High-Energy-Density Fuels for Advanced Propulsion: Design and Synthesis 3527823778, 9783527823772

Citation preview

High‐Energy‐Density Fuels for Advanced Propulsion

High‐Energy‐Density Fuels for Advanced Propulsion Design and Synthesis

Ji‐Jun Zou Xiangwen Zhang Lun Pan

Authors Ji‐Jun Zou

Tianjin University School of Chemical Engineering and Technology 92 Weijin Road Tianjin 300072 China Xiangwen Zhang

Tianjin University School of Chemical Engineering and Technology 92 Weijin Road Tianjin 300072 China Lun Pan

Tianjin University School of Chemical Engineering and Technology 92 Weijin Road Tianjin 300072 China Cover Image: © Philip Steury /Getty Images

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 . © 2020 Wiley‐VCH Verlag GmbH & Co. KGaA, 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: 978‐3‐527‐34669‐1 ePDF ISBN: 978‐3‐527‐82376‐5 ePub ISBN: 978‐3‐527‐82377‐2 oBook ISBN: 978‐3‐527‐82378‐9 Typesetting  SPi Global, Chennai, India Printing and Binding

Printed on acid‐free paper 10 9 8 7 6 5 4 3 2 1

We would like to dedicate this book to Professor Zhentao Mi, who launched the research on advanced fuel and chemical propellant in Tianjin University as early as in the 1980s. He established a top‐level lab and was the leader of the lab until retiring in 2007. The lab was certificated as Key Laboratory for Advanced Fuel and Chemical Propellant of the Ministry of Education in 2008. We would also like to dedicate this book to the 125th anniversary of Tianjin University. Tianjin University, founded in 1895 as Peiyang University, is the oldest institution of higher education in the modern history of China. The authors of this book obtained degrees from this university and continue their professional careers in this university.

vii

Contents Preface  xiii About the Authors  xv Acknowledgments  xvii 1 Introduction  1 Ji‐Jun Zou

­Reference  2

3

Development History and Basics of Aerospace Fuels  5 Xiangwen Zhang and Tinghao Jia

2.1 ­Introduction  5 2.2 ­General Properties and Requirements of Aerospace Fuels  6 2.2.1 Density  7 2.2.2 Low‐Temperature Fluidity  8 2.2.2.1 Viscosity 8 2.2.2.2 Freezing Point  10 2.2.3 Thermal Oxidation Stability  11 2.2.4 Prediction of Jet Fuel Performance  12 2.3 ­Development of Aerospace Fuels  12 2.3.1 Aviation Gas Turbine Engine Fuels (Petroleum Fuels)  12 2.3.2 Development of Russian Aerospace Fuels  15 2.3.3 High‐Thermal‐Oxidative‐Stability Fuels  15 2.3.4 Current Fuels  17 2.3.5 Future Fuels  19 2.4 ­High‐Energy‐Density Fuels  21 2.4.1 RJ‐4  21 2.4.2 RJ‐5 and Related Fuels  22 2.4.3 JP‐10, JP‐9, and RJ‐7  22 2.4.4 Strained and Diamondoid Fuels  25 2.4.5 Gelled Fuels  26 2.5 ­Non‐petroleum Fuels  27 2.5.1 F‐T Fuels  28 2.5.2 Bio‐aviation Fuels  28 2.5.3 Perspectives  31

viii

Contents

­References  3

33

Design and Synthesis of High‐Density Polycyoalkane Fuels  39 Ji‐Jun Zou and Chengxiang Shi

3.1 ­Introduction  39 3.2 ­Cycloaddition  40 3.2.1 Reaction Pathway  40 3.2.2 Cycloaddition Catalysts  44 3.3 ­Hydrogenation  50 3.3.1 Hydrogenation of Dicyclopentadiene  50 3.3.1.1 Hydrogenation Mechanism  50 3.3.1.2 Hydrogenation Catalysts  51 3.3.1.3 Hydrogenation Kinetics  54 3.3.2 Hydrogenation of Tricyclopentadiene  67 3.3.2.1 Hydrogenation Mechanism  67 3.3.2.2 Hydrogenation Catalysts  69 3.3.2.3 Hydrogenation Kinetics  70 3.4 ­Isomerization  74 3.4.1 Isomerization of Tetrahydrodicyclopentadiene  74 3.4.2 Isomerization of Tetrahydrotricyclopentadiene  81 3.5 ­Other Reactions and Procedures  90 3.5.1 Alternative Isomerization–Hydrogenation Synthesis  90 3.5.2 One‐Step Synthesis of exo‐Tetrahydrodicyclopentadiene  95 ­References  97 4

Design and Synthesis of High‐Density Diamondoid Fuels  101 Lun Pan and Jiawei Xie

4.1 ­Introduction  101 4.2 ­Synthesis of Alkyl Diamondoids via Acid‐Catalyzed Rearrangement  102 4.3 ­Synthesis of Alkyl Diamondoids via IL‐Catalyzed Rearrangement  112 4.3.1 Rearrangement of Tetrahydrotricyclopentadiene  114 4.3.2 Rearrangement of Tetrahydrodicyclopentadiene  120 4.3.3 Rearrangement of Other Polycycloalkanes  127 4.3.4 Rearrangement of Biomass‐Derived Hydrocarbons  134 4.4 ­Synthesis of Alkyl Diamondoids via Zeolite‐Catalyzed Rearrangement  135 4.5 ­Alkylation and Other Chemical Synthesis Methods  138 4.6 ­Basic Properties of Alkyl Diamondoids  142 ­References  144 5

Design and Synthesis of High‐Energy Strained Fuels  149 Ji‐Jun Zou, Junjian Xie, Yakun Liu, and Chi Ma

5.1 ­Introduction  149 5.2 ­Quadricyclane Fuel  149 5.2.1 Properties and Synthesis of Quadricyclane  149 5.2.2 Homogeneous Photosensitizers  152

Contents

5.2.2.1 Triplet Sensitizer  152 5.2.2.2 Transition‐Metal‐Compound‐Based Sensitizer  153 5.2.3 Heterogeneous Photocatalysis  155 5.2.3.1 Zinc and Cadmium Oxides and Sulfides  155 5.2.3.2 Modified Zeolites  155 5.2.3.3 Metal‐Doped TiO2  156 5.2.3.4 Ti‐Containing MCM‐41  161 5.2.3.5 Combination of Metal Doping and Framework Ti Species  164 5.2.3.6 Mechanism of Heterogeneous Photocatalysis  167 5.2.4 Utilization of Quadricyclane  168 5.3 ­Cyclopropane Fuel  170 5.3.1 Organometallic Carbenoid‐Mediated Cyclopropanation  170 5.3.1.1 Zinc Carbenoid‐Mediated Cyclopropanation  171 5.3.1.2 Samarium Carbenoid‐Mediated Cyclopropanation  174 5.3.1.3 Lithium Carbenoid‐Mediated Cyclopropanation  175 5.3.1.4 Metallic Aluminum Carbenoid‐Mediated Cyclopropanation  177 5.3.2 Transition Metal Carbene‐Mediated Cyclopropanation  181 5.3.2.1 Diazomethane System  183 5.3.2.2 Copper Catalytic System  185 5.3.2.3 Other Transition Metal Catalyst Systems  187 5.3.3 Other Cyclopropanation Methods  190 5.3.4 Fuel Synthesis and Mechanism  190 5.3.4.1 Cyclopropanation of endo‐DCPD with Monomeric IZnCH2I in Gas Phase  193 5.3.4.2 Cyclopropanation of endo‐DCPD with Monomeric IZnCH2I in Diethyl Ether Solvent  197 5.3.4.3 Cyclopropanation of endo‐DCPD with (ICH2)2Zn in Diethyl Ether Solvent  201 5.4 ­Spiro and Caged Fuels  202 5.4.1 Spiro‐Fuels  203 5.4.2 PCU Monomer, Dimers, and Derivatives  209 5.4.2.1 PCU Monomer  209 5.4.2.2 PCU Dimers  210 5.4.2.3 PCU Derivatives  214 5.4.3 Cubane and Derivatives  218 5.4.4 Other Caged Fuels  222 ­References  224 6

Design and Synthesis of High‐Density Fuels from Biomass  241 Ji‐Jun Zou and Genkuo Nie

6.1 ­Introduction  241 6.2 ­Carbon‐Increasing Reaction Strategies  244 6.2.1 Chain and Ring Increasing by Hydroxyalkylation and Alkylation  244 6.2.1.1 Synthesis of Branched Monocyclic Hydrocarbons by Hydroxylalkylation and Alkylation  250 6.2.1.2 Synthesis of Branched Monocyclic Hydrocarbons by Alkylation  252

ix

x

Contents

6.2.1.3 Synthesis of Branched Multicyclic Hydrocarbons by Alkylation  254 6.2.2 Chain and Ring Increasing by Aldol Condensation  256 6.2.2.1 Synthesis of Branched Monocyclic and Multicyclic Hydrocarbons by Aldol Condensation  256 6.2.2.2 Catalyst Design in the Synthesis of Bi‐ to Tetra‐Five/Six‐Membered Ring Hydrocarbons  260 6.2.3 Ring Increasing by Diels–Alder Cycloaddition  260 6.2.3.1 Synthesis of Multicyclic Hydrocarbons Using Terpinenes  262 6.2.3.2 Synthesis of Branched Multicyclic Hydrocarbons Using 2‐MF  265 6.2.3.3 Synthesis of Branched Monocyclic Hydrocarbons Using Diacetone Alcohol  267 6.2.3.4 Synthesis of JP‐10 Using Furfuryl Alcohol  267 6.2.4 Ring Increasing by Oligomerization  267 6.2.4.1 Synthesis of Multicyclic Hydrocarbons Using Pinene  269 6.2.4.2 Synthesis of Multicyclic Hydrocarbons Using Cyclenes  271 6.2.5 Ring Increasing by Combined Reactions  272 6.2.5.1 Robinson Annulation  272 6.2.5.2 Reductive Coupling  274 6.2.5.3 Guerbet Reaction  275 6.2.6 Fused Cycle Constructing by Skeleton Rearrangement  275 6.2.7 Integrated Reaction Strategies  277 6.2.7.1 Dual‐Bed Catalyst System  278 6.2.7.2 One‐Pot Reaction  279 6.2.7.3 Multistep Coupling Reaction  280 6.2.7.4 Cellulose Co‐conversion with Polyethylene via Catalytically Combined Processes  283 ­References  283 7

Design and Synthesis of Nanofluid Fuels  291 Lun Pan, Xiu‐Tian‐Feng E, Jinwen Cao, and Kang Xue

7.1 ­Introduction  291 7.2 ­Synthesis and Properties of Nanofluid Fuels  292 7.2.1 Single‐Step Methods  293 7.2.1.1 Physical Methods  293 7.2.1.2 Chemical Methods  299 7.2.2 Two‐Step Methods  303 7.3 ­Methods to Evaluate Stability of Nanofluids  305 7.3.1 Sedimentation Photograph Capturing  305 7.3.2 Sedimentation Balance Method  305 7.3.3 Centrifugation Method  305 7.3.4 ζ‐Potential Measurement  306 7.3.5 UV–Vis Spectrophotometer  308 7.3.6 Light Scattering Method  310 7.3.7 Three‐Omega Method  310 7.4 ­Approaches to Enhance Stability of Nanofluids  310 7.4.1 Mechanical Mixing  311 7.4.2 pH Control  312

Contents

7.4.3 Surfactants  313 7.4.4 Surface Modification  313 7.5 ­Typical High‐Energy Nanofluid Fuels  315 7.5.1 Boron‐Based Nanofluids  315 7.5.1.1 Preparation of Stable Boron‐in‐Jet Fuel Nanofluids  316 7.5.1.2 Dispersion of Boron‐Based Nanofluids  317 7.5.2 Aluminum‐Based Nanofluids  320 7.6 ­Physical Properties of Nanofluid Fuels  322 7.6.1 Density and Energy  322 7.6.2 Viscosity  323 7.6.3 Surface tension  328 7.6.4 Latent Heat of Vaporization  329 7.6.5 Combustion Characteristics  331 7.6.6 Evaporation Characteristics  337 7.7 ­Formulation and Synthesis of Gelled Fuels  341 7.7.1 Gel Formulation  341 7.7.2 Gel Preparation and Gelation Mechanism  346 7.8 ­Rheological Behavior  348 7.9 ­Atomization Behavior  352 7.10 ­Combustion Behavior  356 ­References  361 8

Design and Synthesis of Green Hypergolic Ionic Liquid Fuels  377 Xiangwen Zhang and Yong‐Chao Zhang

8.1 ­Introduction  377 8.2 ­Development History of Hypergolic Ionic Liquids  378 8.3 ­Physicochemical Properties of Hypergolic Ionic Liquids  379 8.3.1 Thermal Properties  379 8.3.2 Density  380 8.3.3 Viscosity  380 8.3.4 Heat of Formation  380 8.3.5 Ignition Delay Time  381 8.3.6 Specific Impulse  382 8.4 ­Hypergolic Ionic Liquids  382 8.4.1 Hypergolic Ionic Liquids Based on Dicyanamide Anions  382 8.4.2 Hypergolic Ionic Liquids Based on Nitrocyanamide Anions  397 8.4.3 Hypergolic Ionic Liquids Based on Boronium‐Based and B─H Bond‐ Rich Anions  402 8.4.4 Hypergolic Ionic Liquids Based on Other Anions  421 ­References  431 9

Combustion Properties of Fuels and Methods to Improve Them  437 Lun Pan and Xiu‐Tian‐Feng E

9.1 ­Introduction  437 9.2 ­Typical Equipment Used in Combustion Experiment  439 9.2.1 Rapid Compressor  439

xi

xii

Contents

9.2.2 Shock Tube  441 9.2.2.1 Heated Shock Tube  441 9.2.2.2 Aerosol Shock Tube  441 9.2.3 Hot Plate  446 9.2.4 Laser Ignition  447 9.2.5 Constant‐Volume Strand Burner  447 9.3 ­Combustion and Ignition Characters  450 9.3.1 Ignition Probability  450 9.3.2 Ignition Temperature  450 9.3.3 Ignition Delay Time  453 9.3.4 Combustion Rate  455 9.4 ­Methods to Enhance Ignition and Combustion  458 9.4.1 Effect of NP Concentration on Ignition and Combustion  458 9.4.2 Effect of Surfactants or Dispersants on Ignition and Combustion  461 9.4.3 Effect of Nanoparticle Characteristics on Ignition and Combustion  462 9.5 ­Combustion Mechanism of Nanofluid Fuels  464 ­References  470 Index  475

xiii

Preface Aerospace technology is one key to improve the life quality of human being and extend the capability of exploring space, and high‐energy‐density fuels can specifically boost the performance of aerocrafts. Although most of the high‐energy‐density fuels are hydrocarbons, their molecular structures and compositions are very different from traditional jet fuels and rocket fuels produced directly from petroleum refinery. Therefore such fuels are always synthesized by chemical reactions. The synthesis of high‐energy‐density fuels can be traced back to 1950s, and considerable progress has been achieved especially in the past 20 years. There are many literatures including technical reports and peer‐viewed publications on the synthesis and application of high‐energy‐density fuels. Especially, some reviews have been published by researchers from aerospace industry and focus on the properties of well‐developed fuels, while the synthesis chemistry of fuels, including well‐ developed and underdeveloped fuels, have not been treated at a scientific level. High‐Energy‐Density Fuels for Advanced Propulsion: Design and Synthesis comprehensively and systematically demonstrates the concept, design and synthesis of high‐energy‐density fuels, and its great potential in improving the performance of aerocrafts. The contents range from polycyoalkane fuels, strained fuels, alky‐diamondoid fuels, and hypergolic and nanofluid fuels derived from fossil and biomass, with focus on molecular design, synthesis chemistry, physiochemical properties, and their application. Most of the contents in this book are fresh achievements of the authors’ lab, i.e. Key Laboratory for Advanced Fuel and Chemical Propellant of the Ministry of Education, in the past 20 years. This book will cover the theory and practice of designing, synthesizing, and improving the performance of fuels. We hope it can connect the road from the past, present, and future of fuel chemistry and technology, provide the readers with fundamentals on high‐energy‐density fuels and their potential in advanced aerospace propulsion, and also provide the readers with inspirations for new development of advanced aerospace fuels.

xiv

Preface

This book is directed primarily to fuel chemistry and technology and aerospace propulsion technology and aims to be a definitive reference book for researchers, engineers, and students majoring in chemical science and engineering, mechanical engineering, and aerospace engineering. 4 December 2019

Ji‐Jun Zou Xiangwen Zhang Lun Pan Tianjin, China

xv

About the Authors Dr. Ji‐Jun Zou is a chair professor at School of Chemical Engineering and Technology, Tianjin University, and head of Department of Chemical Technology. He received PhD degrees from School of Chemical Engineering and Technology, Tianjin University, in 2005. He was a visiting scholar at the University of California, Riverside, from 2014 to 2015. He has been devoted to synthesis and application of advanced aerospace fuels for about 15 years. He has authored more than 130 papers and was granted with more than 20 patents. He received several awards including Technological Leading Scholar of Ten Thousand Talent Project, Changjiang Young Scholar, National Science Fund for Excellent Young Scholars, and National Excellent Doctoral Dissertation. He is an associate editor of RSC Advances and editor member of Chinese Journal of Energetic Materials. Dr. Xiangwen Zhang is a professor at School of Chemical Engineering and Technology, Tianjin University, and the director of Key Laboratory of Advanced Fuel and Chemical Propellant of the Ministry of Education. He received his academic degrees from Tianjin University and became a full professor in 2006. He has been devoted to the investigation and development of fuel chemistry including fuel processing technology and reaction engineering for more than 20 years. He has authored/ coauthored more than 300 papers and 30 patents. Dr. Lun Pan is an associate professor at School of Chemical Engineering and Technology, Tianjin University. He received his BS and PhD degrees from the School of Chemical Engineering and Technology, Tianjin University, in 2009 and 2014, respectively. He was a visiting scholar at Georgia Institute of Technology from 2016 to 2017. His research interests mainly focus

xvi

About the Authors

on the synthesis of high‐performance hydrocarbon liquid fuels and the rational design and synthesis of functional catalysts. He has published more than 50 papers and 5 patents. He has been devoted to the investigation and development of fuel chemistry for more than 10 years. He is currently an editor member of Shandong Chemical Industry.

xvii

­Acknowledgments Over the years, many students and colleagues have contributed to the development of the concepts, design, synthesis, and application of high‐energy density fuels included in this book. Professor Li Wang, Professor Qingfa Wang, and Professor Guozhu Liu provide the kind support and cooperation in the research work. This book covers many thesis research works of our graduate students, including Dr. Zhongqiang Xiong, Dr. Enhui Xing, Dr. Lei Wang, Dr. Hong Han, Dr. Jing Kong, Dr. Jiajia Song, Dr. Qiang Deng, Dr. Genkuo Nie, Dr. Xiu‐Tian‐ Feng E, Kai Jiang, Qian Miao, Bin Zhu, Mingyue Zhang, Yi Liu, Di Cao, Na Chang, Yu Zhang, Yan Xu, TingTing Ma, Fang Wang, Peijuan Han, and Zhen Li. Especially, several people made contributions to the writing of this book by checking and polishing the chapters including PhD candidates Yutong Wang, Minhua Ai, Jisheng Xu, Ying Chen, and Shangcong Sun and MS candidates Xuewei Lang and Xiaoting Xu. The work included in this book is partially supported by National Natural Science Foundation of China (20906069, 21222607, U1462119, 21978200, 21676193, 51661145026, 21506156, 21476168), Ministry of Education of China (6141A02033522, 6141A02022507), and the Tianjin Municipal Natural Science Foundation (15CZDJ37300, 16JCQNJC05200).

1

1 Introduction Ji‐Jun Zou Tianjin University, Key Laboratory of Advanced Fuel and Chemical Propellant of the Ministry of Education, Key Laboratory for Green Chemical Technology of the Ministry of Education, Department of Chemical Technology, School of Chemical Engineering and Technology, 92 Weijin Road, Tianjin 300072, China

The aerospace technology has undergone tremendous development since the first flight by Wright Brothers in 1903, which has significantly improved the life quality of human being and extended the capability of space explosion. Nowadays the aerospace vehicles mainly include commercial/military airplanes, missiles, rockets, spaceships, and satellites, which equipped with turbine, turbofan, ramjet, or rocket engines. Most of them apply liquid fuels majorly including hydrocarbon‐based and hydrazine‐based fuels. Hydrocarbons are the sole liquid fuel of airplanes, most missiles, and some rockets, meanwhile hydrazine and its derivatives are solely used for rockets for space exploration. Hydrocarbons have obvious advantages of safety and nontoxicity compared with hydrazine; thus many rockets using hydrocarbons have been used. The primary role of liquid fuels is to provide energy source for propulsion, so the energy density of fuels is critically important because to a large degree it can determine the flight distance and payload of vehicles. Of course, fuels with high‐ energy density are always desirable because they can provide sufficient energy to enhance the flight performance. With the same fuel tank, the utilization of high‐ energy density fuels can extend the flight distance, increase the payload, or increase the cruise endurance. Otherwise, the volume of fuel tank can be reduced when using high‐energy density fuels; thus the more space is accessible for loading or the overall volume of vehicles can be greatly reduced. This is especially important for volume‐limited vehicles like battle planes, unmanned aerial vehicles, tactical missiles, spaceships, and satellites, for which the space for fuels is strictly restricted. Traditional liquid fuels produced from petroleum refinery industry, like widely used jet kerosene and rocket kerosene, possess relatively low energy density ( cyclohexanone (1/5.6) > toluene (1/5.8). Under optimal reaction conditions (benzyl alcohol as solvent, temperature 150 °C, and pressure 900 kPa), a DCPD conversion of 58.0%, a TCPD selectivity of 95.7%, and an exo/endo ratio of 1/5.3 are obtained after an 11 hour reaction. Pressure also influences the cycloaddition ­significantly

Yield of TCPD (%)

50

40

30 100 kPa 500 kPa 700 kPa 900 kPa

20

10 130

135

140

145

150

Reaction temperature (°C)

Figure 3.4  The yield of TCPD under several pressure setpoints at different temperatures. Solid line: 5.5 hours, dashed line: 11 hours. Source: Xiong et al. (2005a). Reproduced with permission of Springer Nature.

43

44

3  Design and Synthesis of High‐Density Polycyoalkane Fuels

(Xiong et al. 2005a, pp. 89–97). Figure 3.4 presents the influences of the pressure on the Diels–Alder reaction at temperatures from 130 to 150 °C. The results indicate that the increase of pressure can speed up the reaction rate, leading to a higher yield of TCPD, especially at lower temperatures. It is noted that the vapor pressure of CPD is higher than that of DCPD, so the in situ formed CPD is easy to evaporate (Palmová et al. 2001, pp. 927–935). However, the reaction rate is more sensitive to the concentration of CPD than that of DCPD as proved by the kinetic analysis. Hence, high pressure can suppress the evaporation and lead to high concentration of CPD, accordingly resulting in the higher yield of TCPD. Zhang et al. conducted the cycloaddition of endo‐DCPD in a continuous‐flow reactor at elevated pressure (>1.2 MPa) (Zhang et  al. 2007c, pp. 2658–2663). Compared with the batch reaction at atmospheric pressure, the continuous‐flow process can significantly accelerate the reaction. A conversion of 82.2% and a TCPD yield of 41.7% are achieved with the reaction temperature of 160 °C, the pressure of 2 MPa, and the residence time of 12 minutes. Moreover, an exo isomer of TCPD is produced due to in situ isomerization of endo‐DCPD to exo‐ DCPD that then reacts with CPD, as shown in Eq. (3.1).

Diels–Alder +

CPD

exo-DCPD Diels–Alder

CPD

exo-DCPD

exo-TCPD

(3.1)

3.2.2  Cycloaddition Catalysts As shown in Figure 3.3, the presence of a catalyst like zeolite gives a totally different product composition (Li et  al. 2010, pp. 2522–2527). Three new adducts, namely, NB‐exo, CP‐exo, and NB‐exo, are produced with the presence of the catalyst, which means the kinetically unpreferred exo‐addition is facilitated by zeolite. Generally, pore size of catalyst is very important in catalytic cycloaddition of DCPD and CPD, because too small pore size inhibits the formation of TCPD. ZSM‐5 is one of the most well‐known acidic catalysts, which has been widely used in the petroleum industry. However, its activity in DCPD cycloaddition is very low, because the reaction can take place only on the external surface due to the diffusion limitation. Thus, it is expected that generating a mesoporous structure may make ZSM‐5 applicable for the cycloaddition of

3.2 Cycloaddition (a)

1 μm

(b)

1 μm

(e)

(f)

20 nm

20 nm

(c)

1 μm

(g)

20 nm

(d)

1 μm

(h)

20 nm

Figure 3.5  SEM and TEM micrographs of (a, e) parent HZSM‐5, (b, f ) Z5–50–0.5–1.0, (c, g) Z5–70–0.5–1.0, and (d, h) Z5–90–0.5–1.0. Source: Deng et al. (2015). Reproduced with permission of Elsevier.

DCPD and CPD. For example, using the unique hierarchically porous HZSM‐5, Deng et al. obtained a high yield of TCPD (Deng et al. 2015, pp. 540–546). The hierarchically porous HZSM‐5 was synthesized through an alkali‐treatment method. The alkali‐treatment temperature (T) ranges from 50 to 90 °C, alkali concentration (C) from 0.25 to 1.25 M, time (t) from 0.5 to 2.0 hours, and the resultant zeolites were named as samples Z5–T–C–t. Figure  3.5 shows the scanning electron microscope (SEM) and transmission electron microscope (TEM) images of the alkali‐treated zeolites. Under alkali modification (Z5–50–0.5–1.0), the crystal structure does not change, and only a few mesopores are formed. Under desilication conditions (Z5–70–0.5–1.0), many etched mesopores with size of about 7 nm are formed (see Figure 3.5f ), but the zeolite still possesses good morphology and crystal structure. Although more mesopores are observed in the severely treated zeolites (Z5–90–0.5–1.0), the crystals are broken, corresponding to the very low crystallinity determined by X‐ray diffraction (XRD). N2 adsorption–desorption isotherms (Figure  3.6) show that the parent sample has an I‐type isotherm of microporous materials with a plateau starting at the relatively low pressures and ending at relative pressure (p/p0) of 0.9. Meanwhile the alkali‐treated zeolite shows an enhanced uptake of N2 at relative pressure ranged from 0.4 to 1.0 accompanied by a hysteresis loop, confirming the formation of mesopores. The pore size distribution embedded in Figure  3.6 clearly shows the alkali‐modification produces some pores centered at 7 nm, and the amount and size of mesopores rise with the intensity of treatment. Specifically, Z5–90–0.5–1.0, Z5–70–1.0–1.0, and Z5–70–0.5–2.0 have obvious mesoporous structures. Figure  3.7 compares the activity of different zeolites under the same reaction conditions. When the catalyst is not added, the TCPD yield is 12.7%. Parent HZSM‐5 and Hβ are not so active in this reaction, and most of TCPD is derived from thermal cycloaddition. This shows that micropores are not suitable for DCPD cycloaddition. Hierarchical Z5–80–0.5–1.0 and mesoporous Al‐MCM‐41 show

45

3  Design and Synthesis of High‐Density Polycyoalkane Fuels 200

Z5 Z5–50–0.5–1.0 Z5–70–0.5–1.0 Z5–80–0.5–1.0 Z5–90–0.5–1.0

180

Volume (cm3/g)

160

dV/dlog D (cm3/g)

0.16 0.12 0.08 0.04

140

0.00

1

10 Pore diameter (nm)

100

120

100

80 0.0

0.2

0.6

0.4

(a)

0.8

1.0

p/p0 200 Z5 Z5–70–0.25–1.0 Z5–70–0.5–1.0 Z5–70–0.75–1.0 Z5–70–1.0–1.0

180

160

dV/dlog D (cm3/g)

0.16

Volume (cm3/g)

46

0.12 0.08 0.04

140

0.00

1

10 Pore diameter (nm)

100

120

100

80 0.0 (b)

0.2

0.6

0.4

0.8

1.0

p/p0

Figure 3.6  N2 adsorption–desorption isotherms along with mesopore size distribution of HZSM‐5 treated with different conditions: (a) temperature, (b) NaOH concentration, and (c) time. Source: Deng et al. (2015). Reproduced with permission of Elsevier.

very similar TCPD yield (over 35%) although the conversion of the former is slightly lower. It is noted that Z5–80–0.5–1.0 has the highest TCPD yield (36.4%) among the three typical hierarchical zeolites (Z5–70–0.5–1.0, Z5–80–0.5–1.0, Z5–90– 0.5–1.0), as shown in Figure 3.7. The cycloaddition mechanism is very ­complicated,

3.2 Cycloaddition 200 Z5 Z5–70–0.5–0.5 Z5–70–0.5–1.0 Z5–70–0.5–1.5 Z5–70–0.5–2.0

180

Volume (cm3/g)

160

140

dV/dlog D (cm3/g)

0.16 0.12 0.08 0.04 0.00

1

10 Pore diameter (nm)

100

120

100

80 0.0

0.2

(c)

0.6

0.4

0.8

1.0

p/p0

Figure 3.6  Continued

and it seems that both Lewis and Brønsted acids can catalyze the reaction. Therefore, the amount of accessible acid sites is very important for the cycloaddition reactions, and more mesopores can promote the cycloaddition process.

0.8

40

0.4

20

0.2

0

0.0

Z5 –5

H β 0– 0. 5– Z5 1.0 –7 0– 0. 5– Z5 1.0 –8 0– 0. 5– Z5 1.0 –9 0– 0. 5– 1.0 Al -M C M –4 1

0.6

Z5

60

W ith ou tc at al ys t

Conversion and yield (%)

80

1.0

Conversion exo-DCPD TCPD TeCPD

Selectivity of catalytic oligomerization

100

Figure 3.7  Activity and selectivity of catalytic cycloaddition of zeolites. Reaction condition: catalyst 10 wt%, no solvent, temperature: 150 °C, time: 5 hours. Source: Deng et al. (2015). Reproduced with permission of Elsevier.

47

3  Design and Synthesis of High‐Density Polycyoalkane Fuels

In addition, Jeon et  al. prepared highly ordered mesoporous material (MMZ‐Hβ) and nanoporous hybrid material (MMZ FER) for DCPD cycloaddition (Kim et al. 2014a, pp. 69–74; Park et al. 2016, pp. 9263–9267). MMZ‐Hβ and MMZ FER were synthesized by using commercially available zeolites as the framework sources. As shown in Figure 3.8, MMZ‐Hβ shows high activity (c. 45.9% conversion of endo‐DCPD) and selectivity (c. 67.5% selectivity of TCPD), which is ascribed to the well‐developed mesopores favorable for the diffusion of reactants, intermediates, and final products. It is noted that the reaction of DCPD and CPD follows the [4+2] addition, but catalyst can promote the formation of considerable [2+2] adduct (Kim et  al. 2014a, pp. 69–74). As early as 1985, Behr and Keim reported the [2+2] cycloaddition in the presence of a homogeneous Pd/acid catalyst (Behr and Keim 1985, pp. 314–315). Recently, Woo et al. (2014, pp. 151–155) further explored the Pd/acid‐catalyzed synthetic route and provided a simple and efficient strategy that affords a single isomer of TCPD in improved yield. TCPD can be produced in good yield (up to 98% from CPD and up to 78% from DCPD) with high selectivity by using the catalytic system comprising a Pd precursor, a phosphine or phosphite ligand,

Conversion of endo-DCPD (%)

50 40 30 20 10

Yield (%)

0 25 20

TCPD TeCPD

15 10 5 0 80

TCPD isomer selectivity (%)

48

60 40 20 0 Without catalyst

Hβ

MMZ-Hβ

Catalysts

Figure 3.8  Comparison of MMZ‐Hβ and zeolite beta in DCPD oligomerization/DCPD oligomer isomerization (reaction condition: catalyst 4.2 wt%, temperature: 150 °C, reaction time: 9 hours). Source: Kim et al. (2014a). Reproduced with permission of Elsevier.

3.2 Cycloaddition

and a carboxylic acid with the appropriate pKa. Figure 3.9 shows that TCPD ­formation with and without the Pd/acid catalysts affords different types of TCPD ­isomer with different selectivity. The most remarkable advantage of this catalytic system is the selectivity for TCPD. As shown in the gas chromatogram in Figure 3.9a,b, TCPD is the major and only product in the product distribution. Moreover, TCPD as a result of the [2+2] addition between CPD and DCPD is the only isomer generated by this method regardless of whether starting with CPD (Figure  3.9a) or DCPD (Figure  3.9b), whereas the typical Diels–Alder cycloaddition of CPD and DCPD with no catalyst generally affords a mixture of TCPD isomers, [4+2] ­isomers being the major components (Figure  3.9c). In addition, the reactions were carried out under ambient conditions, without the nitrogen atmosphere or high‐pressure environment described by existing methods. The overall reaction steps were simplified since the distillation step to separate the single TCPD isomer from other by‐products was not needed. By overcoming the limitations of the existing methods of TCPD synthesis, these Pd/acid catalysts make TCPD more readily available and attractive for potential applications.

100

14.95

(a)

% 4.57 0 100

14.94

(b)

% 4.58

6.94

0 100

6.94

(c)

% 14.70 4.57 0

6.74 5.65

14.21 14.79 7.65

9.65

11.65

13.65

15.65

20.75 17.65

19.65

Time

Figure 3.9  Product distributions of TCPD isomers depending on the reaction methods and conditions: (a) starting from CPD with Pd catalyst, (b) starting from DCPD with Pd catalyst, and (c) starting from DCPD without Pd catalyst. The peaks in the box correspond to TCPD isomers. (The peak at the retention time of 4.57 corresponds to nonane, the internal standard; those at 6.74 and 6.94 indicate DCPD isomers; that at 20.75 corresponds to CPD tetramer.) Source: Woo et al. (2014). Reproduced with permission of John Wiley & Sons.

49

3  Design and Synthesis of High‐Density Polycyoalkane Fuels

3.3 ­Hydrogenation 3.3.1  Hydrogenation of Dicyclopentadiene 3.3.1.1  Hydrogenation Mechanism

The hydrogenation reaction of DCPD is a consecutive process as proved by Zou et al. (2008, pp. 3655–3659), who studied the hydrogenation of pure DCPD over a commercially amorphous Ni‐alloy catalyst, namely, SRNA‐4. There are two hydrogenated intermediates, namely, 8,9‐dihydrodicyclopentadiene (8,9‐DHDCPD) and 3,4‐dihydrodicyclopentadiene (3,4‐DHDCPD). To clarify the reactivity of double bonds in DCPD molecules, a DFT simulation was conducted. The optimized molecular structures at B3LYP/6‐31G* level are illustrated in Figure 3.10a, according to which their thermodynamic properties are calculated. As shown in Table 3.2, the NB bond is longer than the CP bond, and its vibrational frequency is smaller. It clearly shows that the NB‐bond strength is weaker than the CP bond. In addition, Mulliken population analysis shows that the carbon atoms in NB bond have higher charge density and the bond order of the former is smaller than the latter. Consequently, NB bond should be hydrogenated more easily. Table 3.3 shows that the total energy (Etotal) of 8,9‐DHDCPD is lower than that of 3,4‐DHDCPD. Furthermore, the energy gap (ΔE) between LUMO and HOMO is larger for 8,9‐ DHDCPD. These data support that 8,9‐DHDCPD is preferred in thermodynamics, compared with 3,4‐DHDCPD. Therefore, in hydrogenation reaction, the NB bond is preferentially saturated and 8,9‐DHDCPD is the major intermediate. Figure 3.10b shows the concentration of reactants and products varying with the reaction time. Most of DCPD is converted in 50 minutes; correspondingly the concentration of 8,9‐DHDCPD quickly reaches the maximum value. This indicates that the NB bond of DCPD is hydrogenated over SRNA‐4 with great ease. After that, the concentration of 8,9‐DHDCPD declines with the reaction time owing to further hydrogenation reaction. The procedure of 8,9‐DHDCPD to THDCPD takes much longer time than that of DCPD to 8,9‐DHDCPD counterpart, ­indicating

DCPD 8,9-DHDCPD 3,4-DHDCPD THDCPD

6

DCPD

8,9–DHDCPD

5

0.25 0.20

4

0.15

3

0.10

2

0.05

1 0.00

0 3,4–DHDCPD

(a)

0 THDCPD

(b)

Concentration (mol/l)

7

Concentration (mol/l)

50

50

100 150 Time (min)

200

250

Figure 3.10  (a) Optimized molecular structures at B3LYP/6‐31G* level and possible reaction pathways. (b) Concentration of reactant, hydrogenated intermediate and product varying with reaction time (catalyst concentration 11.8 g/l; temperature 383 K; hydrogen pressure 1.5 MPa). Source: Zou et al. (2008). Reproduced with permission of Elsevier.

3.3 Hydrogenation

Table 3.2  DFT simulation results on the properties of the double bond in DCPD computed at B3LYP/6‐31G* level.

Bond type Bond distance (Å)

Mulliken atomic charge

Vibrational frequency (cm−1)

Bond order

NB

1.340

−0.062

−0.036

1649.21

1.346

CP

1.335

−0.024

−0.005

1691.13

1.350

Source: Zou et al. (2008). Reproduced with permission of Elsevier.

Table 3.3  Energy characteristics of hydrogenated intermediates computed at B3LYP/ 6‐31G* level. ΔE (a.u.)

Substance

Etotal (a.u.)

HOMO (a.u.)

LUMO (a.u.)

8,9‐DHDCPD

−389.246

−0.232

0.030

0.262

3,4‐DHDCPD

−389.235

−0.231

0.019

0.250

Source: Zou et al. (2008). Reproduced with permission of Elsevier.

that the hydrogenation rate of CP bond is dramatically low. The highest concentration of 3,4‐DHDCPD appears at about 15 minutes, and the value is much lower relative to that of 8,9‐DHDCPD. There are two possible reasons. First, the formation of 3,4‐DHDCPD cannot compete with that of 8,9‐DHDCPD due to its low reactivity. Second, the formed 3,4‐DHDCPD is rapidly hydrogenated to THDCPD because of the high reactivity of remaining NB bond. Therefore, the hydrogenation of DCPD over SRNA‐4 involves two routes as shown in Figure 3.10b, and the route via 8,9‐DHDCPD is predominant. 3.3.1.2  Hydrogenation Catalysts

Heterogeneous hydrogenation catalysts often come from noble metals or their alloys as they are effective in activating H2 at modest temperature and pressure, but the high cost and limited availability stimulate numerous efforts in seeking a cheap catalyst. Nickel is the cheaper and widely used catalyst for hydrogenation of DCPD, but a high temperature is quite necessary to achieve sufficient conversion due to the low activity of Ni, which unfortunately leads to decreased selectivity for THDCPD due to the decomposition of DCPD to CPD that is subsequently hydrogenated to cyclopentane. Accordingly, a catalyst with good activity at moderate temperatures is more desirable. Recently, great attention has been paid to amorphous alloy Ni catalysts due to their superior catalytic activity in the hydrogenation of organic compounds. They have successfully replaced traditional catalysts in many hydrogenation reactions. Moreover, the quick development of preparation techniques has made some amorphous alloy catalysts commercially available. Zou et al. (2008, pp. 3655–3659) compared the conversion of DCPD and yield of THDCPD over Raney Ni and SRNA‐4 at 383 K, as shown in Figure 3.11. For Raney Ni, the conversion of DCPD is 97.9% in four hours, but the yield of

51

3  Design and Synthesis of High‐Density Polycyoalkane Fuels 100 Raney Ni SRNA-4

80

60

60 40

40 20

20

Yield of THDCPD (%)

80

Conversion of DCPD (%)

52

0

0 0

50

100

150

200

250

Time (min)

Figure 3.11  Comparison of Raney Ni and SRNA‐4 for DCPD hydrogenation (catalyst concentration, 11.8 g/l; temperature, 383 K; hydrogen pressure, 1.5 MPa). Source: Zou et al. (2008). Reproduced with permission of Elsevier.

