell, R., and Beller, M. (2015). Recent progress in carbon dioxide reduction using homogeneous catalysts. In: Carbon Dioxide and Organometallics, 279–304. Springer. 27 Wang, W.-H., Himeda, Y., Muckerman, J.T. et al. (2015). CO2 hydrogenation to formate and methanol as an alternative to photo- and electrochemical CO2 reduction. Chem. Rev. 115: 12936–12973.

 ­Reference

28 Behr, A. and Nowakowski, K. (2014). Chapter seven-Catalytic hydrogenation of carbon dioxide to formic acid. In: Advances in Inorganic Chemistry, vol. 66, 223–258. Elsevier. 29 Schaub, T. and Paciello, R.A. (2011). A process for the synthesis of formic acid by CO2 hydrogenation: thermodynamic aspects and the role of CO. Angew. Chem. Int. Ed. 50: 7278–7282. 30 Schaub, T., Bey, O., Meier, A., Fries, D., and Hugo, R. (2013). Method for producing formic acid by reacting carbon dioxide with hydrogen. WO2013050367A2. 31 Leitner, W., Hintermair, U., and Wesselbaum, S. (2012). CO2 hydrogenation method for producing formic acid. WO Patent 2,012,095,345. 32 Wesselbaum, S., Hintermair, U., and Leitner, W. (2012). Continuous-flow hydrogenation of carbon dioxide to pure formic acid using an integrated scCO2 process with immobilized catalyst and base. Angew. Chem. Int. Ed. 51: 8585–8588. 33 Jens, C.M., Scott, M., Liebergesell, B. et al. (2019). Rh-catalyzed hydrogenation of CO2 to formic acid in DMSO-based reaction media: solved and unsolved challenges for process development. Adv. Synth. Catal. 361: 307–316. 34 Artz, J., Müller, T.E., Thenert, K. et al. (2018). Sustainable conversion of carbon dioxide: an integrated review of catalysis and life cycle assessment. Chem. Rev. 118: 434–504. 35 Scott, M., Blas Molinos, B., Westhues, C. et al. (2017). Aqueous biphasic systems for the synthesis of formates by catalytic CO2 hydrogenation: integrated reaction and catalyst separation for CO2-scrubbing solutions. ChemSusChem 10: 1085–1093. 36 Gunasekar, G.H., Park, K., Jung, K.-D., and Yoon, S. (2016). Recent developments in the catalytic hydrogenation of CO2 to formic acid/formate using heterogeneous catalysts. Inorg. Chem. Front. 3: 882–895. 37 Álvarez, A., Bansode, A., Urakawa, A. et al. (2017). Challenges in the greener production of formates/formic acid, methanol, and DME by heterogeneously catalyzed CO2 hydrogenation processes. Chem. Rev. 117: 9804–9838. 38 Bulushev, D.A. and Ross, J.R. (2018). Heterogeneous catalysts for hydrogenation of CO2 and bicarbonates to formic acid and formates. Catal. Rev. 60: 566–593. 39 Wang, W.-H., Feng, X., and Bao, M. (2018). Chapter Three-Transformation of CO2 to formic acid or formate over heterogeneous catalysts. In: Transformation of Carbon Dioxide to Formic Acid and Methanol, 43–52. Springer. 40 Barbaro, P. and Liguori, F. (2010). Heterogenized Homogeneous Catalysts for Fine Chemicals Production: Materials and Processes, vol. 33. Springer Science & Business Media. 41 Hartley, F.R. (2012). Supported Metal Complexes: A New Generation of Catalysts, vol. 6. Springer Science & Business Media. 42 Zhang, Y., Fei, J., Yu, Y., and Zheng, X. (2004). Silica immobilized ruthenium catalyst used for carbon dioxide hydrogenation to formic acid (I): the effect of functionalizing group and additive on the catalyst performance. Catal. Commun. 5: 643–646. 43 Ying-Min, Y., Yi-Ping, Z., Jin-Hua, F., and Xiao-Ming, Z. (2005). Silica immobilized ruthenium catalyst for formic acid synthesis from supercritical carbon dioxide hydrogenation II: effect of reaction conditions on the catalyst performance. Chin. J. Chem. 23: 977–982. 44 Yu, Y.M., Fei, J.H., Zhang, Y.P., and Zheng, X.M. (2006). MCM-41 bound ruthenium complex as heterogeneous catalyst for hydrogenation I: effect of support, ligand and solvent on the catalyst performance. Chin. J. Chem. 24: 840–844.

175

176

6  Heterogenized Catalyst for the Hydrogenation of CO2 to Formic Acid or Its Derivatives

45 Zhang, Y.-P., Fei, J.-H., Yu, Y.-M., and Zheng, X.-M. (2006). Ruthenium bisphosphine catalyst on functionalized silica:novel efficient catalyst for carbon dioxide hydrogenation to formic acid. Chin. Chem. Lett. 17: 261–264. 46 Yu, Y.-M., Fei, J.-H., Zhang, Y.-P., and Zheng, X.-M. (2006). Immobilized ruthenium catalyst for carbon dioxide hydrogenation. Chin. Chem. Lett. 17: 1097–1100. 47 Zhang, Z., Xie, Y., Li, W. et al. (2008). Hydrogenation of carbon dioxide is promoted by a task-specific ionic liquid. Angew. Chem. Int. Ed. 47: 1127–1129. 48 Zhang, Z., Hu, S., Song, J. et al. (2009). Hydrogenation of CO2 to formic acid promoted by a diamine-functionalized ionic liquid. ChemSusChem 2: 234–238. 49 Xu, Z., McNamara, N.D., Neumann, G.T. et al. (2013). Catalytic hydrogenation of CO2 to formic acid with silica-tethered iridium catalysts. ChemCatChem 5: 1769–1771. 50 McNamara, N.D. and Hicks, J.C. (2014). Capture and conversion with a multifunctional polyethyleneimine-tethered iminophosphine iridium catalyst/adsorbent. ChemSusChem 7: 1114–1124. 51 Kuwahara, Y., Fujie, Y., and Yamashita, H. (2017). Poly(ethyleneimine)-tethered Ir complex catalyst immobilized in titanate nanotubes for hydrogenation of CO2 to formic acid. ChemCatChem 9: 1906–1914. 52 Mori, K., Taga, T., and Yamashita, H. (2017). Isolated single-atomic Ru Catalyst bound on a layered double hydroxide for hydrogenation of CO2 to formic acid. ACS Catal. 7: 3147–3151. 53 Lo, H.K. and Copéret, C. (2019). CO2 hydrogenation to formate with immobilized Ru-catalysts based on hybrid organo-silica mesostructured materials. ChemCatChem 11: 430–434. 54 An, B., Zeng, L., Jia, M. et al. (2017). Molecular Iridium complexes in metal–organic frameworks catalyze CO2 hydrogenation via concerted proton and hydride transfer. J. Am. Chem. Soc. 139: 17747–17750. 55 Wang, S., Hou, S., Wu, C. et al. (2019). RuCl3 anchored onto post-synthetic modification MIL-101(Cr)-NH2 as heterogeneous catalyst for hydrogenation of CO2 to formic acid. Chin. Chem. Lett. 30: 398–402. 56 Wu, C., Irshad, F., Luo, M. et al. (2019). Ruthenium complexes immobilized on an azolium based metal organic framework for highly efficient conversion of CO2 into formic acid. ChemCatChem 11: 1256–1263. 57 Yang, Z.-Z., Zhang, H., Yu, B. et al. (2015). A Tröger’s base-derived microporous organic polymer: design and applications in CO2/H2 capture and hydrogenation of CO2 to formic acid. Chem. Commun. 51: 1271–1274. 58 Park, K., Gunasekar, G.H., Prakash, N. et al. (2015). A highly efficient heterogenized iridium complex for the catalytic hydrogenation of carbon dioxide to formate. ChemSusChem 8: 3410–3413. 59 Hariyanandam, G.G., Hyun, D., Natarajan, P. et al. (2016). An effective heterogeneous Ir(III) catalyst, immobilized on a heptazine-based organic framework, for the hydrogenation of CO2 to formate. Catal. Today 265: 52–55. 60 Bavykina, A.V., Rozhko, E., Goesten, M.G. et al. (2016). Shaping covalent triazine frameworks for the hydrogenation of carbon dioxide to formic acid. ChemCatChem 8: 2217–2221.

 ­Reference

61 Gunasekar, G.H., Park, K., Ganesan, V. et al. (2017). A covalent triazine framework, functionalized with Ir/N-heterocyclic carbene sites, for the efficient hydrogenation of CO2 to formate. Chem. Mater. 29: 6740–6748. 62 Gunasekar, G., Park, K., Jeong, H. et al. (2018). Molecular Rh(III) and Ir(III) catalysts immobilized on bipyridine-based covalent triazine frameworks for the hydrogenation of CO2 to formate. Catalysts 8: 295. 63 Gunasekar, G.H., Shin, J., Jung, K.-D. et al. (2018). Design strategy toward recyclable and highly efficient heterogeneous catalysts for the hydrogenation of CO2 to formate. ACS Catal. 8: 4346–4353. 64 Kann, A., Hartmann, H., Besmehn, A. et al. (2018). Hydrogenation of CO2 to formate over ruthenium immobilized on solid molecular phosphines. ChemSusChem 11: 1857–1865. 65 Shao, X., Yang, X., Xu, J. et al. (2019). Iridium single-atom catalyst performing a quasihomogeneous hydrogenation transformation of CO2 to formate. Chem 5: 693–705. 66 Gunasekar, G.H., Jung, K.-D., and Yoon, S. (2019). Hydrogenation of CO2 to formate using a simple, recyclable, and efficient heterogeneous catalyst. Inorg. Chem. 58: 3717–3723. 67 Gunasekar, G.H. and Yoon, S. (2019). A phenanthroline-based porous organic polymer for the iridium-catalyzed hydrogenation of carbon dioxide to formate. J. Mater. Chem. A 7: 14019–14026. 68 Bao, Z., Chang, G., Xing, H. et al. (2016). Potential of microporous metal–organic frameworks for separation of hydrocarbon mixtures. Energy Environ. Sci. 9: 3612–3641. 69 Kreno, L.E., Leong, K., Farha, O.K. et al. (2012). Metal–organic framework materials as chemical sensors. Chem. Rev. 112: 1105–1125. 70 Yang, D. and Gates, B.C. (2019). Catalysis by metal organic frameworks: perspective and suggestions for future research. ACS Catal. 9: 1779–1798. 71 Rogge, S.M., Bavykina, A., Hajek, J. et al. (2017). Metal–organic and covalent organic frameworks as single-site catalysts. Chem. Soc. Rev. 46: 3134–3184. 72 Rivera-Torrente, M., Filez, M., Hardian, R. et al. (2018). Metal-organic frameworks as catalyst supports: influence of lattice disorder on metal nanoparticle formation. Chem. Eur. J. 24: 7498–7506. 73 Zhu, G. and Ren, H. (2014). Porous Organic Frameworks: Design, Synthesis and Their Advanced Applications. Springer. 74 Artz, J. (2018). Covalent triazine-based frameworks—Tailor-made catalysts and catalyst supports for molecular and nanoparticulate species. ChemCatChem 10: 1753–1771. 75 Kaur, P., Hupp, J.T., and Nguyen, S.T. (2011). Porous organic polymers in catalysis: opportunities and challenges. ACS Catal. 1: 819–835. 76 Ranade, V.V., Chaudhari, R., and Gunjal, P.R. (2011). Trickle Bed Reactors: Reactor Engineering & Applications. Amsterdam: Elsevier https://doi.org/10.1016/ B978-0-444-52738-7.10001-4.

177

179

7 Design and Architecture of Nanostructured Heterogeneous Catalysts for CO2 Hydrogenation to Formic Acid/Formate Kohsuke Mori1,2,3 and Hiromi Yamashita1,3

1 Osaka University, Graduate School of Engineering, Division of Materials and Manufacturing Science, 2-1 Yamadaoka, Suita, Osaka, 565-0871, Japan 2 JST, PRESTO, 4-1-8 Honcho, Kawaguchi, Saitama, 332-0012, Japan 3 Kyoto University, Elements Strategy Initiative for Catalysts Batteries ESICB, 1-30 Goryo-Ohara, Nishikyo-ku, Katsura, Kyoto, 615-8520, Japan

7.1 ­Introduction The use of hydrogen as an ideal clean energy carrier in place of nonrenewable fossil fuels has become of increasing interest, but the safe and economical storage and controllable release of hydrogen remain challenging [1]. A practical strategy is the storage of hydrogen in a liquid phase under ambient conditions [2]. Formic acid (HCOOH) has emerged as one of the most promising hydrogen storage compounds because it has a high gravimetric capacity for hydrogen (4.4 wt%), is relatively nontoxic, and is a nonflammable liquid under ambient conditions [3]. Additionally, the use of formic acid could allow economical CO2mediated hydrogen storage energy cycling via the regeneration of formic acid through the hydrogenation of CO2 [4]. The gas-phase hydrogenation of CO2 to produce formic acid has a positive free energy change (CO2  (g) + H2  (g) → HCOOH  (l), ΔG  =  +33 kJ mol−1) [5]. The reaction proceeds more readily in aqueous solution due to its relatively low activation energy (CO2 (aq) + H2  (aq) → HCOOH (aq), ΔG = −4 kJ mol−1). The addition of a weak base, such as a tertiary amine or alkali/alkaline earth bicarbonate, shifts the thermodynamic equilibrium to the product side (CO2(aq) + H2(aq) + B → HCO2−(aq) + BH+ [B: base], ΔG  =  −35.4 kJ mol−1). Substantial progress has been achieved in developing homogeneous transition metal complexes for formic acid production, with stimulating results obtained based on Ir and Ru metals [3d, 6]. Unfortunately, the advance of heterogeneous catalysts lags considerably in spite of their noticeable practical benefits, especially for industrial applications. To make matters worse, such heterogeneous catalysts still require the use of high catalyst loadings, organic solvents, and elevated gas pressures, and thus the development of reliable heterogeneous systems is highly desired [7]. Metal nanoparticle (NP)-based catalysts are gaining increasing attention for bridging the gap between mononuclear metal complexes and heterogeneous bulk catalysts due to their CO2 Hydrogenation Catalysis, First Edition. Edited by Yuichiro Himeda. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.

180

7  Design and Architecture of Nanostructured Heterogeneous Catalysts for CO2 Hydrogenation

existence in borderline molecular states with discrete quantum energy levels [8]. Their large surface-area-to-volume ratio allows for the effective utilization of expensive metals. The variation in size, composition, morphology, and supports significantly influences their catalytic activities. Additionally, the accurate control of geometric and electronic effects of bimetallic nanoparticles, in which the architectural configuration of two metals as random alloys, segregated or in a core–shell structure, is a key strategy in attaining superior catalytic performance compared to their monometallic counterparts [9]. The interplay of different neighboring metals creates specific new catalytically active sites, which frequently enables the fine-tuning of geometric and electronic properties originating from synergistic alloying effects [10]. Moreover, the replacement of precious noble metal nanoparticles with inexpensive metals contributes to the atomic economy [11]. Thus, the successful synthesis of bimetallic nanoparticles with controllable size, shape, and composition plays a crucial role in designing highly functionalized catalysts. However, insights into the promising design strategy as well as the additional elucidation of the catalytically active species in the supported metal nanoparticles are required. In this section, the state of the art of nanostructured heterogeneous catalysts reported for CO2 hydrogenation to formic acid/formate is summarized by classifying the catalyst types as follows: (i) unsupported bulk metal catalysts, (ii) unsupported metal nanoparticle catalysts, (iii) supported metal catalysts, and (iv) embedded single-atom catalysts. Heterogenized molecular catalysts are the most popular catalyst type reported to date for this target reaction and are summarized in Section 6.

7.2 ­Unsupported Bulk Metal Catalysts The first attempt at the synthesis of formate was performed by Bredig and Carter using a Pd sponge catalyst in 1914 under several sets of conditions [12]. Alkali/alkali metal earth (bi) carbonates were used as the CO2 source (0–30 bar of gaseous CO2 was also added in some cases) in the presence of 30–60 bar of H2. The yield of formate with Na2CO3 was negligible compared with that obtained with KHCO3, and the reaction rate increased not only with the partial pressure of H2 but also with that of CO2. In 1935, Farlow and Adkins employed Raney Ni catalyst in the presence of primary and secondary amines in an ethanol/phenol solvent [13]. For example, 57% yield of formateamine adduct was obtained after six hours at 100 °C in the presence of 2,2,6,6-tetra-Me4-OH piperidine under high pressure (20–40 MPa). Later, Takahashi et al. investigated the reduction of CO2 under hydrothermal conditions (300 °C) by using a mixture of Fe and Ni powders as a catalyst, in which H2O acted as a hydrogen source and CO2 was converted to formic acid accompanied by the formation of H2 gas [14]. Unfortunately, the activity in this catalytic system was quite low (turnover number [TON] = 0.02 at two hours). Fachinetti and coworkers screened the Group 8-11 transition metal blacks, such as Fe, Co, Ni, Cu, Ru, Rh, Pd, Ag, Ir, Pt, and Au, for the reverse reaction, i.e. decomposition of formic acid-amine adducts to CO2 and H2, and found that only Au metal black, which was obtained by reducing a freshly prepared Au(OH)3/H2O slurry with diluted cold alkaline formaldehyde, exhibited satisfactory activity. This result suggested that Au metal black had the potential to boost the formic acid/formate synthesis. As expected, Au metal black was

7.3 ­Unsupported Metal Nanoparticle Catalyst

proven to be active and converted CO2 to formate at 40 °C in triethylamine (NEt3) solution [15]. However, the catalyst was quickly deactivated due to the aggregation of metal particles, as evidenced by the comparison of scanning electron microscope (SEM) images before and after use. These investigations suggest that the low catalytic activity of the unsupported bulk metals can be improved by decreasing the size of the particles, which leads to an increase in the number of active sites on the surface of the catalysts. Recently, unsupported nanoporous nickel (NiNPore) material has attracted immense attention as a catalyst due to its large surface-to-volume ratio, nontoxicity, simple work-up procedure, and high recyclability. NiNPore can be prepared by the selective leaching of Mn from an alloy foil of Ni30Mn70 under electrochemical conditions. Wang et al. reported the first hydrogenation of carbonate (HCO3− and CO32−) to formate using NiNPore [16]. The highest yield of 92.1% was attained with KHCO3 at 200 °C under 6 MPa of H2. Notably, the catalyst was easily recovered and could be recycled at least five times without leaching or loss of activity.

7.3 ­Unsupported Metal Nanoparticle Catalysts The use of bulk metals in catalytic reactions is economically inefficient, since the number of available metal atoms is extremely small, and decreases further under catalytic reaction conditions [17]. Nanoparticles and/or nanocrystals (1–100 nm) are of particular interest in the field of catalysis, because these materials exhibit unique properties on the nanoscale level, and bridge the gap between mononuclear metal complexes and heterogeneous bulk catalysts; such systems are often referred to as “quasihomogeneous systems” [8]. These materials often show exceptional behavior by virtue of their large surface-to-volume ratios, which result in low-coordinated atoms located in defects (terraces, edges, kinks, and vacancies) and higher catalytic reactivity.

7.3.1  Metal Nanoparticles Without Stabilizers Ma et  al. synthesized a nano-Ni catalyst by using a liquid-phase reducing method. This nano-Ni was shown to be a nanochain structure with an average size of approximately 15 nm, and its Brunauer−Emmett−Teller (BET) surface area was determined to be 104.4 m2 g−1, which can provide a sufficient number of active sites. It was demonstrated that nano-Ni was active for CO2 conversion to formic acid at ambient temperature (35 °C) under continuous H2 aeration conditions [18]. The nano-Ni catalyst displayed superior adsorption capacity for both H2 and CO2 compared to that of the commercially available nano-Ni due to the high surface area and the NiO layer formed on the surface of the Ni nanoparticles, which facilitate the selective hydrogenation of CO2 to formate. Unfortunately, this catalytic system still suffers from poor durability. Recently, the reaction with supercritical CO2 was performed by Umegaki et  al. in the presence of trimethylamine using Ru nanoparticles [19]. Metallic Ru nanoparticles were prepared in a methanol solution under solvothermal conditions, which consisted of primary particles with a size of 3–5 nm and secondary particles with a size of 200–240 nm. A maximum TON of 6351 was attained after three hours at 85 °C. It should be noted that the

181

182

7  Design and Architecture of Nanostructured Heterogeneous Catalysts for CO2 Hydrogenation

activity was drastically improved with the addition of an appropriate amount of water to maintain the ruthenium species in the metallic state.

7.3.2  Metal Nanoparticles Stabilized by Ionic Liquids Ionic liquids (ILs) are typically salts that contain at least one organic cation and an organic or inorganic anion, and have melting points below 100 °C. They possess pre-organized structures through hydrogen bonds that induce structural directionality, as opposed to classical salts. Metal nanoparticles dispersed in these fluids are stable and often show high activity for various catalytic reactions [20]. Srivastava reported the in situ generation of Ru nanoparticles by hydrogen reduction of bis(2-methylallyl)(1,5-cyclooctadiene) ruthenium(II) [Ru(COD)(2-methylallyl)2] (COD: 1,5-cyclooctadiene) in the presence of a functionalized IL [DAMI][NTf2] (DAMI: 1,3-di(N,N-dimethylaminoethyl)-2-methylimidazolium; NTf2: bis(trifluoromethylsulfonyl) imide) [21]. The transmission electron microscopy (TEM) image showed small nanoparticles with a narrow size distribution whose average practical size was 2.1(±0.5) nm, and with no significant agglomeration. A TON of 1500 and a turnover frequency (TOF) of 376 h−1 at 100 °C were attained under a total pressure of 40 MPa after five hours in the CO2 hydrogenation. The use of ILs as ligands and solvents also offers easy catalyst isolation. Moreover, [DAMI][NTf2] IL–stabilized Ru nanoparticles can be recycled up to seven times with only a slight loss of their catalytic activity, which was mainly due to the agglomeration of Ru nanoparticles. In the report of Dupont and coworkers, CO2 was selectively hydrogenated into formic acid or hydrocarbons (HCs) in the presence of ferromagnetic RuFe nanoparticles by the proper choice of IL anions (Figure 7.1) [22]. The bimetallic RuFe nanoparticles were prepared by reduction/decomposition of equimolar amounts of [Fe(CO)5] and [Ru(Meallyl)2(COD)] under 18 bar of H2 atmosphere at 150 °C in BMI·NTf2 IL solution (BMI: 1-butyl-3-methyl-1H-imidazol-3-ium). The TEM image showed the formation of irregularly shaped nanoparticles with a mean diameter of 1.7 ± 0.3 nm along with large NTf2–

IL

IL

N

H CO2 CO CO H

N

(CH2))n

IL

IL

CO2 + H2

IL IL

CO2 H H HCO3 CO2H

OAc–

IL

IL IL

N

HCOOH

N

Figure 7.1  Selective hydrogenation of CO2 by RuFe nanoparticles in different ionic liquids. Source: From Qadir et al. [22], © 2018 American Chemical Society. Reprinted with permission of American Chemical Society.

7.3 ­Unsupported Metal Nanoparticle Catalyst

agglomerates between 10 and 25 nm. In a hydrophobic BMI·NTf2 IL containing nonbasic anions, these RuFe nanoparticles formed heavy HCs up to C21 with 12% conversion at 150 °C using a gas mixture (H2/CO2  =  4/1) under 8.5 bar via the reverse water-gas shift (RWGS) reaction and Fischer–Tropsch process. On the other hand, formic acid formation occurred in a hydrophilic BMI·OAc IL (OAc: acetate) containing basic anions at 60 °C using a gas mixture (H2/CO2  =  1/2) under 30 bar, in which a high TON of 400 and a TOF of 23.52 h−1 were attained. The addition of small amounts of water had a substantial effect in enhancing the activity for formic acid formation for the following two reasons: acceleration of the formation of bicarbonate species by the combination of CO2 with water and IL and providing a suitable solvent system with dimethyl sulfoxide (DMSO), which plays a crucial role in stabilizing the formic acid through H bonding.

7.3.3  Metal Nanoparticles Stabilized by Reverse Micelles The catalytic properties are determined not only by the active sites but also by the microenvironment [23]. Various platforms, such as dendrimers, star polymers, and organic and metal–organic nanocapsules, have been used to control the catalytic properties of attached or encapsulated catalysts through a rational microenvironmental engineering strategy. Zhao and coworkers reported that Pd-containing interfacially cross-linked reverse micelles (Pd@ICRMs) could be used for the hydrogenation of bicarbonate and CO2 to formate (Figure 7.2) [24]. The size of the ICRM was 4–5 nm and the Pd nanoparticles inside were ~2 nm. Interestingly, the microenvironment of the Pd nanoparticles could be strictly tuned by the nature of the headgroup of the ICRMs, which greatly influenced the catalytic performance. Quaternary ammonium-based surfactants make it possible to exchange the bromide with bicarbonate and bring it closer to the catalytic center, which worked well for the hydrogenation of bicarbonate. On the other hand, tertiary amine-based surfactants worked better in CO2 hydrogenation because organic amines are commonly used to capture gaseous CO2 reversibly through carbamate formation, which brings CO2 into proximity with

C12H25O C12H25O

Pd

+ N Br–

C12H25O

L

L L

L

L

Pd

L L L

HCO3–/H2

C12H25O C12H25O

L

L H L Pd H H L L L

+ N HCO – 3H

C12H25O

Pd@ICRM(1) or Pd@ICRM(3) Pd@ICRM O O O

versus – +N Br

1

2

O O O

Br–

+ N

3

HN

C12H25O

– HN O Br + O HN O O HN O H N N H

O O

N

C12H25O C12H25O

O L L

L

L Pd L NH L L L L O

CO2/H2

C12H25O C12H25O C12H25O

+

HN O O L H L – H O Pd H NH H L L L O

N

Pd@ICRM(2)

Figure 7.2  Microenvironmental engineering of Pd nanoparticle catalysts using cross-linkable surfactants for the hydrogenation of CO2 and bicarbonate. Source: From Lee et al. [24], © 2017 American Chemical Society. Reprinted with permission of American Chemical Society.

183

184

7  Design and Architecture of Nanostructured Heterogeneous Catalysts for CO2 Hydrogenation

the Pd nanoparticles in the nanospace of the hydrophilic core of the cross-linked reverse micelles. The TON in the CO2 hydrogenation was 204 at 40 °C under a total pressure of 80 bar after 20 hours, which is an order of magnitude higher than that for commercially available Pd/C.

7.4 ­Supported Metal Nanoparticle Catalysts Nanoparticle-based catalysts are key components of catalysts, but unsupported metal nanoparticles often suffer deactivation caused by sintering and/or particle agglomeration during the catalytic reactions. Moreover, their application in liquid suspensions is limited due to difficulties in product separation and catalyst recycling. To overcome the above drawbacks, supported metal nanoparticles with tunable characteristics are the keystone of ongoing investigations toward the development of high-performance catalysts, since it is well known that they show several advantages from the practical viewpoint, such as easy separation, recycling, and mild operating conditions, making them preferable for industrial applications [3b]. A number of support materials with diverse nature have been explored extensively to disperse active metal nanoparticles. These supports endow the nanoparticles with high stability and dispersity, alteration of electronic structure of the nanoparticles, possible metal-support interactions, and creation of unique active sites at the perimeter, which ultimately play a crucial role in their final catalytic behavior [3a, 25]. In spite of the obvious practical advantages of such materials, the number of reports in this category is still limited. In this section, the state of the art in the design and architecture of supported nanoparticle catalysts as well as mechanistic investigations are summarized.