THDCPD is only 2.9%. Expanding the reaction time does not significantly promote the yield. With SRNA‐4 as the catalyst, the hydrogenation reaction can be conducted under moderate conditions (that is, lower catalyst concentration and hydrogen pressure) with high yield. The conversion of DCPD quickly reaches 96.5% in a half‐hour at 383 K. Moreover, the yield of THDCPD continuously increases with the reaction time, and a yield of 71.2% is obtained in four hours. This result clearly testifies that SRNA‐4 has good low‐temperature activity. Furthermore, a series of optimized experiments show that the optimized hydrogen pressure is 1.5 MPa, and the following two‐stage operation is more suitable, 393 K for one hour and then 403 K for four hours, with the yield of THDCPD reaching 98.5%. In addition, skeletal Ni catalysts can also be applied to DCPD hydrogenation in a fixed‐bed reactor. Wang et  al. (2014, pp. 15–20) prepared a supported ­skeletal Ni/Al2O3 catalyst by alkali leaching of calcinated precursors obtained by Ni–Mo–Al alloy powder and pseudo‐boehmite powder. According to Figure  3.12A,B, the sample calcinated at 880 °C has the lowest activity and selectivity of THDCPD than the other samples, which can be owed to the lowest surface area, less active sites, and significant crystallite aggregation according to characterization results. Furthermore, the sample with alloy powder/ pseudo‐boehmite mass ratio of 6/4 exhibits the lowest activity as shown in Figure 3.12C due to its smallest surface area, pore volume, and pore diameter. However, sample 2/8 shows the lowest selectivity of THDCPD (Figure 3.12D) due to less loading of active components. Thus, the precursor shall be calcinated at 860 °C and prepared with alloy powder/pseudo‐boehmite powder mass ratio of 4/6 since the corresponding catalyst shows best compressive strength, appropriate textural properties, activity, and THDCPD selectivity. Subsequently, this catalyst sample was applied to test the stability of catalyst performance, and a successive 1000 hours run experiment was carried out.

98.0 97.5 97.0 96.5 96.0 95.5 95.0 94.5 94.0 93.5 93.0 92.5 92.0 91.5 91.0 90.5 90.0 0

(C)

(a)

(d) (b) (e)

5

10 15 20 25 30 35 40 45 50 55 Reaction time (h)

(h) (i) (g)

(f) (j) 5

10 15 20 25 30 35 40 45 50 55 Reaction time (h)

96 94 92 90 88 86 84 82 80 78 76 74 72 70

(B)

(c) (d)

(b)

(e)

0

5

96 94 92 90 88 86 84 82 80 78 76 74 72 70 68 66 64 62 60 58

(D)

(a)

10 15 20 25 30 35 40 45 50 55 Reaction time (h) (h) (i)

(g) (j)

(f) 0

10 15 20 25 30 35 40 45 50 55 Reaction time (h) 90.0 87.5 85.0 82.5 80.0 77.5 75.0 72.5 70.0 67.5 65.0 62.5 60.0 57.5 55.0 52.5 50.0 47.5 45.0 42.5 40.0

Conversion rate of DCPD (%)

100 99 98 97 96 95 94 93 92 91 90 89 88 87 86 85 84 83 82 81 80

5

Selectivity of THDCPD (%)

Conversion rate of DCPD (%)

(A)

(c)

Selectivity of THDCPD (%)

98.0 97.5 97.0 96.5 96.0 95.5 95.0 94.5 94.0 93.5 93.0 92.5 92.0 91.5 91.0 90.5 90.0 0

Selectivity of THDCPD (%)

Conversion rate of DCPD (%)

3.3 Hydrogenation

–100 0 100 200 300 400 500 600 700 800 90010001100

(E)

Reaction time (h)

Figure 3.12  Patterns of catalytic performance of catalysts. (A) and (B) Catalytic performance of supported skeletal Ni catalysts prepared by precursors with different calcination temperatures: (a) 800 °C, (b) 820 °C, (c) 840 °C, (d) 860 °C, (e) 880 °C. (C) and (D) catalytic performance of supported skeletal Ni catalysts prepared by precursors with various alloy powder/pseudo‐boehmite powder mass ratios: (f ) 2/8, (g) 3/7, (h) 4/6, (i) 5/5, (j) 6/4. (E) stability of supported skeletal Ni catalyst for dicyclopentadiene hydrogenation. Source: Wang et al. (2014). Reproduced with permission of Elsevier.

The stability of catalyst sample is shown in Figure 3.12E. It is observed that the catalyst exhibits high initial activity since conversion of DCPD and selectivity of THDCPD amount to 97% and 87%, respectively. The successive run exhibits a decrease in activity and selectivity due to the fact that the sample experiences the initial catalytic activity period (from 0 to 350 hours). However, during the  next 650 hours reaction time, the conversion of DCPD and s­electivity

53

54

3  Design and Synthesis of High‐Density Polycyoalkane Fuels

of THDCPD remains stable at about 95% and 53%, respectively. This suggests that supported skeletal Ni catalyst shows good stability and it is more active than the traditional Raney Ni catalyst even though selectivity of THDCPD drops to 53% ultimately. There also has been an ongoing effort to find an earth‐abundant, low‐cost, highly active, highly selective, and stable‐in‐air hydrogenation catalyst due to the less hydrogenation activity over nickel‐based catalysts. Recently, Song et  al. (2015, pp. 6594–6599) reported that oxygen‐deficient tungsten oxide, especially WO2.72, is a versatile and efficient catalyst for the hydrogenation of linear olefins, cyclic olefins, and aryl nitro groups, with obvious advantages compared with non‐noble metal nickel catalyst from the aspects of activity and selectivity. Three tungsten oxides with different amounts of oxygen vacancies were prepared. First, WO3 was treated using H2 flow at 350 °C (denoted as WO3‐R350, BET surface area of 171 m2/g) and 60 °C (denoted as WO3‐R60, BET surface area of 173 m2/g) for one hour, respectively. Their bulk crystal structure is not affected as evidenced by XRD (Figure 3.13a), but the surface is partly reduced as indicated by the formation of W5+ and W4+ ions (Figure 3.14a–c), and the surface O/W ratio decreases to 2.87 and 2.95, respectively. WO2.72 (BET surface area of 178 m2/g) was prepared by alcoholysis of WCl6 in ethanol. XRD and high‐resolution TEM image (Figure 3.13a,d) of WO2.72 confirm the monoclinic phase (JCPDS No. 36‐0101), and the SEM image (Figure 3.13c) of WO2.72 shows urchin‐like morphology similar to WO3 (Figure  3.13e,f ), WO3‐R60, and WO3‐R350. The strong absorption tail in the UV‐Vis diffuse reflectance spectrum (Figure 3.13b) undoubtedly confirmed the presence of many oxygen vacancies, in agreement with reported characteristic of WO2.72 (Wu et  al. 2015, pp. 6983–6986). Moreover, its blue color (Figure  3.13b) originated from the intervalence charge transfer transition between W6+/5+ and W5+/4+ is quite stable in air and can remain unchanged for several months. The surface O/W ratio determined by X‐ray photoelectron spectroscopy (XPS) is 2.74 (Figure  3.14d), very close to the bulk composition. As shown in Table 3.4, WO2.72 is a versatile and efficient catalyst for the hydrogenation of linear alkenes, cyclic alkenes, and nitroarenes. Both the conversion of DCPD and the selectivity of THDCPD can reach 100%. The work of Song et al. emphasizes the role of oxygen vacancy in H2 activation and suggests that many cheap metal oxides may be made as hydrogenation catalysts through defect engineering. 3.3.1.3  Hydrogenation Kinetics

It is well known that investigation of the kinetics of chemical reaction to fully understand the reaction essentials is very necessary. The kinetics of DCPD hydrogenation into THDCPD catalyzed by Pd/Al2O3 catalyst in a stirred semi‐ batch reactor has been studied by Liu et al. (2005, pp. 3846–3851). DCPD hydrogenation is a gas–liquid–solid multiphase catalytic reaction. The reaction rate may be influenced by the external mass transfer resistance, which must be excluded by choosing suitable agitation speeds. Figure 3.15 shows the experiments carried out at different agitation speeds with constant temperature and pressure (438.15 K, 3.5 MPa). It is evident that the agitation speed has no effect on the

3.3 Hydrogenation (a)

(b)

WO2.72 Used WO2.72

Used WO2.72 Absorption (a.u.)

Intensity (a.u.)

JCPDS No. 36-0101

WO3 Used WO3

WO2.72

WO3

Used WO3

WO3-R350 WO3-R60 JCPDS No. 43-1035 20

30

(c)

40 2θ (°)

50

60

WO2.72

300

(d)

200 nm (e)

WO3

400

500 600 Wavelength (nm)

700

800

WO2.72

0.38 nm

(f)

200 nm WO3

0.37 nm

200 nm

200 nm

Figure 3.13  (a) XRD patterns, (b) diffuse reflectance spectra, (c, e) SEM images, and (d, f ) TEM images and high‐resolution TEM image (inset) of WO2.72 and WO3. Source: Song et al. (2015). Reproduced with permission of American Chemical Society.

reaction rate when it exceeds 850 rpm, which indicates that there is no external mass transfer resistance. It is necessary to carry out the initial rate study for obtaining some insights into the reaction kinetics. What should be noted is that because the first step reaction is very fast and generally terminates in several minutes, the effects of the catalyst loading and the hydrogen partial pressure on the initial rates were only performed at 363.15 K. The effect of catalyst loading on the initial rate is shown in Figure 3.16, from which the initial rate is found to vary linearly with the catalyst loading, indicating the absence of gas–liquid mass transfer limitation and that

55

3  Design and Synthesis of High‐Density Polycyoalkane Fuels W6+ O/W = 2.99

39

38

37 36 35 35 33 Binding energy (eV)

WO3-R350

40

39

38

32

W6+ W5+ W4+ O/W = 2.87

37 36 35 35 33 Binding energy (eV)

32

40 (b)

39

W6+ W5+ O/W = 2.95

38 37 36 35 35 Binding energy (eV)

33

32

W6+

WO2.72

W5+ W4+ O/W = 2.74

Intensity (a.u.)

40 (a)

(c)

WO3-R60

Intensity (a.u.)

Intensity (a.u.)

WO3

Intensity (a.u.)

56

40 (d)

39

38 37 36 35 35 Binding energy (eV)

33

32

Figure 3.14  Fitted W 4f XPS spectra of (a) WO3, (b) WO3‐R60, (c) WO3‐R350, and (d) WO2.72, respectively. Source: Song et al. (2015). Reproduced with permission of American Chemical Society.

the reaction may involve a single adsorbed species. From Figure 3.17, the first step of hydrogen addition is expected to be first‐order in hydrogen, assuming an elementary reaction. The effect of pressure on DCPD hydrogenation is presented in Figure 3.18. It can be obviously seen that hydrogen pressure strongly impacts the reaction rates of both the first and second steps. For the first reaction step, the reaction rate increases directly with the increase of hydrogen pressure. But for the second reaction step, the influence of hydrogen pressure on the reaction rate is strong at lower pressures, and this influence sharply decreases with the increase of pressure. In detail, the variations of DHDCPD concentrations are more significant when hydrogen pressure increases from 0.76 to 1.26 MPa than from 1.76 to 2.26 MPa. Generally, DCPD hydrogenation is favored at hydrogen pressure higher than 1.26 MPa. The temperature effect is shown in Figure 3.19. As can be seen form the figure, the temperature has a relatively mild effect on the reaction rates of both the first and second steps in comparison with the influence of hydrogen pressure. DCPD hydrogenation is favored at high temperatures, whereas the maximum yield of DHDCPD increases with

3.3 Hydrogenation 0.6

CTHDCPD (mol/l)

0.5 0.4 0.3 0.2 600 rpm 700 rpm 850 rpm 1000 rpm

0.1 0.0

0

10

20

30

40

50

t (min)

Figure 3.15  Effect of agitation speed on the yield of THDCPD (T = 438.15 K, pH2  = 3.5 MPa, C A0  = 0.5133 mol/l, mc = 0.6 g). Source: Liu et al. (2005). Reproduced with permission of American Chemical Society.

0.08

r0 (mol/(l min))

0.06

0.04

0.02

0.00 0.0

0.2

0.4

0.8 0.6 mc (g)

1.0

1.2

1.4

Figure 3.16  Effect of catalyst loading on the initial reaction rate (T = 363.15 K, pH2  = 0.65 MPa, C A0  = 0.65 mol/l). Source: Liu et al. (2005). Reproduced with permission of American Chemical Society.

the decrease of temperature. As can be seen from Figure 3.20, which presents the influences of initial DCPD concentrations on the kinetics, the initial DCPD concentrations have no significant influence on the initial rate of DCPD hydrogenation, indicating that the hydrogenation of DCPD into DHDCPD is zero order with respect to the DCPD concentrations. It is also observed that the time needed to reach the maximum selectivity of DHDCPD increases with the increase of DCPD concentrations.

59

3  Design and Synthesis of High‐Density Polycyoalkane Fuels 0.35 0.30

r0 (mol/(l min))

0.25 0.20 0.15 0.10 0.05 0.00

0

1

2

4

3

pH2 (MPa)

Figure 3.17  Effect of hydrogen pressure on the initial reaction rate (T = 363.15 K, mc = 0.6 g, C A0  = 0.65 mol/l). Source: Liu et al. (2005). Reproduced with permission of American Chemical Society.

0.5 0.4 C (mol/l)

60

0.3 0.2 0.1 0.0

0

20

40

80 60 t (min)

100

120

140

Figure 3.18  Experimental and predicted composition profiles at different hydrogen pressure (T = 373.15 K; C A0  = 0.5133 mol/l; mc = 0.6 g; ( ) experimental at 0.76 MPa; (+) experimental at 1.26 MPa; ( ) experimental at 1.76 MPa; ( ) experimental at 2.26 MPa; ( ) solid line, model predicted at the corresponding pressures). Source: Liu et al. (2005). Reproduced with permission of American Chemical Society.

Furthermore, a Langmuir–Hinshelwood model for the single‐site adsorption of all the organic species was developed. A comparison of all the experimental points and their corresponding values predicted by this model is also plotted in Figure 3.21. As can be seen from the figure, the proposed model describes the experimental data reasonably well with the average absolute error of less than 0.02 mol/l and the average relative error of less than 12.7%. The higher accuracy

3.3 Hydrogenation

0.5

C (mol/l)

0.4 0.3 0.2 0.1 0.0

0

20

40

60 t (min)

80

100

120

Figure 3.19  Experimental and predicted composition profiles at different temperatures ( pH2  = 1.26 MPa; C A0  = 0.5133 mol/l; mc = 0.6 g; ( ) experimental at 358.15 K; (+) experimental at 373.15 K; ( ) experimental at 388.15 K; ( ) experimental at 408.15 K; ( ) solid line, model predicted at the corresponding temperatures). Source: Liu et al. (2005). Reproduced with permission of American Chemical Society. 1.0

C (mol/l)

0.8

0.6

0.4

0.2

0.0

0

20

40

60 80 t (min)

100

120

140

Figure 3.20  Experimental and predicted composition profiles at different initial DCPD concentrations (T = 373.15 K; pH2  = 1.26 MPa; mc = 0.6 g; ( ) experimental at 0.9 mol/l; (+) experimental at 0.7 mol/l; ( ) experimental at 0.5133 mol/l; ( ) experimental at 0.3 mol/l; ( ) solid line, model predicted at the corresponding DCPD concentrations). Source: Liu et al. (2005). Reproduced with permission of American Chemical Society.

of model prediction indicates that this model can be used for the DCPD hydrogenation reactor analysis and design. Then, the hydrogenation over a Pd/Al2O3 catalyst was further studied in a quasi‐adiabatic trickle‐bed reactor (TBR) under steady‐state mode (Liu et al. 2006, pp. 8807–8814; 2008, pp. 4991–5002). The effects of operation parameters

61

3  Design and Synthesis of High‐Density Polycyoalkane Fuels 1.0

0.8

Ccalc (mol/l)

62

0.6

0.4

0.2

0.0 0.0

0.2

0.4

0.6

0.8

1.0

Cexp (mol/l)

Figure 3.21  Comparison between the predicted and experimental concentrations for all the experimental points. Source: Liu et al. (2005). Reproduced with permission of American Chemical Society.

including liquid hourly space velocity (LHSV), H2 pressure, inlet liquid concentration, and inlet temperature were investigated systematically in terms of DCPD conversion, THDCPD yield, hydrogenation rate, and axial temperature profile to get a mathematical model. LHSV is one of the most important operating parameters that influence the residence time of reactants, catalyst wetting efficiency, and, thus, the reactor performance. Figure  3.22a depicts the effect of LHSV on the TBR performance under steady‐state operations (Liu et  al. 2006, pp. 8807–8814). When the LHSV value increases from 5.86 to 14.65 h−1, the DCPD conversion decreases from 92% to 75%, and the THDCPD yield decreases from 71% to 54%, owing to the short resident time of liquid reactants. However, the global hydrogenation rate improves by almost 100%. Catalyst particles are partially wetted at lower LHSV values because of capillary forces, which leads to an increase in the reaction rates ascribed to a direct transfer of hydrogen to the surface of catalyst particles already wetted internally. Hence, the longer residence time of reactants and the higher reaction rate brought the higher conversion. When the LHSV value increases, an increase in the wetted fraction is expected to retard the reaction rate, whereas an increase in the external mass transfer coefficients will enhance it, resulting in an improvement of the catalyst utilization and, thus, higher overall reaction rates. Figure  3.22b provides axial temperature profiles of the catalyst bed at different LHSV values. When the LHSV value increases, the maximum temperature (Tmax) of the catalyst bed increases significantly, from 395 to 438 K, which is a result of the enhancement of global hydrogenation rates at higher LHSV values. Rajashekharam et al. (1998, pp. 787–805) also observed a similar tendency – that a higher liquid flow rate leads to a lower conversion but higher global hydrogenation rate and maximum temperature rise for the hydrogenation of 2,4‐dinitrotoluene. A higher overall reaction rate at higher LHSV releases

3.3 Hydrogenation 70

100

80 50

60

40

RH (kmol/m3 h)

XDCPD or YTHDCPD (%)

60

XDCPD YTHDCPD RH 40

5.86

30

8.79 11.72 LHSV (h−1)

(a)

14.65

450

T (K)

420

390 5.86 h–1

360

8.79 h–1 11.72 h–1 14.65 h–1

330 0.0 (b)

0.2

0.4

0.6

0.8

1.0

ζ

Figure 3.22  (a) Effect of the liquid flow rate on the reactor performance and (b) axial temperature profiles. Conditions: C0 = 1.06 mol/l, T0 = 343.15 K, and p = 1.0 MPa. Points represent experimental results, and the line represents simulated results. Source: Liu et al. (2006). Reproduced with permission of American Chemical Society.

more heat, which cannot be removed to the liquid phase by the solid–liquid heat transfer; thus, the temperature of the catalyst bed significantly increases. When the bed temperatures are sufficiently high to break the quasi‐adiabatic condition, the redundant heats are released to the environment. This also explains why the steady‐state reactor model does not give a satisfactory simulation of the temperature profiles at higher LHSV values. Chou et al. (1997, pp. 175–182) reported that both the conversion of DCPD to DHDCPD and the increase of temperature decrease as the LHSV value increases. The possible reason leading to this difference is the fact that when LHSV value increases, DCPD conversion and the global reaction rate as well as the overall amount of released heat decrease.

63

3  Design and Synthesis of High‐Density Polycyoalkane Fuels

The reactor performance is also dependent on liquid concentration because of its influence on the liquid–solid mass transfer rate and reaction rate. As shown in Figure  3.23a, the reactor performance, in terms of DCPD conversion, THDCPD yield, and global hydrogenation rate, is plotted against liquid concentration. When the liquid concentration increases from 0.25 to 1.35 mol/l, the DCPD conversion decreases from 99% to 85%, and the THDCPD yield decreases from 83% to 65%, but the global hydrogenation rate increases by 100%. As a result of the increase in the global hydrogenation rate, the temperature changes of catalyst surface, which leads to Tmax, also improves dramatically, from 385 to 420 K (Figure 3.23b). Chou et al. (1997, pp. 175–182) also observed a similar trend for the hydrogenation of DCPD into DHDCPD. In general, the model prediction matches well with the experimental results. When the concentrations are less than 1.0 mol/l, the simulated results of the 110

XDCPD or YTHDCPD (%)

90 40

80 70

30 XDCPD YTHDCPD RH

60 50 0.4

0.6

(a)

1.0

0.8

1.2

1.4

RH (kmol/m3 h)

50

100

20

C0 (mol/l) 440 420 400

T (K)

64

380 0.52 mol/l 0.78 mol/l 1.06 mol/l 1.35 mol/l

360 340 (b)

0.0

0.2

0.4

0.6

0.8

1.0

ζ

Figure 3.23  (a) Effect of liquid concentration on the reactor performance and (b) axial temperature profiles. Conditions: T0 = 348.15 K; p = 1.0 MPa; liquid hourly space velocity (LHSV) = 8.79 h−1. Points represent experimental results, and the line represents simulated results. Source: Liu et al. (2006). Reproduced with permission of American Chemical Society.

3.3 Hydrogenation

temperature profiles are in accordance with the experimental data, but when greater than 1.0 mol/l, the significant deviations may be caused by the extrapolation of Peng–Robinson equation to the temperature exceeding 400 K at the higher bed temperature. For the exothermic reactions, the inlet temperature of the liquid phase is an essential operation parameter that influences the performance and the axial temperature profiles of the TBR. Figure 3.24 provides the variations of DCPD conversion, THDCPD yield, global hydrogenation rate, and axial temperature profiles with the inlet temperature. In general, an increase in the inlet temperature leads to a significant increase in the reactor performance. Figure  3.24a presents that when the inlet temperature increases from 319.15 to 379.15 K, the DCPD conversion, THDCPD yield, and global hydrogenation rate increase 30%, 32%, and 48%, respectively. The inlet temperature influences not only the maximum temperature increase but also the axial temperature profiles. Interestingly, when the inlet

XDCPD or YTHDCPD (%)

80

(a)

60

(b)

50

(c)

XDCPD YTHDCPD RH

40

20

320

340

360

380

40

RH (kmol/m3 h)

60

100

30

T0 (K)

(a) 480 440

T (K)

400 360 319.15 K 333.25 K 365.15 K 379.15 K

320 280 (b)

0.0

0.2

0.4

0.6

0.8

1.0

ζ

Figure 3.24  (a) Effect of inlet temperature on the reactor performance and (b) axial temperature profiles. Conditions: C0 = 0.78 mol/l, p = 1.0 MPa, and LHSV = 8.79 h−1. Points represent experimental results, and the line represents simulated results. Source: Liu et al. (2006). Reproduced with permission of American Chemical Society.

65

3  Design and Synthesis of High‐Density Polycyoalkane Fuels

temperature exceeds 365 K, the rate of temperature rise increases sharply, which leads to the temperature at ξ = 0.5 approaching the maximum temperature, as shown in Figure 3.24b. This is also the reason for the performance improvement at higher inlet temperatures. On the other hand, this temperature increase will also result in the higher partial pressure of the solvent, and, consequently, the partial pressure of hydrogen is lowered slightly compensating for the effect of the temperature increase on the reaction rate, which leads to a mild temperature increase at ξ > 0.5. As a result of the two previously described aspects, Tmax appears in the middle of the catalyst bed at higher inlet temperatures. Generally, the operating pressure is an important operating parameter that affects TBR performance significantly. Figure 3.25 shows the influence of hydrogen pressure on the reactor performance and the axial temperature profiles. As shown in Figure 3.25a, DCPD conversion increases by 20% when the hydrogen

XDCPD or YTHDCPD (%)

80

60

60 XDCPD YTHDCPD RH

40 1.0

1.2

1.4

(a)

1.6

1.8

2.0

40

RH (kmol/m3 h)

80

100

20

p (MPa) 450 Tmax 420

T (K)

66

390 1.00 MPa 1.25 MPa 1.50 MPa 1.75 MPa 2.00 MPa

360 330 (b)

0.0

0.2

0.4

0.6

0.8

1.0

ζ

Figure 3.25  (a) Effect of hydrogen pressure on the reactor performance and (b) axial temperature profiles. Conditions: C0 = 1.06 mol/l, T0 = 333.15 K, and LHSV = 11.72 h−1. Points represent experimental results, and the line represents simulated results. Source: Liu et al. (2006). Reproduced with permission of American Chemical Society.

3.3 Hydrogenation

pressure increases from 1.0 to 2.0 MPa while the global hydrogenation rate increases by about 50%. In general, the main reason for reactor improvement via increasing pressure is that the increment of pressure increases the hydrogen concentration in the gas phase and, thus, the solubility (or concentration) of hydrogen in the solvent, which improves the impetus of mass transfer from gas to liquid and, thus, enhances gas–liquid mass transfer rates and overall reaction rates. Aside from this reason, for the system with a volatile liquid phase, it is known that the boiling point of the solvent is increased and the amount of solvent evaporations is reduced under the higher pressures, which improves the partial pressure of hydrogen and, thus, the reaction rates. Meanwhile, the lesser the solvent evaporation, the more the reaction heat can be used to increase temperature. Both effects result in an enhancement in reaction rate and a steeper temperature profile. As a result of the hydrogenation rate increase, the location of Tmax also shifts from the bottom to the middle of the catalyst bed (also from ξ = 1.0 to ξ = 0.57), as shown in Figure 3.25b. The results clearly illustrate that the temperature increase of the catalyst bed exceeds 50 K and even approaches 100 K in several runs. Therefore, it is indispensable for the process design of DCPD hydrogenation to endo‐THDCPD to avoid the local hotspot and prevent runaway of the bed temperature, which provides an opportunity for the application of unsteady‐state operation. Subsequently, unsteady‐state operation of a TBR was investigated using a multistep exothermic reaction, hydrogenation of DCPD in the presence of Pd/Al2O3 catalyst (Liu et al. 2008, pp. 4991–5002). The influences of five operation strategies on the reactor performance were symmetrically studied compared with the steady‐ state operation, including ON‐OFF and PEAK‐BASE modulations of the liquid flow rate or concentrations. It is indicated that modulation of liquid flow rate improves hydrogenation rate up to 10–20% with higher temperature amplitude (c. 15 K) and modulations of liquid concentrations less than 5% and 10 K. A novel operation strategy of TBR, hybrid modulation of liquid flow rate and concentration was proposed. The performance enhancement under the hybrid modulation (15%) is higher than the PEAK‐BASE modulation of the single parameter but the temperature amplitude less than 3 K. It should be noted that this study has examined only few typical operating conditions for different modulation strategies. Notwithstanding its limitation, this study does suggest a promising modulation strategy for exothermic reactions. A mathematic model incorporated nonlinear flow behavior, partial wetting, enthalpy balance, and phase equilibria, and three pathways of hydrogen transfer from gas to solid surface were developed based on gas and liquid plug flow model. The developed model was tested successfully to demonstrate its capability in describing the performance predictions under these conditions. The simulated results are in qualitative agreement with the experimental results except for slight deviation under liquid flow rate modulation. 3.3.2  Hydrogenation of Tricyclopentadiene 3.3.2.1  Hydrogenation Mechanism

Similar to DCPD, TCPD also has two unsaturated bonds in NB ring (NB bond) and CP ring (CP bond), but the stereo‐hindrance effect makes it more difficult to

67

3  Design and Synthesis of High‐Density Polycyoalkane Fuels

be hydrogenated than DCPD, making the hydrogenation process more complex. A DFT calculation at B3LYP/6‐31G* level has been conducted to study the hydrogenation pathway of TCPD (Wang et al. 2009c, pp. 912–917). The bond parameter shows that the NB bond is more reactive than the CP bond, the same trend as DCPD. Total energy calculation also shows 12,13‐dihydrotricyclopentadiene (12,13‐DHTCPD) is the preferred intermediate in thermodynamics, as shown in Figure 3.26 (Zhang et al. 2018, pp. 95–125). Zou et al. (2007a, pp. 209–215) also investigated the hydrogenation of TCPD and analyzed the hydrogenation products and reaction pathways. Gas chromatography–mass spectrometry (GC‐MS) analysis shows that the hydrogenated products include DHTCPD with the bond in NB ring saturated and completely saturated THTCPD. Figure 3.27 shows that the concentrations of reactants and 12

12

12

13

13 6 THTCPD

12

13

5 12,13-DHTCPD

5

13 6

6 5 5,6-DHTCPD

6 5 TCPD

ΔH°/(kJ/mol)

Figure 3.26  Enthalpy difference for stepwise hydrogenation of TCPD. Source: Zhang et al. (2018). Reproduced with permission of Elsevier.

0.15 Concentration (mol/l)

68

TCPD DHTCPD THTCPD

0.10

0.05

0.00

0

40

80

120

160

200

240

Time (min)

Figure 3.27  Concentration of reactant (TCPD), hydrogenated intermediate (DHTCPD), and product (THTCPD) involving with reaction time. Source: Zou et al. (2007a). Reproduced with permission of Elsevier.

3.3 Hydrogenation

products vary with the reaction time. The concentration of TCPD decreases quickly and conversion of more than 95% is obtained within 10 minutes. Correspondingly, the concentration of DHTCPD reaches the highest value during this period and begins to decline after that. The concentration of THTCPD always increases with the increase of reaction time. This tendency is a characteristic of consecutive reaction. Thus, the hydrogenation of TCPD takes place according to Eq. (3.2). H2

TCPD

H2

DHTCPD

THTCPD

 (3.2)

The first step is easy to take place and different catalysts show similar activities. However, the second step is difficult to occur and different catalysts exhibit different activities. Therefore, the yield of THTCPD obtained from the second step is used to assess the hydrogenation activity of different catalysts. 3.3.2.2  Hydrogenation Catalysts

As mentioned in Section  3.3.1.2, amorphous Ni is an excellent catalyst for the hydrogenation of DCPD. Similarly, an amorphous Pd‐based catalyst exhibits excellent catalytic activity and thermal stability in the hydrogenation of TCPD (Zou et  al. 2007a, pp. 209–215). As described in this literature, the catalysts (Pd‐B/γ‐Al2O3) were pre‐calcined at different temperatures before KBH4 reduction. Inductively coupled plasma (ICP) analysis in Figure 3.28 shows that this pretreatment can increase the amount of Pd on the prepared Pd‐B/γ‐Al2O3 catalyst. For the catalyst without pre‐calcination, the amount of Pd is only 0.291 wt%, which is dramatically lower than the initial defined amount of 0.4 wt%. This means that the interaction between impregnated Pd2+ ions and support is not strong enough to anchor Pd particles on the support surface. Pre‐calcinations at the proper temperature can enhance the interaction between Pd2+ ions and support and thus stably anchor the metal particles on the support surface. The amount of Pd is increased to 0.361 wt% after pre‐calcination at 300 °C, but higher pre‐calcination temperature does not improve the effect anymore. Table  3.5 shows the activities of catalysts with pre‐calcination at different temperatures. The catalyst without pre‐calcination leads to a yield of THTCPD as 60.6%, whereas those with pre‐calcination exhibit a yield above 73.0%. Therefore, the pre‐calcination can significantly improve the hydrogenation activity. This is reasonable because pre‐ calcination can increase the amount of Pd on the resulting catalyst. To evaluate the activity of per Pd atom, the turnover frequency (TOF) of Pd atoms was ­calculated as shown in Table 3.5. It is found that pre‐calcination at 200 °C leads to the highest TOF value, indicating that the interaction between metal and support can enhance the hydrogenation activity of amorphous Pd. Nevertheless, pre‐­ calcination at higher temperature conversely results in lower TOF value, suggesting that too strong interaction between metal and support may suppress the hydrogenation activity.

69

3  Design and Synthesis of High‐Density Polycyoalkane Fuels 0.4 300°

400°

500°

200° Amount of Pd (wt%)

70

0.3

none

0.2

0.1

0

Figure 3.28  Amount of Pd on Pd‐B/γ‐Al2O3 catalysts prepared with pre‐calcination at different temperatures. Source: Zou et al. (2007a). Reproduced with permission of Elsevier. Table 3.5  Hydrogenation activity of fresh Pd‐B/γ‐Al2O3 catalysts with different pre‐calcination temperatures. Pre‐calcination temperature (°C)

Conversion of TCPD (%)

Selectivity of THTCPD (%)

Yield of THTCPD (%)

TOF (s−1)

0

97.3

62.3

60.6

8.3

200

97.0

73.6

71.4

8.9

300

96.9

75.8

73.5

8.2

400

96.9

75.1

72.8

8.1

500

96.5

76.7

74.0

8.2

Source: Zou et al. (2007a). Reproduced with permission of Elsevier.

3.3.2.3  Hydrogenation Kinetics

The kinetics of TCPD hydrogenation over the amorphous Pd‐B/γ‐Al2O3 catalyst was studied in a stirred semi‐batch reactor (Zou et al. 2007b, pp. 4415–4420). The reaction is a consecutive reaction with 12,13‐DHTCPD as the intermediate, which is in agreement with the calculation results. Experiments over a range of catalyst concentrations and hydrogenation pressures were performed to analyze the initial hydrogenation rate, with the attempt to get some insights into the reaction kinetics. Because the hydrogenation process of TCPD to DHTCPD is very fast, the reactions for initial rate analysis were conducted at relatively low temperature (353 K). To obtain the initial rate of the first step (r1,0), the concentration–time data of TCPD were fitted by the first‐order decay model from which the initial rates were calculated by differentiation at time t = 0. The concentration–time data of THTCPD were fitted with the sigmoidal (Boltzmann) model to obtain the initial rate of second step (r2,0). It can be found that both initial rates (r1,0 and r2,0) exhibit a first‐order dependence on the catalyst

3.3 Hydrogenation

concentration. This not only confirms that the reaction is free of mass transfer limitation but also indicates that the reaction involves a single adsorbed species. Figure 3.29 presents the linear relationships between the initial reaction rates and the pressure of hydrogen. It is evident that both steps are first order with respect to the hydrogen pressure. Because the concentration of hydrogen dissolved in the liquid phase is linearly dependent on the pressure, the result suggests that it is hydrogen dissolved in the liquid phase that participates in the hydrogenation reaction. Detailed hydrogenation experiments were carried out with different initial TCPD concentrations (0.2–0.4 mol/l), reaction temperatures (383–413 K), and hydrogen pressures (2.0–3.5 MPa). Further information on the reaction kinetics could be obtained through the assessment of the effects for these three ­parameters on the hydrogenation reaction. Figure 3.30 presents the influence of temperature on the reaction. In the studied range, the TCPD concentration decays quickly, 0.05

r1,0 (mol/(gcal min))

0.04 0.03 0.02

0.01

0.00 0.0

0.5

1.0

1.5 pH2 (MPa)

2.0

2.5

0.5

1.0

1.5 pH2 (MPa)

2.0

2.5

(a) 0.0008

r2,0 (mol/(gcal min))

0.0006

0.0004

0.0002

0.0000 0.0 (b)

Figure 3.29  Effect of hydrogen pressure on initial reaction rate of (a) the first step and (b) second step (mc = 1.6 g/l; T = 353 K; [TCPD]0 = 0.3 mol/l). Source: Zou et al. (2007b). Reproduced with permission of American Chemical Society.

71

3  Design and Synthesis of High‐Density Polycyoalkane Fuels 0.30 DHTCPD

Concentration (mol/l)

0.25

THTCPD

0.20 383 K 393 K 403 K 413 K

0.15 0.10 0.05 0.00

TCPD 0

100

200 t (min)

300

400

Figure 3.30  Experimental and predicted concentration−time curves at different reaction temperatures (mc = 5.3 g/l; pH2  = 2.5 MPa; [TCPD]0 = 0.3 mol/l; solid line, model predicted). Source: Zou et al. (2007b). Reproduced with permission of American Chemical Society.

and a conversion of 98% is obtained within 10 minutes, again indicating that the first hydrogenation step takes place easily. The increase of temperature can improve the reaction rate, but the effect is finite. The second step takes much longer time, indicating that this reaction occurs with difficulty because of the steric hindrance of the CP ring. The concentration of DHTCPD decreases more quickly, and the concentration of THTCPD correspondingly rises more quickly at a higher temperature. The maximum DHTCPD concentration decreases with the increase of reaction temperature, suggesting that high temperature is more favorable to the second‐step reaction. Figure 3.31 depicts the influence of hydrogen pressure on the hydrogenation reaction. Increasing the pressure can accelerate the reaction, and the effect on 0.30 DHTCPD

0.25 Concentration (mol/l)

72

THTCPD

0.20

2.0 MPa 2.5 MPa 3.0 MPa 3.5 MPa

0.15 0.10 0.05 0.00

TCPD 0

100

200

300 t (min)

400

500

Figure 3.31  Experimental and predicted concentration–time curves at different hydrogen pressures (mc = 5.3 g/l; T = 413 K; [TCPD]0 = 0.3 mol/l; solid line, model predicted). Source: Zou et al. (2007b). Reproduced with permission of American Chemical Society.

3.3 Hydrogenation

the second step is more obvious. Specifically, the concentration of DHTCPD decreases, and the concentration of THTCPD increases more quickly when the hydrogen pressure increases. The maximum concentration of DHTCPD is not influenced by the change of hydrogen pressure. This indicates that the hydrogen pressure has an equal effect on both steps, consistent with the result of initial rate studies. Figure 3.32 presents the effect of initial concentration of TCPD on the hydrogenation reaction. It can be obviously seen that the initial concentration of TCPD has no significant effect on the rate for TCPD hydrogenation to DHTCPD. Thus, the first step reaction is zero order with respect to the TCPD concentration. The maximum concentration of DHTCPD increases with the initial concentration of TCPD, but the time needed to reach the maximum value also increases. The initial rate studies and kinetic experiments show three important tendencies. First, the reaction involves a single adsorbed species. Second, the reaction is first order with respect to the hydrogen pressure. Third, the first step is zero order with respect to the initial concentration of TCPD. To analyze the dependence of the second step on the concentration of DHTCPD, the concentration–time data of THTCPD in Figure 3.32 were fitted with the sigmoidal model, and the reaction rate was obtained by the differentiation method. Figure 3.33 presents the relationship between ln r2 and ln[DHTCPD]. The reaction shows a tendency from first to zero order with respect to [DHTCPD]. This proves that it is DHTCPD absorbed on the catalysts surface, not free DHTCPD in the liquid phase, that takes part in the reaction. It also shows that the reaction kinetics cannot be explained by a simple power‐law empirical model. Actually, the average relative error of this model is higher than 43.0%. Hence, an Eley–Rideal kinetic model involving single‐site adsorbed organic species and dissolved hydrogen is formulated on the basis of observed tendencies. The model can accurately fit the experimental data and correctly explain the observed tendencies. And the activation energies for the first and second step reactions are 11.11 and 34.71 kJ/mol, respectively.

Concentration (mol/l)

0.36 DHTCPD

0.30

0.4 mol/l 0.3 mol/l 0.2 mol/l

0.24 THTCPD

0.18 0.12 0.06 0.00

TCPD 0

80

160

240 t (min)

320

400

Figure 3.32  Experimental and predicted concentration–time curves at different initial TCPD concentrations (mc = 5.3 g/l; T = 403 K; pH2  = 2.5 MPa; solid line, model predicted). Source: Zou et al. (2007b). Reproduced with permission of American Chemical Society.

73

3  Design and Synthesis of High‐Density Polycyoalkane Fuels

–4.5

In r2

74

–6.0

–7.5

–3.3

–3.0

–2.7 In(DHTCPD)

–2.4

–2.1

Figure 3.33  Plot of ln r2 vs. ln[DHTCPD] (mc = 5.3 g/l; T = 403 K; pH2  = 2.5 MPa; [TCPD]0 = 0.3 mol/l). Source: Zou et al. (2007b). Reproduced with permission of American Chemical Society.