7.4.1  Metal Nanoparticles Supported on Carbon-Based Materials Carbon is unquestionably the most widely studied support material among the vast variety of materials hitherto explored. Generally speaking, the strengths of carbon materials can be related to their surface resistance to both basic and acidic media, thermal stability, tunable pore structure, multiple macroscopic shapes, controllable polarity, and cost effectiveness. Such exceptional properties provide unlimited possibilities toward the design of efficient heterogeneous catalysts by deposition of active metal nanoparticles [26]. Currently, graphene and its derivatives, such as graphene oxide (GO) and reduced graphene oxide (rGO), are widely utilized as catalyst supports and even as metal-free catalysts due to their exceptional physical, chemical, electronic, and mechanical properties [27]. In an early study, Stalder et al. examined the catalytic efficiencies of a series of commercially available supported catalysts (Pd/C, Pd/BaSO4, and Pd/γ-Al2O3) in the hydrogenation of NaHCO3 to formate at 0.1-MPa H2 pressure [28]. All Pd catalysts displayed higher activity compared with the unsupported Pd black, demonstrating the importance of the dispersion onto the supports. Pd/C was found to be the most active (TON = 115) among those investigated, more than two times higher than that of Pd/γ-Al2O3. Sasson and coworkers studied the Pd/C-catalyzed hydrogenation of sodium bicarbonate by using H2O as a hydrogen source, where kinetic and mechanistic features were also examined [29]. Although the

7.4 ­Supported Metal Nanoparticle Catalyst

efficiency of the hydrogenation of sodium bicarbonate was rather low, they were the first to propose the idea of a chemical carrier for hydrogen storage using sodium formate salts. Later, Su, Lin, et al. performed a systematic investigation into the hydrogenation of various bicarbonate salts to formate by using activated carbon (AC) with a porous structure as a support [30]. The Pd/C attained a TON of 1769 after 20 hours under a H2 pressure of 2.75 MPa after 15 hours, while the activities of the Ru, Rh, Pt, and Ni metals supported on AC were quite low. AC showed superior performance to other supports such as Al2O3, CaCO3, and BaSO4 owing to its hydrophobic nature, in which H2 could be locally enriched in the carbon channels or on the surface of the carbon and thus overcome the low solubility of H2 in water. The higher dispersion of Pd nanoparticles on AC compared with the others is another explanation for its efficient activity. They also emphasized the importance of the protonation of carbonate ions in the reactivity. The order of TON corresponds to the order of bicarbonate concentration in aqueous solution (NH4HCO3 [0.92 m] > KHCO3 [0.89 m] > NaHCO3 [0.61 m]). Pd nanoparticles supported on r-GO were employed by Cao and coworkers [31]. The highest TON of 7088 was obtained in the hydrogenation of KHCO3 at 100 °C for 32 hours under 40 bar of H2 pressure. Screening of the Pd loading (1, 2, and 5 wt%) revealed that the lowest Pd loading was superior to those of the higher ones in spite of their similar particle sizes. They explained this difference in activity from the viewpoint of the degree of lattice expansion in the Pd crystals. The lattice strain (ε = 2.32) of 1 wt% Pd/r-GO was substantially larger than 0.31 of 5 wt% Pd/r-GO, which caused an isomorphic effect on the electronic structure of Pd and ultimately resulted in the high activity [32]. This study also demonstrates that the r-GO support can facilitate the creation of highly strained Pd nanoparticles. Nguyen, Lee, et al. reported the synthesis and characterization of PdNi alloy nanoparticles on a carbon nanotube-graphene (CNT-GR) support, and its first application as a heterogeneous catalyst for direct CO2 hydrogenation to formic acid in pure water without a base additive [33]. The yield of formic acid was 1.92 mmol with a TON of 6.4 and a TOF of 1.2 × 10−4 s−1 at 40 °C under 50 bar (H2/CO2 = 1). Alloying Pd with Ni gave a significant enhancement in catalytic activity compared to the monometallic Pd and Ni catalysts, suggesting a synergistic alloying effect. The CNT-GR composites can suppress the stacking of GR and the bundling of CNT, and thus improve the dispersion of Pd–Ni alloy particles, which is responsible for good stability under the reaction conditions. They also proposed a reaction pathway for the direct synthesis of formic acid (Figure 7.3). By alloying, an electron transfer from Ni to Pd atoms occurs to produce electron-rich Pd and electron-deficient Ni. Pd atoms are active for the dissociative H2 adsorption, while CO2 is adsorbed on the Ni atoms through its O atoms. Successive reactions between H on Pd and adsorbed CO2 lead to the formation of adsorbed HCOOH on Ni atoms. This mechanism clearly highlights the necessity for bimetallic surfaces for the selective formation of formic acid by the hydrogenation of CO2.

7.4.2  Metal Nanoparticles Supported on Nitrogen-Doped Carbon Nitrogen atoms are the most successfully investigated doping heteroatom for carbon materials due to their similar electronegativity and electron affinity to carbon, and such materials have recently become promising for both metal-free catalysis and as a support for

185

186

7  Design and Architecture of Nanostructured Heterogeneous Catalysts for CO2 Hydrogenation

δ+ C

δ– O

δ– δ+ HH

δ– Pd

δ+ Ni

δ– O H

–

δ+ Ni

δ+ C O H O

–

H

C O O

Pd

Ni

H Ni

Pd Ni

Ni H

H

δ– Pd

C O O H

δ+ Ni

Ni

H H δ– Pd

C O O

Ni δ+ Ni

H H

H C O O

C O O

–

Ni

Pd Ni

Ni

Pd Ni

Figure 7.3  A possible reaction pathway for CO2 hydrogenation over a PdNi bimetallic surface. Source: From Nguyen et al. [33], © 2015 Royal Society of Chemistry. Reprinted with permission of Royal Society of Chemistry.

metal-based catalysts [34]. The importance of nitrogen-doped carbons in catalysts can be traced back to 1926, when Rideal and Wright revealed the beneficial effect of nitrogen atoms on the surface of charcoal in the oxalic acid oxidation [35]. Their use has been conventionally linked to electrochemical applications, but their versatile applications as heterogeneous catalysts as well as photocatalysts have been increasing exponentially during the last decade [36]. Among various nitrogen-doped carbon materials, polymeric structured graphitic carbon nitride (g-C3N4) has received increasing attention owing to its advantageous features, including high thermal and chemical stability and easy preparation from nitrogen-abundant cheap organic precursors, such as urea, cyanamide, and melamine [37]. g-C3N4 has also attracted immense interest as a semiconducting photocatalyst due to its appropriate band structure with an optical band gap of 2.7 eV [38]. The prominent disadvantage of original g-C3N4 is its low surface area, typically below 10 m2 g−1. To tailor the structural properties and interlayer interactions of g-C3N4, nanostructural designs, such as mesoporous structures, nanosheets, nanotubes, and nanospheres, were attempted [39]. Lee, Yoon, et al. developed Pd-supported mesoporous g-C3N4 (mpg-C3N4) by the simple impregnation of Pd(NO3)2·2H2O, followed by reduction with H2 [40]. TEM analysis revealed highly dispersed Pd nanoparticles with an average size of 1.7 nm on mpg-C3N4 without any agglomeration. The formation of such uniform Pd nanoparticles likely originated from the preferential intermolecular interactions of the Pd2+ precursors with nitrogen atoms located at mpg-C3N4. This hypothesis was explored by density functional theory (DFT) calculations of three melm units (circled in Figure 7.4a). The optimized structure possessed three different types of nitrogen sites (N, N′, and N″) and showed a corrupted structure with a dihedral angle of N–C1–N′–C2 of c. 22° (Figure 7.4b). Natural bond orbital charge analyses found that the sp2-hybridized N and N″ sites (δ: −0.517, −0.525, and −0.529) were more

7.4 ­Supported Metal Nanoparticle Catalyst

negatively charged than that of N′ (δ: −0.422 and −0.429). Molecular orbital analyses revealed that highest occupied molecular orbitals (HOMOs) localized over the sp2-hybridized N and N″ sites were located inside the three melm units, and had more diffused orbitals than the N′ sites (Figure 7.4c). This suggests the easy interaction of Pd2+ ion precursors

N N N N N

N N

N

N

N

N

N

N

N

N N

N

N

N

N

N N

N N

N

N N

N

N

N

N

N

N

N

N

N N

N

N

N

N

N N

N N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N N

N N

N

N

N N

N

N

N N

N

N N

N

N

N

N

N

N N

N N

N

N

N N

N N

N

N

N

N

N

N N

N N

N N

N

N N

N

N

N

N N

N

N N

N N

N

N N

N

N

N

N

N N

N N

N

N

N N

N

N

N

N N

N

N

N

N N

N N

N N

N

N N

N

N N

N

N

N

N

N

N

N N

N N

N

N

N N

N N

N

N

N

N N

N

N

N

N N

N

N N

N N

N

N N

N N

N

N

N N

N N

N

N N

N

N

N

N

N N

N

N

N

N N

N

N

N

N

N

N N

N

N

N

N N

N

N

N N

N

N

N

N

N

N N

N N

N

N

N N N

N

N N

N N

N N

N N

N

N

N

N

N

N

N N

N

N

N

N

N

N N

N

N N

(a)

C2 C1

N′

N′′ N N

(b)

N′′ N′ N N

(c)

Figure 7.4  (a) Molecular structure of graphitic carbon nitride (circled part is the model unit for mpg-C3N4), (b) DFT-optimized structure using a truncated C3N4 unit, and (c) the calculated HOMOs. Source: From Lee et al. [40], © 2014 Royal Society of Chemistry. Reprinted with permission of Royal Society of Chemistry.

187

188

7  Design and Architecture of Nanostructured Heterogeneous Catalysts for CO2 Hydrogenation

with N or N″ sites. The possible interaction between Pd and N atoms was also validated by N K-edge X-ray absorption near-edge spectra (XANES). Moreover, the formed Pd NPs could be stabilized with nitrogen atoms at C3N4, which was electronically promoted by density transfer from the nitrogen atoms. Pd/mpg-C3N4 catalyzed the CO2 hydrogenation to formate in the presence of NEt3. The present system was significantly influenced by the relative ratio of CO2 and H2. Under the optimized condition (CO2: 13 bar; H2: 27 bar), 4.74 mmol of formate adduct was obtained at 150 °C, which is more than two times higher than that of commercial Pd/C. The abundant lone-pair electrons at the nitrogen atoms likely acted as basic sites that readily adsorbed CO2 during the reaction. Park, Lee, et al. reported the direct formic acid synthesis under neutral conditions with Pd supported g-C3N4 in a batch reactor at 40 °C under 50 bar of H2/CO2 = 1 [41]. The average particle size was estimated to be c. 4.4 nm by TEM. The activity of Pd/g-C3N4 was 12 times higher than that of Pd/CNT with a similar Pd particle size. In the CO2-temperature programmed desorption (TPD) analysis, the desorbed CO2 amount of Pd/g-C3N4 was 1.18 μmol g−1, which was substantially higher than that of Pd/CNT (0.03 μmol g−1), suggesting that g-C3N4 is a much more effective support material for the improved activity of HCOOH synthesis. Wang et  al. compared the activity of Pd catalysts supported on N-free and N-doped mesoporous carbon [42] with mean diameters of 3.1 and 2.4 nm for the conversion of KHCO3 at 80 °C under 6 MPa of H2. They found that the N-doping significantly influenced the activity, affording 70% conversion of bicarbonate to formate within two hours compared with only 45% for the N-free catalyst. Koh et al. also employed ultra-small Pd nanoparticles (c. 1.6 nm) supported on nanoporous N-doped carbon, and confirmed the positive effect of the N-doping for the hydrogenation of NaHCO3 at 40 °C under 40 bar of H2 [43]. Several reaction parameters, including bases, reaction temperature, hydrogen pressure, and concentration of bicarbonate in the hydrogenation reaction, were investigated by Shao et al. using Pd/g-C3N4 with a mean particle size of 2.4 nm [44]. Among the bases examined, the efficiency of NH4HCO3 was higher than that of KHCO3 and NaHCO3. This result is in accordance with the order of the bicarbonate ion content in the aqueous solution (NH4HCO3 [92%] > KHCO3 [89%] > NaH CO3 [61%]). A similar tendency was also confirmed by Su, Lin, et al. [30]. They proposed a possible reaction pathway at the metal/support interface, as shown in Figure 7.5. Initially, the HCO3− easily interacts with g-C3N4 via the formation of O─H⋯N hydrogen bonds. Subsequently, H2 activated over Pd NPs attacks the positively polarized carbon in HCO3− with the simultaneous formation of the formate intermediate. The electron density transfer from the semiconductor C3N4 to Pd nanoparticles in the heterojunction also plays an important role in the formation of electronically enriched metallic Pd. Nitrogen-rich graphitic carbon-stabilized cobalt nanoparticles were also reported by Patel, Khan, et al. as an effective heterogeneous catalyst for the hydrogenation of CO2 to formate [45]. The catalyst was prepared by the pyrolysis of impregnated Co(OAc)2·4H2O precursor on cucurbit[6]uril (CB[6]). Cucurbit[n]urils are macrocyclic molecules made of glycoluril (═C4H2N4O2═) monomers linked by methylene bridges (─CH2─), where n is the number of glycoluril units. During the pyrolysis process, the CB[6] framework was converted into the nitrogen-rich graphitic carbon support and cobalt nanoparticles were simultaneously formed. The synthesized catalyst showed a TON of 82 265 in 24 hours with 1-M KOH solution at 120 °C under 62 bar pressure (CO2/H2 = 1).

7.4 ­Supported Metal Nanoparticle Catalyst O –O

HO

H

H

C O e

O–

–O

C Hx Pd

mpg-C3N4

HO

O

Hx Pd

mpg-C3N4

H

C HO

+ H2O

C

O

O

–

Hx Pd mpg-C3N4

Figure 7.5  Possible reaction mechanism for bicarbonate hydrogenation over Pd/C3N4. Source: From Shao et al. [44], © 2016 American Institute of Chemical Engineers. Reprinted with permission of John Wiley and Sons.