3.4 ­Isomerization 3.4.1  Isomerization of Tetrahydrodicyclopentadiene The product from the hydrogenation of DCPD, i.e. endo‐THDCPD, is in solid state with a melting point of c. 79 °C, so it cannot be used as a liquid HED fuel (Sibi et al. 2012, pp. 976–983). Interestingly, its isomer, exo‐THDCPD, possesses a very low freezing point of −79 °C. Therefore, the isomerization of endo‐ THDCPD to exo‐THDCPD is needed. Originally, the isomerization is conducted by using a Brønsted acid like sulfuric acid as catalyst, and now the most used catalyst is Lewis acid such as AlCl3 (Cristol et al. 1960, pp. 2351–2356). But the usage of these catalysts causes many environmental problems, so more environmentally benign catalysts have been explored. Common solid acid like zeolites have been studied, and among the zeolites tested, HY‐type zeolites with larger pores are the most active catalysts (Xing et al. 2007, pp. 589–593; Zhang et al. 2007a, pp. 3059–3063). It is found that the suppression of medium acid and increase of weak acid by fluorine modification can promote the selectivity of exo‐THDCPD. Under the optimal reaction conditions, the conversion of endo‐ THDCPD is 94.0% with exo‐THDCPD selectivity of 98.4% (Zhang et al. 2007a, pp. 3059–3063). Acidic ionic liquids (ILs) are also used to catalyze the isomerization of endo‐THDCPD to exo‐THDCPD. The strong acidity and polarization make the reaction take place easily at room temperature. Huang et  al. (2011, pp. 1012– 1017) investigated the effect of cations in chloroaluminate ILs on the isomerization of endo‐THDCPD and obtained >93.0% conversion and >99.2% selectivity. Table 3.6 summarizes the results concerning the effect of varying the quaternary ammonium salt in the preparation of the chloroaluminate IL catalysts. The quaternary chloride salts pyridine hydrochloride (PHC), 1‐butyl‐4‐ methylpyridinium chloride (BMPC), triethylamine hydrochloride (TEAC), 1‐butyl‐3‐methylimidazolium chloride (BMIC), 1‐hexyl‐3‐methylimidazolium

3.4 Isomerization

Table 3.6  Effect of various cations in IL composition on endo‐THDCPD isomerization.a Selectivity (%) [NR4]Cl

PHC BMPC TEAC BMIC HMIC OMIC HDMIC PHB BMIB

Reaction time (h)

Conversion (%)

exo‐THDCPD

Adamantane (ADM)

1

39.4

99.7

0.3

6

95.9

99.4

0.6

1

22.2

100

0

6

66.5

99.9

0.1

1

36.9

100

0

6

93.0

99.9

0.1

1

31.9

100

0

6

80.8

100

0

1

24.2

100

0

6

64.8

100

0

1

21.0

100

0

6

66.6

100

0

1

24.8

100

0

6

73.5

100

0

1

49.9

100

0

6

98.1

99.2

0.8

1

36.0

100

0

6

88.5

100

0

a) Reaction conditions: mole fraction of AlCl3/[NR4]Cl = 0.60/0.40, mole ratio of [NR4]Cl/endo‐ THDCPD = 1/12.8, reaction temperature 50 °C. Source: Huang et al. (2011). Reproduced with permission of Elsevier.

chloride (HMIC), 1‐octyl‐3‐methylimidazolium chloride (OMIC), and 1‐ hexadecyl‐3‐methylimidazolium chloride (HDMIC) were conjoined with AlCl3 to form the catalysts. All these catalysts are effective in catalyzing the isomerization of endo‐THDCPD to the exo isomer, and their catalytic activities follow the order PHC > TEAC > BMIC > HDMIC > HMIC ≥ BMPC ≥ OMIC. It has been reported that infrared spectra of ILs in acetonitrile solution can be used to gauge the Lewis acidities of the ILs (Yang and Kou 2004, pp. 226–227). Pure acetonitrile shows characteristic CN stretching vibration bands at 2292 and 2252 cm−1. The addition of a Lewis acid to acetonitrile results in part of the ­acetonitrile associating with the Lewis acid, which shifts the stretching vibration bands to higher wavenumbers. The relative intensity of these new bands in comparison with the original bands may be used to estimate the acidity. A steady increase in the intensities of the absorptions at 2305 and 2335 cm−1 with the increasing mole fraction of AlCl3 in PHC/AlCl3 IL in acetonitrile is observed, as shown in Figure 3.34A. These observations confirm that a minimum mole ­fraction of AlCl3 is required in order to form an acidic IL to complex with acetonitrile to

75

3  Design and Synthesis of High‐Density Polycyoalkane Fuels 9

10 (a)

7

4 (d)

2

(e)

Absorbance

6 (c)

(g)

8

8 (b) Absorbance

76

6 5

(h) (i) (j)

4 3

(k)

2

(l)

1

(m) (f) 0 0 2400 2350 2300 2250 2200 2400 2350 2300 2250 2200 (A) Wavenumber (cm–1) (B) Wavenumber (cm–1)

Figure 3.34  (A) FT‐IR spectra of acetonitrile + PHC/AlCl3 (20 : 1, w/w) with different mole fractions of AlCl3 in the PHC/AlCl3 of: (a) 0.40; (b) 0.50; (c) 0.60; (d) 0.65; (e) 0.70; and (f ) 0.75. (B) FT‐IR spectra of acetonitrile + IL/AlCl3 (mole fraction of AlCl3 = 0.60) (10 : 1, w/w) for the ILs: (g) TEAC; (h) PHC; (i) BMIC; (j) BMPC; (k) OMIC; (l) HDMIC; and (m) HMIC. Source: Huang et al. (2011). Reproduced with permission of Elsevier.

produce the new set of stretching vibration absorptions. Any further increase in the mole fraction of AlCl3 in the IL leads to a corresponding increase in the acidity as well as the relative intensity of the new set of stretching vibration bands. Using the same method, Huang et al. recorded the Fourier–transform infrared spectroscopy (FT‐IR) spectra of the IL catalysts used in this study. The results in Figure 3.34B show that the relative absorbance related to Lewis acidity decreases in the following order: TEAC ≈ PHC > BMIC > BMPC ≈ OMIC ≈ HDMIC > HMIC. The absorption for HDMIC is very weak since the molar volume of HDMIC is very large, and this leads to a dilution of the AlCl3 present in the equilibrium mixture. These results suggest that the relative Lewis acidity follows the order TEAC  ≈ PHC > BMIC > HDMIC > BMPC ≈ OMIC ≈ HMIC. Apparently, the order of relative Lewis acidities measured by the FT‐IR is in good agreement with the observed order of catalytic activities. The source of Lewis acid activity in an IL catalyst may best be understood by considering the formation of the equilibrium mixtures that constitute the IL, as shown in Eq. (3.3), where n and m were defined by the initial AlCl3/R4NCl ratio. The precise distribution of the equilibrium mixture depends on the molar ratio of AlCl3 to the quaternary ammonium salt as well as the respective equilibrium constants between AlCl3 with Cl−, [AlCl4]−, and [Al2Cl7]−. When Al/N is equal to one, only the quaternary ammonium ions and [AlCl4]− are present. On adding more AlCl3 or making the Al/N ratio greater than one, besides [AlCl4]−, [Al2Cl7]−, and [Al3Cl10]−, start to form. At a sufficiently high Al/N ratio, [Al2Cl7]− and [Al3Cl10]− will be in equilibrium with AlCl3, and then a catalytically active amount of AlCl3 will be present in the equilibrium mixture. Accordingly, the higher the

3.4 Isomerization

Al/N ratio, the greater the amount of AlCl3 present in the equilibrium mixture. It is believed that the AlCl3 in the equilibrium mixture is primarily responsible for the Lewis acidity and catalytic activity: nAlCl 3 mR 4 NCl

oAlCl 4

pAl 2Cl 7

qAl 3Cl10

rNR 4

sAlCl 3 (3.3)

For instance, the AlCl3 in the equilibrium mixture is responsible for forming a complex with acetonitrile and generating one, and only one, new set of FT‐IR peaks. If other Lewis acids besides AlCl3 were present in the IL catalyst, then one would expect more than one new set of CN stretching absorptions to appear. The FT‐IR analysis of pure AlCl3 in acetonitrile was done and showed the absorptions at 2305 and 2335 cm−1 and confirmed that the absorption of chloroaluminate ILs contributed from AlCl3. A large molar volume of the quaternary ammonium salt, as in the case of HDMIC, leads to dilution of the AlCl3 and results in a lower FT‐ IR response and smaller relative peaks, and hence the acidity of the IL is underestimated. For this reason, the relative absorption is a good but not perfect measure of Lewis acidity and the corresponding catalyst activity. The catalytic activities of the IL catalysts prepared from pyridine hydrobromide (PHB)/AlCl3 and 1‐butyl‐3‐methylimidazolium bromide (BMIB)/AlCl3 were compared. As shown in Table 3.7, PHB/AlCl3 shows higher catalytic activity than BMIB/AlCl3, which is consistent with the activity of PHC/AlCl3 being higher than that of BMIC/AlCl3. In addition, the isomerization rate by using PHB is higher than that with PHC. And a similar trend is obtained for BMIB and BMIC. Thus, it can be concluded that the bromochloroaluminate ILs have higher acidic strengths than the chloroaluminate analogues.

Table 3.7  Effect of concentration of ILs on endo‐THDCPD isomerization.a Selectivity (%) Mole ratio of BMIC/endo‐THDCPD

1/128 1/25.6 1/12.8 1/6.4 1/1.28

Reaction time (h)

Conversion (%)

exo‐THDCPD

ADM

1

3.1

100

0

3

4.3

100

0

1

9.2

100

0

3

22.0

100

0

1

31.9

100

0

3

58.2

100

0

1

46.3

100

0

3

76.2

100

0

1

98.9

100

0

3

99.2

99.8

0.2

a  Reaction conditions: mole fraction of AlCl3/BMIC = 0.60/0.40, reaction temperature 50 °C. Source: Huang et al. (2011). Reproduced with permission of Elsevier.

77

3  Design and Synthesis of High‐Density Polycyoalkane Fuels

Table  3.7 summarizes the results relating to the isomerization of endo‐ THDCPD using different amounts of BMIC/AlCl3 catalyst with 0.60 mol fraction of AlCl3 at 50 °C. It is evident that the rate of isomerization is affected by the amount of chloroaluminate IL. Under optimized reaction conditions, high conversion of 98.9% and 100% selectivity in favor of exo‐THDCPD is achieved after reaction for one hour by using a molar ratio of BMIC/endo‐THDCPD of 1 : 1.28. Almost simultaneously, Wang et al. (2012, pp. 164–169) checked several 1‐n‐ alkyl‐3‐methylimidazolium halides and metal chlorides to catalyze the isomerization of endo‐THDCPD and found that 1‐butyl‐3‐methylimidazolium [BMIM] Cl/AlCl3 shows good performance owing to its stronger acidity. The types of acid species (anions) in IL are dependent on the mole fraction of AlCl3 (denoted as x) in IL as follows: BMIM Cl AlCl 3



AlCl 4



AlCl 3

Al 2 Cl 7



BMIM Al 2 Cl 7

AlCl 3

AlCl 4





Al 3 Cl10



Since these species have different acid strength (the order is [Al3Cl10]− > [Al2Cl7]− >  [AlCl4]−), x value obviously shows significant influence on isomerization, as shown in Figures 3.35 and 3.36. When x ≤ 0.50, anion in IL is [AlCl4]− without any acidity, and the IR vibration bands of IL/CH3CN mixture are identical to those of pure acetonitrile. Accordingly, the isomerization reaction of endo‐THDCPD does not happen. When 0.50  V; thus, the thermal preference follows the reverse order V > II > IV > I. For the CP adducts, the thermal preference is VII > VI > III. Therefore, the reactions prefer exo configurations, consistent with the proposed pathways.

II I V IV

III

VI

VII

Figure 3.38  Possible isomerization pathways of THTCPD. Source: Wang et al. (2009b). Reproduced with permission of American Chemical Society.

81

3  Design and Synthesis of High‐Density Polycyoalkane Fuels 50

M052X/6-31g(d) M052X/6-311g(d,p) B3LYP/6-31g(d)

Relative energy (kJ/mol)

I IV

40 30

III II

20

V VI

10 VII

0

Figure 3.39  The relative energy of THTCPD isomers calculated using different methods. (The total energy of VII calculated using M052X/6‐31g(d), M052X/6‐311g(d,p), and B3LYP/6‐31g(d), is −584.455305, −584.59943735, and −584.496874 a.u., respectively.) Source: Wang et al. (2009b). Reproduced with permission of American Chemical Society.

Some other experimental runs were conducted to disclose the reaction route of each isomer. First, it is found that the concentrations of I and II do not change during the reaction when the reaction temperature was set at −5 °C, indicating that the temperature is too low to isomerize. However, the isomerization of isomer III takes place easily at this temperature, and two products are formed. As shown in Figure 3.40, the lumped composition of the reactant and two products keeps a constant concentration, proving that there are no other components involved in the reaction. The concentration–time curves indicate that the reaction is a typical consecutive process with an intermediate. According to previous analysis, the intermediate is determined as isomer VI and the final product is VII. And the reaction finishes in 120 minutes and is irreversible. 14 12 Composition (%)

82

10 III VI VII III + VI + VII

8 6 4 2 0 0

20

40 60 Time (min)

80

100

Figure 3.40  Composition–time profiles for the isomerization of III (temperature: −5 °C, AlCl3 amount: 3%, solvent: 1,2‐dichloroethane, THTCPD concentration: 50%). Source: Wang et al. (2009b). Reproduced with permission of American Chemical Society.

3.4 Isomerization 75

Composition (%)

60 I IV I + IV

45 30 15 0

0

20

40

60

(a)

80 100 Time (min)

120

140

160

180

Composition (%)

8

6

II V II + V

4

2

0 0 (b)

20

40

60

80 100 Time (min)

120

140

160

180

Figure 3.41  Composition–time profiles for the isomerization of (a) I and (b) II (temperature: 0 °C, AlCl3 amount: 5%, solvent: 1,2‐dichloroethane, THTCPD concentration: 50%). Source: Wang et al. (2009b). Reproduced with permission of American Chemical Society.

When the reaction temperature was set at 0 °C, the transformation (III → VI → VII) finished in a very short time (about 10 minutes). Figure 3.41 shows that the concentrations of I and II gradually decrease, conforming that they are also transformed into other configurations. In this case, two additional components are formed, which are supposed to be isomers IV and V according to the aforementioned theoretical analysis. If the reaction proceeds via the I → II → V route, the concentration of isomer II should go up first and then decline, like the case of isomer VI. However, this phenomenon is not observed, so this route is excluded. The transformation of isomer I most likely takes place through the pathway I → IV → V. If so, intermediate IV should show a volcano‐shaped concentration−time curve. Nevertheless, neither of the newly formed compounds shows this characteristic. Actually, both newly formed compounds increase continuously during the reaction. This suggests

83

3  Design and Synthesis of High‐Density Polycyoalkane Fuels

that the expected consecutive isomerization of I does not occur. Therefore, the most possible reaction pathway is I → IV and II → V. As evidence, Figure  3.40 depicts the consumed I and II are compensated by a formed product. Although the amounts of I + IV and II + V slightly vary, they are still within the limit of the experimental error. Figure 3.41 also presents that I and II are not completely transformed, suggesting that the I → IV and II → V reactions are reversible. The experimental results indicate that the expected I → II and IV → V routes scarcely occur under the present reaction conditions. The reaction was conducted at 100 °C for a prolonged period, yet no endo to exo isomerization of the cyclopropyl fragment was observed. This suggests that the endo‐cyclopropyl fragment of THTCPD is extremely difficult to convert into the exo configuration, although it is possible in thermodynamics. It is noted that the carbocation formed on the cyclopropyl fragment may be a dead end, which may not function as the isomerization intermediate (McCaulay 1959, pp. 6437–6443). Or maybe the transformation of the cyclopropyl fragment needs an energetically unfavorable intermediate or transition state. Figure 3.42 further compares the reaction rate of the three isomerization reactions at different temperatures. The reaction of isomer III is much faster than the other two reactions. In 30 minutes, the conversion of III is 41% at −5 °C, and a 100% conversion is obtained at 0 °C. For I and II, the conversion is less than 60% at 15 °C, obviously indicating that higher temperature and longer reaction period are necessary. The unique structure of III, which contains two norbornyl rings, makes it easy to isomerize, whereas the cyclopropyl ring in I and II decelerates the transformation of the adjacent norbornyl ring. It can be seen that the cyclopropyl structure in THTCPD molecules is the major factor hindering the isomerization reaction. The influences of the reaction conditions including temperature, catalyst amount, and solvent were studied to establish a suitable operation. It seems not

90

I II III

75 Conversion (%)

84

60 45 30 15 0 –5

0

5 Temperature (°C)

10

15

Figure 3.42  Conversions of I, II, and III at different temperatures (AlCl3 amount: 3%, solvent: 1,2‐dichloroethane, THTCPD concentration: 50%, reaction time: 30 minutes). Source: Wang et al. (2009b). Reproduced with permission of American Chemical Society.

3.4 Isomerization

necessary to study the three isomers separately, so they are regarded as whole. First, an experimental run was conducted at the highest temperature in the chosen range (50 °C) and the highest catalyst amount (8%). Results indicate that there are no by‐products. Hence, the selectivity of the isomerization reaction is regarded as 100%. Figure  3.43 presents the influence of temperature on the reaction. It can be seen that the reaction rate is accelerated with the temperature increasing and the time to reach equilibrium is shortened. It takes more than three hours to reach equilibrium when the temperature is below 15 °C. The time is decreased to 90 minutes when the temperature is higher than 40 °C. However, the equilibrium conversion is also reduced. Especially, the equilibrium conversion is dramatically lowered at 50 °C. This is because the equilibrium constant of the exothermic reaction decreases with the increase of temperature. Figure  3.44 presents the effect of the catalyst amount on the reaction. The reaction rate is increased with the increasing amount of AlCl3, but the equilibrium is not influenced. The equilibrium is reached within 160 minutes when the catalyst amount increases to 5%. Because this reaction time is acceptable for practical application, it seems not necessary to further increase the catalyst amount. The acid‐catalyzed isomerization of alkanes proceeds via an ionic chain mechanism, in which the solvent may strongly influence the reaction. Figure 3.45 compares the influence of some solvents. With the absence of solvent, no isomerization occurred even at 100 °C. The conversions are extremely low when non‐halide solvents such as cyclohexane with weak polarity and ethyl acetate with medium polarity are used, so the polarity of the solvent seems not a significant factor for the reaction. AlCl3 is hardly soluble in cyclohexane, ethyl acetate, and THTCPD itself. Solid AlCl3 is not a strong Lewis acid; thus, a suitable solvent for AlCl3 is very necessary. Halohydrocarbons can greatly improve the reaction due to the ability to dissolve AlCl3. Specifically, 1,2‐dichloroethane (EC) is the most active one among the solvents 100

Conversion (%)

80 60 0 °C 15 °C 30 °C 40 °C 50 °C

40 20 0

0

20 40

60

80 100 120 140 160 180 200 220 Time (min)

Figure 3.43  Effect of temperature on the isomerization of THTCPD (AlCl3 amount: 5%, solvent: 1,2‐dichloroethane, THTCPD concentration: 50%). Source: Wang et al. (2009b). Reproduced with permission of American Chemical Society.

85

3  Design and Synthesis of High‐Density Polycyoalkane Fuels 90

Conversion (%)

75 60

3% 5% 8%

45 30 15 0

0

20

40

60

80 100 120 140 160 Time (min)

180

Figure 3.44  Effect of the AlCl3 amount on the isomerization of THTCPD (temperature: 15 °C, solvent: 1,2‐dichloroethane, THTCPD concentration: 50%). Source: Wang et al. (2009b). Reproduced with permission of American Chemical Society.

VI

90 75 Conversion (%)

86

V

60 45 30 15 I

II

III

IV

0

Figure 3.45  Effect of solvent on the isomerization of THTCPD (temperature: 15 °C, THTCPD concentration: 50%): I, cyclohexane; II, ethyl acetate; III, chlorobenzene; IV, toluene; V, chloroform; VI, 1,2‐dichloroethane. Source: Wang et al. (2009b). Reproduced with permission of American Chemical Society.

used owing to the well dissolution of AlCl3. Besides, this solvent may be beneficial to form the carbocation transition state. Figure 3.46 depicts the effect of the concentration of THTCPD on the reaction. There is an optimal concentration for the highest equilibrium conversion. On one hand, the reaction requires enough solvent to dissolve the catalyst and stabilize carbocation. On the other hand, a low reactant concentration lowers the reaction rate. Therefore, a balance between the amount of solvent and the concentration of reactant should be determined for the reaction. Therefore, the optimal reaction conditions are as follows: temperature of 15 °C, AlCl3 amount of 5% by weight, solvent of 1,2‐dichloroethane, and THTCPD

3.4 Isomerization

Conversion (%)

80

60 33% 50% 67% 75%

40

20

0

0

20

40

60

80 100 120 140 Time (min)

160

180

Figure 3.46  Effect of the reactant concentration on the isomerization of THTCPD (temperature: 15 °C, AlCl3 amount: 5%). Source: Wang et al. (2009b). Reproduced with permission of American Chemical Society.

c­ oncentration of 50%. And an equilibrium conversion of 87% can be obtained in 160 minutes. The activation energies of the forward and reverse reactions are 18.27 and 41.21 kJ/mol for I → IV and 11.95 and 29.64 kJ/mol for II → V, ­respectively. The starting material is solid with a melting point of 35 °C. After isomerization, the resultant mixture is a colorless and transparent liquid with a freezing point below −40 °C. The density of the mixture is 1.04 g/ml, and the volumetric energy content is 44.1 MJ/l, which are both higher than those of JP‐10. Chloroaluminate IL catalyst has been also used for the isomerization of THTCPD owing to its adjustable acidity, non‐combustibility, and ease of recycling. Kim et al. conducted the isomerization of THTCPD by using three chloroaluminate ILs including TEAC, BMIC, and PHC (Kim et  al. 2014b, pp. 109–114). Figure  3.47 shows the effect of reaction time on the isomerization of THTCPD. The product of THTCPD isomerization using various IL catalysts consists of diamondoids as well as exo‐THTCPD. The conversions and yields of exo‐THTCPD using PHC and TEAC‐based IL catalysts, respectively, drastically increase and reach conversions of 100% as a function of reaction time. The isomerization reaction progressed slowly in the case of the BMIC‐based IL catalyst. This might be due to the higher Lewis acidic character of the PHC and TEAC‐based catalysts compared with those of the BMIC‐based catalyst. Among the IL catalysts used, the PHC‐based IL catalyst showed the best catalytic performance in terms of conversion and exo/endo ratio. Commercial zeolites, which can be served as efficient catalysts for the isomerization of DCPD and THDCPD, have been recently investigated for the isomerization of THTCPD by Cho et al. (2018, pp. 399–406). And it is concluded that the catalytic performances were significantly affected by the microchannel or pore size, acid strength, and acid type of each catalyst. The NH3‐TPD (temperature programmed desorption) profiles of several aluminosilicate catalysts are shown in Figure 3.48. The total amount of desorbed NH3 corresponds to the total acid amounts of each catalyst, and the desorbed peak position corresponds to the acid strength of each catalyst. The total acid amount of these catalysts follow the order ZSM‐5(40)

87

3  Design and Synthesis of High‐Density Polycyoalkane Fuels 140

140 [TEAC]/AlCl3 conversion [PHC]/AlCl3 conversion

100

100 Yield (%)

120

[TEAC]/AlCl3 exo-THTCPD

80

[BMlC]/AlCl3 exo-THTCPD

80

[PHC]/AlCl3 exo-THTCPD

60

60

40

40 [TEAC]/AlCl3 diamondoid

20

[BMlC]/AlCl3 diamondoid [PHC]/AlCl3 diamondoid

0 0

1

2

3

6 Time (h)

9

16

Conversion (%)

[BMCl]/AlCl3 conversion

120

20 0

24

Figure 3.47  Effect of reaction time on THTCPD isomerization. Reaction condition: reaction temperature: 80 °C, 1,2‐dichloroethane (EC): reactant ratio 1.5, IL content: 150 wt%, x = 0.67 (reactant: 20 g, EC: 30 g). Source: Kim et al. (2014b). Reproduced with permission of Elsevier.

Al-MCM-41 HY(30) 2Cu-HY(30) MOR ZSM-5(40)

Intensity (a.u.)

88

100

200

300

400

500

600

Temperature (°C)

Figure 3.48  NH3‐TPD profiles of several aluminosilicate catalysts. Source: Cho et al. (2018). Reproduced with permission of Elsevier.

(0.19 mmol NH3/g) > HY(30) (0.15 mmol NH3/g) > MOR (0.12 mmol NH3/g) > Al‐ MCM‐41 (0.11 mmol NH3/g). When Cu was loaded on HY, the amount of acid was slightly increased (0.15 mmol NH3/g → 0.17 mmol NH3/g), but the desorption peak corresponding to strong acid strength disappeared. As for the peak positions of NH3 desorption of each catalyst, ZSM‐5(40) was much higher than HY(30), and the former had a stronger acid strength. In the cases of HY(30) and ZSM‐5(40), two

3.4 Isomerization

peak positions were observed, corresponding to the weak (low temperature peak) and strong (high temperature peak) acid strengths, respectively, whereas the MOR, Al‐MCM‐41, and 2Cu‐HY(30) only had one peak position (low‐temperature peak) corresponding to the weak acid strength. The conversion rate of endo‐THTCPD and the exo‐THTCPD yield over these aluminosilicate catalysts are presented in Figure 3.49, and the sequence of conversion rate follows the order HY(30) > Al‐MCM‐41 > MOR > ZSM‐5(40). Considering the pore diameters and THTCPD molecule size (Figure 3.50), it was found that the catalyst performance strongly depends on its pore dimensions because of the molecule diffusion effect. Only some catalysts (HY(30) and Al‐MCM‐41), which had micro‐ or meso‐channel sizes over the specific dimensions, 0.74 × 0.74 nm, the size of a THTCPD molecule, exhibit high conversion rates. However, despite the 100

Conversion/yield (%)

90

100

endo-THTCPD conversion exo-THTCPD yield

90

80

80

70

70

60

60

50

50

40

40

30

30 33

20 10 0

2

ZSM-5(40)

9

MOR

10

14

Al-MCM-41 Catalysts

20

HY(30)

0

Figure 3.49  endo‐THTCPD conversion and exo‐THTCPD yield over ZSM‐5(40), MOR, Al‐ MCM‐41, and HY(30) (reaction temperature: 275 °C; reaction time: 6 hours). Source: Cho et al. (2018). Reproduced with permission of Elsevier.

0.76 nm

Figure 3.50  The molecular size of THTCPD measured by Chem3D pro8.0. Source: Cho et al. (2018). Reproduced with permission of Elsevier.

89

3  Design and Synthesis of High‐Density Polycyoalkane Fuels

smaller pore size than Al‐MCM‐41, the HY(30) catalyst showed the highest conversion rate and yield. The HY catalyst had two types of NH3‐desorbed peaks, which could be interpreted as weak and strong acid characteristics, respectively. Otherwise, Al‐MCM‐41 and MOR had only one peak corresponding to weak acid characteristics. Although the ZSM‐5(40) catalyst had lots of strong acid sites, as shown in Figure 3.48, its small microchannel size (0.54 × 0.56 nm) did not allow the THTCPD molecules to approach the active sites of the inner pores, resulting in the lowest activity.

3.5 ­Other Reactions and Procedures 3.5.1  Alternative Isomerization–Hydrogenation Synthesis Natural DCPD has two stereoisomers, endo and exo isomer, with the former about 99.5% and the latter only 0.5%. Hence, the synthesis of exo‐THDCPD (JP‐10) usually involves two steps: first, hydrogenation of endo‐DCPD to yield a solid endo‐THDCPD and then catalytic isomerization to produce exo‐THDCPD in the liquid phase. Recently, it was found that endo‐DCPD can be isomerized to exo‐DCPD. In this way, the resulted exo‐DCPD can be hydrogenated to JP‐10 directly (Eq.  (3.4)). For example, Zhang et  al. investigated the isomerization of endo‐DCPD by thermal treatment at evaluated temperature and pressure (Zhang et al. 2007b, pp. 673–676). Figure 3.51 presents the effect of temperature on the isomerization reaction. At low temperatures like 100 °C, no isomerization reaction occurs. Because the first step of the isomerization reaction is the dissociation of endo‐DCPD molecules, which needs higher temperature. As increasing the temperature, the conversion of endo‐DCPD increases quickly. Meanwhile the 100

30

60 20 40 10

20

0

100

120

140 160 Temperature (°C)

180

200

Yield or selectivity (%)

40 X (endo-DCPD) Y (exo-DCPD) S (exo-DCPD)

80 Conversion (%)

90

0

Figure 3.51  Effect of reaction temperature on the isomerization of endo-DCPD. Pressure: 4 MPa, reaction time: 13.2 minutes, X: conversion, Y: yield, S: selectivity. Source: Zhang et al. (2007b). Reproduced with permission of Elsevier.

3.5  Other Reactions and Procedures 80 150°C X (endo-DCPD) 150°C S (exo-DCPD)

Conversion (%)

75

160°C X(endo-DCPD) 160°C S(exo-DCPD)

60

60

40

45 30

Selectivity (%)

90

20 15 0 0

1

2 3 Pressure (MPa)

4

5

0

Figure 3.52  Effect of reaction pressure on the isomerization of endo-DCPD (reaction time: 13.2 minutes). Source: Zhang et al. (2007b). Reproduced with permission of Elsevier.

highest selectivity (30.3%) for exo isomers is achieved at 140 °C, and the maximum yield (18.3%) is obtained at 160 °C. The amount of oligomers increases gradually with a further increment of temperature. Nonetheless, when the temperature is above 180 °C, exo‐DCPD will be reversibly decomposed into CPD, resulting a dramatically decrease of the selectivity and yield of exo‐DCPD. Figure 3.52 shows the effect of pressure on the reaction. There is no obvious reaction when the pressure is below 0.5 MPa. The reaction is maximized with the pressure as 2 MPa after which the reaction is stable, while the selectivity of exo‐DCPD slightly declines. Since the vapor pressure of CPD is higher than that of endo‐DCPD (Palmová et al. 2001, pp. 927–935), sufficient pressure is necessary to maintain the sufficient concentration of CPD in the solution to facilitate the recombination of CPD to exo‐DCPD.

Isomerization

Hydrogenation

 (3.4) Acidic zeolites like Hβ, HY, HUSY, HZSM‐5, and mesoporous Al‐MCM‐41 can also catalyze the isomerization of endo‐DCPD (Deng et al. 2015, pp. 540– 546; Han et al. 2009, pp. 84–88; Zou et al. 2012, pp. 79–85). For instance, Han et al. (2009, pp. 84–88) reported the isomerization of endo‐ to exo‐DCPD in the liquid phase over several commercial zeolites and found that acidic zeolites effectively facilitate the isomerization reaction. Among the zeolites tested, the activity order is Hβ > HY > HUSY > HZSM‐5 ≈ H‐mordenite. β and Y‐type zeolites exhibit higher activity because of their large three‐dimensional channels. The surface passivation of Hβ confirms that the reaction proceeds in the inner

91

3  Design and Synthesis of High‐Density Polycyoalkane Fuels

channels. These results suggest that the porous structure is critical for the catalytic isomerization of endo‐DCPD. Thermogravimetric measurements verify that HY is deactivated very quickly because the strong acidity induces serious coke formation, but this phenomenon is much less over β zeolites. Evaluation of Hβ with different SiO2/Al2O3 ratios indicates that both the weak Lewis acid concentration and isomerization activity of zeolites monotonically decrease with the SiO2/Al2O3 ratio. Thus, a weak acid of Hβ, especially a weak Lewis acid, is responsible for the isomerization reaction. The Hβ calcinated at 500 °C provides the highest activity due to the complete removal of template residues, generation of large amounts of weak Lewis acid, and good crystal structure. The Hβ concentration was also studied: the optimal value is 15 wt%. Zeolites deactivated due to coke deposition can be regenerated by calcination in air flow at 500 °C, and there was no obvious activity loss after four times’ regeneration. Then, Zou et al. (2012, pp. 79–85) synthesized Al‐grafted mesoporous MCM‐41 with different Si/Al ratios, which shows higher activity in the isomerization of endo‐DCPD and better coke tolerance capability. There are two active C═C bonds in DCPD molecule, so some cycloaddition side reactions will occur along with the isomerization. One is the [4+2] Diels–Alder cycloaddition of DCPD with its in situ cracked monomer CPD to generate TCPD. The second is the further [4+2] addition of TCPD with CPD to form tetracyclopentadiene (TeCPD). Direct [2+2] cycloaddition of two DCPD molecules is also a possible pathway toward TeCPD. Under thermal reaction conditions, only the [4+2] Diels–Alder cycloaddition occurs, and the formation of exo‐DCPD is negligible. In the presence of an acidic catalyst, a considerable amount of exo‐DCPD is formed. Figure 3.53 compares the activity of Al‐MCM‐41(30) and Hβ with approximately equal Si/Al ratio of 70 60 Conversion and selectivity (%)

92

endo-DCPD conversion exo-DCPD selectivity

50 40 30 20 10 0 No catalyst

Hβ

Al-MCM-41

Figure 3.53  Comparison of Al‐MCM‐41(30) and Hβ in catalytic isomerization. Reaction condition: catalyst 5 wt%, no solvent temperature 150 °C, time 3 hours. Source: Zou et al. (2012). Reproduced with permission of Elsevier.

3.5  Other Reactions and Procedures

30. Al‐MCM‐41 (30) shows higher activity than Hβ, with the conversion and selectivity increasing by 60% and 52%, respectively. The better performance of Al‐MCM‐41 could be attributed to its mesoporous structure that allows the reactant and products (including by‐products with large molecular size) to diffuse freely and provides better tolerance for coke deposition. The isomerization of endo‐DCPD occurs over weak Lewis acid sites, whereas moderate Lewis acid sites lead to the [2+2] cycloaddition of two DCPD molecules. As illustrated in Figure 3.54, the amount of Lewis acid is dominated by Al content. The sample with Si/Al ratio of 8 (Al concentration of 1.24 mmol/g) shows the highest activity as shown in Figure 3.55, due to the maximum amount of weak Lewis acid. The homogeneous [4+2] cycloaddition between DCPD and CPD can be avoided by choosing low reaction temperatures. Adding inert solvent and properly increasing the catalyst dosage can further improve the isomerization, with the highest conversion of 70% and selectivity of 85% obtained. The catalyst shows acceptable stability, and the activity can be easily restored by calcination. ZSM‐5 is one of the most well‐known acidic catalysts, which has been widely used in the petroleum industry because of the strong acidity and low price. Unfortunately, as abovementioned, its activity in DCPD isomerization and cycloaddition is very low, because the reaction can take place only on the external surface due to the diffusion limitation. Thus, it is expected that generating a mesoporous structure may make ZSM‐5 applicable for the isomerization and cycloaddition with a low cost. Accordingly, Deng et al. (2015, pp. 540–546) prepared hierarchical porous HZSM‐5 by using the alkali‐treatment method and obtained high activity and selectivity in both the isomerization and cycloaddition of endo‐DCPD, increasing the product types and improving the low‐­temperature 100 Weak Moderate

Lewis acid amount (µmol/g)

80

Total

60

40

20

0

0.0

0.5

1.0

1.5

2.0

Al concentration (mmol/g)

Figure 3.54.  Distribution of Lewis acid sites of Al‐MCM‐41. Source: Zou et al. (2012). Reproduced with permission of Elsevier.

93

3  Design and Synthesis of High‐Density Polycyoalkane Fuels 90 exo-DCPD TeCPD

Conversion and yield (%)

75

Conversion

60

45

30

15

0 0.0

1.0 1.5 0.5 Al concentration (mmol/g)

2.0

Figure 3.55  The activity of Al‐MCM‐41 with different Si/Al ratios. Reaction condition: catalyst: 15 wt%, reactant concentration: 30 wt%, temperature: 140 °C, time: 5 hours. Source: Zou et al. (2012). Reproduced with permission of Elsevier.

properties of the obtained fuel. As demonstrated by the time‐dependent product distribution in Figure  3.56, in the beginning, the exo‐DCPD yield gradually increases, reaching the maximum after four hours and then decreases, whereas the TeCPD yield continues to go up. This indicates that TeCPD is formed through the catalytic [2+2] cycloaddition of exo‐DCPD. Figure 3.57 shows the activities of 80

Conversion and yield (%)

94

Conversion exo-DCPD TeCPD

60

40

20

0

0

1

2 3 Reaction time (h)

4

5

Figure 3.56  Products distribution of isomerization using hierarchical porous HZSM‐5 (Z5–70–0.5–1.0). Reaction conditions: catalyst: 30 wt%, solvent: 67 wt%, temperature: 130 °C. Source: Deng et al. (2015). Reproduced with permission of Elsevier.

3.5  Other Reactions and Procedures

Conversion and yield (%)

80

Conversion exo-DCPD TeCPD

Al-MCM-41 Z5–70–0.5–1.0

60 Hβ 40

Z5 20

0

Figure 3.57  The activity of various zeolites in catalytic isomerization. Reaction conditions: catalyst: 30 wt%, solvent: 67 wt%, temperature: 130 °C, time: 4 hours. Source: Deng et al. (2015). Reproduced with permission of Elsevier.

several zeolites in the isomerization reaction. Hβ is more active than the parent HZSM‐5, and mesoporous Al‐MCM‐41 shows the highest activity. However, the exo‐TCPD yield over Al‐MCM‐41(42.1%) is just a little higher than that over Hβ (35.3%), because many exo‐DCPD molecules are further converted to TeCPD through [2+2] cycloaddition. This suggests that mesopores favor not only the isomerization but also the cycloaddition and a balance is necessary to obtain optimal isomerization yield and selectivity. The hierarchical HZSM‐5 shows obviously higher isomerization yield (50.6%), although the conversion is not the highest. Therefore, suitable mesopores can make the active sites in the micropores accessible for isomerization and suppress the cycloaddition process indicating the synergetic effect of mesopores and micropores in mass diffusion and shape selectivity. 3.5.2  One‐Step Synthesis of exo‐Tetrahydrodicyclopentadiene The above two‐step synthesis of JP‐10 via hydrogenation and isomerization (Eq. (3.4)) involves two kinds of catalysts or two processes. It is possible to combine the synthesis in one step if a bifunctional catalyst can be developed. For example, Wang et  al. (2013, pp. 6339–6347) reported a one‐step continuous‐ flow‐phase hydrogenation–isomerization of DCPD to exo‐THDCPD over Ni‐ supported catalysts in a fixed‐bed reactor. For each run, the fixed‐bed reactor‐I (part 14, as shown in Figure 3.58) and the fixed‐bed reactor‐II (part 21, as shown in Figure 3.58) were initially heated to the desired reaction temperature (T1 and/ or T2), followed by heating of the preheater to the desired temperature (T3 at 80 °C). T3 was generally lower than T1 and T2 in the present study. T1 and T2 were measured using a thermocouple inserted into the fixed‐bed reactor‐I and fixed‐ bed reactor‐II. The catalysts combine functions of Ni/γ‐Al2O3 (for h ­ ydrogenation) and Hβ (for isomerization), and the process displays a high efficiency and low environmental impact and provides 100% conversion of DCPD and an ­average

95

3  Design and Synthesis of High‐Density Polycyoalkane Fuels 6

4

6 7

8

5

11

10

5

1

15

12

5

13

14 15

15 6 7 4

8

21

6

5

16

5 5

2

00 17

9 19

4

18 5 5 3

Figure 3.58  Schematic diagram of the configuration for the one‐step, continuous‐flow catalytic hydrogenation–isomerization process. 1, H2 cylinder; 2, N2 cylinder; 3, liquid collector; 4, filter; 5, on–off valve; 6, pressure gauge; 7, mass‐flow gas meter; 8, one‐way valve; 9, gas mixer; 10, safety valve; 11, preheater; 12, micro constant‐flow pump; 13, liquid storage tank; 14, fixed‐bed reactor‐I; 15, temperature controller; 16, condenser tube; 17, condensation pump; 18, gas–liquid separator; 19, one‐way valve; 20, back‐pressure regulator; and 21, fixed‐bed reactor‐ II. Source: Wang et al. (2013). Reproduced with permission of American Chemical Society.

100

100

80

80

60

60

40

40

20

20

0

0

25

50

75 100 125 150 175 200 225 250 Time-on-stream (h)

DCPD conversion (%)

exo-THDCPD selectivity (%)

96

0

Figure 3.59  The selectivity for exo‐THDCPD and DCPD conversion for the one‐step continuous hydrogenation/isomerization of DCPD over 25 wt% Ni/γ‐Al2O3 and 15 wt% Ni/Hβ catalysts in a fixed‐bed reaction system under the conditions of P, 1.0 MPa; H2/DCPD, 10; LHSV, 1 h−1; and average T, 150 °C. Source: Wang et al. (2013). Reproduced with permission of American Chemical Society.

­  References

yield of 70% of exo‐THDCPD for 200 hours (Figure 3.59). This novel route has potential applications in the production of JP‐10 fuel on an industrial scale.

­References Behr, A. and Keim, W. (1985). Novel trimerization of cyclopentadiene with a homogeneous, bifunctional palladium‐acid catalyst system. Angewandte Chemie International Edition 24: 314–315. Cho, Y.H., Kim, C.H., Lee, S.H. et al. (2018). Isomerization of endo‐ to exo‐ tetrahydrotricyclopentadiene over alumino‐silicate catalysts. Fuel 221: 399–406. Chou, S.H., Chen, S.C., Tan, C.S. et al. (1997). Hydrogenation of dicyclopentadiene in trickle‐bed reactor. Journal of the Chinese Institute of Chemical Engineers 28: 175–182. Chung, H.S., Chen, C.S.H., Kremer, R.A. et al. (1999). Recent developments in high‐energy density liquid hydrocarbon fuels. Energy Fuels 13 (3): 641–649. Cristol, S.J., Seifert, W.K., and Soloway, S.B. (1960). Bridged polycyclic compounds. X. The synthesis of endo and exo‐1,2‐dihydrodicyclopentadienes and related compounds. Journal of the American Chemical Society 82 (9): 2351–2356. Deng, Q., Zhang, X.W., Wang, L. et al. (2015). Catalytic isomerization and oligomerization of endo‐dicyclopentadiene using alkali‐treated hierarchical porous HZSM‐5. Chemical Engineering Science 135: 540–546. Edwards, T. (2003). Liquid fuels and propellants for aerospace propulsion: 1903–2003. Journal of Propulsion and Power 19 (6): 1089–1107. Gunter, Z. (1968). Dimerization process. US Patent 3,377,398, filed 15 April 1966 and issued 9 April 1968. Han, H., Zou, J.‐J., Zhang, X.W. et al. (2009). endo‐ to exo‐Isomerization of dicyclopentadiene over zeolites. Applied Catalysis A: General 367 (1–2): 84–88. Hirooka, S., and Torii, M. (1985). Fuel composition. US Patent 4,507,516, filed 14 March 1984 and issued 26 March 1985. Hitosh, Y., and Matsuno, M. (1988). High‐density liquid fuel. US Patent 4,762,092, filed 24 February 1987 and issued 9 August 1988. Huang, M.Y., Wu, J.C., Shieu, F.S. et al. (2010). Isomerization of endo‐ tetrahydrodicyclopentadiene over clay‐supported chloroaluminate ionic liquid catalysts. Journal of Molecular Catalysis A: Chemical 315 (1): 69–75. Huang, M.Y., Wu, J.C., Shieu, F.S. et al. (2011). Preparation of high energy fuel JP‐10 by acidity‐adjustable chloroaluminate ionic liquid catalyst. Fuel 90 (3): 1012–1017. Kim, J., Han, J., Kwon, T.S. et al. (2014a). Oligomerization and isomerization of dicyclopentadiene over mesoporous materials produced from zeolite beta. Catalysis Today 232: 69–74. Kim, S.G., Han, J., Jeon, J.K. et al. (2014b). Ionic liquid‐catalyzed isomerization of tetrahydrotricyclopentadiene using various chloroaluminate complexes. Fuel 137: 109–114. Krishnamachary, S., Mohan, S.K., Desingou, J. et al. (2014). Polycyclic alkanes based high density hydrocarbon fuels preparation and evaluation for LFRJ application. RSC Advances 4 (85): 45407–45414.

97

98

3  Design and Synthesis of High‐Density Polycyoalkane Fuels

Li, Y.H., Zou, J.‐J., Zhang, X.W. et al. (2010). Product distribution of tricyclopentadiene from cycloaddition of dicyclopentadiene and cyclopentadiene: a theoretical and experimental study. Fuel 89: 2522–2527. Liu, G.Z., Mi, Z.T., Wang, L. et al. (2005). Kinetics of dicyclopentadiene hydrogenation over Pd/Al2O3 catalyst. Industrial and Engineering Chemistry Research 44: 3846–3851. Liu, G.Z., Mi, Z.T., Wang, L. et al. (2006). Hydrogenation of dicyclopentadiene into endo‐tetrahydrodicyclopentadiene in trickle bed reactor: experiments and modeling. Industrial & Engineering Chemistry Research 45 (26): 8807–8814. Liu, G.Z., Zhang, X.W., Wang, L. et al. (2008). Unsteady‐state operation of trickle‐ bed reactor for dicyclopentadiene hydrogenation. Chemical Engineering Science 63 (20): 4991–5002. McCaulay, D.A. (1959). Mechanism of acid‐catalyzed isomerization of the hexanes. Journal of the American Chemical Society 81 (24): 6437–6443. Norton, R.V., Frank, P.J. and Fisher, D.H. (1981). Process for preparing high density fuels. US Patent 4,277,636, filed 11 August 1980 and issued 7 July 1981. Palmová, I., Kosek, J., Schöngut, J. et al. (2001). Experimental and modeling studies of oligomerization and copolymerization of dicyclopentadiene. Chemical Engineering Science 56 (3): 927–935. Park, E., Yim, J.H., Han, J. et al. (2016). Synthesis of tricyclopentadiene from dicyclopentadiene over nanoporous hybrid material. Journal of Nanoscience and Nanotechnology 16 (9): 9263–9267. Rajashekharam, M.V., Jaganathan, R., and Chaudhari, R.V. (1998). A trickle‐bed reactor model for hydrogenation of 2,4 dinitrotoluene: experimental verification. Chemical Engineering Science 53: 787–805. Sibi, M.G., Singh, B., Kumar, R. et al. (2012). Single‐step catalytic liquid‐phase hydroconversion of DCPD into high energy density fuel exo‐THDCPD. Green Chemistry 14 (4): 976–983. Song, J.J., Huang, Z.F., Pan, L. et al. (2015). Oxygen‐deficient tungsten oxide as versatile and efficient hydrogenation catalyst. ACS Catalysis 5 (11): 6594–6599. Wang, L., Zhang, X.W., Zou, J.‐J. et al. (2009a). Synthesis and blending of high‐ density hydrocarbon fuels with density beyond 1.0 g cm−3. Chinese Journal of Energetic Materials 17 (2): 157–160. Wang, L., Zhang, X.W., Zou, J.‐J. et al. (2009b). Acid‐catalyzed isomerization of tetrahydrotricyclopentadiene: synthesis of high‐energy‐density liquid fuel. Energy Fuels 23 (5): 2383–2388. Wang, L., Zou, J.‐J., Kong, J. et al. (2009c). Hydrogenation of tricyclopentadiene to prepare high‐energy‐density fuel: reaction pathway. CIESC Journal 60: 912–917. Wang, L., Zou, J.‐J., Zhang, X.W. et al. (2012). Isomerization of tetrahydrodicyclopentadiene using ionic liquid: green alternative for Jet Propellant‐10 and adamantane. Fuel 91 (1): 164–169. Wang, W., Chen, J.G., Song, L.P. et al. (2013). One‐step, continuous‐flow, highly catalytic hydrogenation–isomerization of dicyclopentadiene to exo‐ tetrahydrodicyclopentadiene over Ni‐supported catalysts for the production of high‐energy‐density fuel. Energy Fuels 27 (11): 6339–6347.

­  References

Wang, Y.L., Luo, G.H., Xu, X. et al. (2014). Preparation of supported skeletal Ni catalyst and its catalytic performance on dicyclopentadiene hydrogenation. Catalysis Communications 53: 15–20. Woo, J.O., Park, J.E., Han, J. et al. (2014). Selective synthesis of tricyclopentadiene from dicyclopentadiene with homogeneous Pd catalysts. Applied Organometallic Chemistry 28 (3): 151–155. Wu, R., Zhang, J.F., Shi, Y.M. et al. (2015). Metallic WO2‐carbon mesoporous nanowires as highly efficient electrocatalysts for hydrogen evolution reaction. Journal of the American Chemical Society 137 (122): 6983–6986. Xing, E.H., Zhang, X.W., Wang, L. et al. (2007). Greener synthesis route for Jet Propellant‐10: the utilization of zeolites to replace AlCl3. Green Chemistry 9 (6): 589–593. Xiong, Z.Q., Mi, Z.T., and Zhang, X.W. (2005a). Study on the oligomerization of cyclopentadiene and dicyclopentadiene to tricyclopentadiene through Diels– Alder reaction. Reaction Kinetics and Catalysis Letters 85 (1): 89–97. Xiong, Z.Q., Mi, Z.T., Zhang, X.W. et al. (2005b). Development of synthesized high‐ density hydrocarbon fuels. Progress in Chemistry 17 (2): 359–367. Yang, Y.L. and Kou, Y. (2004). Determination of the Lewis acidity of ionic liquids by means of an IR spectroscopic probe. Chemical Communications (2): 226–227. Zhang, X.W., Miao, Q., Zou, J.‐J. et al. (2007a). Zeolite catalytic isomerization of endo‐THDCPD to exo‐THDCPD. Journal of Chemical Industry and Engineering 58: 3059–3063. Zhang, X.W., Jiang, K., Jiang, Q. et al. (2007b). Novel endo‐ to exo‐isomerization of dicyclopentadiene. Chinese Chemical Letters 18 (6): 673–676. Zhang, X.W., Jiang, K., Zou, J.‐J. et al. (2007c). Continuous oligomerization of dicyclopentadiene at elevated pressure for synthesis of high‐energy‐density fuel. Journal of Chemical Industry and Engineering 58 (10): 2658–2663. Zhang, X.W., Jiang, Q., Xiong, Z.Q. et al. (2008). Diels–Alder addition of dicyclopentadiene with cyclopentadiene in polar solvents. Chemical Research in Chinese Universities 24 (2): 175–179. Zhang, X.W., Pan, L., Wang, L. et al. (2018). Review on synthesis and properties of high‐energy‐density liquid fuels: hydrocarbons, nanofluids and energetic ionic liquids. Chemical Engineering Science 180 (28): 95–125. Zou, J.‐J., Xiong, Z.Q., Wang, L. et al. (2007a). Preparation of Pd‐B/γ‐Al2O3 amorphous catalyst for the hydrogenation of tricyclopentadiene. Journal of Molecular Catalysis A: Chemical 271 (1–2): 209–215. Zou, J.‐J., Xiong, Z.Q., Zhang, X.W. et al. (2007b). Kinetics of tricyclopentadiene hydrogenation over Pd‐B/γ‐Al2O3 amorphous catalyst. Industrial and Engineering Chemistry Research 46 (13): 4415–4420. Zou, J.‐J., Zhang, X.W., Kong, J. et al. (2008). Hydrogenation of dicyclopentadiene over amorphous nickel alloy catalyst SRNA‐4. Fuel 87 (17–18): 3655–3659. Zou, J.‐J., Xu, Y., Zhang, X.W. et al. (2012). Isomerization of endo‐dicyclopentadiene using Al‐grafted MCM‐41. Applied Catalysis A: General 421: 79–85.

99

101

4 Design and Synthesis of High‐Density Diamondoid Fuels Lun Pan and Jiawei Xie Tianjin University, Key Laboratory of Advanced Fuel and Chemical Propellant of the Ministry of Education, Key Laboratory for Green Chemical Technology of the Ministry of Education, Department of Chemical Technology, School of Chemical Engineering and Technology, 92 Weijin Road, Tianjin 300072, China

4.1 ­Introduction The diamond consisting of repeating units of 10 carbon atoms forms a beautiful three‐dimensional array and has fascinated chemists since 1913 (Bragg and Bragg 1913, p. 557; Fort and Schleyer 1964, p. 277). Besides, its weight in carats, the color, the perfection of the stone, and its value as a material for chemical research are shining in recent decades (Schwertfeger, Fokin, and Schreiner 2008, p. 1022). Later, it was recognized that a series of saturated hydrocarbons (e.g. adamantane, diamantane, and triamantane, generally called diamondoids) have the subunit excised from the diamond crystal lattice and the dangling carbon bonds are terminated with hydrogen (HáIa, Landa, and Hanuš 1966, p. 1045; Marchand 2003, p. 52). Initially, adamantane (from the Greek for diamond) was first isolated from the petroleum of Hodonin in Czechoslovakia (Fort and Schleyer 1964, p. 277). Since its uniquely high melting point, a single compound of adamantane could be crystallized by cooling the fractionated petroleum steam distillates to low temperature (Fort and Schleyer 1964, p. 277). Coincidently, diamantane (formerly called congressane) was found in the high‐boiling fractions of the crude oil of Hodonin as well (Dahl, Liu, and Carlson 2003, p. 96; HáIa, Landa, and Hanuš 1966, p. 1045). The synthesis of these diamondoids remained stagnant until Schleyer (1957, p. 3292) first reported the rearrangement of tetrahydrodicyclopentadiene (THDCPD) to adamantane using aluminum bromide or aluminum chloride with reflux overnight as well as diamantane. It also has been achieved at Princeton in 1965 by aluminum halide‐catalyzed isomerization of a mixture of norbornene [2+2] photodimers (Cupas, Schleyer, and Trecker 1965, p. 917). Unfortunately, adamantane and diamantane (including other diamondoid homologs) suffer from their high melting point (>200 °C), making it unsuitable as a significant component of jet fuels. Notably, the introduction of alkyl chains on the diamondoid core is of great interest due to the fact that the alkyl High-Energy-Density Fuels for Advanced Propulsion: Design and Synthesis, First Edition. Ji-Jun Zou, Xiangwen Zhang, and Lun Pan. © 2020 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2020 by Wiley-VCH Verlag GmbH & Co. KGaA.

102

4  Design and Synthesis of High‐Density Diamondoid Fuels

diamondoids, possessing satisfactory viscosity characteristics and high thermal stability, are showing great potential as a feasible component of jet fuels (Harvey et al. 2016, p. 10171). Hence, the research on synthetic approaches to alkyl diamondoids has continued so far. The alkyl diamondoids with high energy density, superior low‐temperature fluidity, and thermal stability will open up brand‐new fields in fuel chemistry (Chung et al. 1999, p. 641; Heneghan et al. 1996, p. 170; Zhang et al. 2018, p. 95).

4.2 ­Synthesis of Alkyl Diamondoids via Acid‐ Catalyzed Rearrangement It was found that many saturated polycyclic hydrocarbons with 10 or more carbon atoms rearrange to the thermodynamically most stable diamondoid isomers. Generally, the driving force of the rearrangement is the relief of considerable strain, and this is as easily accomplished by the destruction of the molecule as by its isomerization (Gund et al. 1974, p. 101). Strongly acidic catalyst systems are required, e.g. conventional acids and chloroaluminate ionic liquids (ILs), prodding the system toward equilibrium by repeatedly forming and re‐forming cationic intermediates. In 1957, Schleyer and Donaldson (Schleyer 1957, p. 3292; Schleyer and Donaldson 1960, p. 4645), who were studying the facile AlCl3‐catalyzed isomerization of endo‐THDCPD to its exo isomer, observed a small amount of adamantane as a white crystalline substance collected in the head of a fractionating column at the end of a distillation of an isomerization mixture. This crystalline substance has been proved to be an adamantane structure later. Meanwhile, scientists have developed some modifications of the reaction since then. For example, while conducting the reaction in a hydrogen atmosphere at 40 atm pressure with catalyst AlCl3 + HCl, adamantane was obtained in about 40% yield with small amount of trans‐decalin. Substitutes HF‐BF3 gave 30% yield of adamantane, and aluminosilicate catalysts, in the gas phase at 450–475 °C; endo‐ THDCPD gave 6–13% adamantane and numerous acyclic, monocyclic, and bicyclic aliphatic and aromatic by‐products. Notably, under the catalysis of milder catalysts, e.g. concentrated H2SO4, although capable of giving the isomerization endo to exo, gave no adamantane (Scheme 4.1; Schleyer and Donaldson 1960, p. 4645). AlCl3

1. AlCl 3 + HCl, 40 atm H2, 40% 2. HF + BF3, 30% 3. Aluminosilicate catalysts, 30% 4. H2SO4, none

Scheme 4.1  Isomerization and rearrangement of tetrahydrodicyclopentadiene. Source: Adapted from Schleyer and Donaldson (1960).

4.2  Synthesis of Alkyl Diamondoids via Acid‐Catalyzed Rearrangement

The aluminum chloride‐catalyzed rearrangement of alkanes, proceeding through carbonium ions, differs from ordinary carbonium ion rearrangements in that the ions are formed reversibly and repeatedly. As mentioned above, the driving force for the adamantane rearrangement may be found in the relief of the considerable strain inherent in the bicyclo[2.2.1]heptane system (Bedford et al. 1963, p. 3823). However, no entirely satisfactory mechanism for the transformation is presently available. Schleyer and Donaldson (1960, p. 4645) have proposed a mechanism for the adamantane rearrangement (Scheme 4.2), merely as an illustration. The 2,6‐alkyl migration, A–B, has no direct precedent and is therefore subject to some suspicion. The other rearrangement steps, C–D and E–F, should be quite favorable, since they would proceed with relief of strain. Whatever the detailed mechanism may actually be, it is clear that a very large number of carbonium ions and rearrangement steps are possible. Some of these may be nonproductive, but there may also be more than one pathway leading to adamantane.

E

F

A

B

D

C

G

Scheme 4.2  Proposed mechanism for adamantane rearrangement. Source: Adapted from Schleyer and Donaldson (1960).

Analogously, methyl‐substituted THDCPD is the feasible precursor of alkyl adamantane (Scheme 4.3; Schleyer and Nicholas 1961, p. 305). It was found that every methyl‐substituted THDCPD examined gave a mixture of 1‐ and 2‐ methyladamantane and that the ratio of the two isomers varied relatively little with starting material. At equilibrium, at 25 °C, there is present about 98% of the 1‐methyl isomer, which is more stable by virtue of having no repulsive interactions. A methyl group in the 2‐position of adamantane is equatorial to one of the fused cyclohexane rings, but axial to another, so it is destabilized. In the 1‐position, on the other hand, the methyl is equatorial to three rings and is attached to a quaternary carbon. Given sufficient time to attain equilibrium, it would yield the same mixture of 1‐ and 2‐methyladamantanes. The above thermodynamic results are capable of further generalization. A substituent will normally be more stable at the 1‐position rather than the 2‐position,

103

104

4  Design and Synthesis of High‐Density Diamondoid Fuels

H3C CH3

CH3 2-Methyladamantane

CH3

H3C

CH3

CH3 1-Methyladamantane

Scheme 4.3  Rearrangement of methyl‐substituted tetrahydrodicyclopentadiene. Source: Adapted from Schleyer and Nicholas (1961).

and additional groups will also prefer attachment at other available bridgehead positions (Scheme  4.4). For example, the most stable dimethyladamantane (DMA) is the 1,3‐derivative, and 1,3,5‐trimethyladamantane and 1,3,5,7‐ tetramethyladamantane should also be preferred. Thus, alkyl adamantane is produced by isomerization of the appropriate norbornane derivative (e.g. tetramethylenenorbornane, tetrahydrodicyclohexadiene, dimethyltetrahydrocyclopentadiene; Schleyer and Nicholas 1961, p. 305).

CH3

CH3

H3C H3C CH3

Scheme 4.4  Rearrangement of norbornane derivative. Source: Adapted from Schleyer and Nicholas (1961).

It is clear that the adamantane isomerization process is capable of considerable extension. In fact, all strained tricyclic, saturated hydrocarbons having 10 or more carbon atoms thus far investigated have rearranged at least in part to adamantane derivatives. C11 tricyclics gave methyladamantanes, C12 tricyclics gave DMA, and C13 and higher tricyclics also gave adamantane products. The yields of adamantane are dramatically increased by additional carbons in the precursors. In particular, tetrahydrotricyclopentadiene (THTCPD) (C15) has analogous

4.2  Synthesis of Alkyl Diamondoids via Acid‐Catalyzed Rearrangement

molecular structure to THDCPD (precursor of adamantane), showing great potential as precursor to produce alkyl diamondoids (Xie et  al. 2019, p. 652). Both the starting (endo‐THTCPD), intermediates (exo‐THTCPD), and resulting hydrocarbons (methyldiamantane [MDAM, C15H22], methyl‐1,2‐tetramethyleneadamantane [MTAM, C15H24], and methyldiethyladamantane [MEAM, C15H26]) are mixtures of many configurations, and the methyl group or ethyl group may be positioned on different carbon atoms of the adamantane or diamantane core that cannot be separated by vacuum distillation. Since the mixture is acceptable for fuel‐related application, no further work was performed to obtain individual isomers. In previous research (Wang et al. 2009, p. 2383), no diamondoids were found in the acid‐catalyzed isomerization of endo‐THTCPD (temperature: 0–50  °C, solvent: 1,2‐dichloroethane). Surprisingly, at high temperature (>140 °C), the rearrangement is quite favorable, with abundant alkyl diamondoids (MDAM, MTAM, and MEAM) produced, as shown in Scheme 4.5. Meanwhile, the mass spectra and infrared spectra of alkyl diamondoids are consistent with the characteristic of the diamondoid structure (Gund et  al. 1974, p. 101; Mckervey 1974, p. 479; Silva et al. 2013, p. 125; Waltman and Ling 1980, p. 2189; Warren, Schneider, and Janoski 1968, p. 115). Diamondoids possess the unique rigid but strain‐free ring system, composed of fused chair cyclohexane rings. However, no entirely satisfactory mechanism for the skeletal rearrangement is presently available.

endo-THTCPD k3

k1

exo-THTCPD k2

Alkyl-diamondoids CH3

C15H24

CH3

CH3 (C2H5)2

C15H26

C15H22

Scheme 4.5  Isomerization and rearrangement pathway of THTCPD. Source: Xie et al. (2019). Reproduced with permission of Elsevier.

Figure 4.1 shows the evolution of THTCPD isomers during the AlCl3 catalytic reaction. Original THTCPD (endo) is conformational isomerized/skeletal rearranged to exo isomers and alkyl diamondoids, while the transformation of exo‐ THTCPD shows a volcano‐shaped tendency, indicating that the exo‐THTCPD is one intermediate in the synthesis of alkyl diamondoids. And MTAM is the dominant product and accounts for more than 85% of the diamondoids, whereas MEAM and MDAM are minor products. In order to optimize the reaction conditions, some experimental runs were conducted (see Table 4.1). Several acids (AlCl3, AlBr3, FeCl3, CF3SO3H, H2SO4, HPW, H3PO4) were tested, and only AlCl3, AlBr3, and CF3SO3H show catalytic activity. Overall, AlCl3 is a better choice for practical application with high content of 60.8% diamondoids in liquid phase. As the reaction temperature ranges from

105

4  Design and Synthesis of High‐Density Diamondoid Fuels 100

H3C

endo-THTCPD exo-THTCPD MDAM MTAM MEAM

80 Content (%)

106

MTAM H3C

60

(C2H5)2

MEAM

40

H3C

20 0 0

5

10 15 Reaction time (h)

20

MDAM

Figure 4.1  Product distribution vs. time in rearrangement of endo‐THTCPD. Reaction condition: 15 g endo‐THTCPD, 8 wt% AlCl3, 180 °C. Source: Xie et al. (2019). Reproduced with permission of Elsevier.

140 to 200 °C, the content of diamondoids increases obviously from 18.8% to 65.8%. However, higher temperature causes severe coking (black precipitate), resulting in a decline of carbon yield from 80.5% to 70.8%. While increasing the temperature from 180 to 200 °C, the activity acceleration rate slows down (carbon yield of diamondoids, 45.8–46.6%). The solvent has an effect on this reaction, as shown in Table  4.2. Adding polar solvent (halohydrocarbon, 1,2‐dichlorobenzene) or nonpolar solvent (paraffin, n‐dodecane) reduces the coke formation (carbon yield increases from 78.2% to more than 90%). Unfortunately, the reaction rate is suppressed in the presence of these solvents. It can be rationalized because the solvents dilute the concentration of the acid. Overall the use of solvent cannot improve the reaction but increases the separation cost. In general, catalyst dosage has a significant impact on the catalytic reaction. However, the dosage change of AlCl3 shows unexpected effect on the rearrangement of THTCPD (Table 4.1). The carbon yield of diamondoids did not change much (45.8% to 44.8%) when an increasing amount of AlCl3 (8–10.7 wt%) is added to the system. It is notable that there is an obvious decrease in the carbon yield of diamondoids (36.2%) with further increase of the catalyst dosage (13.3 wt%). The reason may be that adding excess catalyst (>8 wt%) at one time causes intensive reaction and produces viscous tar (owing to the local violent reaction) covering the catalyst surface, leading to the irreversible deactivation of catalyst. To avoid this side reaction, the reaction was conducted in two steps. As shown in Figure 4.2a, with the reaction time extending from 11 to 23 hours, the content of diamondoids slightly rises from 60.1% to 67.4% (solid line), due to the catalyst deactivation. Hence, new catalyst was added to the product of the first 11 hour reaction. Interestingly, both the initial reactant (endo‐THTCPD) and intermediate (exo‐THTCPD) are converted nearly completely to diamondoids. While using exo‐THTCPD as the feedstock, the same result was obtained (Figure 4.2b). More importantly, the complete transformation of THTCPD indicates that the rearrangement process is irreversible, and a considerable high yield can be obtained

4.2  Synthesis of Alkyl Diamondoids via Acid‐Catalyzed Rearrangement endo-THTCPD exo-THTCPD Diamondoids

100

80 Content (%)

Content (%)

80

exo-THTCPD Diamondoids

100

60 40 20

60 40 20

0

0 0

5

(a)

10

15

20

0

5

(b)

Reaction time (h)

10

15

20

Reaction time (h)

Figure 4.2  Product distribution in two‐step rearrangement of (a) endo‐THTCPD and (b) exo‐THTCPD. Reaction condition: 15 g THTCPD, 180 °C. For solid line, 8 wt% AlCl3 was added at 0 hour; for dash line, 8 wt% AlCl3 was added at 0 h and another 5.3 wt% AlCl3 was added at 11 hours. Source: Xie et al. (2019). Reproduced with permission of Elsevier.

by optimizing the reaction conditions. And the catalyst dosage was optimized in the second step (see Table 4.3). Notably, with the increasing dosage up to 5.3 wt%, the concentration of diamondoids in liquid phase can reach 91.8%. However, the increasing dosage at second step (from 2.7 to 5.3 wt%) slightly reduces the carbon yield (70.1% to 60.5%). Overall, an appropriate catalyst dosage (8 wt% + 2.7 wt%) is adopted in view of high carbon yield of diamondoids (60.0%), which exhibits the best performance among other catalytic rearrangement routes. As mentioned above, the isomerization and rearrangement of THTCPD exhibit typical irreversible characteristics, and the reactions are temperature dependent. Therefore, a pseudo‐first‐order irreversible reaction is proposed as Scheme 4.5. The kinetic equations are expressed as

dC A / dt

k1

k3 C A (4.1)

Table 4.3  Effect of catalyst dosage on the two‐step rearrangement of endo‐THTCPD.

First AlCl3 dosagea (wt%)

Second AlCl3 dosageb (wt%)

Concentration of different hydrocarbons in liquid phase (%)

Diamondoids

exo‐ THTCPD

endo‐ THTCPD

Carbon yield (%)

Carbon yield of diamondoids (%)

8

—

60.8

33.5

5.2

75.3

45.8

8

2.7

85.6

13.6

0.4

70.1

60.0

8

4.0

89.7

9.4

0

66.1

59.3

8

5.3

91.8

7.6

0

60.5

55.5

Reaction condition: 15 g endo‐THTCPD, 180 °C. a  The catalyst was added at 0 hour. b  The catalyst was added at 11 hours. Source: Xie et al. (2019). Reproduced with permission of Elsevier.

109

4  Design and Synthesis of High‐Density Diamondoid Fuels

dCB / dt

k1C A

k2CB (4.2)

dCC / dt

k 3C A

k2CB (4.3)

where CA, CB, and CC represent the concentrations of endo‐THTCPD, exo‐ THTCPD, and diamondoids, respectively. The integration of Eqs.  (4.1)–(4.3) gives C A

CA0e

CB

k1

CC

k1 k3 t

(4.4)

k1C A 0 e k3 k2

CA0 1 e

k2 t

e

k1 k3 t

k1

k1 k3 t

(4.5)

k1 e k3 k2

k2 t

e

k1 k3 t

(4.6)

where CA0 represents the initial concentration of endo‐THTCPD. To determine the apparent reaction kinetic rate constants, a series of experiments were conducted at temperature of 433.15, 423.15, 413.15, and 403.15 K, as shown in Figure 4.3. Because temperature has great influence on both the reaction rates, the kinetic models can fit the experimental data very well with high values of correlative coefficient (R2), as shown in Table 4.4. Almost all the correlative coefficients reach up to more than 0.96, which guarantees the accuracy of the fittings. The apparent reaction rate constant k3 is the lowest compared with k1 and k2 under the same temperature, and the reaction rates increase with elevated temperature, suggesting that the reaction rate of endo‐THTCPD to diamondoids is the slowest among other two routes. On the basis of the first‐order rate constants at different temperatures, the activation energy can be derived from the Arrhenius equation (Eq. (4.7)) because –0.3

1.4

433.15 K 423.15 K 413.15 K 403.15 K

–0.4 1.2

–0.5

CC (mmol/g)

–0.6

ln(CA/CA0)

110

–0.7 –0.8 –0.9

–1.1 –1.2

(a)

1.0

1.5

0.8 0.6 0.4

403.15 K 413.15 K 423.15 K 433.15 K

–1.0

1.0

0.2 2.0

2.5

Time (h)

3.0

3.5

0.0

(b)

1.0

1.5

2.0

2.5

3.0

3.5

Time (h)

Figure 4.3  Fitting results of (a) logarithm of CA/CA0 and (b) Cc in THTCPD rearrangement (scatter: experimental data; line: fitting result) vs. time at different temperature. Source: Xie et al. (2019). Reproduced with permission of Elsevier.

4.2  Synthesis of Alkyl Diamondoids via Acid‐Catalyzed Rearrangement

Table 4.4  Estimated apparent rate constants and corresponding correlative coefficients at different temperatures. k1 (h−1)

k2 (h−1)

k3 (h−1)

R12 a

R22 b

433.15

2.25 × 10−1

2.04 × 10−1

4.89 × 10−2

0.999

0.999

423.15

1.23 × 10

−1

−1

3.63 × 10−2

0.999

0.998

413.15

8.01 × 10

−2

−2

5.70 × 10

−2

3.14 × 10

0.995

0.997

403.15

4.22 × 10−2

3.02 × 10−2

2.41 × 10−2

0.962

0.999

Temperature (K)

1.25 × 10

a

 Linear fitting was used based on Eq. (4.4) to determine the value of k1 + k3.  Nonlinear fitting (the least‐square method) was used based on Eq. (4.6) to determine the value of k1 and k2, respectively. Source: Xie et al. (2019). Reproduced with permission of Elsevier.

b

pre‐exponential factor (A) and activation energy (Ea) are constant for given product. The natural log of the reaction rate constant (ln k) was plotted as a function of inverse temperature (an Arrhenius plot) to obtain the apparent activation energy (Ea) (see Figure  4.4). The calculated activation energy (Ea) and corresponding correlative coefficients (R2) based on the Arrhenius plots are recorded in Table 4.5. All the values of R2 are higher than 0.98, indicating good fittings of the kinetic data. The respective apparent activation energies for the rearrangement are 79.1 kJ/mol for endo‐THTCPD to exo‐THTCPD, 94.7 kJ/mol for exo‐ THTCPD to diamondoids, and 32.9 kJ/mol for endo‐THTCPD to diamondoids (Table 4.5). The order of apparent activation energies manifests that the reaction of endo‐THTCPD to diamondoids is easier to take place while exo‐THTCPD to diamondoids is a little harder.

ln k1

–1.6

ln k1 = –9.5125 × 1000/T + 20.4463

–2.0 –2.4

ln k2

–2.8 –3.2 –1.6 –2.0 –2.4 –2.8 –3.2 –3.6 –3.0

ln k2 = –11.3877 × 1000/T + 24.7439

ln k3 = –3.9544 × 1000/T + 6.0830

ln k3

–3.2 –3.4 –3.6 –3.8 2.30

2.32

2.34

2.36

2.38

2.40 3

2.42

2.44

2.46

2.48

2.50

–1

1/T × 10 (K )

Figure 4.4  Arrhenius plots for the rearrangement of THTCPD. Source: Xie et al. (2019). Reproduced with permission of Elsevier.

111

112

4  Design and Synthesis of High‐Density Diamondoid Fuels

Table 4.5  The apparent activation energies and corresponding correlative coefficients based on the Arrhenius plots for the rearrangement of THTCPD. Ea (kJ/mol)

R2

endo‐THTCPD to exo‐THTCPD

79.1

0.995

exo‐THTCPD to diamondoids

94.7

0.995

endo‐THTCPD to diamondoids

32.9

0.983

Route

Source: Xie et al. (2019). Reproduced with permission of Elsevier.

ln k

Ea RT

ln A (4.7)

4.3 ­Synthesis of Alkyl Diamondoids via IL‐Catalyzed Rearrangement The difficulties with conventional aluminum chloride catalysis are catalyst deactivation and the formation of numerous by‐products that complicate the isolation of alkyl diamondoids. ILs, a subset of molten salts with melting points at or below 100 °C, have gained wide recognition as novel solvents for various applications especially as a medium for organic synthesis and catalysis. The distinct advantages of ILs enabled them to receive extensive attention from both academia and industry. ILs are composed entirely of ions, which suggests that they are high‐temperature, corrosive, viscous media. Meanwhile, ILs can be liquid at temperature as low as −96 °C and always be frequently colorless, fluid, and easy to handle. ILs are not intrinsically “green” (some are extremely toxic), but they can be designed to be environmentally benign, with large potential benefits for sustainable chemistry. Besides their advantages with respect to environmental impact, a number of critical aspects of ILs make them an interesting medium in which to study chemical syntheses. A large number of cations and anions allow a wide range of physical and chemical characteristics to be achieved, including volatile and nonvolatile systems, and thus the terms “designer” and “task‐specific” ILs have been developed (Amarasekara 2016, p. 6133; Amde, Liu, and Pang 2015, p. 12611; Pârvulescu and Hardacre 2007, p. 2615; Rogers and Seddon 2003, p. 792; Wang et al. 2017, p. 7113; Zhang, Song, and Han 2017, p. 6834). In any process, the steps following reaction are often at least important for the overall environmental impact and economics. In this respect, ILs have been shown to have a significant advantage over conventional solvents for homogeneously catalyzed reactions. In these cases, the IL may be used in “biphasic catalysis” or the catalyst can be entrapped or “immobilized,” allowing extraction/ distillation of the organic product and the IL/catalyst system reused. Specifically, chloroaluminate IL has been used as acid catalyst in many reactions, such as isomerization, alkylation, and Friedel–Crafts reaction (Huang et al. 2010, p. 69; Huang et al. 2009, p. 1747; Liu et al. 2008, p. 189; Rui et al. 2008,

4.3  Synthesis of Alkyl Diamondoids via IL‐Catalyzed Rearrangement

p. 8205; Snelders and Dyson 2011, p. 4048; Sun and Shi 2011, p. 9339; Wang et al. 2011, p. 1342; Wang et al. 2012, p. 164; Yoo et al. 2004, p. 511). The phase separation from nonpolar hydrocarbons and recycling ability make IL‐based operation suitable for continuous‐flow reaction system. There are already some pilots or industrial process involving IL publicly being announced, and new continuous reactors are developed specifically to facilitate the mixing of different phases, separation of product, and recycling of catalyst inside the reactor system (Ladnak et  al. 2007, p. 719; Olivier‐Bourbigou, Magna, and Morvan 2010, p. 1; Wasserscheid and Eichmann 2001, p. 309). Figure 4.5 depicts the FT‐IR spectra of 1‐butyl‐3‐methylimidazolium chloride (BMIC)‐based IL/CH3CN mixtures. Pure acetonitrile has two bands around 2292 and 2253 cm−1 assigned to CN stretching vibrations. The vibrations shift to higher frequencies once acetonitrile is absorbed on Lewis acid species of IL, and it is known that higher wavenumber corresponds to stronger acidity (Yang and Kou 2004, p. 226). ILs containing CuCl, ZnCl2, or FeCl3 have weak Lewis acidity, and IL with AlCl3 possesses the strongest acidity. It is widely known that the rearrangement activity is closely connected with IL’s acidity. It is natural that only this chloroaluminate IL is suitable for the diamondoid‐formation reaction because it requires stronger acidity. The types of acid species (anions) in IL are dependent on the mole fraction of AlCl3 (denoted as x) in IL as follows:

AlCl 4 Al 2 Cl 7

AlCl 3 AlCl 3

cation Al 2 Cl 7

Al 2 Cl 7

(4.8)

(4.9)

Al 3 Cl10

(4.10)

[BMIM]Cl/AlCl3

Absorbance (a.u.)



cation Cl AlCl 3

[BMIM]Cl/ZnCl2 [BMIM]Cl/FeCl3 [BMIM]Cl/CuCl

Pure acetonitrile

2500

2400

2300

2200

2100

Wavenumber (cm–1)

Figure 4.5  FT‐IR spectra of [BMIM]Cl/MC ln(x = 0.64) ILs. Source: Wang et al. (2012). Reproduced with permission of Elsevier.