Recently, further modification of the g-C3N4 was attempted with the aim of improving the potential ability of this type of support material. Pd nanoclusters supported on Schiff base modifying graphitic carbon nitride (Pd/u-CN100), which was synthesized using urea and terephthalaldehyde (TPAL) as precursors, was reported by Zhou et al. [46]. The resulting sheet materials exhibited a TOF of 98.9 h−1 at 100 °C under 7 MPa of pressure (CO2/H2 = 1) in the mixture of ethanol/NEt3, which is higher than that of the unmodified ones despite the formation of slightly larger Pd nanoparticles. The grafting of a suitable amount of TPAL resulted in sufficient Schiff base groups and various nitrogen species, which allows for a large promotion in material surface area and volume, as well as for the electronic modification of Pd nanoparticles. More recently, Mondelli et al. reported the tuning of edge-defect sites of g-C3N4 by both bottom-up and top-down strategies with the aim of increasing the desired basic sites [47]. The synthetic approaches are summarized in Figure  7.6a. Nanosheet materials (ECN-x, x = duration of the thermal treatment in hours) with interparticle mesoporosity were synthesized from bulk carbon nitride (BCN) by thermally driven exfoliation. C-enriched BCN was prepared by adding variable amounts of 2,4,6-triaminopyridine (TAP) to the reaction mixture (BCN-x, x = TAP/dicyandiamide mass ratio), which was further subjected to thermal delamination to afford C-enriched ECN. By adding SiO2 as a hard template for the direct polymerization of cyanamide followed by removal by acid etching, mesoporous carbon nitride (MCN-x, x = SiO2/cyanamide mass ratio) was obtained. The activity was evaluated in the base-free formic acid production by CO2 hydrogenation at 40 °C under 5 MPa of pressure (CO2/H2 = 1). Good correlation between the metal time yield (MTY) of formic acid and the concentration of basic sites (CB) determined by CO2-TPD analysis was found (Figure 7.6b). These results clearly confirmed that the basicity of the support is most relevant for the reaction.

7.4.3  Metal Nanoparticles Supported on Al2O3 It has been suggested that ruthenium hydride species, which are generated from [Ru− OH2]2+ by subsequent hydrogenation, are a potentially active species in the hydrogenation of CO2 to formic acid with ruthenium complexes under acidic conditions in water [48]. Additionally, a theoretical study indicated that the introduction of hydroxyl groups with

189

7  Design and Architecture of Nanostructured Heterogeneous Catalysts for CO2 Hydrogenation

Dicyandiamide

Cyanamide

Direct polymerization + BCN

Hard templating C-enriched BCN Thermal delamination

+ ECN

(a)

C-enriched ECN

MCN

800 2Pd/ECN-1h MTY (µmolFA S–1 molpd–1)

190

600

3Pd/ECN-1h 1Pd/ECN-1h

0.5Pd/ECN-1h

400

0.5Pd/BCN

200

0.5Pd/ECN-1h-0.02 0.5Pd/BCN-0.02

0.5Pd/MCN-0.5

0 (b)

0

5

10 CB (µmol m–2)

15

20

Figure 7.6  (a) Synthetic approaches to tailor the basic properties of g-C3N4. BCN, MCN, and ECN stand for bulk, mesoporous, and exfoliated carbon nitride, respectively. (b) Dependence of the metal time yield of formic acid upon CO2 hydrogenation over carbon nitride–supported Pd catalysts on the concentration of basic sites. Source: From Mondelli et al. [47], © 2018 John Wiley and Sons. Reprinted with permission of John Wiley and Sons.

strong donation ability would improve the catalytic efficiency in the hydrogenation of CO2 to formic acid in KOH solution [49]. Motivated by the above results, Hao et al. hypothesized that the hydroxyl group on the surface of the support plays an important role in enhancing the activity [50]. Tervo verify this assumption, the authors synthesized Ru hydroxide catalysts on the surface of support materials, such as γ-Al2O3, AC, and MgO, with different hydroxyl group contents; γ-Al2O3 possesses abundant hydroxyl groups, while the number of hydroxyl groups is limited in AC and MgO does not have any hydroxyl groups. The appearance of a peak at 1856 cm−1, attributed to the interaction between the hydroxyl group and the Ru component, was observed in the Fourier transform infrared spectroscopy (FT-IR) spectrum of Ru/γ-Al2O3, confirming the formation of Ru-OH species. The catalytic activity in the CO2 hydrogenation was consistent with the amount of hydroxyl groups. Ru/γ-Al2O3 displayed a TON of 91 at 80 °C under 13.5 MPa of pressure (CO2/H2 = 1.7) in

7.4 ­Supported Metal Nanoparticle Catalyst

the mixture of ethanol/NEt3, whereas the activity of Ru/activated was moderate (TON = 10) and Ru/MgO was inactive under the same reaction conditions. It was also reported that the highly dispersed Ru hydroxide was the active species, while crystalline RuO2 species, which were formed at high Ru content or pH conditions of the solution during catalyst preparation, significantly retarded the catalytic reaction. Based on these observations, they proposed that the Ru hydroxide is initially hydrogenated to form the hydride species, which undergoes the insertion of CO2 to form the Ru-formate intermediate. Finally, substitution with H2O regenerates the Ru hydroxide species accompanied by the formation of formic acid. Later, Liu et al. performed the deposition of Ru hydroxide species on γ-Al2O3 nanorods with high surface area as well as plenty of hydroxyl groups for the enhancement of activity  [51]. Expectedly, Ru/γ-Al2O3 nanorods exhibited higher catalytic activity compared with the previous Ru/γ-Al2O3. Pidko and coworkers studied the formation of formate using Au-supported catalysts with different supports in NEt3/EtOH solution at 70 °C under 40 bar of pressure (CO2/H2 = 1) [52]. They reported that Au/Al2O3 showed the highest TON of 215 after 20 hours compared with Au catalysts supported on TiO2, ZnO, CeO2, hydrotalcite (HT), and CuCe2O4. Since the colloidal Au nanoparticles showed no activity under the same reaction conditions, this clearly suggests that metal-support interactions play a crucial role in the Au-catalyzed hydrogenation of CO2. X-ray photoelectron spectroscopy (XPS) analysis showed the Au 4f7/2 binding energy peak at 83.3 eV, which is lower than that of bulk metallic Au0 (84.0 eV), suggesting the presence of a metallic state. The cyanide-leaching test also verified the importance of the metallic Au0 as the catalytically active species in the CO2 hydrogenation. In the FT-IR spectrum of Au/Al2O3 treated with N,N-dimethylformamide (DMF)/NEt3 in the absence of H2 and CO2, several bands attributed to DMF (νC═O 1664 cm−1), NEt3 (νC─H 2979, 2935 cm−1), and carbonates on γ-Al2O3 (1576, 1463, and 1387 cm−1) were observed (Figure 7.7A(a)). After the reaction, an additional sharp peak was observed at 1619 cm−1 for the spent catalysts, which was assignable to the surface-bound formate species (νCOO) (Figure  7.7A(b)). The temperature-dependent TOF and dynamic equilibrium modeling revealed that the apparent activation energy using Au/Al2O3 has nearly a zero-order dependence, which is presumably associated with the compensating effects of the adsorption and desorption of reactants. Upon consideration of the above experimental results, a reasonable mechanism for CO2 hydrogenation over Au/Al2O3 was proposed (Figure 7.7B). The reaction is initiated by a heterolytic dissociation of H2 at the Au/support interface accompanied by the desorption of the solvent or bases, affording surface hydroxyl and metal hydride species. Next, CO2 adsorbs to produce the surface bicarbonate species, then attack of the bicarbonate by the Au-hydride gives an adsorbed formate species. Surface formates and bicarbonates observed by FT-IR are the key intermediates in the present catalytic cycle.

7.4.4  Metal Nanoparticles Supported on TiO2 Titanium dioxide (TiO2) is one of the most conventionally used support materials for catalytically active metal nanoparticles, which is surely due to its good mechanical properties and low price. In particular, the strong metal−support interaction (SMSI) is an important

191

1619

7  Design and Architecture of Nanostructured Heterogeneous Catalysts for CO2 Hydrogenation

1576 1463 1387

2935

1664

2979

192

(b)

(a)

(A)

3500

3000

2500

2000

1500

Wavenumber (cm–1) [BH]+[HCOO]–

CO2 + H2 + B: Formate desorption/adsorption

Hydrogen dissociation/recombination L

[BH]+[HCOO]–

Au

0

O

H

H L

B: O

Au0

O H O

Hydride attack/abstraction (B)

H

H H O

Au0

Au0

H O H C O O

O

L

C

O

= DMF or NEt3

Figure 7.7  (A) FT-IR spectra of Au/Al2O3 catalyst treated with (a) DMF/NEt3 in the absence of H2 and CO2 and (b) DMF/NEt3 in the presence of H2 and CO2. (B) Proposed mechanism for the CO2 hydrogenation over Au/Al2O3 catalyst. Source: From Filonenko et al. [52], © 2016 Elsevier. Reprinted with permission of Elsevier.

characteristic, which was first described by Tauster et al. to explain the unusual chemisorption of H2 and CO and catalytic properties of TiO2-supported platinum-group metal nanoparticles after high-temperature reduction under H2 atmosphere [53]. It is well accepted that the interactions between metals and supports, i.e. metal−support interactions, are of great importance in determining their catalytic activities and selectivities [54].

7.4 ­Supported Metal Nanoparticle Catalyst

As mentioned in Section  7.2, Fachinetti and coworkers demonstrated that Au metal black was active for the conversion of CO2 to formate at 40 °C in triethylamine (NEt3) solution after the screening of the Group 8-11 transition metal blacks [15]. Simultaneously, they also reported that the aggregation of metal particles quickly deactivated the catalyst during the reaction. In an effort to overcome this drawback, commercial TiO2-supported Au nanoparticles (AUROlite, Au: 1 wt%, Au particle size c. 2–3 nm) were employed for this conversion at 40 °C under 180 bar of pressure (CO2/H2 = 1). When the gas pressure dropped to 130 bar after the gas absorption, the pressure was restored to 180 bar. AUROlite showed high stability for 37 days, and a total of 1.326 kg of formate-amine adduct was obtained. No other organic byproducts were detected by nuclear magnetic resonance (NMR) analysis, but the accumulation of 9% of CO was observed, which was presumably formed by the Au-catalyzed RWGS (CO2 + H2O → CO + H2O) reaction. PdAg alloy nanoparticles supported on TiO2 were also reported as an efficient selective hydrogenation catalyst of CO2 at 100 °C under relatively low pressure conditions (2.0 MPa, CO2/H2  =  1) [55]. Other PdAg catalysts supported on different supports, such as MgO, CeO2, and Al2O3, and layered double hydroxide (LDH), were found to be less active. Interestingly, PdAg/TiO2 exhibited considerably higher activity compared with the corresponding PdCu, PdZn, PdAu, and PdNi catalysts supported on TiO2. The catalytic activity of PdAg/TiO2 was significantly affected by the composition of the PdAg alloy nanoparticles. A volcano-type variation in activity with a maximum at 70% Ag proportion was observed, which suggests not only the formation of a uniform PdAg alloy structure on TiO2, but also a synergistic alloying effect between Pd and Ag. The co-reduction of Pd and Ag precursors conventionally affords random PdAg alloy nanoparticles (PdAg/TiO2) due to their complete solid solubility and the similar reduction potentials of Pd and Ag ions. The precise tuning of the surface composition of PdAg nanoparticles was performed to investigate the effect of surface-exposed active Pd atoms in alloy NPs. By applying a surface engineering approach via the successive reduction of metal precursors, Pd@Ag/TiO2 with a PdcoreAgshell structure and Ag@Pd/TiO2 with an AgcorePdshell structure were synthesized. In the XPS spectra, the Pd 3d peaks of all PdAg samples were shifted to lower binding energies than those of Pd/TiO2, and this shift was decreased in the order of Pd@Ag/TiO2 > PdAg/TiO2 > Ag@Pd/TiO2. Thus, the Pd atoms in the PdAg NPs were clearly electron enriched by the charge transfer from Ag atoms owing to the net difference in ionization potential between the two metals. A similar tendency in the electronic state of the Pd species was observed in FT-IR experiments using CO as a probe molecule. Figure 7.8a shows a comparison of catalytic activities in the CO2 hydrogenation. Pd@Ag/ TiO2 exhibited an elevated TON (2496) based on the total quantity of Pd employed despite the low density of surface-exposed Pd atoms. As expected, a maximum TON value of 14 839 was obtained from Pd@Ag/TiO2 based on the quantity of surface Pd atoms, which was determined by pulsed CO adsorption measurements. This TON value is more than 10 times higher than that for the monometallic Pd/TiO2. Moreover, a good correlation between the TON based on surface Pd atoms and the Pd 3d5/2-binding energy determined by XPS analysis was observed (Figure 7.8b). In this study, the enhancement of activity by alloying was well evidenced based on DFT calculations, employing Pd22, Pd11Ag11, and Pd6Ag16 clusters as models for monometallic Pd and alloy nanoparticles (Figure 7.9). The CO2 hydrogenation over Pd22 is initiated by the

193

20 000

Based on surface exposed Pd Based on all employed Pd

15 000

14 839

10 000 7126

6072

5000

(a)

0

367

1213

Pd/TiO2

1993

2166

2496

Ag@Pd/TiO2 PdAg/TiO2 Pd@Ag/TiO2

TON based on surface Pd atoms

7  Design and Architecture of Nanostructured Heterogeneous Catalysts for CO2 Hydrogenation

TON

194

15 000

10 000

5000

(b)

0

Pd@Ag/TiO

Ag@Pd/TiO2 PdAg/TiO2

Pd/TiO2 335.1 335.3 334.9 Pd 3d5/2 binding energy (eV)