113

4  Design and Synthesis of High‐Density Diamondoid Fuels

Since these species have different acid strength (the order is [Al3Cl10]− > [Al2Cl7]− > [AlCl4]−), x value obviously shows significant influence on acidity. Besides acetonitrile, FT‐IR analysis using pyridine as a probe molecule was conducted to estimate the Lewis acidity of the IL catalysts (Kim et al. 2014, p. 109). Figure 4.6 shows the FT‐IR spectra of three kinds of IL catalysts, each with different molar fractions of AlCl3 in the range of 0.60–0.67. The peak near 1450 cm−1 is assigned to pyridine groups that are coordinated to Lewis acid sites, while the peak near 1540 cm−1 is indicative of pyridinium ions from Brønsted acid sites. The characteristic peak for these Brønsted acid sites is observed at 1538 cm−1 for all the IL catalysts. The wavenumber of the Lewis acid sites shifted from 1438 cm−1 for pure pyridine to 1449 cm−1 for the BMIC‐based IL catalyst (x = 0.60). Meanwhile, the wavenumber of the peak for pyridines coordinated to Lewis acid sites gradually increased as a function of molar fraction of AlCl3 to 1454 cm−1 for BMIC‐based IL catalyst (x = 0.67). In general, the predominant anionic species should be Al2Cl7−, when the molar faction of AlCl3 (x) is 0.67. In contrast, the major anionic species should be a mixture of AlCl4− and Al2Cl7−, when the molar fraction of AlCl3 (x) is 0.60. The increase of wavenumber could be attributed to an increase of acidity as a function of x. Interestingly, the peak location also shifted as a function of the type of cationic species, even though the molar fraction of AlCl3 remained constant. The wavenumber of the Lewis acid peaks for pyridine hydrochloride (PHC), triethylamine hydrochloride (TEAC), and BMIC‐based ILs (x  = 0.67) were 1457, 1456, and 1454 cm−1, respectively. This implies that Lewis acidity can be controlled using different types of cationic species as well as by altering the molar fraction of AlCl3. 4.3.1  Rearrangement of Tetrahydrotricyclopentadiene As mentioned above, THTCPD shows great potential as precursor to produce alkyl diamondoids (C15) under the catalysis of AlCl3. Likewise, when IL is used as catalyst,

1538

Pure pyridine

1449

(a) x = 0.60

1454

(b) x = 0.64

1454

(d) x = 0.60

[PHC]/AICI3

1438

1538

Absorbance

Pure pyridine

[TEAC]/AICI3

1438

1538

1454 1455

1438

Pure pyridine

Absorbance

[BMIC]/AICI3

Absorbance

114

1452 (g) x = 0.60

1456 (e) x = 0.64

1456

(h) x = 0.64

1457 (c) x = 0.67

(f) x = 0.67

(i) x = 0.67

1650 1600 1550 1500 1450 1400 1350 1650 1600 1550 1500 1450 1400 1350 1650 1600 1550 1500 1450 1400 1350

(A)

Wavenumber (cm–1)

(B)

Wavenumber (cm–1)

(C)

Wavenumber (cm–1)

Figure 4.6  FT‐IR spectra of pure pyridine, (A) (a) pyridine + BMIC‐based IL catalyst (x = 0.60), (b) pyridine + BMIC‐based IL catalyst (x = 0.64), (c) pyridine + BMIC‐based IL catalyst (x = 0.67), (B) (d) pyridine + TEAC‐based IL catalyst (x = 0.60), (e) pyridine + TEAC‐based IL catalyst (x = 0.64), (f ) pyridine + TEAC‐based IL catalyst (x = 0.67), (C) (g) pyridine + PHC‐based IL catalyst (x = 0.60), (h) pyridine + PHC‐based IL catalyst (x = 0.64), (i) pyridine + PHC‐based IL catalyst (x = 0.67). Source: Kim et al. (2014). Reproduced with permission of Elsevier.

4.3  Synthesis of Alkyl Diamondoids via IL‐Catalyzed Rearrangement

alkyl diamondoids can be generated. Figure 4.7 shows the total ion chromatogram of these mixtures obtained by GC‐MS analysis. The starting material is composed of three THTCPD isomers (I–III) (see Figure 4.8). Theoretical computation has shown the endo‐fragments of THTCPD can be converted to thermodynamically stable exo configurations. Since the isomerization is a carbocation process, the two endo‐fragments (endo‐norbornyl and endo‐cyclopropyl) of THTCPD should be converted stepwisely via possible pathways of I → IV → V, I → II → V, and III → VI. When IL is used as catalyst, isomer IV disappears, and the amount of isomer V increases, indicating that the endo to exo isomerization of cyclopropyl fragments happens smoothly through I → IV → V and/or I → II → V route. In addition, a new isomer VII is formed and its structure will be elucidated later. IR spectra shown in Figure 4.9 also support the formation of diamondoids in IL catalytic rearrangement. The spectrum of AlCl3 catalytic product (low‐temperature

I

THTCPD With AlCl3

III II

With IL

IV VI

VI 12

13

V

VII

14 15 Residence time (min)

I

V

16

Figure 4.7  Total ion chromatogram of product derived from AlCl3 and IL catalytic reactions (I–VII: C15H22 isomers; △: C15H24; □: C15H26). Source: Wang et al. (2011). Reproduced with permission of American Chemical Society.

CH3 II

CH3 (C2H5)2

V

CH3 I

III

IV

VI

Figure 4.8  Isomerization/rearrangement pathway of THTCPD. Source: Wang et al. (2011). Reproduced with permission of American Chemical Society.

115

116

4  Design and Synthesis of High‐Density Diamondoid Fuels

1374 1360 1346 1453

With IL With AlCl3 THTCPD

1600

1500

1400

1300

1200

Wavenumber (cm–1) Figure 4.9  Infrared spectra of product derived from AlCl3 (low‐temperature catalysis) and IL catalytic reactions. Source: Wang et al. (2011). Reproduced with permission of American Chemical Society.

catalysis) is similar to that of starting material, confirming that they have the same skeleton despite of different stereo‐configurations. However, the product of IL catalysis shows distinctly different characteristic bands, indicative of significantly skeletal changes. The bands around 1453 and 1360 cm−1 are the asymmetric and symmetric bending of bridge methylene group, and that around 1346 cm−1 is the symmetric bending of bridgehead methine in adamantane unit, respectively (Warren, Schneider, and Janoski 1968, p. 115). The band around 1374 cm−1 is the symmetric bending of methyl substituent. The rearrangement of polycyclic hydrocarbons is a complex thermodynamic controlled carbocation process in which the most stable isomer with lowest free energy should be the final product. Many researchers have found that alkanes containing three and more cycles tend to rearrange to adamantane‐containing skeleton. The most typical case is that tricyclo[5.2.1.02,6]decane can be isomerized to adamantane in the presence of strong acid or superacid (Olah and Farooq 1986, p. 5410; Schleyer and Donaldson 1960, p. 4645). Similarly, the rearrangement of tetracyclo[6.5.1.02,7.09,13]tetradecane and pentacyclo[8.2.1.14,7.02,9.03,8] tetradecane finally produces 1,2‐tetramethyleneadamantane that is the most stable and thermodynamically preferred molecule among C14H22 isomers, as confirmed by both theoretical calculation and experiment (Farooq et  al. 1988, p. 2840; Gund et al. 1974, p. 101; Lerman 1991, p. 736; Osawa et al. 1982, p. 1923). THTCPD molecules have structure very similar to these two polycyclic tetradecanes. The only difference is that isomers I, II, IV, and V have one additional bridge methylene group connecting the 3,6‐position of tetracyclo[6.5.1.02,7.09,13] tetradecane, while isomers III and VI possess a five‐membered ring connecting the two norbornane fragments instead of four‐membered ring of pentacyclo[8.2.1.14,7.02,9.03,8]tetradecane. It naturally leads to the conclusion that THTCPD can also be rearranged to 1,2‐tetramethyleneadamantane derivates. Figure  4.10 shows the evolution of isomers during the IL catalytic reaction. Original THTCPD (I + II + III) are converted to exo isomers (IV + V + VI) in

4.3  Synthesis of Alkyl Diamondoids via IL‐Catalyzed Rearrangement I + II + III IV + V + VI MDAM C15H24

Contribution (wt%)

100 80

C15H26 Byproduct

60 40 20 0 0

50

100

150

200

250

300

350

Time (min)

Figure 4.10  Product distribution vs. time in IL catalytic rearrangement of THTCPD (temperature: 80 °C; IL (x = 0.67)/THTCPD molar ratio: 1/1). Source: Wang et al. (2011). Reproduced with permission of American Chemical Society.

30 minutes, indicating that the conformational transformation occurs very quickly. However, the skeletal rearrangement is not so fast and the slow increase of diamondoids can be seen in the product. C15H24 is the dominant product and accounts for more than 90% of the diamondoids, whereas C15H26 and C15H22 (MDAM) are minor products. According to the abovementioned analysis, THTCPD undergoes both conformational isomerization and skeletal rearrangement, as illustrated in Figure 4.8. In IL catalytic reaction, the endo to exo isomerization of both norbornyl and cyclopropyl fragments (I → II → V, I → IV → V, and III → VI) proceeds readily. Moreover, considerable skeletal rearrangements happen with MTAM as the primary and dominant product. MEAM is formed when the six‐membered ring fused on adamantane unit of MTAM is broken. It is not clear whether MDAM is formed directly from THTCPD or the former two alkyl adamantanes. In addition, some by‐products such as 2‐pentene, methylcyclopentane, cyclohexane, methylcyclohexane, bicyclo[3.3.0]octane, 1‐ethylcyclohexene, bicyclo[4.3.0]nonane, and tricyclo[5.2.1.02,6]decane are formed through the dissociation of polycyclic structure, which also provide extra H necessary for the formation of diamondoids via hydrogen abstraction and transfer. Compared with CF3SO3H catalytic reaction, the IL catalytic rearrangement has many advantages: much higher activity in the skeletal rearrangement of THTCPD, less dosage of catalyst, no necessity of solvent, no corrosive chemicals, and more environmentally benign. The outstanding performance of IL may be attributed to two points. First, the acidity of chloroaluminate anions in IL is strong enough to catalyze the reaction that is usually catalyzed by superacid. Second, IL itself works as a novel solvent to provide strong polar and electrostatic environment that greatly stabilizes the carbonium and/or intermediate (Cui et al. 2006, p. 1571).

117

4  Design and Synthesis of High‐Density Diamondoid Fuels

Figure 4.11 shows the effect of AlCl3 molar composition on the rearrangement. When x ≤ 0.50, skeletal rearrangement does not happen because [AlCl4]− anion in IL has no acidity. When 0.50  0.67. This indicates that polycyclic hydrocarbons are prone to cracking when the acidity is too strong. Figure 4.12 shows the effect of temperature on the rearrangement. The reaction can take place at room temperature, and the conversion is threefold when the temperature ranges from 20 to 60 °C. This tendency continues at higher temperature, but the acceleration rate slows down. The selectivity of C15H24 does not change much, although a decreasing tendency is observed in the range studied. It is notable that there is an obvious decrease in the selectivity of MDAM with the increase of temperature, hinting that high temperature is not preferred for the formation of MDAM. Figure 4.13 shows the effect of IL dosage on the rearrangement. The reaction is promoted with the increase of IL dosage because the concentration of acidic anion is increased, accompanied with slight decrease in the selectivity of C15H24. When the molar IL/THTCPD ratio exceeds 1.0, the conversion of THTCPD no longer increases. The effect of some additives and solvents on rearrangement is also studied, as shown in Table 4.6. The conversion of THTCPD is promoted when trace amount of H2O (0.5 wt%) is introduced, but the formation of C15H24 is suppressed dramatically. Presence of trace water in chloroaluminate IL may generate some 100

Conversion or selectivity (%)

118

80 60

CTHTCPD SMDAM

40

SC15H24 SC15H26

20

Sbyproduct

0

0.55

0.60

0.65 (x) AlCl3 in IL

0.70

Figure 4.11  Effect of AlCl3 molar composition on IL catalytic rearrangement of THTCPD (temperature: 80 °C; time: 3 hours; IL/THTCPD molar ratio: 1/1). Source: Wang et al. (2011). Reproduced with permission of American Chemical Society.

4.3  Synthesis of Alkyl Diamondoids via IL‐Catalyzed Rearrangement

Conversion or selectivity (%)

80

60

CTHTCPD SMDAM SC15H24

40

SC15H26 Sbyproduct

20

0 20

40 60 80 Reaction temperature (°C)

100

Figure 4.12  Effect of temperature on IL catalytic rearrangement of THTCPD (time: 3 hours; IL (x = 0.67)/THTCPD molar ratio: 1/1). Source: Wang et al. (2011). Reproduced with permission of American Chemical Society.

Conversion or selectivity (%)

100 80 CTHTCPD

60

SMDAM SC15H24

40

SC15H26 Sbyproduct

20 0 0.5

1.0 1.5 IL/THTCPD molar ratio

2.0

Figure 4.13  Effect of IL dosage on IL catalytic rearrangement of THTCPD (temperature: 80 °C; time: 3 hours; IL (x = 0.67)). Source: Wang et al. (2011). Reproduced with permission of American Chemical Society.

Brønsted acid, and the conjugation of Brønsted acid and Lewis acid produces superacid with the Hammett acidity of −18, noting that of CF3SO3H superacid is −14.1 (Smith et al. 1989a, p. 525; Smith et al. 1989b, p. 5075; Wasserscheid and Keim 2000, p. 3772). It is very possible that superacid species result in cracking of C15H24. General, many rearrangements have to be conducted in solvent to enhance the formation and stabilization of carbonium. However, the presence of solvent is deleterious to the IL catalytic reaction, because IL itself already provides a perfect solvent environment that will be disturbed by additional solvent.

119

120

4  Design and Synthesis of High‐Density Diamondoid Fuels

Table 4.6  Effect of additive/solvent on IL catalytic rearrangement of THTCPD.

Additive/solvent

CTHTCPD (%)

SC15 H24

(%)

SMDAM (%)

SC15 H26

(%) Sbyproduct (%)

—

66.8

87.0

5.5

3.6

3.9

0.5 wt% H2O

85.5

76.5

6.1

3.8

13.6

50 wt% chloroform

61.2

86.1

5.1

4.7

4.1

50 wt% dichloroethane

63.6

85.7

5.3

5.2

3.8

50 wt% methylbenzene

58.8

86.6

5.1

4.8

3.6

Source: Wang et al. (2011). Reproduced with permission of American Chemical Society.

4.3.2  Rearrangement of Tetrahydrodicyclopentadiene Originally exo‐THDCPD was synthesized through isomerization of endo‐ THDCPD using Brønsted acid like sulfuric acid, now Lewis acid like AlCl3 is used for industrial batch operation (Cristol, Seifert, and Soloway 1960, p. 2351; Norton and Howe 1981; Schneider, Ware, and Janoski 1978). Adamantane was initially isolated in small amount from petroleum mixtures; then Schleyer reported a simple synthesis route via endo‐THDCPD isomerization catalyzed by large amount of AlCl3 that is nowadays applied on industrial scale (Schleyer 1957, p. 3292; Schleyer and Donaldson 1960, p. 4645). AlCl3‐based processes inevitably bring up many environmental concerns like corrosion of equipment, troublesome product separation, no recyclability of catalyst, and thus catalyst disposal problems. It should be noted that the reaction conditions to produce exo‐THDCPD and adamantane are dramatically different. The former occurs easily under mild conditions, but the latter happens only under severe conditions. Firstly, several 1‐n‐ alkyl‐3‐methylimidazolium halides were checked, as shown in Table  4.7. The conversion of endo‐THDCPD and selectivity of exo‐THDCPD do not change with the types of cation. As to reaction toward adamantane, the selectivity of adamantane decreases very slightly as the alkyl group of cation becomes longer. Overall, the effect of cation on isomerization is negligible, and it is the anion that determines the activity. Figure 4.14 shows that proper temperature is necessary for isomerization reaction. There is a considerable increase in the conversion when temperature rises from 40 to 50 °C, and it does not increase with temperature anymore. On the contrary, the selectivity of exo‐THDCPD drops with reaction temperature due to the formation of adamantane. For synthesizing exo‐THTCPD, the optimal temperature is 50 °C. Table 4.8 shows that the catalyst dosage needed for the reaction greatly depends on the mole fraction of AlCl3, that is, higher the x value is, smaller the dosage is. It is worth noting that both the conversion and selectivity exceed 98% with 3.3 mol% IL (x = 0.67) used. From Table 4.8, one may note that, when dosage of IL is small, the conversion of endo‐THDCPD is very low that cannot be promoted by expanding reaction time, suggesting that some acid species are deactivated or poisoned. Impurities in the reactant like oxides and absorbed water may be detrimental for the

4.3  Synthesis of Alkyl Diamondoids via IL‐Catalyzed Rearrangement

Table 4.7  Isomerization of THDCPD using different acidic ILs. Cation

Anion

Cendo or Cexo (%)

Sexo (%)

SAD (%)

endo‐THDCPD → exo‐THDCPDa CuCl

[BMIM]Cl

0

FeCl3

[BMIM]Cl

5.2

100

0

ZnCl2

[BMIM]Cl

8.7

100

0

AlCl3

[BMIM]Cl

98.2

98.2

1.8

AlCl3

[HMIM]Cl

97.6

98.5

1.5

AlCl3

[OMIM]Cl

96.8

99.1

0.9

exo‐THDCPD → adamantaneb AlCl3

[BMIM]Cl

100

35.8

64.9

AlCl3

[HMIM]Cl

100

33.5

63.5

AlCl3

[OMIM]Cl

100

30.4

62.8

a

 Reaction conditions: IL (x = 0.60) 10 mol%; temperature: 50 °C; time: 0.5 hour.  Reaction conditions: IL (x = 0.67) 100 mol%; temperature: 80 °C; time: 4 hours. Source: Wang et al. (2012). Reproduced with permission of Elsevier.

b

Conversion or selectivity (%)

100 80 Cendo

60

Sexo SAD

40 20 0

40

50

60 70 80 Temperature (°C)

90

100

Figure 4.14  Effect of temperature on endo to exo isomerization of THDCPD (reaction conditions: IL (x = 0.67) 3.3 mol%; time 0.5 hour). Source: Wang et al. (2012). Reproduced with permission of Elsevier.

catalyst. So it was purified by vacuum distillation, and this pre‐purification does significantly improve the reaction. The conversion is increased from 19.8% to 45.5% when 5 mol% IL (x  =  0.60) is used and from 23.4% to 47.5% when 2.8 mol% IL (x = 0.67) is used (see Table 4.8). Using purified endo‐THDCPD as reactant, the reuse of IL was conducted in two ways. In first operation, the IL layer was directly used as catalyst for another run. As shown in Figure 4.15, the activity quickly decreases with the increase of recycling times, indicating gradual deactivation of IL. The reaction produces

121

4  Design and Synthesis of High‐Density Diamondoid Fuels

Table 4.8  Effect of mole fraction of AlCl3 and dosage of ILs on endo to exo isomerization of THDCPD.

x(AlCl3) in IL

IL dosage relative to endo‐THDCPD (mol%)

Cendo (%)

Sexo (%)

SAD (%)

99.1

0.39

0.55

100

98.8

0.55

25

93.5

100

0

0.55

16.7

23.7

100

0

0.55

10

17.1

100

0

0.6

100

99.5

90.2

3.7

0.6

14.3

98.8

97.9

0.3

0.6

10

98.2

98.1

0

0.6

6.3

0.6

5

19.8

100

0

0.6

5a

45.5

100

0

0.67

51.2

100

100

90.2

5.6

14.3

99.1

97.0

1.1

0.67

10

98.8

98.3

0.4

0.67

5

98.5

98.4

0

0.67

3.3

98.3

98.7

0

0.67

2.8

23.4

98.9

0

0.67

2.8

100

0

0.67

b

47.5

100

0

a

 Reaction conditions: temperature: 50 °C; time: 0.5 h.  Reactant is purified via distillation. Source: Wang et al. (2012). Reproduced with permission of Elsevier. b

100 Conversion or selectivity (%)

122

80 60 40

Cendo-(used IL is treated) Sexo-(used IL is treated)

20

Cendo-(used IL is untreated) Sexo-(used IL is untreated)

0

1

2

3 Reuse times

4

5

Figure 4.15  Reuse of IL in endo to exo isomerization of THDCPD (reaction conditions: IL (x = 0.67) 3.3 mol%; temperature 50 °C; time 0.5 hour). Source: Wang et al. (2012). Reproduced with permission of Elsevier.

4.3  Synthesis of Alkyl Diamondoids via IL‐Catalyzed Rearrangement

some polymerized by‐products, which is very considerable in adamantane‐formation reaction (see below). Some of them may enter the IL layer and be transferred into polymers in subsequent runs. These polymers called as acid‐soluble oil in reference (Berenblyum, Katsman, and Karasev 2006, p. 128) will accumulate and finally deactivate chloroaluminate species. To prevent the formation of polymers, the used IL was heated under vacuum after each run to remove residues with relatively low boiling point. In this case, IL can be used for four runs without obvious activity loss, indicating the treatment retards deactivation. However, in the fifth run, the activity becomes very low again, probably because the treatment cannot completely eliminate the formation and accumulation of polymers in IL. In addition, leaching of chloroaluminate into hydrocarbon layer may become obvious after several runs, which lowers the concentration of active species and thus decreases the activity. In endo to exo isomerization of THDCPD, a phenomenon worth noting is that adamantane appears with trace amount under some conditions, i.e. high mole faction of AlCl3 in IL, high catalyst dosage, and high reaction temperature. One may expect that adjusting the reaction conditions will produce considerable adamantane. Under these conditions, isomerization of endo‐ to exo‐THDCPD takes place almost quantitatively in less than 10 minutes, and then adamantane appears, suggesting that adamantane is formed via isomerization of exo‐ THDCPD. Figure 4.16 shows that the conversion of exo‐THDCPD goes up with reaction time and reaches 56% in 15 hours, whereas the selectivity of adamantane slowly decreases from 75.1% to 60.8%. It is clear that the reaction rate of adamantane forming is much lower compared with that of endo to exo isomerization, and the formation of by‐products is inevitable. Although the starting material is endo‐THDCPD, reaction in this stage is actually the exo‐ THDCPD → adamantane one, so the activity is assessed using the conversion of exo‐THDCPD and selectivity of adamantane.

Conversion or selectivity (%)

80 70 60 50

Cexo

40

SAD SDHN

30

SPoly

20 10 0 0

2

4

6

8 10 Time (h)

12

14

16

Figure 4.16  Isomerization of exo‐THDCPD to adamantane with prolonged reaction time (reaction conditions: IL (x = 0.67) 100 mol%, temperature 80 °C). Source: Wang et al. (2012). Reproduced with permission of Elsevier.

123

4  Design and Synthesis of High‐Density Diamondoid Fuels

Transformation from endo‐ to exo‐THDCPD is easy because it only undergoes conformational isomerization. However, isomerization from endo‐THDCPD to adamantane is a complex, thermodynamics‐controlling, and much slow skeletal rearrangement because it involves hydride abstractions and multiple Wagner– Meerwein rearrangements. In this process the conversion from [exo‐THDCPD]+ to [adamantane]+ is the rate‐determining step. Along with the rearrangement, decalin is also formed through disproportionation, and other by‐products form through polymerization. The reaction pathway is briefly illustrated in Figure 4.17. As seen in Figure 4.18, the dosage of IL has to be very high to give considerable conversion. The conversion of exo‐THDCPD linearly increases with IL dosage, along with a slight decrease in the selectivity. Figure 4.19 shows that acidity is a crucial factor for adamantane synthesis and the conversion of exo‐THDCPD increases quickly with mole faction of AlCl3. The conversion of exo‐THDCPD reaches 92.3% when x = 0.71. Unfortunately, strong acidity also induces considerable side reactions, as a result the selectivity of adamantane decreases from 77.6% to 55.2%, with the yield being 50.9%. Isomerization of endo‐THDCPD to exo‐THDCPD and adamantane using various acid catalysts is listed in Table 4.9. All existing operations offer high yield of

endo-THDCPD

Adamantane

[Adamantane]+

[endo-THDCPD]+

[exo-THDCPD]+

Polymers

[Decalin]+

Polymers

exo-THDCPD

Decalin

Figure 4.17  Reaction pathway of THDCPD isomerization. Source: Wang et al. (2012). Reproduced with permission of Elsevier.

70 Conversion or selectivity (%)

124

60

Cexo SAD SDHN

50 40 30 20 10 0 20

40

60 80 IL dosage (mol%)

100

Figure 4.18  Effect of IL dosage on isomerization of exo‐THDCPD to adamantane (reaction conditions: IL (x = 0.67); temperature 80 °C; time 4 hours). Source: Wang et al. (2012). Reproduced with permission of Elsevier.

4.3  Synthesis of Alkyl Diamondoids via IL‐Catalyzed Rearrangement

Conversion or selectivity (%)

100 80 60 40

Cexo SAD SDHN

20 0 0.55

0.60

0.65

0.70

0.75

x(AlCl3) in IL

Figure 4.19  Effect of mole fraction of AlCl3 in IL on isomerization of exo‐THDCPD to adamantane (reaction conditions: IL (x = 0.67) 100 mol%; temperature: 80 °C; time: 4 hours). Source: Wang et al. (2012). Reproduced with permission of Elsevier.

exo‐THDCPD. CF3SO3H‐based operation can be conducted at room temperature, but the dosage of catalyst and solvent is very high. As to AlCl3‐based reaction, high temperature and long reaction time, and/or large amount of catalyst and solvent, are necessary, whereas zeolite catalytic reaction has to be conducted at significantly high temperature. IL‐based route gives high yield (>97%) in shorter reaction time (0.5 hours), and there is no solvent used. In addition, liquid–liquid biphasic reaction makes product separation and catalyst recycling easy. If IL is replaced by same amount (10.8 wt%) of AlCl3 or CF3SO3H, the yield of exo‐THDCPD will be lower (about 91%). Considering the four‐time recycling, the average catalyst consumption (2.7 wt%) is actually very low. The reaction temperature is just a little higher than room temperature that can be easily controlled; also extra energy for heating or cooling is saved. For exo‐THDCPD production, industrial AlCl3 catalytic operation is in batch process because the catalyst cannot be properly separated and recycled. The phase separation and recycling ability make IL‐based operation suitable for continuous‐flow reaction system. However, some technical problems have to be solved for scaling up. Fortunately, there are already some pilots or industrial process involving IL publicly announced, and new continuous reactors are developed specifically to facilitate the mixing of different phases, separation of product, and recycling of catalyst inside the reactor system (Ladnak et al. 2007, p. 719; Olivier‐Bourbigou, Magna, and Morvan 2010, p. 1; Qiao et al. 2004, p. 61; Wasserscheid and Eichmann 2001, p. 309), which can be applied in the present case. Imidazolium salt is a little expensive, which makes the IL‐based operation less attractive in economics. Since the activity of IL is totally decided by anions, it is expected that cheaper reagent can be used to reduce cost. As shown in Table 4.9, when trialkylammonium chloride (Et3NHCl) whose cost is comparable with AlCl3 is applied to replace [RMIM]Cl, the yield of exo‐THDCPD is still very high (96.1%).

125

4.3  Synthesis of Alkyl Diamondoids via IL‐Catalyzed Rearrangement

4.3.3  Rearrangement of Other Polycycloalkanes Analogous to the classical acid‐catalyzed synthesis of adamantane, the rearrangement of polycyclic hydrocarbons with more than 10 carbon is a facile route to synthesize alkyl adamantanes. Choosing different precursors with different carbon number can produce alkyl adamantanes with different alkyl chain number. Notably, dimethyl‐ and trimethyltricyclo[5.2.1.02,6]decane (DMTCD and TMTCD) with 12 and 13 carbon are ideal precursors to produce alkyl adamantanes (C12–C13). Both DMTCD and TMTCD have many stereoisomers because the methyl group may be anchored on different carbon atoms. Figure 4.20 shows the distribution of products varying with the reaction time. It can be seen that almost all the starting materials are converted in 20 minutes. During this period, conformational isomers are the major products, and their summed concentration reaches the maximum value. This indicates that the conformational isomerization happens very quickly in the presence of IL. After that, the amount of

Product distribution (wt%)

60

MTCDs Stereoisomers 1,3-DMAM 1,4-DMAM 1-EAM

40

1,3,5-TMAM 1-M-3-EAM 1,3,4-TMAM Others

20

0 0

1

2

3

4

5

6

Time (h) H2 CH3

+

Stereoisomers CH3 CH3 DMTCD

CH3

1,3-DMAM

1-EAM

1,4-DMAM H2 CH3

(CH3)2

+

Stereoisomers CH3 (CH3)2 TMTCD

1,3,5-TMAM

1-M-3-EAM

1,3,4-TMAM

Figure 4.20  Product distribution and pathways in IL catalytic rearrangement of MTCDs (reaction conditions: temperature: 80 °C; x(AlCl3): 0.67; IL/MTCDs: 0.5). Source: Ma et al. (2013). Reproduced with permission of American Chemical Society.

127

128

4  Design and Synthesis of High‐Density Diamondoid Fuels

stereoisomers begins to decline, and some alkyl adamantanes appear with a tendency of increase, indicating the occurrence of skeletal rearrangement from stereoisomers to alkyl adamantanes. After reaction of five hours, alkyl adamantanes account for 80% of the products, along with some undefined products formed via fragmentation, disproportionation, and polymerization reactions. Six alkyl adamantanes are determined by GC‐MS analysis, including 1,4‐ DMAM, 1,3‐DMAM, 1‐EAM, 1,3,4‐TMAM, 1,3,5‐TMAM, and 1‐M‐3‐EAM (see Figure  4.20 for the structures). The former three are generated from DMTCD, whereas the later three are from TMTCD. As shown in Figure 4.20, at the beginning (one hour), 1,4‐DMAM and 1,3,4‐TMAM are the major alkyl adamantanes in the two parallel reactions, respectively. But they decrease to very small quantity in the later stage. On the contrary, 1,3‐DMAM and 1,3,5‐TMAM increase with the prolonging of reaction time and finally become the major products. This indicates that 1,4‐DMAM and 1,3,4‐TMAM are the intermediate products in the skeletal rearrangement, which are, respectively, converted to 1,3‐ DMAM and 1,3,5‐TMAM with prolonged reaction time. After four hours, the amount of products keeps stable, except the continuous increase of 1,3‐DMAM, suggesting a direct route from DMTCD to 1,3‐DMAM. As stated in Section 4.1, the driving force for the formation of diamondoids is the thermodynamic preference. Theoretic calculations were conducted to assess the total energy of the related compounds. The energy gaps (also the gaps of enthalpies of formation) of products relative to the corresponding reactants were shown in Table 4.10. endo‐3,9‐DMTCD and endo‐3,4,9‐TMTCD were selected as the benchmarks of DMTCDs and TMTCDs because they are the most abundant. It can be seen that the starting compounds have the highest energy and the alkyl adamantanes have lower energy, suggesting that these diamondoids are Table 4.10  Energy gaps (ΔE, kJ/mol) between reactants and the corresponding products. Abbreviation

exo–endo‐TCDD

a

ΔEa

ΔEb

ΔEc

ΔEd

—

—

−159.3

—

1,3‐DMAM

−236.3

—

−3402.5

—

1,4‐DMAM

−224.6

—

−3390.8

—

1‐EAM

−214.5

—

−3380.7

—

2‐EAM

−197.0

—

−3363.2

—

1,3,5‐TMAM

—

−165.6

—

—

1‐M‐3‐EAM

—

−146.2

—

—

1,3,4‐TMAM

—

−145.1

—

—

1,2‐TMAM

—

—

—

−102.8

 Represent the ΔE relative to endo‐3,9‐DMTCD.  Represent the ΔE relative to endo‐3,4,9‐TMTCD. c  Represent the ΔE relative to exo–exo‐TCDD. d  Represent the ΔE relative to TCTD. Source: Ma et al. (2013). Reproduced with permission of American Chemical Society. b

4.3  Synthesis of Alkyl Diamondoids via IL‐Catalyzed Rearrangement

favored in thermodynamics. Based on the calculated energy gaps, the thermodynamic preference for the rearrangement of DMTCD and TMTCD is 1,3‐ DMAM > 1,4‐DMAM > 1‐EAM > 2‐EAM > DMTCD and 1,3,5‐TMAM > 1‐M‐3‐ EAM > 1,3,4‐TMAM > TMTCD, respectively. This confirms that 1,3‐DMAM and 1,3,5‐TMAM are two most favored products in the rearrangement. And 1,4‐DMAM and 1,3,4‐TMAM are prone to transfer to more stable 1,3‐DMAM and 1,3,5‐TMAM via methyl transfer, respectively. For the rearrangement of DMTCD, only a few 1‐EAM are formed because they are not preferred in thermodynamics when compared with DMA. As for the rearrangement of TMTCD, considerable 1‐M‐3‐EAM is generated because it is more preferred than 1,3,4‐ TMAM. The order of abundance of product is in good agreement with the computed result, showing that the rearrangement is thermodynamically controlled. The reaction pathway based on the experimental and computational results is shown in Figure 4.20. Figure 4.21 shows the effect of acid strength (AlCl3 fraction), temperature, and IL dosage on the rearrangement. The acidity order of anions is acidic [Al3Cl10]− > acidic [Al2Cl7]− > neutral [AlCl4]−. A conversion of 80% is obtained with x = 0.64, and it is increased to about 100% at x = 0.7. In the range of x = 0.64– 0.67, the selectivity of thermodynamically favored products (1,3‐DMAM and 1,3,5‐TMAM) increases quickly, whereas that of other alkyl adamantanes decreases. The distribution of product does not change much when x is beyond 0.67, indicating that the acidity of [Al2Cl7]− ions is strong enough to catalyze the skeletal rearrangement. Also the temperature shows a significant effect on the reaction. There is an obvious increase in the conversion from 71% to 99% when the temperature rises from 60 to 80 °C. The total selectivity (SAMs) of alkyl adamantanes reaches the highest value of 77% at 80  °C, but it shows a downside trend with the further increase of temperature. Specifically, the selectivity of 1,3‐ DMAM declines considerably. As to the effect of IL dosage, enough IL is necessary to ensure considerable yield of alkyl adamantanes in acceptable reaction period. When the IL/MTCD ratio goes up from 0.25 to 0.5, the conversion and SAMs reach 99% and 78%, respectively. Overall, at mild reaction conditions, the selectivity of 1,3‐DMAM and 1,3,5‐TMAM increases, but that of 1,4‐DMAM and 1,3,4‐TMAM is reduced when increasing the temperature, acid strength, or IL dosage, as a result of thermodynamic control. Tetracyclo[6.2.1.13,6.02,7]dodecane (TCDD) (C12) is a mixture of two stereoisomers (exo–exo 96.7 wt% and exo–endo 2.4 wt%) (see Figure 4.22). Similar to the case of MTCDs, TCDD firstly undergoes configurational isomerization from exo–exo‐ to exo–endo‐TCDD and then rearrangement to DMA (1,3‐ and 1,4‐DMAM) and ethyladamantanes (1‐ and 2‐EAM) (see Figure  4.22). For DAM, 1,3‐DMAM always increases during the reaction, with small amount of 1,4‐DMAM formed. For ethyladamantanes, 2‐EAM is formed in large amount but then disappears, along with the increase of 1‐EAM. When the reaction time exceeds three hours, the amount of 2‐EAM decreased is equal to that of 1‐EAM formed. This clearly shows the ethyl transfers from 2‐EAM to 1‐EAM. It is worth noting that although DMTCD and TCDD have distinctly different molecular structures, they give the same alkyl adamantanes in the reaction. This testifies that the rearrangement of polycyclic hydrocarbons will finally lead to

129

4  Design and Synthesis of High‐Density Diamondoid Fuels

Conversion or selectivity (%)

100 80 60 40 20 0 0.64

0.66

0.68

0.70

0.72

x(AlCl3)

(a) Conversion or selectivity (%)

100 80 60 40 20 0 60

80

(b)

100

120

Temperature (°C)

100

Conversion or selectivity (%)

130

80 60 40 20 0 0.2

(c)

0.4

0.6

0.8

1.0

IL/MTCDs molar ratio

Figure 4.21  Effects of (a) AlCl3 fraction, (b) reaction temperature, and (c) ionic liquid dosage on the IL‐catalytic rearrangement of MTCDs (■: conversion; ●: S1,3‐DMAM; ▲: S1‐EAM; △: S1,4‐DMAM; □: S1,3,5‐TMAM; ○: S1,3,4‐TMAM; ★: S1‐M‐3‐EAM; ☆: SAMs; reaction conditions: time: 5 hours; x(AlCl3): 0.7; temperature: 80 °C; IL/MTCDs: 0.5). Source: Ma et al. (2013). Reproduced with permission of American Chemical Society.

thermodynamically favored diamondoids, although via different reaction channels with different barriers. Therefore, some simple polycyclic hydrocarbons can be used to synthesize diamondoids.

4.3  Synthesis of Alkyl Diamondoids via IL‐Catalyzed Rearrangement

Product distribution (wt%)

100 1,3-DMAM 1-EAM 1,4-DMAM 2-EAM exo–endo-TCDD exo–exo-TCDD Others

80 60 40 20 0 0

1

2

3

4

5

6

7

8

Time (h) +

Heat

H2 exo–endo-TCDD

exo–exo-TCDD

+

1,3-DMAM

+

1-EAM

1,4-DMAM

+

2-EAM

Figure 4.22  Product distribution and pathways in IL catalytic rearrangement of TCDD (reaction conditions: temperature: 40 °C; x(AlCl3): 0.67; IL/TCDD: 2). Source: Ma et al. (2013). Reproduced with permission of American Chemical Society.

It is interesting that the distribution of the four alkyl adamantanes changes significantly with reaction conditions (see Table  4.11). At mild conditions, considerable ethyladamantanes are formed, especially when the IL/TCDD ratio is below 2. In this range, the summed selectivity of 1‐EAM and 2‐EAM keeps stable (48–59%) with the increase of temperature or AlCl3 fraction. And much more 2‐EAM is formed at low temperature and low AlCl3 fraction. For example, two extreme cases appear under conditions of 50  °C, IL/ TCDD = 0.5, x = 0.67, and 60 °C, IL/TCDD = 1.5, x = 0.67, with the selectivity of 2‐EAM and 1‐EAM being 53.6% and 0.9% and 3.1% and 47.4%, respectively. When IL/TCDD > 2, DAM increase quickly with the increase of temperature and AlCl3 fraction. The selectivity of 1,3‐DMAM is 65.7% under conditions of 60  °C, IL/TCDD  =  4, x  = 0.67; meanwhile the summed selectivity of ethyladamantanes is just 1.3%. It is also noted that considerable amount of 1,4‐DMAM is formed under mild conditions, but very less under harsh conditions. So diamondoids less preferred in thermodynamics are predominant under mild conditions, whereas thermodynamically most preferred compounds become major products only under harsh conditions. It suggests that the barrier energy for thermodynamically most feasible reaction channels is very high. This result also hints a possibility to tune the distribution of diamondoids by adjusting the reaction conditions.

131

4.3  Synthesis of Alkyl Diamondoids via IL‐Catalyzed Rearrangement

Treatment of TCTD with AlCl3 IL gives 1,2‐TMAM as the only diamondoid, along with minor conformational stereoisomers. As shown in Figure 4.23, almost all the TCTD molecules are converted, and the concentration of 1,2‐TMAM reaches 85% in one hour. Both the configurational isomerization from TCTD to stereoisomers and further skeletal rearrangement to 1,2‐TMAM are fast and easy in the presence of IL. Calculation data in Table 4.10 also shows that the energy of 1,2‐TMAM is lower than that of TCTD, so it is preferred in thermodynamics. Figure 4.24 shows that the reaction conditions have significant effect on the conversion, but the selectivity is almost unchanged. When x increases from 0.6 to 0.67, there is a quickly rise in the conversion from 73% to 98%, so the acidity of [Al2Cl7]− is strong enough to catalyze the skeletal rearrangement of TCTD. Increasing the temperature or IL dosage can also improve the reaction. As a result, the conversion can be as high as 99%, with the selectivity of 1,2‐TMAM above 90%. The recyclability of IL was also explored in two ways for the rearrangement of TCTD. In one operation, new reactant was directly added in the reaction mixture of previous run. As shown in Figure  4.25, the conversion and selectivity decrease slowly, because the accumulated polymerized by‐products and water deactivate the IL. In another operation, the upper hydrocarbon layer was removed, and the bottom IL layer was used for subsequent run. It can be seen that the conversion and selectivity decrease very quickly, due to the loss of IL dissolved in the hydrocarbon layer. This result suggests that the incomplete phase separation leading to the loss of IL is a major problem for recycling.

Product distribution (wt%)

100 80

1,2-TMAM TCTD Stereoisomers

60 40 20 0

0

40

80

120

160

Time (min)

+

H2

Heat

TCTD

Stereoisomers 1,2-TMAM

Figure 4.23  Product distribution and pathways in IL catalytic rearrangement of TCTD (reaction conditions: temperature: 60 °C; x(AlCl3): 0.67; IL/TCTD: 0.125). Source: Ma et al. (2013). Reproduced with permission of American Chemical Society.

133

4  Design and Synthesis of High‐Density Diamondoid Fuels

Conversion or selectivity (%)

100

C S1,2-TMAM

80

60

40

20

0.60

0.65

0.70

0.75

x(AlCl3)

(a)

100

Conversion or selectivity (%)

134

S1,2-TMAM, 60 °C C70 °C S1,2-TMAM, 70 °C

60

C80 °C S1,2-TMAM, 80 °C

40 0.06

(b)

C60 °C

80

0.08

0.10

0.12

0.14

0.16

IL/TCTD molar ratio

Figure 4.24  Effects of (a) AlCl3 fraction and (b) ionic liquid dosage on the IL‐catalytic rearrangement of TCTD (reaction conditions: time: 3 hours; x(AlCl3): 0.67; temperature: 60 °C; IL/TCTD: 0.125). Source: Ma et al. (2013). Reproduced with permission of American Chemical Society.

4.3.4  Rearrangement of Biomass‐Derived Hydrocarbons Cedarwood oil is an intriguing starting material for advanced biofuel development. It is primarily composed of the tricyclic sesquiterpenes α‐cedrene, β‐ cedrene, and thujopsene along with a significant quantity of cedrol. α‐Cedrol is commercially available, and dehydration of α‐cedrol with p‐toluenesulfonic acid yielded α‐cedrene in excellent yield, while hydrogenation provided analytically pure cedrane (Figure 4.26; Harrison and Harvey 2017, p. 467). It was of interest to isomerize cedrane to a mixture of adamantanes (mainly composed of 1‐ethyl‐3,5,7‐trimethyladamantane) by reaction with a strong Lewis acid catalyst. Initially cedrane was allowed to react with fresh AlCl3 under an inert atmosphere. Interestingly, the reaction was extremely sluggish with only traces of adamantanes observed by GC‐MS after 24 hours. However,

4.4  Synthesis of Alkyl Diamondoids via Zeolite‐Catalyzed Rearrangement

Conversion or selectivity (%)

100 C S1,2-TMAM

80 60 40 20 0

1

2

3

4

Reuse times Figure 4.25  Reuse of IL in catalytic rearrangement of TCTD (solid line: the upper layer not removed; dashed line: the upper layer was removed after each run; reaction conditions: time: 3 hours; x(AlCl3): 0.67; temperature: 60 °C; IL/TCTD: 0.125). Source: Ma et al. (2013). Reproduced with permission of American Chemical Society.

OH –H2O α-Cedrene

Cedrol

H2, cat. AlCl3

ETMA

Cedrane

Figure 4.26  Conversion of α‐cedrol to 1‐ethyl‐3,5,7‐trimethyladamantane (ETMA). Source: Harrison and Harvey 2017. Reproduced with permission of Royal Society of Chemistry.

it was noticed that an aliquot exposed to air generated the adamantanes quite rapidly. This difference in reactivity suggested that adventitious water was necessary for the reaction to proceed. Addition of ~5 mol% water promoted the reaction and cedrane underwent >90% conversion to a mixture of adamantanes (Figure 4.27).

4.4 ­Synthesis of Alkyl Diamondoids via Zeolite‐Catalyzed Rearrangement Most of above methods are not free from essential disadvantages, such as drastic conditions, the use of anhydrous aluminum halides or IL as catalysts, formation of a lot of wastes, and poor yields of the target product. Skeletal

135

4  Design and Synthesis of High‐Density Diamondoid Fuels

Abundance

136

7

Et-adamantanes

7.4

7.8

8.2

8.6

9

9.4

9.8

10.2

10.6

11

Time (min)

Figure 4.27  GC chromatogram of ETMA mixture. Source: Harrison and Harvey 2017. Reproduced with permission of Royal Society of Chemistry.

isomerization of tetracyclododecane was performed at 250–300 °C (reaction time 10–30  hours) in the presence of granular binder‐free Y‐zeolite (Figure  4.28). Samples of Y‐zeolite with a degree of Na+/H+ exchange of 0.40–0.97 were tested in an amount of 50–120 wt% with respect to tetracyclododecane. The most efficient catalyst for the rearrangement of IL into 1,3‐ DMA was Y‐zeolite with a degree of Na+/H+ exchange of 0.97; it ensured the maximum conversion of tetracyclododecane and the highest yield of 1,3‐ DMA. The results of the isomerization of compound are given in Table 4.12. The best yield of 1,3‐DMA was 67%, the conversion of tetracyclododecane being 90% (300 °C, 10 hours). Apart from 1,3‐DMA, the reaction mixture contained isomeric 1,4‐ and 1,2‐DMAs and 1‐ethyladamantane (EA) (Khusnutdinov et al. 2013, p. 1273). The isomerization of perhydroacenaphthene into 1,3‐DMA was catalyzed by Y‐zeolite with a degree of Na+/H+ exchange of 0.97 (Figure 4.28) (Khusnutdinov et  al. 2013, p. 1273). The reaction was carried out in hexane at 300–320 °C (reaction time 3–15 hours; catalyst concentration 50–100 wt%). The results were collected in Table 4.13. After 10 hours at 300 °C, the major product was 1,3‐DMA, the conversion of perhydroacenaphthene was complete, and the maximum yield of 1,3‐DMA was 65%. As follows from the data in Table 4.13, the yield of 1,3‐DMA is strongly determined by the amount of the catalyst. The best yield of 1,3‐DMA was obtained at a substrate‐to‐catalyst weight ratio of 1 : 1. Reduction of the amount of the catalyst from 100 to 50 wt% led to almost twofold reduction of the yield of 1,3‐DMA. As in the isomerization of TCDD, the reaction mixture also contained isomeric alkyl‐substituted adamantanes in an overall yield of 10–15%. Like the catalysis by aluminum halides, the isomerization in the presence of Y‐zeolite follows a carbenium ionic mechanism and involves intramolecular rearrangements with rupture of C─C bonds. The rate‐determining step in the

138

4  Design and Synthesis of High‐Density Diamondoid Fuels Y-NaH, 250–300 °C 10–30 h Tetracyclo[6.2.1.13,6 .02,7]dodecane

1,3-Dimethyladamantane

Y-NaH, 300 °C 10 h Perhydroacenaphthene

1,3-Dimethyladamantane

Figure 4.28  Synthesis of alkyl diamondoids catalyzed by zeolite. Source: Adapted from Khusnutdinov et al. (2013, schemes 1 and 2). Table 4.13  Synthesis of 1,3‐dimethyladamantane by skeletal isomerization of perhydroacenaphthene over Na/H‐Y zeolite. Weight ratio perhydroacenaphthene– catalyst–hexane

Temperature (°C)

Reaction time (h)

Yield (%)

100 : 50 : 50

300

10

34

100 : 50 : 50

320

7

30

100 : 50 : 100

300

5

38

100 : 50 : 100

320

4

32

100 : 100 : 50

320

4

55

100 : 100 : 50

300

15

63

100 : 100 : 50

300

10

65

100 : 100 : 50

320

7

61

100 : 100 : 0

300

7

53

100 : 100 : 100

300

7

56

100 : 100 : 100

300

5

55

100 : 100 : 100

300

3

58

Source: Khusnutdinov et al. (2013). Reproduced with permission of Springer.

isomerization of alkyl adamantanes is the formation of tert‐adamantyl cation. The rearrangements are accompanied by reduction of the internal molecular strain and increase of the degree of branching.

4.5 ­Alkylation and Other Chemical Synthesis Methods Alkylation is a facile and efficient route to functionalize adamantane with alkyl chains for reducing high melting point of the parent compound (270 °C) by more than 300 °C (Harrison, Rosenkoetter, and Harvey 2018, p. 7786). The diversity of the alkylation feedstock makes this story fascinating. A variety of alkyl sources

4.5  Alkylation and Other Chemical Synthesis Methods

and catalysts have been studied for the alkylation of adamantane, e.g. olefins, alcohols, alkyl halides, and alkanes. Superacid‐catalyzed (e.g. CF3SO3H) alkylation of adamantane with lower olefins (ethene, propene, and butene) gives alkyl adamantane that was formed by following proposed pathways: adamantylation of olefins by adamantyl cation formed through hydride abstraction from adamantane by alkyl cations (generated by the protonation of the olefins) and direct σ‐alkylation of adamantane by the alkyl cations via insertion into the bridgehead C─H bond of adamantane through a pentacoordinate carbonium ion (Olah et al. 1985, p. 7541). Besides, Lewis acids (Shen and Wu 1993) and alkyl peroxides (Tabushi and Fukunishi 1975) show significant activity on the alkylation of adamantane with alkenes or alcohols. However, in the presence of strong acid, which is necessary for adamantane activation, alkenes will rapidly isomerize and polymerize, with a low yield of alkyl adamantane. Alkyl halides are less effective as alkylating agents, and the alkylation of adamantane and its derivatives with RX (X = Cl, Br) catalyzed by AlCl3 proceeds nonselectively and is often accompanied by halogenation. Efforts have been made to search for convenient preparative methods of selective introduction of alkyl substituents into adamantane molecule (Figure 4.29) (Khusnutdinov et al. 2006, p. 159). The reaction of adamantane with alkyl halides was explored in the presence of metal‐complex catalysts based on platinum, rhodium, palladium, and ruthenium compounds, which are known to activate C─H and C─Hal bonds. It was found that, in the variety of examined catalysts, the ruthenium complexes exhibited noticeable activity in the alkylation of adamantane with alkyl halides. For example, the reaction of adamantane with chloro‐, bromo‐, and iodoalkanes in the presence of ruthenium compounds (RuCl3, Ru(OH)Cl3, Ru(PPh3)3Cl2) taken in a molar ratio of [Ru]:[adamantane]:[RX] = 1 : 100 : 100 in a CH2Cl2 medium resulted in the formation of alkyl adamantanes with a total yield of 70–98%. The process is highly regioselective: alkylation proceeds exclusively at the bridgehead (tertiary) carbon atom of adamantane, with the yield of 1‐alkyl‐ and 1,3‐dialkyladamantanes depending on the length and branching of the alkyl radical in a halogen alkane and on the reaction time. The activity of C1–C3 alkyl halides in this reaction is practically the same: they alkylate adamantane with three hours at 150 °C giving mono‐alkyl adamantane with quantitative yields. An increase in the length of the alkyl radical in RX gradually decreases the activity of alkyl halide. Adamantane alkylation with alkyl bromides and alkyl iodides proceeds more rapidly and more readily; however, the yield of alkyl adamantanes is considerably lower because of the formation of high‐molecular‐mass compounds. A GC‐MS analysis and titration of the reaction mixture with a NaOH solution showed that the reaction mixture contained hydrogen halides (HCl, HBr, HI) along with alkylated adamantanes, thus indicating that alkylation proceeds via the substitution of alkyl Figure 4.29  Alkylation of adamantane.

Alkenes Alcohols Alkyl halides Acid catalysis

R

139

140

4  Design and Synthesis of High‐Density Diamondoid Fuels

for the bridgehead hydrogen of adamantane. An increase in alkyl halide concentration increases the degree of adamantane alkylation. For example, a twofold excess of RX results in replacement of two bridgehead hydrogen atoms in the adamantane molecule and the formation of 1,3‐dialkyladamantanes. Notably, the production of alkyl‐diamondoid jet fuels requires high throughput method that does not rely on the use of stoichiometric organometallic reagents, can be conducted under moderate conditions, and selectively generates fuel fractions that have applications as high‐energy‐density jet fuels. Fortunately, adopting alkanes as the alkyl source addresses this problem. Concentrated sulfuric acid (Moore 1972), alumina catalysts (Frid et al. 1978, p. 8), and Lewis acids (Harrison, Rosenkoetter, and Harvey 2018, p. 7786) all promote the alkylation of adamantane with paraffins (hexane, pentane, heptane, and nonane) to alkyl adamantane. Notably, Harvey (Harrison, Rosenkoetter, and Harvey 2018, p. 7786) first investigated suitability of mixtures of alkyl diamondoids as high‐performance fuels synthesized by Lewis acid‐catalyzed cracking of intermediate chain length alkanes in the presence of adamantane (see Figure  4.30). Calculations suggest that cracking of paraffins takes place near the central carbon atoms due to the higher proton affinities of those atoms. Therefore, heptane cracks to form primarily propene and isobutane, while nonane cracks to form primarily C4 and C5 hydrocarbons. Hence, the primary components of the fuel mixture derived from adamantane and heptane (HA) were 1‐ethyl‐3‐methyladamantane (E) and 1‐propyladamantane (PrA) (F). Significant quantities of 1‐methyladamantane (B), 1,3‐ DMA (C), and 1,3,5‐trimethyladamantane (D) were also observed. When nonane was used as the alkyl source, the product distribution was shifted to higher molecular weight. In contrast to HA, virtually no 1‐methyladamantane, 1,3‐ DMA, or 1,3,5‐trimethyladmantane was observed (see Figure 4.31). Several alkyl diamondoids (e.g. 1‐pentyladamantane (PA), 2‐butyladamantane (BA), 2‐PrA, and 2‐EA) can be synthesized on a preparative scale, derived from diamondoid derivatives (Harvey et al. 2016, p. 10171). The first adamantane, PA, was selected with the goal of maximizing the cetane number through incorporation of a relatively long alkyl chain while keeping the viscosity and boiling point as low as Rx

AIBr3 Nonane

Heptane AIBr3

C13–C15

Ry

C11–C13

Figure 4.30  Synthesis of alkyl‐adamantane fuels via alkylation with AlBr3. Source: Harrison, Rosenkoetter, and Harvey (2018). Reproduced with permission of American Chemical Society.

4.5  Alkylation and Other Chemical Synthesis Methods E Relative abundance

F

7

D A

G

B

m/z = 220 (C16H28)

C

7.5

8

8.5

9

9.5

10

10.5

11

11.5

12

Time (min)

B

A

C

D

E

F

Relative abundance

192 (C14H24)

7

206 (C15H26)

206 (C15H26)

Ad 8

9

10

11

12

13

Time (min)

Figure 4.31  Gas chromatogram of the fuel mixture prepared via alkylation of adamantane with heptane (HA) and nonane (NA). Source: Harrison, Rosenkoetter, and Harvey (2018). Reproduced with permission of American Chemical Society.

possible. 1‐Adamantane carboxylic acid was converted to n‐butyl‐1‐adamantanylketone, which was then reduced under Wolff–Kishner conditions to generate PA (Figure 4.32). After evaluation of PA, an alkyl adamantane with a shorter hydrocarbon chain was targeted to potentially improve low‐temperature fluidity and increase the density. BA was prepared on a preparative scale by reaction of commercially available 2‐adamantanone with n‐BuLi, followed by dehydration of the resulting alcohol and, finally, hydrogenation to obtain the saturated product (Figure  4.33). The synthesis of EA was straightforward and similar to that of BA. Commercially available 2‐ethyl‐2‐adamantanol was dehydrated and hydrogenated to yield EA. To complete the series of short‐chain 2‐alkyl adamantanes and further elucidate the effect of the alkyl chain length on the DCN of these fuels, 2‐PrA was synthesized from 2‐adamantanone (Figure 4.34). Although 2‐PrA can be synthesized by reaction of 2‐adamantanone and propyl lithium, propyl lithium is not commercially available and is typically prepared as a dilute solute in cyclopentane.

141

142

4  Design and Synthesis of High‐Density Diamondoid Fuels O

CO2H 2nBuLi

N2H4 KOH

PA

Figure 4.32  Synthesis of 1‐pentyladamantane. Source: Harvey et al. (2016). Reproduced with permission of American Chemical Society. O nBuLi

HO

pTsOH Toluene

Pd/C, H2

BA

Figure 4.33  Synthesis of 2‐butyladamantane from 2‐adamantanone. Source: Harvey et al. (2016). Reproduced with permission of American Chemical Society. HO

HO

O BrMgCHCHCH3

Pd/C, H2

+

pTsOH Toluene

Pd/C, H2

Pd/C, H2 2-PrA

Figure 4.34  Synthesis of 2‐propyladamantane. Source: Harvey et al. (2016). Reproduced with permission of American Chemical Society.

4.6 ­Basic Properties of Alkyl Diamondoids The properties of the mentioned alkyl diamondoids are listed in Table 4.14. All alkyl diamondoids are colorless and transparent liquid with adequate low‐temperature performance and high density. Hence, alkyl‐diamondoid fuels are expected to show good performance in propulsion application.

144

4  Design and Synthesis of High‐Density Diamondoid Fuels

­References Amarasekara, A.S. (2016). Acidic ionic liquids. Chemical Reviews 116 (10): 6133–6183. Amde, M., Liu, J.‐F., and Pang, L. (2015). Environmental application, fate, effects, and concerns of ionic liquids: a review. Environmental Science & Technology 49 (21): 12611–12627. Bedford, A.F., Beezer, A.E., Mortimer, C.T. et al. (1963). Heats of combustion and molecular structure. Part IX. Bicyclo[2,2,1]heptane, 7‐oxabicyclo[2,2,1]heptane, and exo‐ and endo‐2‐methyl‐7‐oxabicyclo[2,2,1]heptane. Journal of the Chemical Society: 3823–3828. Berenblyum, A.S., Katsman, E.A., and Karasev, Y.Z. (2006). The nature of catalytic activity and deactivation of chloroaluminate ionic liquid. Applied Catalysis A: General 315: 128–134. Bragg, W.H. and Bragg, W.L. (1913). The structure of the diamond. Nature 91: 557. Chung, H.S., Chen, C.S.H., Kremer, R.A. et al. (1999). Recent developments in high‐energy density liquid hydrocarbon fuels. Energy & Fuels 13: 641–649. Cristol, S.J., Seifert, W.K., and Soloway, S.B. (1960). Bridged polycyclic compounds. X. The synthesis of endo and exo‐1,2‐dihydrodicyclopentadienes and related compounds. Journal of the American Chemical Society 82 (9): 2351–2356. Cui, S., Lu, B., Cai, Q. et al. (2006). Highly selective synthesis of diphenylmethane with acidic ionic liquids. Industrial & Engineering Chemistry Research 45 (5): 1571–1574. Cupas, C., Schleyer, P.v.R., and Trecker, D.J. (1965). Congressane. Journal of the American Chemical Society 87 (4): 917–918. Dahl, J.E., Liu, S.G., and Carlson, R.M.K. (2003). Isolation and structure of higher diamondoids, nanometer‐sized diamond molecules. Science 299 (5603): 96–99. Farooq, O., Morteza, S., Farnia, F. et al. (1988). Superacid‐catalyzed near‐ quantitative isomerization of C4n+6H4n+12 (n = 1–3) polycyclic precursors to diamondoid cage hydrocarbons promoted by 1‐haloadamantanes and sonication. Journal of Organic Chemistry 53 (12): 2840–2843. Fort, R.C. and Schleyer, P.v.R. (1964). Adamantane: consequences of the diamondoid structure. Chemical Reviews 64 (3): 277–300. Frid, T.Y., Solov’yev, V.N., Zaikin, V.G. et al. (1978). Alkylation on adamantane with iso‐octane in the presence of alumina catalysts. Petroleum Chemistry U. S. S. R. 18 (1): 8–14. Gund, T.M., Osawa, E., Williams, V.Z. et al. (1974). Diamantane. I. Preparation of diamantane. Physical and spectral properties. Journal of Organic Chemistry 39 (20): 101–107. HáIa, S., Landa, S., and Hanuš, V. (1966). Isolation of tetracyclo[6.3.1.02,6.05,10] dodecane and pentacyclo[7.3.1.14,12.02,7.06,11]tetradecane (diamantane) from petroleum. Angewandte Chemie International Edition 5 (12): 1045–1046. Harrison, K.W. and Harvey, B.G. (2017). Renewable high density fuels containing tricyclic sesquiterpanes and alkyl diamondoids. Sustainable Energy & Fuels 1 (3): 467–473.

­  References

Harrison, K.W., Rosenkoetter, K.E., and Harvey, B.G. (2018). High density alkyl diamondoid fuels synthesized by catalytic cracking of alkanes in the presence of adamantane. Energy & Fuels 32: 7786–7791. Harvey, B.G., Harrison, K.W., Davis, M.C. et al. (2016). Molecular design and characterization of high cetane alkyl diamondoid fuels. Energy & Fuels 30 (12): 10171–10178. Heneghan, S.P., Zabarnick, S., Ballal, D.R. et al. (1996). JP‐8+100: the development of high‐thermal‐stability jet fuel. Journal of Energy Resources Technology 118 (3): 170–179. Huang, M.‐Y., Wu, J.‐C., Shieu, F.S. et al. (2009). Isomerization of exo‐ tetrahydrodicyclopentadiene to adamantane using an acidity‐adjustable chloroaluminate ionic liquid. Catalysis Communications 10 (13): 1747–1751. Huang, M.‐Y., Wu, J.‐C., Shieu, F.‐S. et al. (2010). Isomerization of endo‐ tetrahydrodicyclopentadiene over clay‐supported chloroaluminate ionic liquid catalysts. Journal of Molecular Catalysis A: Chemical 315 (1): 69–75. Khusnutdinov, R.I., Schchadneva, N.A., Malikov, A.I. et al. (2006). Alkylation of adamantane with alkyl halides catalyzed by ruthenium complexes. Petroleum Chemistry 46 (3): 159–163. Khusnutdinov, R.I., Kislitsina, K.S., Khazipova, A.N. et al. (2013). Synthesis of 1,3‐ dimethyladamantane by skeletal rearrangement of C12H18 and C12H20 hydrocarbons over Na/H‐Y‐zeolite. Russian Journal of Organic Chemistry 49 (9): 1273–1278. Kim, S.‐G., Han, J., Jeon, J.‐K. et al. (2014). Ionic liquid‐catalyzed isomerization of tetrahydrotricyclopentadiene using various chloroaluminate complexes. Fuel 137: 109–114. Ladnak, V., Hofmann, N., Brausch, N. et al. (2007). Continuous, ionic liquid‐ catalysed propylation of toluene in a liquid–liquid biphasic reaction mode using a loop reactor concept. Advanced Synthesis & Catalysis 349 (4–5): 719–726. Lerman, B.M. (1991). Skeletal rearrangements of cage compounds with medium rings. Russian Chemical Reviews 60 (4): 736–763. Liu, Y., Hu, R., Xu, C. et al. (2008). Alkylation of isobutene with 2‐butene using composite ionic liquid catalysts. Applied Catalysis A: General 346 (1–2): 189–193. Ma, T., Feng, R., Zou, J.‐J. et al. (2013). Ionic liquid catalytic rearrangement of polycyclic hydrocarbons: a versatile route to alkyl‐diamondoid fuels. Industrial & Engineering Chemistry Research 52 (7): 2486–2492. Marchand, A.P. (2003). Diamondoid hydrocarbons–delving into nature’s bounty. Science 299: 52–53. Mckervey, M.A. (1974). Adamantane rearrangements. Chemical Society Reviews 3 (4): 479–512. Moore, R.E. (1972). Reaction of paraffins with adamantane compounds. US Patent 3,646,233, filed 14 October 1970 and issued 29 February 1972. Navrátilová, M. and Sporka, K. (2000). Synthesis of adamantane on commercially available zeolitic catalysts. Applied Catalysis A: General 203 (1): 127–132. Norton, R.V. and Howe, S.C. (1981). Production of high energy fuel. US Patent 4,270,014, filed 30 April 1979 and issued 26 May 1981.

145

146

4  Design and Synthesis of High‐Density Diamondoid Fuels

Olah, G.A. and Farooq, O. (1986). Chemistry in superacids. 7. Superacid‐catalyzed isomerization of endo‐ to exo‐trimethylenenorbornane (tetrahydrodicyclopentadiene) and to adamantane. Journal of Organic Chemistry 51 (26): 5410–5413. Olah, G.A., Farooq, O., Krishnamurthy, V.V. et al. (1985). Superacid‐catalyzed alkylation of adamantane with olefins. Journal of the American Chemical Society 107: 7541–7545. Olivier‐Bourbigou, H., Magna, L., and Morvan, D. (2010). Ionic liquids and catalysis: recent progress from knowledge to applications. Applied Catalysis A: General 373 (1): 1–56. Osawa, E., Tahara, Y., Togashi, A. et al. (1982). Fused and bridged tetracyclic C13 and C14 adamantanes. Synthesis of methyl‐2,4‐ethano‐, 1,2‐trimethylene‐, 2,4‐ trimethylene‐, and 1,2‐tetramethyleneadamantane. Journal of Organic Chemistry 47: 1923–1932. Pârvulescu, V.I. and Hardacre, C. (2007). Catalysis in ionic liquids. Chemical Reviews 107 (6): 2615–2665. Qiao, C.Z., Zhang, Y.F., Zhang, J.C. et al. (2004). Activity and stability investigation of [BMIM][AlCl4] ionic liquid as catalyst for alkylation of benzene with 1‐ dodecene. Applied Catalysis A: General 276 (1): 61–66. Rogers, R.D. and Seddon, K.R. (2003). Ionic liquids–solvents of the future? Science 302 (5646): 792–793. Rui, Z., Meng, X., Liu, Z. et al. (2008). Isomerization of n‐pentane catalyzed by acidic chloroaluminate ionic liquids. Industrial & Engineering Chemistry Research 47 (21): 8205–8210. Schleyer, P.v.R. (1957). A simple preparation of adamantane. Journal of the American Chemical Society 79 (12): 3292. Schleyer, P.v.R. and Donaldson, M.M. (1960). The relative stability of bridged hydrocarbons. II. endo‐ and exo‐Trimethylenenorbornane. The formation of adamantane. Journal of the American Chemical Society 82 (17): 4645–4651. Schleyer, P.v.R. and Nicholas, R.D. (1961). Further examples of the adamantane rearrangement. Tetrahedron Letters 2 (9): 305–309. Schneider, A., Ware, R.E., and Janoski, E.J. (1978). Isomerization of endo‐ tetrahydrodicyclopentadiene to a missile fuel diluent. US Patent 4,086,284, filed 3 September 1976 and issued 25 April 1978. Schwertfeger, H., Fokin, A.A., and Schreiner, P.R. (2008). Diamonds are a chemist’s best friend: diamondoid chemistry beyond adamantane. Angewandte Chemie International Edition 47 (6): 1022–1036. Shen, D.‐M. and Wu, M.M. (1993). Method for making diamondoid lubricant base stock by alkylation with an olefin in the presence of a Lewis acid. US Patent 5,345,020, filed 3 June 1993 and issued 6 September 1994. Silva, R.C., Silva, R.S.F., de Castro, E.V.R. et al. (2013). Extended diamondoid assessment in crude oil using comprehensive two‐dimensional gas chromatography coupled to time‐of‐flight mass spectrometry. Fuel 112: 125–133. Smith, G.P., Dworkin, A.S., Pagni, R.M. et al. (1989a). Brønsted superacidity of hydrochloric acid in a liquid chloroaluminate. Aluminum chloride‐1‐ethyl‐3‐ methyl‐1H‐imidazolium chloride. Journal of the American Chemical Society 111 (2): 525–530.

­  References

Smith, G.P., Dworkin, A.S., Zingg, S.P. et al. (1989b). Quantitative study of the acidity of HCl in a molten chloroaluminate system (AlCl3/1‐ethyl‐3‐methyl‐1H‐ imidazolium chloride) as a function of HCl pressure and melt composition (51.0–66.4 mol% AlCl3). Journal of the American Chemical Society 111 (14): 5075–5077. Snelders, D.J.M. and Dyson, P.J. (2011). Efficient synthesis of β‐chlorovinylketones from acetylene in chloroaluminate ionic liquids. Organic Letters 13 (15): 4048–4051. Sun, Y. and Shi, L. (2011). Removal of trace olefins from aromatics at room temperature using pyridinium and imidazolium ionic liquids. Industrial & Engineering Chemistry Research 50 (15): 9339–9343. Tabushi, I. and Fukunishi, K. (1975). Process of producing alkyl adamantanes. US Patent 3,928,480, filed 25 November 1974 and issued 23 December 1975. Waltman, R.J. and Ling, A.C. (1980). Mass spectrometry of diamantane and some adamantane derivatives. Canadian Journal of Chemistry 58: 2189–2195. Wang, L., Zhang, X.W., Zou, J.‐J. et al. (2009). Acid‐catalyzed isomerization of tetrahydrotricyclopentadiene: synthesis of high‐energy‐density liquid fuel. Energy & Fuels 23 (5): 2383–2388. Wang, L., Zou, J.‐J., Zhang, X.W. et al. (2011). Rearrangement of tetrahydrotricyclopentadiene using acidic ionic liquid: synthesis of diamondoid fuel. Energy & Fuels 25 (4): 1342–1347. Wang, L., Zou, J.‐J., Zhang, X.W. et al. (2012). Isomerization of tetrahydrodicyclopentadiene using ionic liquid: green alternative for Jet Propellant‐10 and adamantane. Fuel 91 (1): 164–169. Wang, B., Qin, L., Mu, T. et al. (2017). Are ionic liquids chemically stable? Chemical Reviews 117 (10): 7113–7131. Warren, R.W., Schneider, A., and Janoski, E.J. (1968). Separation and identification of bridgehead‐substituted methyl and ethyl adamantanes. Applied Spectroscopy 22 (2): 115–120. Wasserscheid, P. and Eichmann, M. (2001). Selective dimerisation of 1‐butene in biphasic mode using buffered chloroaluminate ionic liquid solvents–design and application of a continuous loop reactor. Catalysis Today 66 (2): 309–316. Wasserscheid, P. and Keim, W. (2000). Ionic liquids—new “Solutions” for transition metal catalysis. Angewandte Chemie International Edition 39 (21): 3772. Xie, J., Zhang, X., Xie, J. et al. (2019). Acid‐catalyzed rearrangement of tetrahydrotricyclopentadiene for synthesis of high density alkyl‐diamondoid fuel. Fuel 239: 652–658. Xing, E., Zhang, X., Wang, L. et al. (2007). Greener synthesis route for Jet Propellant‐10: the utilization of zeolites to replace AlCl3. Green Chemistry 9 (6): 589–593. Yang, Y.‐L. and Kou, Y. (2004). Determination of the Lewis acidity of ionic liquids by means of an IR spectroscopic probe. Chemical Communications 10 (2): 226–227. Yoo, K., Namboodiri, V.V., Varma, R.S. et al. (2004). Ionic liquid‐catalyzed alkylation of isobutane with 2‐butene. Journal of Catalysis 222 (2): 511–519. Zhang, X., Qiu, L., Tao, R. et al. (1999). Studies on the AlCl3 catalyst systems for the synthesis of adamantane. Petrochemical Technology 28 (8): 546–548.

147

148

4  Design and Synthesis of High‐Density Diamondoid Fuels

Zhang, Z., Song, J., and Han, B. (2017). Catalytic transformation of lignocellulose into chemicals and fuel products in ionic liquids. Chemical Reviews 117 (10): 6834–6880. Zhang, X., Pan, L., Wang, L. et al. (2018). Review on synthesis and properties of high‐energy‐density liquid fuels: hydrocarbons, nanofluids and energetic ionic liquids. Chemical Engineering Science 180: 95–125.

149

5 Design and Synthesis of High‐Energy Strained Fuels Ji‐Jun Zou, Junjian Xie, Yakun Liu, and Chi Ma Tianjin University, Key Laboratory of Advanced Fuel and Chemical Propellant of the Ministry of Education, Key Laboratory for Green Chemical Technology of the Ministry of Education, Department of Chemical Technology, School of Chemical Engineering and Technology, 92 Weijin Road, Tianjin 300072, China

5.1 ­Introduction Liquid fuels are widely used in aerospace vehicles with turbine, turbofan, ramjet, rocket, combined engine, etc. The propulsion performance of the engine is deeply dependent on the properties of the applied fuels, for which the most important features are density and volumetric energy content (Zhang et al. 2018, p. 95). For the volume‐limited aerospace vehicles, such as military mission and space exploration, their fuel tanks must be designed as small as possible to restore enough space for ordinance, electronics, and other components (Gao and Shreeve 2011, p. 7377). With this consideration, high‐energy density (HED) liquid fuel has been synthesized and used, because it has higher density and volumetric net heat of combustion (NHOC) than conventional distilled fuel and can provide more propulsion energy to extend the flight range and/or increase the payload. Most of the traditional HED fuels are cyclic hydrocarbons containing one or more five‐ or six‐membered rings to afford high density. Actually, the three‐ or four‐membered rings and caged fuels have largely distorted C─C bonds, and the strain energy of these strained rings is much higher than that of cyclopentane and cyclohexane, which endows higher NHOC than common cyclic hydrocarbons (Dunitz and Schomaker 1952, p. 1703; Morris, Quintana, and Harvey 2019, p. 1646).

5.2 ­Quadricyclane Fuel 5.2.1  Properties and Synthesis of Quadricyclane Traditional high‐density liquid fuels are mainly synthesized using petroleum‐ based feedstocks, which are consisted of multi‐cyclic hydrocarbons to afford high density. For example, JP‐10, RJ‐4, RJ‐7, and HDF‐T1 are typical polycyclic or bridged‐cyclic hydrocarbons. However, these cyclic fuels are normally High-Energy-Density Fuels for Advanced Propulsion: Design and Synthesis, First Edition. Ji-Jun Zou, Xiangwen Zhang, and Lun Pan. © 2020 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2020 by Wiley-VCH Verlag GmbH & Co. KGaA.

150

5  Design and Synthesis of High‐Energy Strained Fuels

c­ onstructed by five‐ or six‐membered rings, while the strained and caged fuels have several three‐ or four‐membered rings that are in metastable state and contain extra strain energy. The caged hydrocarbons with high energy, structural tension, and density are very promising to serve as high‐density fuels or fuel additives. For instance, Table 5.1 listed four typical strained and caged hydrocarbons, which all contain large energy in the strained and caged molecules. However, more strain energy leads to much more difficulty in formation; therefore, cubane and tetrahedrane are not yet available by a feasible synthetic process. Both quadricyclane (QC) and dihydrobenzvale are liquids at room temperature, a feature highly desirable for aviation purposes. Nevertheless, dihydrobenzvale is limited by its proclivity to rearrange to isomeric benzene. Since QC is stable and readily available commercially, it is a very promising high‐density fuel. Furthermore, the other mostly studied strained and caged liquid fuels involve diamondoid fuels and cyclopropane‐based fuels. Normally, QC (tetracyclo[3.2.0.02,7.04,6]heptane) is synthesized via photoisomerization of norbornadiene (NBD, [2.2.1]bicycloheptadiene). Initially, the photoisomerization of NBD to QC has been regarded as a potential way for solar energy accumulation (Dubonosov, Bren, and Chernoivanov 2002, p. 917; Bren et al. 1991, p. 451; Bach, Schilke, and Schlegel 1996, p. 4845). When 1 mol of NBD is transformed into QC, 89 kJ of solar energy could be stored in form of strain energy. Then, over some catalysts or thermal conditions, QC can release the stored strain energy in form of heat and inverse to NBD again (Figure 5.1a). QC is a strained and caged hydrocarbon with a high density of 0.982 g/cm3 (Dubonosov, Bren, and Chernoivanov 2002, p. 917; Bren et al. 1991, p. 451; Pan et al. 2010, p. 8526; Zou et al. 2010, p. 439; Pan et al. 2015a, p. 959), which contains 2 three‐membered rings, 1 four‐membered ring, and 2 five‐membered ring structures (Figure 5.1b). The bond angles in the three‐membered ring and four‐membered ring are in the range of 59.9–90.0°, much smaller than the ­normal 109.5°. With such strained and caged structure, QC possesses NHOC of 44.35 MJ/kg, much higher than the currently used high‐energy jet fuel JP‐10 (42.10 MJ/kg; Figure  5.1c). Importantly, QC maintains good low‐temperature properties (e.g. −40 °C viscosity: 0.03 Pa s; Pan et al. 2015a, p. 959), which is a very promising high‐density fuel. Traditional carbon–carbon coupling methods often have poor atom economy and are carried out under rather harsh conditions. Photochemical carbon–carbon coupling reactions are intrinsically advantageous, because activation is obtained Table 5.1  The typical strained and caged hydrocarbons. Fuel

Formula

Cubane

Tetrahedrane

Quadricyclane

Dihydrobenzvale

C8H8

C4H4

C7H8

C6H6

693.9

560.7

392.9

334.7

Molecule structure Strain energy (kJ/mol)

5.2  Quadricyclane Fuel

.1°

QC

1.550

c2 8

1.5 18

60

°

Density (g/cm3)

hv Photosensitizer

98.6°

3° 4.

c4

c5

NBD

8 51 1. 110.7°

c6

8 51

1.

10

(c)

c7

8 1.51

1.515

c5

18

° 60.1

c3

c4

1.5

59.9°

c

c1

c6 1.550

1.550

c2

° 90.0 .0 1 °° 990 0. 0° 90.0 1.515

1.51

c2

1.515

c3

110.7

c1

3°

(b)

Heat

4.

NBD

10

(a)

c3

Freezing Net Energy point content (°C) (MJ/kg)

QC

0.982

–44

44.35

JP-10

0.941

–79

42.10

QC

Figure 5.1  (a) NBD⇄QC for solar‐light utilization, (b) molecular structures of QC from different views, and (c) the properties of QC. Source: (b,c) Pan et al. (2014a). Reproduced with permission of Royal Society of Chemistry.

by the absorption of photons, which leave no residue, whereas most thermal catalytic methods involve toxic/polluting reagents as the catalyst itself or required additives, which must be eliminated from the end products. Photoreactions occur under unparalleled mild conditions and in many cases involve deep‐seated chemical transformations occurring in high yields and with high selectivity. This allows the researcher to design a shorter synthetic sequence with respect to alternative multistep methods, again in accord with the key postulates of green chemistry (Fagnoni et  al. 2007, p. 2725). Many photochemical carbon–carbon coupling methods have been developed, such as intermolecular addition onto double or triple C─C bonds (Mella et al. 1998, p. 81), intramolecular addition onto C═C bonds (Hintz et al. 1998, p. 1583), cycloaddition (including [2+2], [3+2], and [4+2] cycloaddition; Bauld 1989, p. 5307; Mattay, Trampe, and Runsink 1988, p. 1991), and radical coupling (Künneth et al. 1993, p. 209). Among them, cycloaddition is the most effective way to increase the energy density of fuels. The synthesis of high‐tension four‐membered rings has been easily ­ accomplished by [2+2] cycloaddition of alkenes after an initial photocatalyzed oxidation step as illustrated in Scheme  5.1. Sensitizer (S) was regenerated by back electron transfer (BET) from S− to the cyclobutane radical cation. Alternatively, this cation oxidized the initial alkene and initiated a short‐chain process (Fagnoni et al. 2007, p. 2725).

R

+ S*

R

S + R

R

R

R

R

R

or S

Scheme 5.1  Example of [2+2] cycloaddition.