Figure 7.8  (a) Comparison of the catalytic activities of a series of supported PdAg catalysts with different surface compositions and Pd/TiO2 during CO2 hydrogenation. (b) Relationship between the TON for CO2 hydrogenation based on surface-exposed Pd atoms (as determined by CO pulse adsorption) and the Pd 3d-binding energy (as determined by XPS). Source: From Mori et al. [55], © 2018 American Chemical Society. Reprinted with permission of American Chemical Society.

dissociation of H2 to form a metal-hydride species via TSI/II (transition state) with a barrier of 13.9 kcal mol−1 (step 1). Next, the adsorption of HCO3− to produce intermediate III (step 2) is followed by the attack of the H atom on the C atom of HCO3− via TSIII/IV, with a barrier of 77.4 kcal mol−1 (step 3). Finally, formate is produced accompanied by H2O to regenerate the initial active species (step 4). The activation energies for step 1 using Pd11Ag11 and Pd6Ag16 clusters were 11.9 and 11.0 kcal mol−1, respectively, which were similar to that obtained with Pd22. On the other hand, the reduction of HCO3− via TSIII/IV occurs with barriers of 58.7 and 46.2 kcal mol−1 for Pd11Ag11 and Pd6Ag16, respectively. These results show that the rate-determining step is step 3, and further demonstrate the importance of the low Pd/Ag ratio of the PdAg alloy nanoparticles in boosting the rate-determining step. This enhanced activity can be explained by considering the electronic state in reaction intermediate III (Figure 7.9b). Mulliken atomic charges of Pd atoms decrease in the order of −0.115 (Pd22) > −0.168 (Pd11Ag11) > −0.216 (Pd6Ag16), which consequently decrease the electronegativity of the dissociated hydride species on the Pd atoms. In contrast, the electronic charges of the C atoms in the adsorbed HCO3− are almost constant for all models, with their positive charges intact. Thus, the hydride species on the Pd6Ag16 with more negative charge easily attack the C atoms of the adsorbed HCO3−, whereas a higher activation energy is necessary in the reaction between the less negative hydride species on the Pd22 and the positively charged C atoms. It can be concluded that isolated and electron-rich Pd atoms created with the aid of neighboring Ag atoms explains the enhanced activity, which provides advanced insights into the architecture of catalytically active sites for CO2 hydrogenation to formic acid.

7.4.5  Metal Nanoparticles Supported on Surface-Functionalized Materials In addition to the composition, the surface properties of the support materials are another crucial factor for determining the catalytic activity, because they frequently influence the growth of metal nanoparticles, alter the electronic state owing to the SMSI effect, and create a unique active site at the perimeter [25b, 25c]. The modification of silica or carbon

7.4 ­Supported Metal Nanoparticle Catalyst OH HH Step 1

Step 2

I

II

HH O

C O–

Step 3

III

HO H C H O O–

Step 4

HCOO– + H2O

IV

V

(a) TSIII/IV

C: 0.800

46.7

H: 0.543

Relative energy (kcal mol–1)

H: 0.470 Pd22 I

TSI/II

Pd: –0.115

IV

13.9

0.0

2.1

II –14.0

III –30.7

TSIII/IV 46.2

Pd11Ag11 TS I/II I

11.9

0.0

IV

II 1.5

V –5.3

9.5

III

Pd= –0.168

V –3.4

–12.5 TSIII/IV Pd6Ag16 I 0.0

11.0

C: 0.784 H: 0.231

29.4

TSI/II II –4.1

III –16.8

C: 0.775 H: 0.430

IV –1.9

Pd= –0.216

H: 0.162

V –5.6

(b)

Figure 7.9  (a) Possible reaction mechanism for CO2 hydrogenation to formic acid. (b) Potential energy profiles and representative Mulliken atomic charges in the reaction intermediate III as determined by DFT calculations for Pd22, Pd11Ag11, and Pd6Ag11 cluster models. Source: From Mori et al. [55], © 2018 American Chemical Society. Reprinted with permission of American Chemical Society.

supports with organic reagents has also been widely utilized for the design of functionalized surfaces. It is well known that the existence of suitable surface functional groups in the vicinity of active metal centers exerts an advantageous influence on catalytic activity, and also provides unique catalytic functions, including cooperative action by several sites and steric control of the reaction intermediate [56]. In addition, the tuning of hydrophilic/ hydrophobic properties of the support materials can result in dramatic improvements in the accessibility of reactants in the liquid phase. Bimetallic PdAg nanoparticles supported on a series of amine-grafted mesoporous silica Santa Barbara Amorphous (SBA-15) were investigated for the conversion of CO2 into formic acid [57]. As illustrated in Figure 7.10a, the pristine SBA-15 was modified using various silane-coupling reagents containing primary, secondary, and tertiary amine groups (Amine-x, x = 1–5). Pd and Ag (Pd 3 wt%; molar ratio of Pd/Ag = 1) were impregnated in samples then reduced with NaBH4. For example, highly dispersed PdAg nanoparticles with a mean diameter of 3.9 nm were observed in mesoporous channels without aggregation for PdAg/SBA-15-Amine-5. This modification showed a positive effect on the catalytic performance in a 1.0-M NaHCO3 aqueous solution under a total pressure of 2.0 MPa (H2/CO2 = 1)

195

7  Design and Architecture of Nanostructured Heterogeneous Catalysts for CO2 Hydrogenation (i)

R

EtO Si EtO

OEt

Modification

OH O

Si

OH Si

O

O

Si

O

O SBA-15

O

Si O

Impregnation

OH

(iii)

O

R

EtO

(ii) Pd(NH3)4Cl2 AgNO3

Si

O

NaBH4 Reduction

O Si

O

O

OH Si

O

O

O

O

PdAg/SBA-15-Amine-x NH2 NH

N

O

Si

EtO

EtO O

O

Si

NH2

NH2

N

EtO

O

O

EtO

EtO Si

O

O

Si

O

O

Si

O

Si Si O O O O O

Si Si O O O O O

Si Si O O O O O

Si Si O O O O O

Si Si O O O O O

Amine-1

Amine-2

Amine-3

Amine-4

Amine-5

(a) NH2 NH2

EtO O

700

Si

NH

O

Si Si O O O O O

EtO

Amine-5

600 TOF in the CO2 hydrogenation

196

O

300 200

O

Si Si O O O O O

500 400

Si

N

Amine-4

NH2

EtO

EtO O

Si

N

O

EtO

Si Si O O O O O

O

Amine-3

O

100

Si

O

Si

O

Si Si O O O O O

Amine-2 O

Si Si O O O O

Amine-1 0 19 (b)

19.5

20

20.5

21

21.5

Shift of v(C=O) from free HCOOH (cm–1)

Figure 7.10  (a) Schematic illustration for the preparation of PdAg/SBA-15-Amine-x (x = 1–5). (b) The differences in the shift of the ν(C═O) band and catalytic activity in the CO2 hydrogenation. Source: From Mori et al. [57], © 2017 Royal Society of Chemistry. Reprinted with permission of Royal Society of Chemistry.

7.4 ­Supported Metal Nanoparticle Catalyst

at 100 °C. The phenylamine-grafted PdAg/SBA-15-Amine-5 was the most active catalyst at three times higher than that of the monometallic Pd catalyst. Moreover, the recovered catalyst could be recycled at least three times without significant loss of its inherent activity. Upon treatment with HCOOH, the ν(C═O) band in the FT-IR spectra of all PdAg/ SBA-15-Amine-x shifted toward shorter wavenumbers in comparison with the free carbonyl group at 1720 cm−1. Notably, the differences in the shift of the ν(C═O) band were well correlated with the catalytic activity for CO2 hydrogenation (Figure 7.10b). It can be concluded that the O⋯H─N hydrogen bonding in the acid–base pair is crucial, and moderate interaction with the weakly basic phenylamine group is suitable for attaining high catalytic activity. Another important role of the grafted amine groups is the CO2 adsorption property originating from their basicity, which increases the CO2 concentration in the vicinity of the active center. Motivated by the above results, highly dispersed PdAg nanoparticles supported on phenylamine-functionalized mesoporous carbon were further developed and assessed as an efficient heterogeneous catalyst for the conversion of CO2 into formic acid [58]. The mesoporous carbon (BET surface area = 209 m2 g−1) was oxyfunctionalized with an aqueous nitric acid solution, and then further modified with p-phenylenediamine. The concentration of grafted amine functional groups on the amine-MSC (mesoporous carbon) was determined to be approximately 0.57 mmol g−1. Finally, PdAg nanoparticles were deposited by impregnation, followed by chemical reduction. The average diameter of the PdAg nanoparticles was 1.2 nm, which was smaller than that of the unmodified mesoporous carbonsupported PdAg nanoparticles, whose diameter was 2.6 nm, demonstrating that amine functionalization resulted in the formation of smaller NPs. A TON of 839 was obtained in the hydrogenation of CO2 to formate using a 1.0 M aqueous NaHCO3 solution at pH = 8.5 under a total pressure of 2.0 MPa (H2/CO2 = 1) at 100 °C, corresponding to a TON of 3227 based on the number of surface Pd atoms calculated from CO adsorption. In addition, this material promoted the dehydrogenation of formic acid with a TOF of 5638 h−1 for evolved H2 based on the total amount of Pd, corresponding to a TOF value of 21 686 h−1 based on the quantity of surface Pd atoms. In the CO2 hydrogenation under basic aqueous conditions, the in situ–generated bicarbonate (HCO3−) frequently serves as the real substrate for the formation of formate. However, the hydrogenation of bicarbonate is disfavored and a more enhanced hydrogenation ability is needed, because bicarbonate species are thermodynamically more stable than CO2 and formate [5]. Liu et al. used a Schiff-base modified Au nanocatalyst (Au/SiO2Schiff) for realizing the direct catalytic conversion of CO2 [59]. Au/SiO2-Schiff was prepared by aldimine condensation of (3-aminopropyl)trimethoxysilane with formaldehyde, followed by chemical reduction of the deposited Au precursor. Most Au existed in the form of nanoparticles with the size of ~1.5 nm, but a smaller fraction of single atom Au species was also observed. In contrast to the completely inert behavior of Au/SiO2 without surface modification, Au/SiO2-Schiff exhibited substantially higher activity in H2O/methanol solution containing NEt3 at 90 °C under 80 bar of a CO2/H2 mixture, achieving a TON of 14 470 after 24 hours. A possible catalytic process for Au/SiO2-Schiff was proposed as shown in Figure 7.11. The NEt3 additive acts as a reservoir for gaseous CO2 via a carbamate zwitterion intermediate. This could migrate and be transferred to the gold–Schiff base interface while retaining its carbamate zwitterionic nature, and then be hydrogenated by the active

197

198

7  Design and Architecture of Nanostructured Heterogeneous Catalysts for CO2 Hydrogenation CO2(g) Capture O +

N

C

[BH]+ [HCOO]– –

B:

O O

Formate desorption

–

Activation at interface

C

H2C +

O

N H

H

Au

H2

n

tio Activa

SiO2-Schiff

Figure 7.11  Proposed synergistic mechanism for the hydrogenation of CO2 to formate by Schiff-base modified Au nanocatalyst (Au/SiO2-Schiff). Source: From Liu et al. [59]. Licensed under CC by 4.0. Public Domain.

H species at the gold–Schiff base interface. The authors also noted the beneficial effect of the electron-rich gold surface induced by the electron donation from nitrogen groups, because it could provide a more negative hydride species and boost the rate-determining nucleophilic attack to the carbon center of CO2. Recently, poly(acrylic acid) (PAA)-grafted polyethylene (PE) membranes modified with polyethylenimine (PEI) were employed for CO2 capture and in situ enzymatic CO2 hydrogenation to formic acid for the first time by Wang et al. [60] The surface amino group density can be controlled by several factors, including the grafting ratio of PAA, modification time and temperature, and PEI concentration. Under optimal conditions, the modification with PEI significantly facilitated the enzymatic CO2 hydrogenation, and the initial reaction rate was accelerated 23.6 times.

7.5 ­Embedded Single-Atom Catalysts The downsizing of supported metal nanoparticles to clusters, or eventually to isolated single atoms dispersed or anchored on supports, is a key challenge in the creation of novel heterogeneous catalysts, and offers unrivaled opportunities to maximize atom efficiency [61]. In contrast to heterogenized molecular catalysts, directly embedded single-atom catalysts on metal oxides and defects in graphene have also led to geometric and electronic effects owing to their particular locations and chemical bonds between metals and the associated interfaces [62]. For example, Co porphyrin complexes have been introduced into a graphene-like carbon matrix while maintaining their single-atom structure even after high-temperature heat treatment without the formation of Co NPs. These complexes act as active centers for various organic transformations [63]. Recent outstanding progress in single-atom catalysts has evoked an exciting research field in the architecture of heterogeneous catalysts, but their

7.5  ­Embedded Single-Atom Catalyst

use in CO2 hydrogenation has not been extensively explored, presumably due to their limited stability. Significant progress has been made by using homogeneous catalysts for formic acid synthesis by hydrogenation from CO2. The use of electron-rich metal centers with electrondonating organic ligands is responsible for attaining high catalytic activity [64]. Based on this design strategy, a single-atom electron-rich Ru species was created on a solid surface by utilizing LDH (M2+1−xM3+x(OH)2)x+(An−)x/n·mH2O, where M2+ and M3+ are di- and trivalent metal cations and An− represents interlayer anions such as CO32− as a promising support with its abundant basic OH groups with defined locations and high CO2 adsorption properties [65]. Treatment of Mg10Al2(OH)24CO3 with RuCl3·nH2O in an aqueous NaOH solution readily afforded Ru/LDH. High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) indicated the formation of an isolated single-atomic structure without agglomeration. Additionally, there were no peaks due to metallic Ru─Ru or contiguous Ru─O─Ru bonds in the Fourier transform-extended X-ray absorption fine structure (FT-EXAFS) spectrum (Figure  7.12). Upon consideration of the curve-fitting results, it can be concluded that the single-atom Ru in an octahedral coordination geometry surrounded by one hydroxyl and two water ligands was grafted onto a triad of oxygen atoms of the basic hydroxyl groups on the LDH surface. Ru/LDH acted as an effective heterogeneous catalyst in a basic aqueous solution containing NaHCO3 under a total pressure of 2.0 MPa (H2/CO2 = 1) at 100 °C, affording a TON of 461 at 24 hours. The use of a RuCl3·nH2O precursor solution did not result in efficient catalysis. Other heterogeneous Ru catalysts, such as Ru/MgO, Ru/Mg(OH)2, Ru/Al2O3, and Ru/Al(OH)3, were found to be less active. The easily recovered catalyst could be recycled at least three times. Characterization by Ru K-edge X-ray absorption fine structure (XAFS) and TEM analysis also showed that the catalyst remained virtually unchanged even after the reaction without the formation of Ru nanocrystallites. The prominent catalytic performance of Ru/LDH can be explained by an electron-rich single-atom Ru species, which was attained by the special placement of electron-donating hydroxyl groups with an ordered arrangement on the LDH surface. Notably, the catalytic activities of a series of Ru catalysts

H2O Magnitude (a.u)

Ru/LDH

O

OH Ru O

OH2 O

RuCl3 RuO2 Ru foil 0

1 2 3 4 5 6 Interatomic distance (Å)

Single-atom Ru Mg2+

Al3+

OH–

H2O

CO32–

Figure 7.12  Ru K-edge FT-EXAFS spectra and schematic illustration of Ru/LDH. Source: From Mori et al. [65], © 2017 American Chemical Society. Reprinted with permission of American Chemical Society.

199

7  Design and Architecture of Nanostructured Heterogeneous Catalysts for CO2 Hydrogenation

corresponded well to the Ru 3p-binding energy determined by XPS analysis (Figure 7.13a). A series of LDHs with different basicities could be easily synthesized by changing the M2+/ M3+ components and ratios. The optimal ratio for the Mg–Al–LDHs investigated was Mg2+/ Al3+ = 5. The catalytic activity correlated well with differences in CO2 adsorption capacity originating from different levels of basicity in the LDHs (Figure  7.13b). The high CO2 adsorption property of the hydroxyl sites was another important factor in boosting activity owing to the concentration effect on the surface of the LDH support. The CO2 hydrogenation reaction by Ru/LDH is initiated by heterolytic dissociation of H2 in the periphery of the Ru center and LDH, followed by the insertion of CO2 and isomerization via the attack of hydride H species onto C atoms. The formed Ru η1-formate intermediate is regenerated accompanied by the formation of formic acid. The activation energy (Ea) with Ru/LDH was 54.3 kJ mol−1, which was similar to that of [Cp*Ir(bpy)(OH2)]SO4 (51.4 kJ mol−1) [64], but differed from that of Au/Al2O3 (74.0 kJ mol−1) [52], indicating the involvement of a single-site reaction pathway. The reaction rate (R) can be expressed by R k PH2 1.95 PCO2 1.71 (k: rate constant) at 100 °C. This suggests that the rate-determining step is the formation of Ru-hydride due to the acceleration of the CO2 insertion step by the surface hydroxyl ligands. Later, Maru, Shukla, et  al. reported that ruthenium-hydrotalcite (Ru/HT), which was prepared by the coprecipitation method, showed high activity for CO2 hydrogenation [66]. The maximum TON of 11 389 was attained in a H2O/methanol solution (5 : 1 = x/v) at 60 °C under 60 bar of CO2/H2, and the catalyst could be recycled without significant loss of activity. Similarly, they also reported rhodium-hydrotalcite (Rh/HT) as an efficient heterogeneous catalyst for this target reaction [67]. Research on the development of single-site catalysts for the CO2 hydrogenation is in progress but lags behind that on metal nanoparticle catalysts. Some theoretical investigations have been reported, which not only help in understanding the possible reaction 500

500 Ru/LDH (Mg2+/Al3+ = 5)

(Mg2+/Al3+ = 5) 400

300

200

Mg(OH)2

Mg2+/Al3+ = 4 Mg2+/Al3+ = 6

300

Ni2+/Al3+ = 5 200 Mg2+/Fe3+ = 5 100

Al2O3

Al(OH)3 0 485

MgO

RuCl/LDH (Mg2+/Al3+ = 5)

100

(a)

TON based on RU

400 TON based on RU

200

Zn2+/Al3+ = 5 0

484.5

484

483.5

483

482.5

Ru 3p binding energy/eV

482

0

(b)

0.2 0.4 0.6 CO2 adsorption (mmolg–1)

0.8

Figure 7.13  (a) Relationship between TON-based on Ru during CO2 hydrogenation and Ru 3p-binding energy determined by XPS spectra, and (b) relationship between TON based on Ru during CO2 hydrogenation and CO2 adsorption capacity originating from the basicity of various Ru samples. Source: From Mori et al. [65], © 2017 American Chemical Society. Reprinted with permission of American Chemical Society.

7.5  ­Embedded Single-Atom Catalyst

Energy (kcal mol–1)

mechanisms, but also provide new insights as well as rational design strategies for searching for efficient single-site catalysts [68]. Sirijaraensre and Limtrakul demonstrated that Cu embedded on defects in graphene sheets, denoted as Cu-dG, shows efficient activity for the conversion of CO2 into formic acid based on DFT calculations [69]. The energy profile is shown in Figure 7.14. Since CO2 hydrogenation without activating H2 has a high activation energy (Ea = 34.6 kcal mol−1), the heterolytic cleavage of the hydrogen molecule via TS1b with a barrier of 19.7 kcal mol−1 is an unavoidable step for the CO2 conversion, where the hydride and the proton are coordinated on the Cu atom and a carbon atom at the defect site of graphene, respectively (1NT1b). The H2-activated Cu-dG can facilitate the hydrogenation of CO2 to form the HCOO-Cu/H-dG species (INT2b) via TS2b with a barrier of 13.6 kcal mol−1. Finally, the dissociation of the second hydrogen molecule on the bidentate HCOO-Cu species produces formic acid with an activation energy of 11.6 kcal mol−1, which is kinetically favorable over the direct protonation from the hydrogenated site of graphene. Sredojević et al. also performed DFT calculations to study the prospects of single-atom Ru and Cu embedded on defects in graphene sheets (Ru-dG and Cu-dG) [70]. The direct hydrogenation of CO2 with protons from the H2 molecule is hindered due to large energy barriers, i.e. higher than 35 kcal mol−1 in both models. It was found that the alternating mechanism, which is initiated by the H2 dissociation at the single metal atom followed by the CO2 adsorption, was favorable for Ru-dG. The activation energy for the above step was 20.0 TS1b

10.0 0.0 –10.0

CO2 + H2 + Cu-dG

Ea = 19.7 TS2b

H2-ads INT1b

–20.0

Ea = 13.6

CO2/INT1b

–30.0 INT2b

–40.0

TS1b

Cu-H2 = 1.51 Å 2.34

1.52

2.35

1.51 2.20 1.08

1.71

3.08

1.51 2.23 1.08

INT2b

TS2b 1.19

CO2–/INT2b

INT1b

1.53 1.60 2.16 1.07

1.27 1.26 2.02

2.12 2.43 1.07

Figure 7.14  Energy profile for CO2 hydrogenation to formic acid on Cu-dG. Distances are in Å. Source: From Sirijaraensre and Limtrakul [69], © 2016 Elsevier. Reprinted with permission of Elsevier.

201

202

7  Design and Architecture of Nanostructured Heterogeneous Catalysts for CO2 Hydrogenation

determined to be 18.8 kcal mol−1, which corresponds to the desorption energy for formic acid. On the other hand, the Cu-dG favors the pathway via proton transfer from an additional H2 to coadsorbed CO2, where the highest energy barrier was below 20 kcal mol−1. They conclude that Ru-dG is a more promising candidate as a single-atom catalyst due to its robustness compared with Cu-dG, since adsorbed H species markedly decrease Cu binding at the vacancy sites. Esrafili and Nejadebrahimi conducted a theoretical exploration of the reaction mechanism of CO2 hydrogenation over a single Co atom incorporated into nitrogen-doped graphene [71]. Due to the strong hybridization between the Co-3d and N-2p states near the Fermi level, a single Co atom is quite stable at a mono- or di-vacancy defective nitrogendoped graphene (CoN3-Gr). The strong hybridization between the H2-σ states and empty Co-3d orbitals of CoN3-Gr enabled the efficient activation of H2 and CO2 molecules. The hydrogenation of CO2 is initiated by the coadsorption of H2 and CO2, followed by the formation of a formate intermediate with an activation energy of 0.31 eV. The HCOO moiety is converted to HCOOH with a barrier of 0.51 eV. These results suggest that CoN3-Gr would be a promising catalyst for hydrogenation of CO2 even at ambient temperature.

7.6 ­Summary and Conclusions The present chapter summarizes recent approaches of diverse forms of heterogeneous metal catalysts for CO2 hydrogenation into formic acid/formate, with a focus on the design and architecture of their nanostructures. The supremacy of transition metal-based catalysts, such as Ru, Pd, and Au, is highlighted by recapping the achievements made by using monometallic and bimetallic systems, while emphasizing the importance of the active phase features, in terms of nanoparticle size, electronic features and composition, types of support and their functionalization. The understanding of the reaction mechanism has also led to significant progress in the development of active heterogeneous catalysts for this target reaction. Despite the efforts made by the research community during the last decade, there are some aspects still lacking that should be tackled in future investigations. One of the principal issues is insufficient stability and durability under the reaction conditions, which must be improved to meet practical application criteria while preserving the surface characteristics of the nanoparticles. In light of the breakthroughs achieved by modulating the metal-support interactions using N-doped carbon materials, it could be envisaged that the functionalization of carbon materials with other doping candidates (i.e. boron, sulfur, and phosphorus) could be useful for the fine modulation of supported nanoparticle features as well as for preventing the nanoparticles from sintering and leaching. Concerning the architecture of metallic active sites, single-atom catalysts are emerging as superior among the commonly explored catalysts. However, progress lags behind that for metal nanoparticle catalysts in spite of some promising theoretical investigations. Further investigation is needed to design reliable catalysts that meet practical application criteria in terms of efficiency, cost, and reusability, and it can be foreseen that tunable features and versatility of nanostructured heterogeneous catalysts will open new avenues for the conversion of CO2 into formic acid/formate.

 ­Reference

­References  1 Eberle, U., Felderhoff, M., and Schüth, F. (2009). Angew. Chem. Int. Ed. 48: 6608–6630.  2 (a) Fuku, K., Hayashi, R., Takakura, S. et al. (2013). Angew. Chem. Int. Ed. 52: 7446–7450. (b) Dalebrook, A.F., Gan, W., Grasemann, M. et al. (2013). Chem. Commun. 49: 8735–8751. (c) Staubitz, A., Robertson, A.P.M., and Manners, I. (2010). Chem. Rev. 110: 4079–4124.  3 (a) Mori, K., Tanaka, H., Dojo, M. et al. (2015). Chem. Eur. J. 21: 12085–12092. (b) NavlaniGarcía, M., Mori, K., Kuwahara, Y., and Yamashita, H. (2018). NPG Asia Mater. 10: 277–292. (c) Singh, A.K., Singh, S., and Kumar, A. (2016). Catal. Sci. Technol. 6: 12–40. (d) Mellmann, D., Sponholz, P., Junge, H., and Beller, M. (2016). Chem. Soc. Rev. 45: 3954–3988. (e) Grasemann, M. and Laurenczy, G. (2012). Energy Environ. Sci. 5: 8171–8181; (f) Eppinger, J. and Huang, K.-W. (2017). ACS Energy Lett. 2: 188–195.  4 Hull, J.F., Himeda, Y., Wang, W.H. et al. (2012). Nat. Chem. 4: 383–388.  5 Moret, S., Dyson, P.J., and Laurenczy, G. (2014). Nat. Commun. 5: 4017.  6 (a) Joó, F. (2008). ChemSusChem 1: 805–808. (b) Enthaler, S., von Langermann, J., and Schmidt, T. (2010). Energy Environ. Sci. 3: 1207–1217. (c) Azua, A., Sanz, S., and Peris, E. (2011). Chem. Eur. J. 17: 3963–3967. (d) Kanega, R., Onishi, N., Szalda, D.J. et al. (2017). ACS Catal. 7: 6426–6429. (e) Sordakis, K., Tang, C., Vogt, L.K. et al. (2018). Chem. Rev. 118: 372–433.  7 (a) Gunasekar, G.H., Park, K., Jung, K.-D., and Yoon, S. (2016). Inorg. Chem. Front. 3: 882–895. (b) Álvarez, A., Bansode, A., Urakawa, A. et al. (2017). Chem. Rev. 117: 9804–9838. (c) Bulushev, D.A. and Ross, J.R.H. (2018). Chem. Rev. 60: 566–593.  8 Mori, K. and Yamashita, H. (2010). Phys. Chem. Chem. Phys. 12: 14420–14432.  9 (a) Liao, F., Lo, T.W.B., and Tsang, S.C.E. (2015). ChemCatChem 7: 1998–2014. (b) Gao, F. and Goodman, D.W. (2012). Chem. Soc. Rev. 41: 8009–8020. 10 (a) Tedsree, K., Li, T., Jones, S. et al. (2011). Nat. Nanotechnol. 6: 302–307. (b) Mori, K., Naka, K., Masuda, S. et al. (2017). ChemCatChem 9: 3456–3462. (c) Masuda, S., Mori, K., Sano, T. et al. (2018). ChemCatChem 10: 3526–3531. 11 (a) Mori, K., Kondo, Y., and Yamashita, H. (2009). Phys. Chem. Chem. Phys. 11: 8949–8954. (b) Mori, K., Yoshioka, N., Kondo, Y. et al. (2009). Green Chem. 11: 1337–1342. 12 Bredig, G. and Carter, S.R. (1914). Ber. Disch. Chem. Ges. 47: 541–545. 13 Farlow, M.W. and Adkins, H. (1935). J. Am. Chem. Soc. 57: 2222–2223. 14 Takahashi, H., Liu, L.H., Yashiro, Y. et al. (2006). J. Mater. Sci. 41: 1585–1589. 15 Preti, D., Resta, C., Squarcialupi, S., and Fachinetti, G. (2011). Angew. Chem. Int. Ed. 50: 12551–12554. 16 Wang, T., Ren, D., Huo, Z. et al. (2017). Green Chem. 19: 716–721. 17 Wilcoxon, J.P. and Abrams, B.L. (2006). Chem. Soc. Rev. 35: 1162–1194. 18 He, C.-S., Gong, L., Zhang, J. et al. (2017). J. CO2 Util. 19: 157–164. 19 Umegaki, T., Enomoto, Y., and Kojima, Y. (2016). Catal. Sci. Technol. 6: 409–412. 20 Dupont, J., Fonseca, G.S., Umpierre, A.P. et al. (2002). J. Am. Chem. Soc. 124: 4228–4229. 21 Srivastava, V. (2014). Catal. Lett. 144: 1745–1750. 22 Qadir, M.I., Weilhard, A., Fernandes, J.A. et al. (2018). ACS Catal. 8: 1621–1627. 23 Das, S., Brudvig, G.W., and Crabtree, R.H. (2008). Chem. Commun. 44: 413–424. 24 Lee, L.-C., Xing, X., and Zhao, Y. (2017). ACS Appl. Mater. Interfaces 9: 38436–38444.