R

R

151

152

5  Design and Synthesis of High‐Energy Strained Fuels

The most extensively studied [2+2] cyclodimerization involved phenyl(aryl) vinyl ethers (Ilyas and De Mayo 1985, p. 5093; Draper et al. 1984, p. 6222). The corresponding cyclobutane derivatives were obtained from parent phenyl vinyl ether in  30% yield when photocatalyzed by 1,4‐dimethyl terephthalate in MeCN (Kuwata, Shigemitsu, and Odaira 1973, p. 3803). Five‐membered ring compounds can be achieved through the photoinduced electron transfer induced ring opening of three‐membered ring derivatives (oxiranes, azirines, or cyclopropanes) followed by reaction with a double bond (Albrecht et  al. 1994, p. 219). Photochemical [4+2] cycloaddition can construct a six‐membered ring structure, which can also improve the fuel density. Moreover, photocatalytic [4+2] cycloaddition was generally preferable to thermally induced Diels–Alder reactions, such as those initiated by ground state acceptors such as ­triarylaminium salts (e.g. Ar3N+ SbF6−), in which a large excess of dienophile was required and underwent competitively isomerization or polymerization under these conditions. Actually, the direct photoinduced valence isomerization of NBD to produce QC is very difficult, because the optical absorption of NBD is c. 230 nm, which can only absorb the ultraviolet (UV) region of the light. Therefore, the utilization of photosensitizers is necessary to use high‐wavelength light and improve the isomerization efficiency. The photosensitizers can be divided into two types: homogeneous photosensitizers and heterogeneous photocatalysts. 5.2.2  Homogeneous Photosensitizers 5.2.2.1  Triplet Sensitizer

The homogeneous photosensitizers contain triplet sensitizer and transition‐ metal‐compound‐based sensitizer (Bren et  al. 1991, p. 451; Dubonosov, Bren, and Chernoivanov 2002, p. 917). For triplet sensitization, the energy of the triplet state (ET) of NBD is c. 257 kJ/mol, which is very high energy; therefore, only a few of sensitizers with higher ET than 257 kJ/mol can fit the classical triplet–triplet transfer process (Cuppoletti et al. 1999, p. 11253). As shown in Scheme 5.2, irradiation is normally carried out in the absorption band of sensitizer. First, it is turned into the singlet state by absorbing photons and then into the triplet state 3S by intersystem crossing. The triplet energy transfers from 3S to a ground state NBD molecule to generate the energy‐rich 3 NBD, followed by its adiabatic isomerization into 3QC and then relaxation to the ground state (QC) (Hammond et al. 1964, p. 2532; Wang et al. 2011, p. 1342). Hammond et  al. first reported the photosensitized of NBD to QC through triplet sensitizers such as benzophenone or acetophenone (Hammond et  al. 1964, p. 2532). Furthermore, many photosensitizers have been searched for isomerization, most of which are carbonyl compounds, including propiophenone, benzophenone, acridinone, Michler’s ketone, etc. (Gorman, Hamblett, and McNeeney 1990, p. 145; Ristić et  al. 1992, p. 7). Among them, the most efficient ones are in the order of Michler’s ketone, acetophenone, and benzophenone (Cahill and Steppel 2004). However, the mixture of sensitizers performed worse than the single sensitizer, perhaps due to the waste of energy in intermediate transfer step. With acetophenone or benzophenone as the sensitizer, the reaction is first order and the energy transfer rate is the rate‐controlling

5.2  Quadricyclane Fuel 3

3S+NB

+S 3 S

3 +S

hv

BCET

3

RET 3NB

= 3

NB+S

3

Q

+S

=

Scheme 5.2  The mechanism of triplet sensitization valence isomerization of norbornadiene to quadricyclane. BCET and RET are short for bond‐coupled electron transfer and return electron transfer, respectively.

step rather than the intersystem crossing. Differently, the order of reaction is definitely neither unity nor zero when using Michler’s ketone as sensitizer, which perhaps suggests that the rate‐controlling step is the intersystem crossing to the triple state activated intermediate (Philippopoulos and Marangozis 1984, p. 458). The effect of temperature is very little for acetophenone and benzophenone photosensitizer, due to their very small differences in ET value from NBD (−ΔET +15.06 to −6.28 kJ/mol), whereas with temperature increase from 38 to 60 °C, the conversion of NBD increases by 24% with Michler’s ketone as sensitizer due to the big −ΔET of −37.66 kcal/mol (Philippopoulos et al. 1983, p. 627; Philippopoulos and Marangozis 1984, p. 458). Quenching experiments using piperylene or 2,5‐dimethyl‐2,4‐hexadiene indicates that triplet excitation transfer from ketones to NBD occurs with rate constants of about 109 l/(mol s). The intermediate of excitation transfer to NBD either have lifetimes of 200 mol of QC is formed per mole of CuCl. The conversion of NBD can be >90%. Moreover, the photoreaction exhibits obvious solvent dependence: the isomerization proceeds efficiently in chloroform and ethanol, while no detectable conversion occurs in acetonitrile. Since CuCl and NBD are essentially transparent at 313 nm in ethanol or chloroform, the photoassisted conversion observed at this wavelength should characterize the ClCu–NBD complex. This conclusion is compelled especially by no detectable complex formation occurring in acetonitrile, a solvent in which photoreactivity is obviously absent. The mechanism for the photoassisted conversion of NBD by CuCl is presented in Scheme 5.3. Irradiation at 313 nm induces charge transfer and then the excited state of the ClCu–NBD π complex. The 3d electron of Cu(I) transfer to the lowest unoccupied π molecular orbital (ψ3) of NBD, or in the opposite sense, the electron transfer from the highest occupied π molecular orbital (ψ2) of NBD to the Cu(I). Both cases can weaken C2–C3 and C5–C6 and strengthen C2–C6 and C3–C5 (Schwendiman and Kutal 1977a, p. 719; 1977b, p. 5677). The electronically excited complex may then split between relaxation pathways that lead to NBD and QC. Importantly, no NBD dimers are produced during the photoreaction, because the single coordination sites on the metal [Cu(I)] can effectively inhibit dimerization (Schwendiman and Kutal 1977b, p. 5677). 4

5 6

1

3

CuCl

NBD

hv

CuCl

2

Ψ2

Q

Ψ3

Scheme 5.3  Proposed mechanism of the CuCl photoassisted isomerization of NBD to QC and representation of the highest filled (ψ2) and lowest unfilled (ψ3) π molecular orbits on NBD.

Furthermore, other transition‐metal‐compound‐based sensitizers, such as L3CuX (L = Ph3P, MePh2P; X = Cl, Br, I; Fife, Moore, and Morse 1985, p. 7077), orthometalated complex (2,2′‐bipyrid‐3‐yl‐C3,N′)bis(2,2‐bipyridine‐N,N′) iridium(III) (Grutsch and Kutal 1986, p. 3108), Cu(I) based dinuclear

5.2  Quadricyclane Fuel

complexes [Cu2L2(μ‐NBD), where L = 2‐methyl‐8‐oxoquinolinato, 2‐methyl‐ 5,7‐dichloro‐8‐oxoquinolinato, 4‐oxoacridinato, 2‐(2‐oxo‐3,5‐di‐tert‐butyl phenyl)benzotriazole] (Franceschi et  al. 1997, p. 4099), and Rh(III) diimine complexes [Rh(phen)33+ and Rh(phi)2(phen)3+, (phen = 1,10‐phenanthroline, phi = 9,10‐phenanthrenequinone diimine)] (Sluggett, Turro, and Roth 1997, p. 8834), have been applied to the photoisomerization of NBD. Importantly, these sensitizers with complex ligands can expand the optical absorption range to visible light to enhance solar‐light utilization efficiency (Rosi et al. 1999, p. 1520), but the high cost limits their practical applications. Besides, some metal complexes including carboxamide‐ and sulfonamide‐ linked polystyrene‐anchored cobalt(II) tetraarylporphyrins are also highly active catalysts for the conversion of QC to NBD. Unfortunately, these cobalt complex catalysts gradually become inactivated by partial oxidation during reaction, and the activity can be partially restored by treatment with a strong reducing agent such as titanium(III). When the catalyst is finely powdered, the apparent activity of these catalysts increases, indicating that the reaction is controlled by diffusion (King and Sweet 1979, p. 385). 5.2.3  Heterogeneous Photocatalysis 5.2.3.1  Zinc and Cadmium Oxides and Sulfides

Heterogeneous photocatalysts have been extensively used in photocatalytic processes such as degradation of pollutants, hydrogen generation, solar cell, etc. (Pan et  al. 2011, p. 10000; 2012a, p. 12782; 2013b, p. 252; 2014b, p. 71; 2015b, p. 7226; Chen and Mao 2007, p. 2891; Zhang et  al. 2017, p. 241). Considering the easy purification of product and reuse of catalyst, heterogeneous photocatalysts are also attractive for photoisomerization. In fact, semiconductors and zeolites were found to be active for the photoisomerization of NBD. Lahiry et al. firstly reported that NBD can be isomerized over semiconductors like ZnO, ZnS, and CdS in the presence of air (Lahiry and Haldar 1986, p. 71). ZnO does not exhibit any side reaction such as the formation of dimer of NBD, while ZnS and CdS will produce a little sulfur due to photolysis of catalyst. In particular, all catalysts will catalyze the conversion process only in the presence of oxygen, so oxygen plays a vital role in the reaction process. The reason maybe that the photoinduced electron produced on absorption of photons by semiconductor convert O2 into ·O2− while the positive semiconductor hole produced may draw electrons from NBD, forming the diradicaloid of NBD and then converting into QC. The excess of semiconductor hole may combine with ·O2−, forming semiconductor and O2. But in this mechanism, the cation radical of NBD was not considered, and more detailed mechanisms still need to be explored. 5.2.3.2  Modified Zeolites

Then, Ghandi et al. found that Y‐zeolites exchanged with heavy atoms including K+, Cs+, and Tl+ ions can catalyze the intramolecular addition of some dienes like NBD and afford the corresponding triplet products. This result could be explained because the porous structure decreases the mobility of diene within

155

5  Design and Synthesis of High‐Energy Strained Fuels

the zeolite, which in turn increases the energy transfer between the diene and heavy atom cation (Ghandi, Rahimi, and Mashayekhi 2006, p. 56). Similarly, Gu and Liu compared La‐, Cs‐, Zn‐, and K‐exchanged Y‐zeolites for the photoisomerization of NBD, and the activity order is as shown in Figure  5.2: LaY (t) > LaY > CsY > ZnY > KY (Gu and Liu 2008, p. 143). This is attributed to the heavy atom effect induced by spin–orbit coupling, which facilitates the transfer of the triplet state to the ground state. Among the above four metals explored, La is the largest atom and enjoys the highest activity. In addition, LaY is the only one to possess Brönsted acid sites, which may also accelerate the valence isomerization of NBD to QC. 5.2.3.3  Metal‐Doped TiO2

Among the photocatalysts, TiO2 is the most widely used owing to its low cost, nontoxicity, chemical inertness, and photostability (Pan et  al. 2012b, p. 1650; 2013a, p. 6593; Wang et al. 2015, p. 2975), so it has been also applied into the photoisomerization of NBD to QC. Although the activity of TiO2 is relatively low due to the low optical absorbance and high charge–hole recombination rate, many methods, such as doping, highly dispersed Ti–O species, etc., have been established to improve its photoactivity (Pan et al. 2015c, p. 576; Zou et al. 2012, p. 41). Doping with metal ions has been regarded as an effective method to improve the charge separation efficiency and further the photoactivity of TiO2 and other semiconductors (Pan et al. 2016a, p. 181; Wang et al. 2017, p. 235). Metal (Cu, Cr, Ce, V, Fe, Zn)‐doped TiO2 was studied for the photoisomerization of NBD (Pan et al. 2010, p. 8526; Zou et al. 2008b, p. 63). The structural parameters of prepared materials characterized by XRD, EDX, XPS, and N2 adsorption are shown in Table  5.2. According to the bulk composition from EDX data and surface 90

LaY (t)

80

LaY

70

CsY

60 Conversion (%)

156

50

ZnY

40

KY

30 20 10 0

1

2

3

4

5

6

7

Reaction time (h)

Figure 5.2  Photocatalytic activity of the zeolites prepared for the photoisomerization of NBD. LaY (t) means the twice exchanged LaY zeolites. Source: Gu and Liu (2008). Reproduced with permission of Springer Nature.

5.2  Quadricyclane Fuel

Table 5.2  Structural characteristics and composition of metal‐doped TiO2. Ti/M ratio Materials

Grain size (nm)

SBET (m2/g)

EDX

XPS

TiO2

21.5

21.5

—

—

Cu–TiO2(15)

19.9

13.1

90.4

3.8

Cr–TiO2(15)

14.7

40.9

20.0

3.0

Ce–TiO2(15)

11.4

64.3

16.9

19.8

V–TiO2(15)

9.9

102.7

19.0

15.6

Fe–TiO2(15)

7.0

120.6

18.5

19.8

Zn–TiO2(100)

8.1

84.9

—

7.1

Source: Pan et al. (2010). Reproduced with permission of American Chemical Society.

composition from XPS data, V, Fe, and Ce are dispersed in the inner part of prepared materials, whereas Cu, Cr, and Zn ions are enriched on the particle surface. Specifically, only a small amount of Cu is introduced into the material. Generally, there are three possible dispersion modes for dopants, namely, substitutional, interstitial, and surface positions. The local structure of dopants ions can be deduced based on their ionic radii, that is, Fe and V ions with radii close to Ti ions in substitutional site, large Ce ions in interstitial positions, and Cu ions with largest radii on the surface. The surface enrichment of relatively small Cr and Zn ions that have comparable radii with Ti ions is a little surprising because they could enter the lattice, but consistent with results reported by other researchers (Zhu, Tao, and Dong 2010, p. 2873; Jing et al. 2006, p. 17860). The reason may be that these ions are originally inside the lattice but diffuse to the surface through oxygen vacancies during the calcination process or the hydrolysis rate of these ions is much slower than that of Ti ions. When metal dopants are dispersed in the substitutional site, some Ti–O–M structures are expected to form, which will cause a shift in the binding energy of Ti species because the difference in Pauling electronegativity can induce electron transfer from Ti to M ions. As shown in Figure  5.3, the XPS signal (binding energy) of Ti is shifted to higher values after doping with V and Fe, while for other doping the shift is not so obvious because the metals are not located in the substitutional sites with no, or only a few, M–O–Ti structures formed. Doping can restrain the growth of particle to some degree no matter what the doping mode is, but the mechanism may be different. Fe and Zn doping produce considerably small particles (see Table 5.2 and Figure 5.4). For the substitutional doping like Fe and V doping, dopants in the lattice can destroy the crystal structure and restrain its growth. For the surface deposition or interstitial mode, like Ce and Zn doping, dopants may prevent the direct contact of TiO2 crystallites and retard them, agglomerating to the big particle. The relative photocatalytic activity of doped TiO2 (k/k0) is shown in Figure 5.5. Except Cu, doping metal ions show positive effect on the photoisomerization of  NBD, among which Zn–TiO2 and Fe–TiO2 are specifically active. The

157

5  Design and Synthesis of High‐Energy Strained Fuels 458.3 464.0 TiO2

c/s

158

Ce–TiO2 Cr–TiO2 Cu–TiO2 Fe–TiO2 V–TiO2

470

465 460 Binding energy (eV)

455

Figure 5.3  Ti 2p XPS spectra of metal‐doped TiO2. Source: Pan et al. (2010). Reproduced with permission of American Chemical Society.

(a)

20 nm

(c)

10 nm

(b)

10 nm

(d)

20 nm

Figure 5.4  TEM images of (a) pure TiO2, (b) Fe–TiO2(15), (c) V–TiO2(15), and (d) Zn–TiO2(100). Source: (b,c) Pan et al. (2010). Reproduced with permission of American Chemical Society. Source: (a,d) Zou et al. (2008b). Reproduced with permission of Elsevier.

5.2  Quadricyclane Fuel 2

k/k0

1.5

1

0.5

0

) ) ) ) 2 0) 5) (15 (15 (15 (15 TiO O 2(1 (10 2 2 2 2 O O O O 2 i i i i i O i T –T –T –T –T –T V– Cr Fe Cu Ce Zn

Figure 5.5  The activity of metal‐doped TiO2 for the photoisomerization of norbornadiene. Source: Zou et al. (2012). Reproduced with permission of IntechOpen.

photoisomerization reaction is a complex process, and the physicochemical properties of photocatalyst such as grain size, type of dopant ions, and their local structure are very important. A small particle is of course desired because it provides large active surface. It has been reported that the surface doping of Zn ions produces many surface OH groups that greatly enhance the intensity of surface photovoltage spectrum and photoluminescence and improve the photoactivity (Jing et al. 2006, p. 17860). As shown in Figure 5.6, the activity of NBD photoisomerization is also closely relative to the concentration of surface OH. However, the role of surface OH seems invalid for the materials with substitutional doping. As shown in Figure 5.7, the activity of Fe‐ and V‐doped TiO2 and

2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0

OH/OH0 k/k0 Pure TiO2

Ti/Zn = 33

Ti/Zn = 100

Ti/Zn = 200

Figure 5.6  Relationship of activity for the photoisomerization of norbornadiene and the relative surface OH concentration of Zn–TiO2. OH, the content of surface OH; OH0, the OH content of pure TiO2. Source: Zou et al. (2012). Reproduced with permission of IntechOpen.

159

160

5  Design and Synthesis of High‐Energy Strained Fuels

1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 (a)

Olattice/Olattice0 k/k0 Ti/Fe = 5 Ti/Fe = 10 Ti/Fe = 15 Ti/Fe = 25 Ti/Fe = 50

1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 (b)

Ti/V = 5

Ti/V = 10 Ti/V = 15 Ti/V = 25 Ti/V = 50

Olattice/Olattice0 k/k0

Figure 5.7  Relationship of activity for the photoisomerization of norbornadiene and the relative lattice oxygen concentration of (a) Fe–TiO2 and (b) V–TiO2. Source: Zou et al. (2012). Reproduced with permission of IntechOpen.

their lattice oxygen concentration, not the surface OH, change in an identical manner, strongly suggesting there is an inherent correlation between the photoisomerization and lattice oxygen. It is still not clear why two doping modes induce contrary results, probably because the reactant molecule is adsorbed on different sites that will be discussed latter. As to the role of substitutional dopants, it has been reported that metal ions in substitutional sites can improve the photoinduced charge transfer and separation (Wang, Yang, and Cao 2009a, p. 20912). It is believed that this process is very likely to occur through the M–O–Ti structure in which the metal dopants mainly serve as charge trapping and transferring center. Taking Fe–TiO2 as an example, the role of Fe is shown as follows. Fe ions temporarily trap photoinduced charges in the neighboring Ti–O moiety: 4 Ti

O2

Fe3

O

Ti 3

Ti 4

O2

Fe2

O

Ti 4 (5.1)

4 Ti

O2

Fe3

O

Ti 3

Ti 4

O2

Fe 4

O2

Ti 3 (5.2)

The trapped charges are transferred to sideward Ti–O species, resulting in separated charges:

5.2  Quadricyclane Fuel 4 Ti

O2

Fe2

O

4 Ti

O2

Fe 4

O2

Ti 4 Ti 3

Ti 3 Ti 4

O2

Fe3

O

Ti 4 (5.3)

O

Fe3

O2

Ti 3 (5.4)

In this way, the charge induced in one Ti–O moiety is quickly transferred to another Ti–O moiety through the Fe–O–Ti structure, thus effectively separating the charge and retarding the recombination (Pan et al. 2010, p. 8526). Furthermore, nanosized TiO2, especially quantum dots (QDs), generally show better performance than the bulk materials because of the highly dispersed photoactive sites and effective separation of photoinduced charges (Pan et  al. 2016b, p. 296). However, nanoparticles tend to aggregate with each other, suppressing their inherent high photoactivity. Loading TiO2 onto porous materials with a large surface area can avoid the aggregation of nanoparticles and also make the recycling of catalysts easy. Besides, transition metal (for example, the abovementioned V) doping with a suitable concentration can not only suppress the growth of TiO2 particles but also inhibit charge recombination. Pan et al. demonstrated that TiO2 QDs can be easily fabricated by combining V doping and loading strategies (Pan et al. 2014c, p. 988). V doping greatly influenced the crystal size of TiO2: trace V doping leads to particle size smaller than 5 nm, while the higher V amount inversely increases the size (Figure 5.8a). Moreover, many Ti3+ defects were simultaneously created in the QDs (Figure 5.8b), which is consistent with other research (Ma et al. 2016, p. 63984). Accordingly, the doped and Ti3+‐defected QDs show high charge‐separation efficiency and high photocatalytic activity in photoisomerization of NBD to QC (Figure 5.8c). 5.2.3.4  Ti‐Containing MCM‐41

Moreover, the local structure of the Ti‐oxide species can be tuned to improve the charge separation efficiency, which can be realized by supporting on porous structure. MCM‐41 has uniform hexagonal mesopores with large internal surface area, exhibiting great potential as the supporting materials of TiO2. It has been reported that incorporating Ti ions into a framework or loading them on the wall of MCM‐41 gives unique photocatalytic activity (Hu et al. 2003, p. 161; 2006, p. 1680). So both Ti‐incorporated and Ti‐grafted MCM‐41 materials were prepared for the photoisomerization of NBD. Grafting TiO2 in the pore of MCM‐41 does not affect the ordered hexagonal structure of support as its XRD patterns in the low‐angle region are identical to MCM‐41 (see Figure  5.9; Zou et  al. 2008a, p. 139). An additional peak corresponding to the (101) reflex of anatase TiO2 is observed at 25.5°, but the intensity is extremely weak, indicating that TiO2 crystallites are highly dispersed in the pore of MCM‐41. Incorporating Ti ions in the MCM‐41 framework slightly impairs the structural integrity of MCM‐41, but ordered structure is well retained, shown by the weakened but obvious diffractive peaks. Also, the cell unit of Ti–MCM‐41 is enlarged because the Ti─O bond distance is longer than the Si─sO bond distance. TEM images in Figure 5.10 further confirm the XRD result. No TiO2 nanoparticles are observed for TiO2–MCM‐41, and its pore structure is identical to MCM‐41, but some linear tubular pores of Ti–MCM‐41 collapse into irregular pores.

161

5  Design and Synthesis of High‐Energy Strained Fuels

Increase of V doping

Ti3+ defected V–TiO2 QDs CB

Ti3+

Ov V4+

Photoisomerization of NBD

EPR

g =1.989

k/h–1

Intensity (a.u./g)

0.6

g = 2.003

VTM30

3.06 ev

0.23 ev VB of quantum TiO2

V–Tio2 (a)

TM

2.95 ev

MCM-41

2.62 ev

3.30 ev

Decrease of V doping

g = 1.961

VTM20

0.4

0.2

VTM10 VTM5

3300

3350 3400 3450 Magnetic field (mT)

(b)

0.0

3500

TM TM30 TM20 TM10 VTM5 V V V Samples

(c)

Figure 5.8  (a) Effects of V doping amount on the crystal size of TiO2. (b) low‐temperature (110 K) EPR spectra, and (c) photoactivity of V‐TiO2/MCM‐41 samples. TM is the undoped sample, and VTMx is V doped TM, with x = Ti/Vmol. Source: Adapted from Pan et al. (2014c, figures 2 and 3).

(100)

Intensity (a.u.)

162

(110) (200)

10 20 30 40 50 60 70 80

MCM-41 TiO2–MCM-41 Ti–MCM-41(70) Ti–MCM-41(50) Ti–MCM-41(30)

2

4

6 2θ (°)

8

10

Figure 5.9  XRD patterns of Ti–MCM‐41 and TiO2–MCM‐41. Source: Zou et al. (2008a). Reproduced with permission of Springer Nature.

5.2  Quadricyclane Fuel (a)

(b)

(c)

Figure 5.10  TEM images of (a) MCM‐41, (b) TiO2–MCM‐41, and (c) Ti–MCM‐41(50). Source: Zou et al. (2008a). Reproduced with permission of Springer Nature.

The nature and coordination of Ti4+ ions were deduced according to the UV–visible (UV–vis) diffuse reflectance spectra shown in Figure  5.11. The absorption peak at 220 nm is ascribed to tetracoordinated Ti, whereas the peak at −270 nm represents species in higher coordination environments (penta‐ or hexacoordinated species). For Ti–MCM‐41, most of the Ti species are dispersed in the framework (Ti–O–Si) when Ti content is low but polymerized Ti species (Ti–O–Ti) present in the case of higher Ti content. TiO2–MCM‐41 contains highly dispersed quantum‐size TiO2 nanodomains, showing the blueshifted absorption compared with bulk TiO2. The overall activity for the photoisomerization of NBD is Ti–MCM‐41(30) >  Ti–MCM‐41(50) > TiO2–MCM‐41 > Ti–MCM‐41(70) > TiO2 in Figure  5.12a. Since the amount of Ti species is different in these materials, the activity based on TiO2 was also calculated to compare the inherent activity of different Ti species, with the order of Ti–MCM‐41(50) ≈ Ti–MCM‐41(70) > Ti–MCM‐41(30) > TiO2–MCM‐41 > TiO2, as shown in Figure 5.12b. Considering the local structure of Ti, it can be seen that TiO2–MCM-41

Absorbance (a.u.)

TiO2

Ti–MCM-41(70) Ti–MCM-41(30)

Ti–MCM-41(50)

200

300

400

500

600

700

800

Wavelength(nm) Figure 5.11  UV–vis diffuse reflectance spectra of Ti–MCM‐41 and TiO2–MCM‐41. Source: Zou et al. (2008a). Reproduced with permission of Springer Nature.

163

164

5  Design and Synthesis of High‐Energy Strained Fuels 1.4 1.2 1 0.8 0.6 0.4 0.2 0 TiO2/MCM-41 Ti–MCM-41(70) (a)

Ti–MCM-41(50) Ti–MCM-41(30)

k/k

0

700 600 500 400 300 200 100 0 TiO2/MCM-41 Ti–MCM-41(70) (b)

Ti–MCM-41(50) Ti–MCM-41(30)

k/k (b 0 a on TiOsed 2)

Figure 5.12  (a) The overall activity and (b) the activity based on TiO2 of Ti–MCM‐41 and TiO2–MCM‐41 for the photoisomerization of norbornadiene. Source: Zou et al. (2012). Reproduced with permission of IntechOpen.

framework Ti species are most active in the photoisomerization of NBD, followed by polymerized species, and bulk TiO2 has the lowest activity. 5.2.3.5  Combination of Metal Doping and Framework Ti Species

Transition‐metal‐incorporated MCM‐41 generally shows high photocatalytic activity due to the high dispersion of photoactive sites and effective separation of electrons and holes (Matsuoka and Anpo 2003, p. 225; Hu et  al. 2007, p. 139; Davydov et  al. 2001, p. 157). Since Ti–MCM‐41 exhibits high activity for the photoisomerization of NBD, it is expected that introducing second transition metal ion into Ti–MCM‐41 may further enhance the activity. So a series of transition‐metal‐incorporated (V, Fe, and Cr) Ti–MCM‐41 were synthesized for the photoisomerization of NBD, with Si/Ti ratio of 30. According to the UV–vis spectra in Figure  5.13, V and Fe ions are well dispersed in the materials, whereas the dispersion of Cr ions is very poor. For V–Ti–MCM‐41(150), V ions are highly dispersed in MCM‐41 framework at the atomic level with tetrahedral coordination, with some species in sixfold (absorption around 370 nm) and higher coordination or even polymerized environments (absorption in >400 nm region) formed with the increase of V content. This tendency is also observed for Fe–Ti–MCM‐41. However, for Cr–Ti–MCM‐41, the absorption at 470 and 610 nm ascribed to poly‐ and bulk Cr2O3 is very intensive. The local structure of Cr ions is also testified by the IR spectra in Figure  5.14. All Cr–Ti–MCM‐41 samples show a band shoulder at

5.2  Quadricyclane Fuel Fe–Ti–MCM-41

V–Ti–MCM-41

Absorbance (a.u.)

Absorbance (a.u.)

c a b d e f

200

a b c d e

300

(A)

400 500 600 Wavelength (nm)

700

200

800

300

(B)

400 500 600 Wavelength (nm)

700

800

Absorbance (a.u.)

Cr–Ti–MCM-41

a b c d e 200

(C)

300

400 500 600 Wavelength (nm)

700

800

Figure 5.13  UV–vis diffuse reflectance spectra of M ((A) V, (B) Fe, and (C) Cr)–Ti–MCM‐41 (a: Si/M = 10, b: Si/M = 33, c: Si/M = 75, d: Si/M = 100, e: Si/M = 150, f: Ti–MCM‐41). Source: Zou et al. (2010). Reproduced with permission of Elsevier. 898

630 570

Transmittance (%, a.u.)

a b c d e

1500

1200

900

Wavenumbers (cm-1)

600

Figure 5.14  IR spectra of Cr–Ti–MCM‐41 (a: Si/M = 10, b: Si/M = 33, c: Si/M = 75, d: Si/M = 100, e: Si/M = 150, f: Ti–MCM‐41). Source: Zou et al. (2010). Reproduced with permission of Elsevier.

880–900 cm−1 assigned to Cr6+ species; specifically, Cr–Ti–MCM‐41(10) has two bands at 630 and 570 cm−1 belonging to extra‐framework Cr2O3 oxides.

165

5  Design and Synthesis of High‐Energy Strained Fuels (100)

(110) (200)

20 40 60 80

Intensity (a.u.)

Si/V=10

Intensity (a.u.)

166

Si/V=33 Si/V=75 Si/V= 100 Si/V= 150 Ti–MCM-41 MCM-41

2θ (°)

Si/Cr =10 Si/Cr=33 Si/Cr =75 Si/Cr =100 Si/Cr =150

(a)

2

4

6 2θ (°)

8

10

(b)

2

4

6 2θ (°)

8

10

(c)

Figure 5.15  XRD patterns of (a) V–Ti–MCM‐41 and (b) Cr–Ti–MCM‐41, and (c) TEM image of Cr–Ti–MCM‐41(10). Source: Zou et al. (2010). Reproduced with permission of Elsevier.

The well‐dispersed V and Fe species show no obvious influence on the ordered structure of prepared materials, but the polymerized Cr species obviously impose a negative effect on the structure (see Figure 5.15). An extreme is observed for Cr–Ti–MCM‐41(10), in which the characteristic diffractive peaks of ordered structure completely disappear and a peak of bulk Cr2O3 appears. In TEM image, this material no longer possesses hexagonal mesoporous structure, but agglomerates of many crystallites. All the materials exhibit higher activity than Ti–MCM‐41 in Figure 5.16, indicating that introducing second metal is beneficial to the photoisomerization. Among the three metals, V incorporation is most effective, Fe incorporation follows, and Cr incorporation is the worst. The photocatalytic activity has nothing to do with the concentration of second transition metal ions, and the improvement in activity should be related to their state of dispersion and local structure. It has been reported that tetrahedrally coordinated M‐oxide moieties dispersed in mesoporous materials can be easily excited under UV and/or visible‐light irradiation to form corresponding charge transfer excited states (Matsuoka and Anpo 2003, p. 225; Yamashita et al. 2001, p. 435):

Mn

O2

hv

Mn

1

O

*

M V , Cr, Fe (5.5)

Then M species can donate an electron to surrounding Ti–O moieties and O− can scavenge an electron from surrounding Ti–O moieties, inducing charge separation in Ti–O species (Davydov et al. 2001, p. 157). Therefore, two different

5.2  Quadricyclane Fuel

2.5 2.0

1.0

k/k0

1.5

0.5

150

V

100

75

Si/M

0.0

Fe 33 10 T i–M

rati

o

Cr

CM

-41

Figure 5.16  The activity of M (V, Fe, and Cr)–Ti–MCM‐41 for the photoisomerization of norbornadiene. Source: Zou et al. (2010). Reproduced with permission of Elsevier.

excitation mechanisms exist in M–Ti–MCM‐41. One is direct excitation of Ti–O moieties by UV irradiation, and the other is indirect excitation via charge transition from [M(n−1)+ − Q−]* species. The second process should be responsible for the high photocatalytic activity of M–Ti–MCM‐41 because of its high efficiency in charge formation and separation. V–Ti–MCM‐41(150) shows specifically high activity because the majority of V ions are highly dispersed in fourfold coordination, which brings up highly efficient excitation of Ti–O species. In addition, the well‐retained ordered structure and high surface area can provide more active sites to enhance the adsorption of NBD molecules. With the increase of V content, the activity is decreased because some fourfold ions are transformed into undesirable highly coordinated species and the damaged structure and small surface area may suppress the adsorption of reactants. The low activity of Cr–Ti–MCM‐41 is due to poorly dispersed chromium ions and dramatically destroyed textural structure. Since some photocatalysts show absorption in the visible‐light region, one may wonder whether they can catalyze the isomerization under visible‐light irradiation. However, there is no observable conversion when the experiment was conducted using visible irradiation (>420 nm). This is different from the case of H2 generation and organic degradation where Cr in that Ti–MCM‐41 is reported to exhibit visible‐light activity (Chen and Mao 2007, p. 2891; Davydov et al. 2001, p. 157; Yamashita et al. 2001, p. 435). These results suggest that the reaction mechanism between the photoisomerization and other photocatalytic reactions maybe very different. 5.2.3.6  Mechanism of Heterogeneous Photocatalysis

For the heterogeneous Ti‐containing photocatalysis, the triplet–triplet isomerization mechanism is not suitable because the vertical triplet energy transfer from Ti‐oxide species to NBD is very difficult. NBD molecules have to be firstly positively charged by photoinduced holes, but the free radical ion isomerization

167

168

5  Design and Synthesis of High‐Energy Strained Fuels

TiO2

QC

CB

e–

hv Ti4+ – [O2–]lattice – Ti4+ – O(H)– + QC

Ti4+ – [O2–]lattice – Ti4+ – O(H)2– … QC+

VB

NBD h+

Ti4+ – [O2–]lattice – Ti4+ – O(H)– … NBD Ti4+ – [O2–]lattice – Ti4+ – O(H)2– … NBD+

Figure 5.17  Photoisomerization of NBD via adsorption–photoexcitation over semiconductor. Source: Zou et al. (2012). Reproduced with permission of IntechOpen.

mechanism is ruled out because the energy of free NBD+ is significantly lower than free QC+. In fact, the transformation of QC to NBD is through the QC+ → NBD+ free radical route (Ikezawa and Kutal 1987, p. 3299). So the photoisomerization of NBD over semiconductors should be an adsorption– photoexcited process, which is very likely through the exciplex (charge transfer intermediate; see Figure 5.17). First, NBD molecule is adsorbed on the photoexcited Ti oxides. Then surface‐trapped hole is transferred to absorbed molecule, and a complex with NBD positively charged is formed. Subsequently, the complex is transformed to structure with QC skeleton. Finally, QC is released into the liquid phase and the charge is recombined through reverse electron transfer. In this case the adsorption and charge transfer are two critical steps. The adsorbing site on different Ti‐containing materials may be different. For Zn–TiO2, surface OH very likely serves as the site because it plays an important role in the reaction, and the excited complex may be TiO2− − OH⋯NBD+. For Fe–TiO2 and V–TiO2, however, the lattice oxygen may work as the adsorbing site with the complex of Ti4+ − [O2−] − Ti4+ − O2−⋯NBD+. Any charge recombination process can deactivate the complex, and the function of dopants and framework Ti species can effectively inhibit the undesired recombination. 5.2.4  Utilization of Quadricyclane Containing very high strained energy and further highly chemical reactivity, QC has been found to be an excellent hypergolic fuel (Pan et al. 2014a, p. 50998). The ignition delay (ID) time of QC–NA (nitric acid) binary system is 98 ms (Table 5.3), and particularly, the ID time of QC–N2O4 binary system is only 29 ms. This superior hypergolicity is very encouraging because the target ID time is approximately 50 ms for the practical applications. As seen in Figure 5.18, the extremely powerful and intense single flames appear upon the contacting of fuel and oxidant. Besides, adding boron (B) or carbon (C) in nanosize into QC fuel as both energetic and ignition additives can efficiently enhance the hypergolic performance of QC. Fuels containing 0.25 wt% nanoparticles show significantly reduced

5.2  Quadricyclane Fuel

Table 5.3  The ID time of QC‐based hypergolic liquid bipropellant. Entry

Fuel Oxidant a

ID time (ms)

1

2

3

4

5

6

QC

QC

QC + C

QC + B

QC + C

QC + B

N2O4

NA

N2O4

N2O4

NA

NA

29

98

27

18

73

68

QC+N2O4

a) As measured from the time the drop hits the oxidizer until the first sign of a flame. Source: Pan et al. (2014a). Reproduced with permission of Royal Society of Chemistry.

t = 28 ms

t = 29 ms

t = 30 ms

t = 0 ms

t = 97 ms

t = 98 ms

t = 99 ms

QC+WFNA

t = 0 ms

Figure 5.18  Ignition process of QC/N2O4 and QC/NA recorded by high‐speed camera. The fuel droplets are marked by green circles. Source: Pan et al. (2014a). Reproduced with permission of Royal Society of Chemistry.

ID time (Table 5.3). Especially in the presence of B, the ID time is decreased from 29 to 18 ms for N2O4 and from 98 to 68 ms for NA, respectively. Pan et al. further estimated the propulsion performance of QC/N2O4 (Pan et al. 2014a, p. 50998). As shown in Table 5.4, the computed specific impulse of QC is higher than unsymmetrical dimethylhydrazine (UDMH) and ethylidenenorbornene (EN) due to the high heat of combustion. Moreover, the density of QC is much higher than UDMH, so its volumetric specific impulsion, important for volume‐limited aerospace vehicles, is 15% higher than that of UDMH. Besides, the vapor pressure of QC (4.5 kPa at 29 °C; Hall, Smith, and Baldt 1973, p. 3197) is much lower than that of UDMH (22.3 kPa at 25 °C; Hess 2005, p. 805), leading to a lower flash exposition risk and harmfulness in the applications. Moreover, QC is easy to greenly synthesize on a large scale, safe to handle, transport, and store, and therefore it is a promising benign and energetic hypergolic fuel.

169

170

5  Design and Synthesis of High‐Energy Strained Fuels

Table 5.4  Calculated propulsion properties of QC, UDMH, and EN. Fuel

QC

UDMH

EN

Heat of formation (kJ/mol)

307.8

84.0

102.3

Heat of combustion (kJ/g)

−44.35

−31.02

−43.01

Density of fuel (g/cm3)

0.982

0.793

0.896

Density of propellant (g/cm3)

1.32

1.14

1.30

Specific impulsion (s)

305

297

298

Volumetric specific impulsion (kg s/l)

402

337

388

Source: Pan et al. (2014a). Reproduced with permission of Royal Society of Chemistry.