203

204

7  Design and Architecture of Nanostructured Heterogeneous Catalysts for CO2 Hydrogenation

25 (a) Qian, X., Kuwahara, Y., Mori, K., and Yamashita, H. (2014). Chem. Eur. J. 20: 15746– 15752.(b) Mori, K., Miyawaki, K., and Yamashita, H. (2016). ACS Catal. 6: 3128–3135.(c) Mori, K., Taga, T., and Yamashita, H. (2015). ChemCatChem 7: 1285–1291. 26 Zhang, B. and Su, D.S. (2015). ChemCatChem 7: 3639–3645. 27 (a) Zhu, S., Wang, J., and Fan, W. (2015). Catal. Sci. Technol. 5: 3845–3858. (b) Navalon, S., Dhakshinamoorthy, A., Alvaro, M., and Garcia, H. (2014). Chem. Rev. 114: 6179–6212. 28 Stalder, C.J., Chao, S., Summers, D.P., and Wrighton, M.S. (1983). J. Am. Chem. Soc. 105: 6318–6320. 29 Wiener, H., Sasson, Y., and Blum, J. (1986). J. Mol. Catal. 35: 277–284. 30 Su, J., Yang, L., Lu, M., and Lin, H. (2015). ChemSusChem 8: 813–816. 31 Bi, Q.-Y., Lin, J.-D., Liu, Y.-M. et al. (2014). Angew. Chem. Int. Ed. 53: 13583–13587. 32 Kasatkin, I., Kurr, P., Kniep, B. et al. (2007). Angew. Chem. Int. Ed. 46: 7324–7327. 33 Nguyen, L.T.M., Park, H., Banu, M. et al. (2015). RSC Adv. 5: 105560–105566. 34 (a) Zhao, Z., Li, M., Zhang, L. et al. (2015). Adv. Mater. 27: 6834–6840. (b) Salinas-Torres, D., Navlani-García, M., Mori, K. et al. (2019). Appl. Catal., A 571: 25–41. 35 Rideal, E.K. and Wright, W.M. (1926). J. Chem. Soc. 129: 1813–1821. 36 Cao, Y., Mao, S., Li, M. et al. (2017). ACS Catal. 7: 8090–8112. 37 Wang, X., Blechert, S., and Antonietti, M. (2012). ACS Catal. 2: 1596–1606. 38 Ong, W.-J., Tan, L.-L., Ng, Y.H. et al. (2016). Chem. Rev. 116: 7159–7329. 39 (a) Mori, K., Tatsumi, D., Iwamoto, T. et al. (2018). Chem. Asian J. 13: 1348–1356. (b) Mori, K., Itoh, T., Kakudo, H. et al. (2015). Phys. Chem. Chem. Phys. 17: 24086–24091. 40 Lee, J.H., Ryu, J., Kim, J.Y. et al. (2014). J. Mater. Chem. A 2: 9490–9495. 41 Park, H., Lee, J.H., Kim, E.H. et al. (2016). Chem. Commun. 52: 14302–14305. 42 Wang, F., Xu, J., Shao, X. et al. (2016). ChemSusChem 9: 246–251. 43 Koh, K., Jeon, M., Chevrier, D.M. et al. (2017). Appl. Catal., B 203: 820–828. 44 Shao, X., Xu, J., Huang, Y. et al. (2016). AIChE J. 62: 2410–2418. 45 Patel, P., Nandi, S., Maru, M.S. et al. (2018). J. CO2 Util. 25: 310–314. 46 Zhou, Y., Huang, Y., Jin, B. et al. (2019). Ind. Eng. Chem. Res. 58: 44–52. 47 Mondelli, C., Puértolas, B., Ackermann, M. et al. (2018). ChemSusChem 11: 2859–2869. 48 Ogo, S., Kabe, R., Hayashi, H. et al. (2006). Dalton Trans. 36: 4657–4663. 49 Ohnishi, Y.-y., Matsunaga, T., Nakao, Y. et al. (2005). J. Am. Chem. Soc. 127: 4021–4032. 50 Hao, C., Wang, S., Li, M. et al. (2011). Catal. Today 160: 184–190. 51 Liu, N., Du, R.J., and Li, W. (2013). Adv. Mater. Res. 821–822: 1330–1335. 52 Filonenko, G.A., Vrijburg, W.L., Hensen, E.J.M., and Pidko, E.A. (2016). J. Catal. 343: 97–105. 53 Tauster, S.J., Fung, S.C., and Garten, R.L. (1978). J. Am. Chem. Soc. 100: 170–175. 54 (a) Chen, P., Khetan, A., Yang, F. et al. (2017). ACS Catal. 7: 1197–1206. (b) Zhao, E.W., Zheng, H., Ludden, K. et al. (2016). ACS Catal. 6: 974–978. 55 Mori, K., Sano, T., Kobayashi, H., and Yamashita, H. (2018). J. Am. Chem. Soc. 140: 8902–8909. 56 Mori, K., Dojo, M., and Yamashita, H. (2013). ACS Catal. 3: 1114–1119. 57 Mori, K., Masuda, S., Tanaka, H. et al. (2017). Chem. Commun. 53: 4677–4680. 58 Masuda, S., Mori, K., Futamura, Y., and Yamashita, H. (2018). ACS Catal. 8: 2277–2285. 59 Liu, Q., Yang, X., Li, L. et al. (2017). Nat. Commun. 8: 1407. 60 Wang, Y., Chen, Y., Wang, C. et al. (2018). ACS Appl. Mater. Interfaces 10: 29003–29009.

 ­Reference

61 (a) Thomas, J.M. (2015). Nature 525: 325–326. (b) Yamashita, H., Mori, K., Kuwahara, Y. et al. (2018). Chem. Soc. Rev. 47: 8072–8096. 62 Hu, P., Huang, Z., Amghouz, Z. et al. (2014). Angew. Chem. Int. Ed. 53: 3418–3421. 63 (a) Banerjee, D., Jagadeesh, R.V., Junge, K. et al. (2014). Angew. Chem. Int. Ed. 53: 4359– 4363. (b) Nakatsuka, K., Yoshii, T., Kuwahara, Y. et al. (2017). Phys. Chem. Chem. Phys. 19: 4967–4974. 64 Wang, W.-H., Hull, J.F., Muckerman, J.T. et al. (2012). Energy Environ. Sci. 5: 7923–7926. 65 Mori, K., Taga, T., and Yamashita, H. (2017). ACS Catal. 7: 3147–3151. 66 Maru, M.S., Ram, S., Shukla, R.S., and Khan, N.-U.H. (2018). Mol. Catal. 446: 23–30. 67 Maru, M.S., Ram, S., Adwani, J.H., and Shukla, R.S. (2017). ChemistrySelect 2: 3823–3830. 68 Esrafili, M.D. and Dinparast, L. (2017). Chem. Phys. Lett. 682: 49–54. 69 Sirijaraensre, J. and Limtrakul, J. (2016). Appl. Surf. Sci. 364: 241–248. 70 Sredojević, D.N., Šljivančanin, Ž., Brothers, E.N., and Belić, M.R. (2018). ChemistrySelect 3: 2631–2637. 71 Esrafili, M.D. and Nejadebrahimi, B. (2019). Appl. Surf. Sci. 475: 363–371.

205

207

8 Heterogeneously Catalyzed CO2 Hydrogenation to Alcohols Nat Phongprueksathat and Atsushi Urakawa Delft University of Technology, Department of Chemical Engineering, Catalysis Engineering, Van der Maasweg 9, Delft, 2629 HZ, The Netherlands

8.1 ­Introduction Global warming, originating from anthropogenic CO2 emission, had started to become a major concern by the end of the twentieth century since the environmental impacts became more significant. Consequently, the CO2 sequestration and utilization technologies have been developed as an approach for global warming mitigation. Among the various technologies being used for recycling CO2, catalytic conversion of CO2 into methanol, ethanol, and higher alcohols is a promising approach in terms of high CO2 conversion rates and highly desired product selectivity. Moreover, these alcohols can be easily stored, transported, or continuously converted into high-octane gasoline and other feedstocks for a variety of chemical and energy industries. In the following, we first discuss more selective and technologically matured routes for alcohol synthesis through CO2 hydrogenation to methanol, and then discuss a more challenging synthesis of higher alcohols. In each section, we highlight suitable catalysts, industrial process configurations, thermodynamic constraints, and recent development.

8.2 ­CO2 Hydrogenation to Methanol – Past to Present 8.2.1  Syngas to Methanol Methanol synthesis by CO2 hydrogenation (8.1) is not a new concept. It was discovered almost in the same period (1925) as methanol synthesis from synthesis gas (8.2), a mixture primarily consisting of H2 and CO [1]. At that time, the synthesis gas used for methanol synthesis had already contained a small amount of CO2  –  up to 10%. The CO2 and H2

CO2 Hydrogenation Catalysis, First Edition. Edited by Yuichiro Himeda. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.

208

8  Heterogeneously Catalyzed CO2 Hydrogenation to Alcohols

obtained as a byproduct of fermentation processes were also used for methanol synthesis by some early plants operated in the United States [2]. However, hydrogenation of pure CO2 appeared to be too costly because CO2 requires more energy for C─O bond dissociation than CO to be converted into methanol or any other fuel [3]. CO2

3H 2  CH3OH H 2O

CO 2H 2  CH3OH CO2

H 2  CO H 2O

(8.1) (8.2)



(8.3)

In an early process, CO2 was even considered harmful for the methanol synthesis catalyst (e.g. ZnO/Cr2O3 and Cu/ZnO) due to its strong adsorption characteristic [4, 5]. Moreover, investigation of CO2 hydrogenation is usually dismissed by the preconception that the carbon dioxide is converted to the monoxide and subsequently follows the usual behavior of CO hydrogenation. It was not until decades later that the promotional effect of CO2 on methanol synthesis was reported by Klier et al. [5]. During methanol synthesis from synthesis gas (Eq. (8.2)), the Cu/ZnO catalysts experienced overreduction by the presence of CO that led to catalyst deactivation, even when the catalyst was pre-reduced with H2 prior to reaction. It was found that a small concentration of CO2 in synthesis gas can reoxidize the CO-reduced Cu catalyst and maintain the catalyst in an active state during the reaction. At higher concentration, however, the reaction rate still suffered from strong adsorption of CO2 and the selectivity of methanol was greatly reduced by the formation of methane, especially when the CO2 concentration is above 10%. Therefore, the maximum reaction rate can be achieved by balancing those promoting and retarding effects. The synthesis rate, depending on carbon dioxide concentration, has a maximum of around 1% CO2 for the commercial Cu/ZnO/Al2O3 catalyst (Imperial Chemical Industries [ICI] catalyst) [6]. After this finding, the role of CO2 in methanol synthesis had been investigated widely on various kinds of Cu/ZnO catalysts.

8.2.2 CO2 to Methanol CO2 hydrogenation has become a subject of considerable interest since CO2 was identified as the main carbon source in methanol synthesis of syngas over the Cu/ZnO/Al2O3 catalyst [9, 10]. The development of a catalytic system for pure CO2 and feed is challenging since water produced during the reaction can adsorb on active site suppressing the catalytic activity and negatively affect the chemical equilibrium. Therefore, the early process was usually designed to eliminate water from the process consisting of two methanol synthesis reactors – a rectifier is simply introduced after the first reactor to remove water and methanol from outlet stream before feeding into the second reactor. The CO produced from the first reactor is claimed to be beneficial for the catalyst since it acts as a water scavenger regenerating the Cu-active sites [11]. An increased methanol production by using such a design approach in a pilot plant was reported by Lurgi (a German engineering company) in 1994 [12]. A similar approach was applied for two-step CO2 conversion in the carbon dioxide hydrogenation to form methanol via a reverse-water-gas-shift reaction (CAMERE) pilot plant, in which the first reactor was designed to produce CO via reverse water–gas shift

8.2 ­CO2 Hydrogenation to Methanol – Past to Present

(RWGS) (8.3) before feeding to the second methanol synthesis reactor (Figure 8.1a) [8]. The traditional Cu/ZnO/Al2O3 catalyst is employed for both RWGS and methanol synthesis reactors. This process proves to have a higher methanol yield than a direct CO2 hydrogenation process, achieving 100 kg of methanol production capacity per day on a pilot scale. H2

Mixer

CO2

RWGS reactor

Compressor

Methanol synthesis Steam drum reactor

Separator Recycle stream

Cooler Water Vent

(a)

Separator

Methanol, water

Sweep gas Syngas Boiling water

Catalyst

HD-SOD membrane

H2

H2O

CO2

H2O CO

H2

CH3OH

CO2 H2O H 2O CO

(b)

Figure 8.1  (Contd.)

CH3OH

Sweep gas, H2O CH3OH H2O Boiling water H2 CO2 CO

209

Vent Compressor Recycle stream

H2 CO2 HP HE HP HE

Steam drum

Steam drum

Steam drum

Air cooler

Air cooler

Air cooler Separator Cooler

(c)

Cooler

Cooler

Methanol, water

Separator

Separator

Methanol, water

Methanol, water

Figure 8.1  Schematic diagrams of the CAMERE process; (a) the original process, (b) with membrane reactors, and, (c) with three-stage reactors. Source: (a, b) From Samimi et al. [7]. © 2018 Elsevier. Reprinted with permission of Elsevier, (c) Based on Joo et al. [8].

8.2 ­CO2 Hydrogenation to Methanol – Past to Present

Generally, the economic feasibility of methanol plants depends on several factors such as the price of oil, electricity, CO2, and byproducts. By the end of 2010, Carbon Recycling International (CRI) company established the first commercial direct CO2 hydrogenation plant in Iceland. The plant was named in honor of Professor George A. Olah. The George Olah plant is capable of producing 3000 tons of methanol per year from recycled CO2 and renewable hydrogen (i.e. hydrogen produced through renewable energy), as depicted in Figure 8.2. The CO2 is recycled from Svartsengi geothermal plant and an aluminum production plant, and hydrogen is generated from water electrolysis using geothermal power [13, 14]. Recently, the CAMERE process has been mathematically modeled in two dimensions and optimized numerically to employ a water perm-selective membrane in the methanol synthesis reactor for water removal (Figure 8.1b) [15]. The inlet temperature and pressure of the RWGS reactor are optimized at 700 °C and 30 bar over Ni/Al12O19 catalyst. The CAMERE process–assisted membrane can achieve 20.8% higher methanol production rate than through the conventional route. In the CAMERE process without membrane, water produced acts as a poison and deactivates the catalyst. Therefore, CAMERE process– assisted membrane, in which the water production is notably reduced, was proposed to increase the catalyst lifetime. Although the CAMERE process has shown potential in prolonging catalyst lifetime, the commercially available Cu/ZnO/Al2O3 catalyst still suffers from low CO2 conversion. Therefore, the concept of three-stage heat exchanger reactors (Figure 8.1c) is proposed by the same author [16]. In the three-stage configuration, the product stream of each reactor is conveyed to a flash drum to remove methanol and water from unreacted H2, CO, and

Iceland

Figure 8.2  The “George Olah Carbon Dioxide to Renewable Methanol Plant” of Carbon Recycling International in Iceland based on local geothermal energy. The first commercial carbon dioxide recycling plant operating in the world. Source: Reproduced with permission from Olah [7]. Copyright 2013, Wiley-VCH Verlag GmbH & Co. KGaA.

211

212

8  Heterogeneously Catalyzed CO2 Hydrogenation to Alcohols

CO2. Then, the gaseous stream enters the top of the next reactor as the inlet feed. Using the same catalyst volume, the CO2 conversion and methanol production rate from the threestage reactor increase c. 50% and 103% compared to the one-stage reactor. These studies show that advanced reaction engineering concepts have the potential to exceed the performance of a more conventional process, although the economic evaluations need to be carefully performed.

8.2.3  Thermodynamic Consideration – Chemical and Phase Equilibria According to Le Châtelier’s principle, the thermodynamic equilibrium of methanol synthesis is mainly governed by fugacity and enthalpy of the reaction. Therefore, the thermodynamically desirable conditions are low temperature due to its exothermicity, and high pressure due to the reduction in the number of molecules during the reaction. The reaction equilibrium calculation is based on minimization of the Gibbs free energy. For decades, the reaction equilibrium of methanol synthesis from synthesis gas has been performed using the Soave–Redlich–Kwong (SRK) equation of state [17]. This equation is one of the most accurate for a nonideal gas. In some cases, the theoretical calculation did not match with experiment results, especially at high pressure and low temperature. It was simply because methanol can condense to liquid at such conditions and the phase equilibrium was not considered in those calculations. Dew-point calculations are important for reactor and catalyst design. Precise predictive knowledge about when methanol condenses should facilitate taking advantages of in situ condensations for methanol synthesis beyond chemical equilibrium. van Bennekom et al. proposed a model based on modified SRK to simultaneously calculate the phase and chemical equilibria of methanol synthesis from synthesis gas [19]. The thermodynamic equilibria are consistent with experimental results performed in a packed bed reactor at 190–280 °C and 200 bar. The conversion values obtained in the experiments are higher than those of one-phase chemical equilibrium, and this suggests that condensation is beneficial for methanol synthesis. Moreover, the visual inspection of phase separation during methanol synthesis of synthesis gas has been demonstrated for the first time by the authors [18]. As shown in Figure 8.3, the liquid condensation was observed from a view cell at 200 °C and 200 bar.

Catalyst

Stirrer

Liquid level

Figure 8.3  Liquid formation in a view cell during methanol synthesis of H2, CO, and CO2 (0.70/0.28/0.02) at 200 °C and 200 bars. Source: Reproduced with permission from van Bennekom et al. [18]. Copyright 2013 Elsevier.

8.2 ­CO2 Hydrogenation to Methanol – Past to Present

To exploit in situ condensation in methanol synthesis, operating conditions may be selected to enhance overall driving force in methanol synthesis, whereas the conversion at phase and chemical equilibria is considerably different from that at which condensation starts. The actual reaction rates result from a tradeoff between the reaction rate and the condensation rate. For example, the reaction rates increase with increasing temperature, while the saturation and condensation at the dew point decrease with increasing temperature. Therefore, a highly active catalyst is crucial for in situ condensation since it allows lower reaction temperature with sufficient conversion for saturating condensable products. Nowadays, conducting in situ condensation is possible not only in methanol synthesis from synthesis gas but also in chemically stable CO2 thanks to the availability of active catalysts. Thermodynamic equilibrium evaluation of methanol synthesis from only CO2 and H2 has been performed by Bansode and Urakawa [21] and Gaikwad et al. [22]. The analysis considers both phases and chemical equilibria, and the calculation was based on the modified-SRK model reported by van Bennekom et  al. [18, 19]. Figure  8.4 shows the CO2 conversion and methanol selectivity derived from such a model over a wide range of pressures and temperatures. The other possible product is CO and only the reactions of Eqs. (8.1)–(8.3) were assumed to take place. The indication of phase condensation and separation in CO2 hydrogenation to methanol at stoichiometric ratio (H2/CO2 = 3) is an abrupt change in CO2 conversion within a narrow temperature range, especially at 100–300 bar (Figure 8.4). For example, the CO2 conversion changes drastically within 230–240 °C at 200 bar, which is close to that reported by van Bennekom et al. On the other hand, the phase condensation does not seem to exist at

100

(e)

(d)

60

(g)

(c)

40 (b) 20 (a)

H2/CO2 = 3

0 100

Methanol selectivity (%)

(f)

CO2 conversion (%)

80

(g) (f)

80

(e) 60

(d) (c)

40 (b) 20 0

(a) 150

200

250

300

Temperature (°C)

350

(g) (f)

80

(e) (d)

60 (c) 40

(b) (a)

20

H2/CO2 = 10

0 100

Methanol selectivity (%)

CO2 conversion (%)

100

(g) (f)

80

(e) (d)

60 (c) 40 (b) (a)

20 0

150

200

250

300

350

Temperature (°C)

Figure 8.4  Equilibrium CO2 conversion and methanol selectivity at different temperatures with initial H2/CO2 mixtures of 3 (left) and 10 (right), and at (a) 10 bar, (b) 30 bar, (c) 100 bar, (d) 200 bar, (e) 300 bar, (f) 400 bar, and (g) 500 bar. Source: From Álvarez et al. [20]. © 2017 American Chemical Society.

213

214

8  Heterogeneously Catalyzed CO2 Hydrogenation to Alcohols

all at H2/CO2 = 10, as thermodynamically expected, although CO2 equilibrium conversion is greatly boosted. The change in CO2 conversion becomes smoother at higher pressure (>400 bar) due to the formation of one dense phase. The visual inspection of phase separation during CO2 hydrogenation to methanol was demonstrated in a high-pressure view cell by Kommoß et al. [23]. In this cell, a mixture of H2, CO2, CH3OH, and H2O was prepared to simulate a product stream outlet from a reactor at 50% CO2 conversion and 67% CH3OH selectivity. The results confirmed an in situ phase separation at 206 °C and 150 bar, which is close to the theoretical expectation (Figure 8.4). However, to take advantage of phase separation at that pressure, the reaction should be operated below 206 °C, which is not practical and kinetically favorable. The optimal temperature to maximize CO2 conversion, CH3OH selectivity, and thus methanol yield is reported to be around 260–280 °C. Therefore, to increase liquid condensation temperature to match the optimum temperature, increasing reaction pressure is suggested. The advantage of high pressure in CO2 hydrogenation to methanol has been exemplified by Gaikwad et al. [22]. About 90% CO2 conversion and >95% methanol selectivity could be attained at 260–280 °C and 442 bar (partial pressure of the reactants) from the stoichiometric feed (H2/CO2 = 3). However, under such high-pressure conditions, a dense phase has formed and limited the internal mass transfer and ultimately decreased the overall reaction rate. Mass transfer limitation inhibits the utilization of the whole catalyst pellet, which can be eliminated by reducing catalyst pellet size. On the other hand, the mass transfer limitation is negligible at 331 bar and results show the almost-full (>95%) CO2 conversion and >98% methanol selectivity at 260 °C. Therefore, when mass transfer limitations are minimized, it is possible to achieve c. 90% methanol yield under continuous operation with unprecedentedly high weight time yield (gram of methanol produced per gram of catalyst per hour). Another example of taking advantage of phase separation by a novel reactor design was reported by Bos and Brilman [24]. The reactor makes use of two temperature zones to shift the chemical equilibrium: high-temperature zone for optimal catalyst activity and lowtemperature zone for full conversion by in situ condensations of a water/methanol mixture. Since the thermodynamic limitations are surpassed in this reactor, a full carbon conversion (>99.5%) and high methanol selectivity (>99.5% on a carbon basis) could be achieved at relatively low temperature and pressure (210 °C and 50 bar). Recently, the concept of in situ water absorption has been proposed to overcome thermodynamic limitation similar to the in situ condensation concept [25]. The thermodynamic analysis has shown that the methanol yield can increase up to >130%, leading to 15% higher methanol productivity. From the process simulation, in situ water sorption allows operation at much milder pressure conditions at 50 bar and 230 °C to produce c. 70% methanol yield, whereas the conventional process achieves only 25%. However, the methanol selectivity is slightly lower than the conventional process, as the RWGS equilibrium seems to be affected more than that of CO2 hydrogenation to methanol. Generally, high-pressure conditions are thermodynamically beneficial for methanol synthesis. At the same time, the reaction rates are enhanced by increased partial pressures of the reactants. The phase condensation allows full CO2 conversion and such a high conversion may offer a possibility to omit the necessity of a recycle stream. Particularly, highpressure shows potential for the combination of methanol synthesis and reforming in

8.2 ­CO2 Hydrogenation to Methanol – Past to Present

supercritical water, where a high-pressure synthesis gas is produced by pressurizing a liquid [19].

8.2.4  Catalyst Developments The first patented catalyst for methanol synthesis process is Cu-based catalysts developed by Lormand in 1925 [1]. Despite higher activity, abundancy, and economic advantages compared to other transition metals at that time (e.g. Pt, Ni, Fe, and Co), these catalysts were still thermally unstable and susceptible to sulfur poisoning – these drawbacks hindered the commercial application of Cu-based catalysts for almost half a century [26]. It was not until 1966 that ICI could discover a new co-precipitation technique for preparation of ternary Cu/ZnO/Al2O3 catalyst and overcome the intrinsic drawbacks previously mentioned. Simultaneously, gas purification processes were improved to produce sulfur-free synthesis gas from natural gas, naphtha, and crude oil, such as the Rectisol process developed by Lurgi. Both cleaner feedstocks and better development ultimately resulted in a great enhancement of methanol synthesis in terms of activity and selectivity, and significantly milden reaction condition toward lower temperature and pressure. This sparked the research trend toward Cu/ZnO based-catalysts ever since and eventually resulted in hundreds of publications. Even in the past 10 years, Cu/ZnO-based catalysts remained one of the most investigated for CO2 hydrogenation, as shown in a statistical breakdown (Figure 8.5). Technically, high Cu surface area, defects, and Cu–ZnO interfaces are well accepted to be required for the high catalytic activity of the Cu/ZnO-based catalysts. Those requirements are directly related to the method of preparation. Among various methods reported in the literature over the past 10 years, it is obvious that co-precipitation is by far the most successful and preferable method for preparation of the Cu/ZnO-based catalysts (Figure 8.6). The main advantage of the co-precipitation method is the ability to produce nanoparticles and porous microstructure of Cu–ZnO through multistep synthesis route. The prominent example is the original synthesis procedure for Cu/ZnO/Al2O3 catalyst developed by ICI shown in Figure 8.7. The initial step of co-precipitation is the formation mineral-like hydroxycarbonate precursors by mixing between metal salt solution (e.g. aqueous nitrates, sulfates, or chlorides of Cu, Zn, and/or Al) and basic precipitating agent (e.g. carbonates, bicarbonates, or hydroxides). The hydroxycarbonate structure depends on Cu and Zn contents as well as precipitating conditions (temperature and pH), and varies from amorphous zincian georgeite to crystallite structure such as copper hydrozincite ((Cu xZn1−x)5(OH)6(CO3)2, when x