5.3 ­Cyclopropane Fuel Currently, high energy density fuels mainly include three types: polycyclic hydrocarbon fuels, adamantane fuels, and high‐tension cage hydrocarbon fuels. The chemical reactions involved in the synthesis of HED fuels include Diels–Alder reactions, hydrogenation reactions, and Simmons–Smith cyclopropanation reaction. Among them, the synthesis of a three‐membered ring with high tension by cyclopropanation is one of the most popular reactions for the synthesis of high‐energy fuels. The reasons are as follows: the cyclopropanyl compounds synthesized by cyclopropanation have high density and high energy caused by ring tension, which is a significantly promising fuel. In 1906, Oswald Silberrad reported for the first time the catalytic decomposition of ethyl diazoacetate (EDA) with a copper catalyst to form diethyl fumarate and diethyl maleate. As a starting point of catalytic carbene reaction, a series of important carbon–carbon bonding reactions, such as cyclopropanation, dipolar addition, C–H, and N–H insertion reactions, have been discovered. The cyclopropanation reaction method generally includes a metal carbene‐mediated cyclopropanation reaction, a transition metal‐catalyzed cyclopropanation reaction by decomposition of a diazo compound, and an ylide cyclopropanation reaction. Among them, the cyclopropanation reaction initiated by the metal carbene‐mediated cyclopropanation reaction and the transition metal‐catalyzed decomposition of the diazo compound is widely used. And commonly used metal carbene bodies include the following four types: zinc carbenoids (for example, Simmons–Smith reagent [Simmons and Smith 1958, p. 5323; 1959, p. 4256], Furukawa reagent [Furukawa, Kawabata, and Nishimura 1966, p. 3353]), lithium carbenoids (Closs and Moss 1964, p. 4042), aluminum carbenoids (Maruoka, Fukutani, and Yamamoto 1985, p. 4412; Maruoka, Sakane, and Yamamoto 1989, p. 176), and samarium carbenoids (Molander, Etter, and Zinke 1987, p. 453; Molander and Harring 1989, p. 3525). The reaction will be described in detail below. 5.3.1  Organometallic Carbenoid‐Mediated Cyclopropanation The earliest cyclopropanation reaction was the Simmons–Smith reaction, which was first reported by Simmons and Smith (1958, p. 5323). It is generally the reaction

5.3  Cyclopropane Fuel

between olefins and metal carbenoid. The initial metal carbenoid is a compound produced by reaction of diiodomethane with Zn–Cu couple. Thereafter, in order to improve the selectivity and yield, a large number of metal carbenoid were explored. Their basic structure of metal carbenoid is RMCH2X, wherein R represents some atoms or groups, M represents metal atoms such as Zn, Sm, Al, Li, etc., and X represents halogen atoms such as Cl, Br, and I. At present, the most widely used is zinc carbenoid, and the most reactive is samarium carbenoid, which can synthesize a large amount of cyclopropane products at a low temperature of −78 °C (Molander, Etter, and Zinke 1987, p. 453; Molander and Harring 1989, p. 3525; Padwa, Dent, and Yeske 1987, p. 3944), while zinc carbenoid needs higher temperature. 5.3.1.1  Zinc Carbenoid‐Mediated Cyclopropanation

The most classic cyclopropanization reaction mediated by zinc carbenoid is Simmons–Smith reaction. In 1958, Simmons and Smith reported the reaction of diiodomethane and Zn–Cu couple with cyclohexene in an ether solvent to form bicyclo[4.1.0]heptane (Simmons and Smith 1958, p. 5323), as shown in Eq. (5.6). In this reaction, diiodomethane first forms an organozinc compound with zinc powder, which is capable of transferring a methylene group to a carbon–carbon double bond in cyclohexene to construct a three‐membered carbon ring. The generation of two new δ bonds and the cleavage of π bonds are generally accomplished in a coordinated manner. The Simmons–Smith cyclopropanation reaction, although lower in yield, was developed as a new method for converting olefinic unsaturated double bonds into three‐membered rings, and various metal carbenoids are gradually explored. +CH2I2



+Zn(Cu)

(C2H5)2O

+ZnI2 +(Cu)

(5.6)

To explain the reaction mechanism, Simmons and Smith proposed a methylene transfer mechanism with a concerted [1+2] addition via a butterfly‐type transition state (TS) (see Scheme 5.4, path A), accompanied by the migration of halide from carbon to the metal atom. Later, a two‐step carbometalation mechanism involving a four‐center transition state was postulated (see Scheme 5.4, path B). It suggests that the metal carbenoid acts as a nucleophile to attack the carbon–carbon double bond and form a new carbon–carbon single bond and carbon–metal bond through a four‐center transition state. An intramolecular SN2 substitution reaction then takes place to give the final product. Assuming that the metal carbenoid is a nucleophile, the electron‐deficient olefin should be more reactive than the electron‐rich olefin; however, this speculation is contrary to the experiment. The mechanism of methylene transfer (Simmons and Smith 1959, p. 4256) suggests that the metal carbenoid has electrophilicity and can be added to the double bond in one step through a three‐center “butterfly” transition state. This mechanism can explain the experiment such as the stereospecificity of olefin addition, the high reactivity of electron‐rich olefins relative to electron‐deficient olefins, and the reaction kinetics reported by Simmons and Smith. Therefore, the methylene transfer mechanism is widely acceptable for cyclopropanation reaction.

171

172

5  Design and Synthesis of High‐Energy Strained Fuels R

X M

M

Path A H C

X R

M

CH2

H

+

H

H2 C

C

H

H

TS (methylene transfer)

C

H

H C

Products

+ H

X

R

CH2

H

H C

C

H

X

H

Path B

H

C

C

H

CH2 H

H

H C

M = Zn, Al, Sm X = halogen or OR

M

CH2

M

X

R

R

C

H

H

H

TS (carbometalation)

Scheme 5.4  Mechanism of cyclopropanation involving carbenoid structures.

In 1968, Furukawa reported that active intermediates prepared from Et2Zn to CH2I2 could rapidly and efficiently convert olefins to cyclopropanes. The electron‐donating group attached to the C═C double bond in the olefin can increase the yield and reaction rate. In addition, nonpolar hydrocarbon solvent can obtain a higher yield. However, polar solvents such as tetrahydrofuran (THF) and diglyme may coordinate with the active intermediate, which hinders the formation of cyclopropane and results in a low yield. It is worthy note that for the cyclopropanation of vinyl ether, the Simmons–Smith reaction can produce cyclopropyl ether due to polymerization of vinyl ether. However, with this new method reported by Furukawa (see Eq. (5.7)), vinyl ether is converted to cyclopropyl ether in a good yield, which perfectly solves the defects of Simmons–Smith method. This is a big improvement over the traditional cyclopropanation reaction. Me C H

CH2I2

H C O

i

C4H9

Et2Zn

H2 C

Me C H

H C O

i

C4H9

(5.7)

The rate and yield of cyclopropanation reaction primarily depend on the structure of zinc carbenoid. By changing the R group, the activity of zinc carbenoid can be enhanced, thereby increasing the reaction rate and yield. In 1998, Shi et  al. (Yang, Lorenz, and Shi 1998, p. 8621) reported a highly active zinc carbenoid prepared by the reaction of diethyl zinc, diiodomethane, and Lewis acid (see Eqs. (5.8) and (5.9)). The cyclopropanation of methylstyrene can be obtained by RXZnCH2I at room temperature. And the zinc carbenoid CF3COOZnCH2I prepared from trifluoroacetic acid has higher activity than other zinc carbenoids, as shown in Figure 5.19. In addition, the method reported by Shi et al. has good reactivity with olefins that are not reactive with traditional zinc carbenoid, such as stilbene.

5.3  Cyclopropane Fuel 100

Conversion (%)

80 60

a b c d e f g

40 20 0 0

5

10

15

20

25

Time (h)

Figure 5.19  Plot of the conversion of trans‐β‐methylstyrene against time (h). The curves presented are (a) No RXH, (b) EtOH or CICH2CH2OH, (c) CI2CHCH2OH, (d) CC13CH2OH, (e) CF3CH2OH, (f ) PhCO2H, and (g) CF3CO2H. Source: Yang, Lorenz, and Shi (1998). Reproduced with permission of Elsevier.

Et 2Zn 2CH2 I2 Et 2Zn RXH

2EtI

ICH2

EtH RXZnEt

2

RXH

Zn CH2 I2

RXZnCH2 I CH3I (5.8)

RXZnCH2 I EtI (5.9)

In 2000, Andre B. Charette et al. reported a new zinc carbenoid ArOZnCH2I (Charette et al. 2000, p. 4539), which is prepared as shown in Scheme 5.5. In both reaction paths, although the order of reagents addition is different, the same zinc carbenoid can be obtained. This metal carbenoid has a strong reactivity to non‐ functionalized olefins (see Scheme  5.6). A high conversion is obtained for unfunctionalized olefin such as cyclohexene, 4‐zincene, or the like. Compared to OZnCH2I

OZnEt

OH

Et2Zn

CH2I2

–40 °C

–40 °C R

R

R

R OH

Zn(CH2I)2 CH2Cl2

–40 °C

Scheme 5.5  Methods for synthesis of ArOZnCH2I.

173

174

5  Design and Synthesis of High‐Energy Strained Fuels OH

OZnI

OZnCH2I

(1) Et2Zn (2) CH2I2 R

R

R1

R1

R

R2

R3

R2

R3

Scheme 5.6  Reaction of ArOZnCH2I with aryl‐substituted alkenes.

the traditional Simmons–Smith method, iodomethylzinc phenoxides have better reactivity and stability to unfunctionalized olefins. 5.3.1.2  Samarium Carbenoid‐Mediated Cyclopropanation

After the discovery of the Simmons–Smith reaction, a great deal of work has been done to improve and develop alternative methods to produce similar active reagents. Many of these Simmons–Smith‐type reagents and related carbenoids are prepared from CH2I2 (or other polyhalomethanes) and a metal atom to form the general RMCH2X structures. In 1987, samarium/mercury amalgam in conjunction with CH2I2 was reported by Molander (Molander and Etter 1987, p. 3942; Molander, Etter, and Zinke 1987, p. 453), which is thought to be one of the most efficient and highly diastereoselective cyclopropanating reagents. Several stereoselective reactions of the Sm/CH2I2 and/or the Sm/CH2ICl reagents have also been reported for cyclopropanation of allylic alcohols by Lautens and Delanghe (1992, p. 798; 1993, p. 5037; 1994, p. 8526; 1995, p. 2474) and Cossy, Blanchard, and Meyer (1998, p. 5728). As so far, the Sm/CH2I2 carbenoid is believed to be one of the most efficient and highly diastereoselective cyclopropanating reagents. Zhao, Wang, and Phillips (2003, p. 15200) conducted a theoretical calculation on the samarium carbenoid ISmCH2I with ethylene. The ISmCH2I carbenoid was found to possess a “samarium carbene complex” character with the structure, properties, and chemical reactivity similarly to previously studied lithium carbenoids (LiCH2X where X  =  Cl, Br, I) but significantly different from the related classical Simmons–Smith carbenoid (IZnCH2I). This explains why the ISmCH2I carbenoid cyclopropanation reaction is able to occur at low temperature. The ISmCH2I carbenoid cyclopropanation reactions can proceed via a methylene transfer pathway or a carbometalation pathway, as shown in Figure 5.20. The effect of THF solvent on the reactions was investigated by using explicit coordination of the solvent THF molecules to the Sm(II) center of ISmCH2I. The barriers for the ISmCH2I/(THF)n (where n  =  0,1,2) carbenoid methylene transfer pathway reactions became progressively lower as additional

5.3  Cyclopropane Fuel 14.2 TS6 10.9 9.7

TS4 TS2

6.1 5.8 5.5

TS5 TS3 TS1

0.0

SM = ISmCHI2/(THF)n + C2H4 a: n=0 b: n=1 c: n=2

Carbometalation

–2.7 RC3 –4.5 RC2 –7.4 RC1

E (kcal/mol)

IM3 –5.5 –7.5 IM2 –10.2 IM1

SM

RC3 –2.7 –4.5 RC2 –7.4 RC1

∼ c-C3H6 + ∼∼ ∼∼ ∼ Sml2/(THF)n c ba –45.9 –48.7 –52.2

Methylene transfer

Reaction coordinate

Figure 5.20  Schematic diagram showing the computed relative energies (in kcal/mol) at the B3LYP/6‐311A and B3LYP/6‐311B (the data in parentheses) levels for reactions of ISmCH2I/ (THF)n (n = 0, 1, 2) with ethylene with the transition states and products energies given relative to the starting materials [SM = ISmCH2I/(THF)n + C2H4 (n = 0, 1, 2)]. Source: Zhao, Wang, and Phillips (2003). Reproduced with permission of American Chemical Society.

THF molecules added with the barrier decreasing from 12.9 kcal/mol for ISmCH2I to 8.8 kcal/mol for ISmCH2I/(THF)2. However, barriers to the carbometalation pathway stay relatively high (>15 kcal/mol). It indicates that the THF solvent can enhance the reactivity of ISmCH2I/(THF)n carbenoids. 5.3.1.3  Lithium Carbenoid‐Mediated Cyclopropanation

Lithium carbenoid is the most effective cyclopropanation reagent among these carbenoid reagents, and some lithium carbenoids even can cyclopropanate olefins efficiently at −110 °C (Closs and Moss 1964, p. 4042; Stiasny and Hoffmann 1995, p. 619). The structure of lithium carbenoid is quite different from zinc carbenoid, samarium carbenoid, and aluminum carbenoid. In the preparation of lithium carbenoid, the α‐hydrogen of halogenated hydrocarbon is acidified by halogen and is easily substituted by lithium atom. However, due to the ionic nature of carbon and lithium atoms, a rapid lithium halide elimination reaction usually occurs after metallization of halogenated hydrocarbon, resulting in poor thermal stability of lithium carbenoid. In order to avoid the elimination of Li–X, halogenated hydrocarbon should be metallized at a low temperature to make the carbon–lithium bond and the carbon–halogen bond of the synthesized lithium carbenoid is relatively stable. For example, Köbrich et al. prepared lithium carbenoid with 1‐chloro‐2,2‐diarylethylenes and tert‐butyllithium, but it is stable only at very low temperature such as −105 °C (Köbrich et al. 1967, p. 41).

175

176

5  Design and Synthesis of High‐Energy Strained Fuels

However, not all lithium carbenoids have poor thermal stability. The first example of Li/halide (Li/Hal) carbenoid, stable at room temperature, has been reported by Le Floch and coworkers (Cantat et al. 2007, p. 5947). An X‐ray analysis has revealed a strong stabilization occurring for the lithium cation provided by the two sulfur atoms and two molecules of diethyl ether that prevent LiCl elimination, thus accounting for thermal stability. Another feature of lithium carbenoids is the ambiphilic behavior, which originates from the coexistence of an electron‐donating and an electron‐withdrawing substituent at the carbon center. As a consequence, lithium carbenoids have nucleophilic as well as electrophilic reactivity (Boche and Lohrenz 2001, p. 697; Braun 1998, p. 430; Maercker 1993, p. 1023). The “chameleon‐like” reactivity becomes evident from the resonance formulas of the carbenoid b in Eq. (5.10): whereas the carbanionic character is expressed by the resonance formula a, the electrophilic character is represented by c. X

X

C

C

Li



X C

Li

a

b

Li

c

(5.10)

In addition, Boche et al. studied the influence of leaving group on the structure and reactivity in the carbenoid series LiCH2X (X  =  Hal, OH) toward a “classic” cyclopropanation reaction (Hermann et  al. 2000, p. 4109). It can be concluded that there is no special halide effect in the reaction of the carbenoids LiCH2Hal with ethene, because the energy of the C─Hal bond cleavage in the transition states is essentially compensated by the energy of the Li─Hal bond formation; in contrast, the higher energy for the reaction of LiCH2OH (chosen as a model system of a Li/OR carbenoid) with ethene results from the high C─OH bond energy. The first reported reaction to lithium carbenoid was discovered by Closs and Closs (1962, p. 431) in 1962, and they noticed that these reactions proceed in a stereoselective manner (see Scheme 5.7). A. J. Macías‐Sánchez et al. reported the cyclopropanation of geraniol and the chemical selectivity of a series of lithium carbenoids (see Scheme 5.8). The research shows variable levels of chemoselectivity when the carbenoid is generated from dihalomethanes (CH2Cl2, CH2Br2, CH2I2) and mainly the cyclopropyl derivative can be obtained at the proximal olefin to the hydroxyl group (see Table 5.5). H Ph

Ph

Br Br

Ph

MeLi Ph

Br

Ph

Br

Ph H

Scheme 5.7  Cyclopropanation reaction of carbenoid.

5.3  Cyclopropane Fuel

n-BuLi, pentane, haloalkane –78 °C (3 h), then, r.t. (18 h)

HO R1

R2

R2

R1

R2

R1

+ HO

HO

H

H

H

Scheme 5.8  Cyclopropanation of geraniol with lithium carbenoid. Table 5.5  Cyclopropanation of geraniol with a set of lithium carbenoids. Product/Sa (yield %)

Entry Haloalkane

1

CH2I2

3′ 2′ HO

(R)

1′

(R)

(33%).

H

2

CH2Br2

3′

3′ 2′ HO

HO

(R)

1′

(R)

CH2Cl2

(16%)

H

R2

Cl

3′

(S) 2′

1′

(R) H

R1

HO

(R)

1′

(49%)

H

3

2′

+

(R) H

H Cl

(R) 3′

+ HO

(R) 3″

(S) 2′

1′

1″

(R) H

R1 = H, R2 = Cl (40%) R1 = Cl, R2 = H (17%)

(S)

2″

H (7%)

a  Yields were evaluated by GC. Source: Adapted from Durán‐Peña et al. (2016, table 1).

5.3.1.4  Metallic Aluminum Carbenoid‐Mediated Cyclopropanation

In 1985, Yamamoto et al. (Maruoka, Fukutani, and Yamamoto 1985, p. 4412; Maruoka, Sakane, and Yamamoto 1989, p. 176) discovered aluminum alkyl carbenoid (see species 1 in Eq. (5.11)). The reaction of aluminum carbenoid with ethylene in a dichloromethane solvent produces a high yield of cyclopropane. The cyclopropanation reactions with these aluminum carbenoids are usually performed by addition of the solution of olefin and 1–2 equiv. of CH2I2 in CH2Cl2 solvent at −40 °C, and high yields of cyclopropanated products can be produced (Maruoka, Fukutani, and Yamamoto 1985, p. 4412; Maruoka, Sakane, and Yamamoto 1989, p. 176). Furthermore, the aluminum carbenoid has a different character in chemoselectivity from other analogous carbenoids

177

5  Design and Synthesis of High‐Energy Strained Fuels

(Maruoka, Fukutani, and Yamamoto 1985, p. 4412; Maruoka, Sakane, and Yamamoto 1989, p. 176; Molander, Etter, and Zinke 1987, p. 453; Molander and Harring 1989, p. 3525). Because of this unusual character of aluminum carbenoids, it is important to better understand the chemical reactivity of the aluminum carbenoid species and their cyclopropanation mechanism(s). C

C

+ RCHI2

HR C

R3′Al C

I C

R′

H C

Al

R

R′ 1

(5.11)

The mechanism of promoting cyclopropanation by carbenoid varies with different carbenoid body. For zinc carbenoids, the methylene transfer mechanism is widely accepted (Bernardi, Bottoni, and Miscione 1997, p. 12300; Boche and Lohrenz 2001, p. 697; Dargel and Koch 1996, p. 877; Fang et  al. 2002, p. 154; Hermann et al. 2000, p. 4109; Nakamura, Hirai, and Nakamura 1998, p. 5844; 2003, p. 2341; Wang, Phillips, and Fang 2002, p. 5901; Zhao, Wang, and Phillips 2002, p. 12903). Samarium carbenoid cyclopropanation reactions have some competitions between the methylene transfer mechanism and the carbometalation mechanism (Stiasny and Hoffmann 1995, p. 619; Wang, Zhao, and Phillips 2004, p. 5512; Zhao, Wang, and Phillips 2003, p. 15200). Zhao et al. (Li et al. 2006, p. 3735) studied the reaction mechanism of aluminum carbenoid by density functional theory. Similarly, the cyclopropanation reaction of aluminum carbenoid with ethylene has two reaction channels: the methylene transfer channel and the carbon metallization channel. The energy barriers for the methylene transfer pathway (11–15 kcal/mol) are significantly smaller than those of the carbon metallization pathway (about 30 kcal/mol; see Figure 5.21). So, the methylene transfer process is more favorable. The report also demonstrated that the methylene transfer transition state corresponds to a three‐centered structure 31.2 30.6 28.9

TS6 TS4 TS2 SM = (CH3)2 AICH2X (X = CI, Br, I) + CH2CH2 RC2 RC3 RC1

–16.8 –19.3 –21.6

–1.5 –1.6 –2.3

IM3 IM2 IM1 Carbometalation

0.0 SM

–1.5 –1.6 –2.3

E (kcal/mol)

178

12.8 10.4 9.0

TS5 TS3 TS1

RC2 RC3 RC1

Methylene transfer

c-C3H6 + (CH3)2AIX (X = CI, Br, I) –28.7 –34.4 –38.9

I-p Br-p CI-p

Reaction coordinate

Figure 5.21  Schematic diagram showing the computed relative energy (in kcal/mol) at the B3LYP/6‐311G** level for reactions of (CH3)2AlCH2X (X = Cl, Br, I) + CH2CH2 with the transition states and products energies given relative to the separated reactants. Source: Li et al. (2006). Reproduced with permission of American Chemical Society.

5.3  Cyclopropane Fuel

similar to that originally suggested by Simmons and Smith (1959, p. 4256) and Moser (1969, p. 1135). Reactant complexes located on the reaction surface appear to be formed without any barrier. Among the (CH3)2AlCH2X (X = Cl, Br, I) series of carbenoids, the (CH3)2AlCH2Cl carbenoid is the most reactive (with a barrier of about 11.3 kcal/mol), while the (CH3)2AlCH2I carbenoid is the least reactive. The relatively lower barrier for the (CH3)2AlCH2Cl carbenoid is mainly due to the following: an increase of its electrophilicity by the halogen and the smaller structural changes that occur in the (CH3)2AlCH2Cl carbenoid as the reaction goes from the reactant to the transition state. As for lithium carbenoids, Hoffmann Stiasny and Hoffmann (1995, p. 619) observed that either the carbometalation pathway or the methylene transfer pathway is feasible, which is also supported by Nakamura, Hirai, and Nakamura (1998, p. 5844; 2003, p. 2341) using computational methods. The metal center of the carbenoid is the most important factor to determine the reaction mechanism for the cyclopropanation reaction. Lithium carbenoid has almost completely ionic bond character in the L─C bond, which makes the process of carbometalation easier, with a barrier height of about 7.7 kcal/mol. The samarium carbenoid has a similar ionic bond nature in the Sm─C bond, so it is also believed to have some competitions between the methylene transfer mechanism (with a barrier of about 5.5 kcal/mol) and the carbometalation mechanism (with a barrier of about 9.7 kcal/mol). The aluminum carbenoid and the zinc carbenoid favor the methylene transfer pathway over carbometalation pathway, because of the more covalent nature of the M─C bond. As shown in Table 5.6, the reaction barriers of the carbometalation pathways are much higher than those of the methylene transfer pathways for both aluminum carbenoid and zinc carbenoid. Holger Hermann et al. (Boche and Lohrenz 2001, p. 697; Hermann et al. 2000, p. 4109) studied the potential energy surface with about 6.6 kcal/mol barrier energy for the reaction between ethylene and LiCH2I. Zhao et al. (Wang, Zhao, and Phillips 2004, p. 5512; Zhao, Wang, and Phillips 2003, p. 15200) used density functional theory calculations to examine the Simmons–Smith reaction and found a barrier height of −20.0 kcal/mol for the IZnCHI2 reagent c­ yclopropanation reaction. The ISmCH2I Table 5.6  Activation energies and experimental conditions for the reactions of various carbenoids with ethylene. Barrier (kcal/mol) Reactive species

LiCH2I

a

Methylene transfer

6.8 (7.5)a

Carbometalation

7.7 (8.5)a

Experimental condition (T, °C)

−78

ISmCH2I

5.5

9.7

−78

(CH3)2AlCH2I

12.8

31.2

−40

IZnCH2I

21.2

36.5

≈25

 Values in parentheses are from single‐point energies computed at the CCSD(T) level using the structures optimized with the B3LYP/6‐311G** method, corrected by ZPE energies from frequency analysis at B3LYP/6‐311G**. Source: Li et al. (2006). Reproduced with permission of American Chemical Society.

179

180

5  Design and Synthesis of High‐Energy Strained Fuels

and (ZnI)2CHI carbenoids were also studied by Zhao et al. (Wang, Zhao, and Phillips 2004, p. 5512; Zhao, Wang, and Phillips 2003, p. 15200). The barrier energies for selected Li, Zn, Sm, and Al carbenoids computed at the same level of theory are given in Table 5.6. It can be observed that the barrier heights for the reactions increase in the following order: LiCH2I (6.8 kcal/mol) ≈ ISmCH2I (5.5 kcal/mol) 99% ee

N2CHCO2R1 1 mol% 3

+

ClCH2CH2Cl, 25 °C

CO2R1

CO2R1

(77% yield) R1 = D-menthyl

97% ee

CH3

CH3

N2CHCO2Et 1 mol% 3

Ph

+

ClCH2CH2Cl, 25 °C CH3

(50% yield)

97% ee

(63:37)

Ph

Ph

CO2Et

97% ee

N2CHCO2R1 1 mol% [Cu(5)]ClO4

(80:20)

CO2Et

7% ee

R1 = dicyclohexylmethyl CO2R1

CH2Cl2, 0 °C 94% ee

(78% yield)

(Trans/cis 95:5)

Scheme 5.14  Enantioselective cyclopropanation with copper complexes of semicorrins and bisoxazolines. Ph

+ EtOOCCHN2

MeO



Rh2(OAc)4

Ph

Et2O

COOEt

MeO

0.1 mol

0.1 mol

94%

(5.18)

Mats Tilset et al. reported a highly cis‐diastereoselective Rh(I) cyclopropanation catalysts (see Scheme 5.15; Rosenberg et al. 2011, p. 2465). The change of activator

5.3  Cyclopropane Fuel

from AgOTf to NaBArf reduces the amount of the catalyst loaded and increases the yield and selectivity of some olefins. It has been found that this catalyst is very efficient and highly cis‐selective in the cyclopropanation reaction between EDA and sterically unacceptable electron‐rich olefins such as cyclic substrate cyclopentene. N N

CO Rh N

Cl

R2

O

+ R1

R2

O

R3

+NaBArf

O

R1 N2

R3 O

Scheme 5.15  Highly cis‐selective cyclopropanation reaction with Rh(I) catalyst.

Hossain and coworkers reported [CpFe(CO)2(THF)](BF4) as catalysts in the cyclopropanation of a variety of olefins, with either EDA or phenyl diazoacetate as the carbene source, to synthesize cyclopropanes in good yields and with high cis selectivity (Seitz and Hossain 1994, p. 7561; Seitz et al. 1992, p. 7755; Seitz, Saha, and Hossain 1993, p. 2604). Optimal conditions for these reactions were nitrogen atmosphere with dichloromethane as the solvent at 40 °C. In 1995, Woo, Kodadek, and coworkers reported several iron(II) and iron(III) porphyrins as cyclopropanation catalysts to produce mainly trans‐cyclopropanes from various terminal alkenes and EDA (Wolf et al. 1995, p. 9194). Most of iron(III) porphyrins had to be reduced to iron(II) by cobaltocene (Wolf et al. 1995, p. 9194) or a mild reducing agent like EDA (Salomon and Kochi 1973, p. 3300) to be efficient. An exception to this was chloro[meso‐tetrakis(pentafluorophenyl)porphyrin] iron(III), which is a room temperature catalyst. Since cobalt(II) complexes have been found to be effective (stereo‐ and enantioselective) in cyclopropanation reactions of olefins (Nakamura et  al. 1978a, p. 3443; 1978b, p. 3449), various cobalt‐based catalytic systems have been developed. So far, the most successful cobalt‐based catalysts are complexes with salen (Niimi et  al. 2000, p. 3647; 2001, p. 79) or porphyrin (Chen, Fields, and Zhang 2004, p. 14718; Chen, Ruppel, and Zhang 2007, p. 12074; Chen and Zhang 2007, p. 5931; Huang et al. 2004, p. 8179; Ruppel et al. 2010, p. 2273; Zhu, Perman, and Zhang 2010, p. 8460; Zhu et al. 2008, p. 5042) ligands developed by the groups of Katsuki and Zhang, respectively. Chiral cobalt–porphyrin cyclopropanation catalysts are unprecedented in their reactivity, stereocontrol, and ability to affect cyclopropanation with (near) stoichiometric amounts of alkenes avoiding ­carbene dimer formation (Doyle 2010, p. 850). Another intriguing feature of cobalt(II) porphyrin systems is their effectiveness in cyclopropanation of electron‐deficient olefins. Zhang et al. reported a general and effective cobalt(II)–porphyrin catalyst for the asymmetric cyclopropanation of various electron‐deficient olefins and achieved a high yield and high stereoselectivity of electrophilic cyclopropane products (Dzik et al. 2010, p. 10891).

189

190

5  Design and Synthesis of High‐Energy Strained Fuels

Palladium catalysts such as Pd3(OAc)6 and PdCl2·(PhCN)2 are considered to be active cyclopropanation catalysts, especially for diazomethane (Paulissen, Hubert, and Teyssie 1972, p. 1465). When Pd3(OAc)6 are used as catalysts, electron‐deficient alkenes are generally more reactive than that of electron‐rich alkenes. For example, the high yields (80%) of cyclopropane were obtained from α,β‐unsaturated aldehyde and diazomethane (Tomilov et  al. 1993, p. 799). Especially, with two palladium catalysts, highly ring‐strained alkenes such as norbornenes were cyclopropanated in a higher yield than many other simple alkenes (Zefirov et al. 1990, p. 7702). In addition, highly regioselective cyclopropanation is particularly remarkable in these palladium systems, such as high E‐selective (E : Z = 98 : 2) cyclopropanation of (E,Z)‐1,5‐cyclodecadiene. Some reliable conclusions can be drawn from the above experimental data. First, the most effective catalysts for preparation of trans isomer with the widest range of reactions are copper‐based catalysts. Second, rhodium‐based catalysts are very effective, but usually produce a lower enantio‐ and diastereomeric ratio. Finally, cobalt‐based catalysts are commonly used for cis‐selective cyclopropanation reactions, but ligands are quite complex. 5.3.3  Other Cyclopropanation Methods In addition to the above two cyclopropanation processes, another common cyclopropanation process is the ylide cyclopropanation process. Ylide is one of the most important active intermediates in organic synthesis, and sulfur and phosphorus ylide can be used to synthesize cyclopropane. Sulfur ylide can react with α,β‐unsaturated ketone to form cyclopropane ketone (Toda and Imai 1994, p. 2673). Importantly, the chiral cyclopropanone can be obtained from chiral amino sulfoxide ylide and α,β‐unsaturated ketone (Johnson and Schroeck 1973, p. 7418). Besides, the reaction of phosphorus ylide with olefin also can synthesize cyclopropane compound (Li, Dai, and Aggarwal 1997, p. 2341). In addition to the above method, in 1989, Kulinkovich (2003, p. 2597) first discovered that a carboxylate and two molecules of ethylmagnesium bromide could be catalyzed by the tetraisopropyl titanium oxide to synthesize a substituted cyclopropanol (see Eq. (5.19)). This Kulinkovich reaction provided a simpler and safer method for the synthesis of cyclopropanol (Haym and Brimble 2012, p. 7649). O EtMgBr, Ti(i-PrO)4 R2 R1

O

RO THF/Et2O,



Ti RO

R1

OH

(5.19)

5.3.4  Fuel Synthesis and Mechanism Through the cyclopropanation and a series of zinc carbenoid derived therefrom, a series of high‐tension organic compounds with three‐membered ring structure can be synthesized, which significantly increase the density of fuel and the heat of combustion, providing more energy to the spacecraft. Scientists in the Soviet Union synthesized liquid hydrocarbon fuel called “syntin” (Jaffe et  al. 2013, p. 3968) that has been used in the second‐stage engine of the Soyuz‐U2 rocket.

5.3  Cyclopropane Fuel

The chemical name of this fuel is 1‐methyl‐1,2‐dicyclopropylcyclopropane, C10H16, and composed of three cyclopropyl rings joining together. It has a positive enthalpy of formation (+199.2 kJ/mol [gas] and +151.9 kJ/mol [liq.]) and an estimated strain energy (from the cyclopropyl rings) of 313.6 kJ/mol. Sherburn et  al. reported a cyclopropane‐based fuel named [n]ivyane synthesized by [n]dendralene, where n represents the number of carbon–carbon double bonds or three‐membered rings (Bojase et al. 2011, p. 229). In this study, all synthesized cyclopropane‐based fuels from [3]ivyane to [8]ivyane have higher energy. The heat of combustion of [6]ivyane is 50.8 ± 2.5 MJ/kg (12.3 ± 0.6 MJ/mol), which is one of the highest reported for hydrocarbon (Domalski 1972, p. 221) and is significantly higher than that of cubane (see Table 5.7). The molar heat of combustion of [6]ivyane is roughly six times that of cyclopropane (2.1 MJ/mol), but Table 5.7  Heats of combustion for selected strained hydrocarbons. Name

Structure

Heat of combustion (MJ/mol−1 at 298 K)

Cyclopropane

2.1 (experimental)

49.7 (experimental)

Cubane

4.7 (experimental)

46.5 (experimental)

Ditetrahedryl

4.9 (estimated)

47.5 (estimated)

Pentaprismane

5.6 (estimated)

43.0 (estimated)

Pentacyclo[5.4.0.02,6.03,10.05,9] undecane

6.1 (estimated)

41.7 (estimated)

Pagodane

10.8 (experimental)

41.6 (experimental)

[6]Ivyane

12.3 (experimental)

50.8 (experimental)

Source: Bojase et al. (2011). Reproduced with permission of Royal Society of Chemistry.

191

192

5  Design and Synthesis of High‐Energy Strained Fuels

the heat of combustion per kg for these two compounds is very similar (cyclopropane = 49.7 MJ/kg; Domalski 1972, p. 221). Despite their strain energy (a cyclopropane ring has strain energy of c.115 kJ/mol; Bach and Dmitrenko 2006, p. 4598), the ivyanes are stable at moderate temperatures. For example, ivyane melts at 95 °C in air without significant decomposition, and [6]ivyane is stable at c. 200 °C demonstrated by differential scanning calorimeter (DSC) experiments. Thus, the thermodynamic strain energies in isolated and covalently connected cyclopropane rings are comparable, and the connected nature of the cyclopropane rings in the ivyane structure does not lead to a decrease in kinetic stability. Han et al. converted norbornene and dicyclopentadiene (DCPD) into a series of cyclopropane‐fused hydrocarbons (see Scheme  5.16) by Simmons–Smith cyclopropanation (Oh et al. 2007, p. 322). All cyclopropanated products have higher density and calorific value, shown in Table  5.8. Compared with JP‐10 and RJ‐4 fuels, tetracyclic compounds 2b and 2c have higher density of 1.02 and 0.99 g/cm3 and combustion heat value of 10.277 and 10.210 kcal/mol, respectively.

Et2Zn, 2,4,6-trichlorophenol 1a

CH2Cl2, –40 °C to r.t.

2a

Et2Zn, 2,4,6-trichlorophenol CH2Cl2, –40 °C to r.t.

2b

1b H2, Pd/C

Et2Zn, 2,4,6-trichlorophenol 1c

CH2Cl2, –40 °C to r.t.

2c

Scheme 5.16  Cyclopropanation of norbornene and dicyclopentadiene.

The cyclopropanation of endo‐DCPD with high density and energy is very attractive in the synthesis of advanced aerospace fuel (Oh et al. 2007, p. 322; Li et al. 2010, p. 2522; Wang et al. 2009b, p. 2383; 2011, p. 1342). In contrast, unlike most previous reaction theories limited to cyclopropanation of simple and symmetric olefins such as ethylene, the low symmetry of endo‐DCPD with two double bonds makes the reaction channel more complex. Zou et al. carried out a computational study on the cyclopropanation of endo‐DCPD with zinc carbenoids by hybrid DFT method (M06), focusing on the effects of diethyl ether solvent and (ICH2)2Zn (Feng et al. 2012, p. 10065). As mentioned in the Simmons–Smith cyclopropanation reaction mechanism, only the methylene transfer mechanism is considered in the endo‐DCPD cyclopropanation reaction. The carbenoid can attack both C═C

5.3  Cyclopropane Fuel

Table 5.8  Physical properties of cyclopropane‐fused hydrocarbons. Heat of combustion No.

Compounds

Density (g/cm3)

Btu/gal

MJ/kg

kcal/mol

1

2a

0.94

139 967

41.493

9.318

2

2b

1.02

154 378

42.182

10.277 10.21

3

2c

0.99

153 373

42.331

7

JP‐10

0.94

142 000

42.106

9.453

8

RJ‐4

0.94

141 000

41.81

9.387

Qnet = 10.479 kcal/g × 4.186 MJ g/kcal/kg − 0.2122 × 11.18 (H wt%) = 41.493 MJ/ kg = 9.913 kcal/g = 9.318 kcal/mol. Source: Adapted from Oh et al. (2007, table 1).

bonds in norbornyl and cyclopentyl rings (denoted as NB and CP, respectively) from different directions, and various primary and complete cyclopropanated compounds can be expected (see Scheme 5.17). Zn CH2l2 P1

P5

P6

P7

P8

Zn CH2l2 P2 Zn CH2l2

Zn

endo-DCPD

CH2l2 P3

P5

P7

P6

P8

Zn CH2l2 P4

Scheme 5.17  The anticipated cyclopropanated compounds of endo‐DCPD with zinc carbenoids.

5.3.4.1  Cyclopropanation of endo‐DCPD with Monomeric IZnCH2I in Gas Phase Primary Cyclopropanation of endo‐DCPD

The optimized structures and the calculated energies of starting materials (SMs), reaction complexes (RCs), transition states (TSs), and product c­ omplexes

193

5  Design and Synthesis of High‐Energy Strained Fuels

(PCs) with zero‐point energies (ZPE) correction on the primary cyclopropanation of endo‐DCPD to form P1–P4 in gas phase are displayed in Figure 5.24. The most stable monomeric IZnCH2I in gas phase possesses the special I1, Zn,

I2 2.855 C1 2.487C2 2.4 87 C3

C2 1

2.30 1 2. 33 2

I

C3

TS2(19.21)

1

C

3.

23

5

I1 TS1(10.43) IZnCH2I+endo-DCPD

E (kcal/mol)

I2

31.13

SM1(0.00)

R2

RC2(–11.92)

18.43 RC1(–8.00)

C H Zn I

R1

1

1C

I1

C1

(a)

C1 C

2

C3 2

I

4

2.18

I2

I2

C

C C3

I

I1

1

2

1

C

C2 2.823 2.633 C3

3

3.

PC2(–35.53)

I2 2.18

2.848 13 8

C2

I1

PC1(–47.76)

2.52

6

Reaction coordination

C2 3 C 1 TS4(10.50) 36 2. I1

2.

87

7

C1 E (kcal/mol)

I2

22.25

C22 .573 2.775 1 C3 50 C 2.4

I2

TS3(8.38) I1

IZnCH2I+endo-DCPD SM2(0.00)

R4

16.17

RC3(–7.79)

C H Zn I

R3

2

I 2.179

RC4(–11.74) 1

C C 2.78 4 2

C2

C3

C2

3

C

C1 I2 PC4(–38.89)

82 C1

1 2.1

I

2.70 9

194

3

C

72 2.6 I2

I1

C2

C1

I2

C3 I1 PC3(–43.01)

I1

(b)

5 2.74

Reaction coordination

Figure 5.24  The reaction profiles of the primary cyclopropanation of (a) NB (R1, R2) and (b) CP (R3, R4) C═C bond of endo‐DCPD with monomeric IZnCH2I (M06/BS level) in gas phase. The bond lengths are given in Å. Energies relative to the starting materials are shown in the parentheses. Source: Feng et al. (2012). Reproduced with permission of American Chemical Society.

5.3  Cyclopropane Fuel

I2, and C1 planar geometry structure (the I2–C1–Zn–I1 dihedral angle is −0.246°) and the length of C1–I2, C1–Zn, and Zn–I1 is 2.183, 1.932, and 2.429 Å, respectively. The Zn–C1–I2, I2–C1–H, Zn–C1–H, and H–C1–H angle is 105.3°, 106.1°, 114.3°, and 109.7°, respectively (see Figure 5.26), indicating that carbon atoms in monomeric zinc carbenoid are almost sp3 hybridization. This is much distinctive from the almost sp2 hybridization in LiCH2I and ISmCH2I (Nakamura, Hirai, and Nakamura 2003, p. 2341; Zhao, Wang, and Phillips 2003, p. 15200). As zinc carbenoid approaches the NB C═C bond, the interaction between the zinc vacant p orbital and C═C π orbital promotes the formation of π‐type complexes (RC1, RC2). The NBO calculations reveal that natural charges of Zn decrease from 0.871 in isolated IZnCH2I to 0.854 and 0.815 in RC1 and RC2, respectively, and the C1 charges decrease respectively from −1.155 to −1.168 and −1.182. This reflects the increment of the electron density of C1 and Zn in π‐type complexes, which stabilizes the π‐type complexes and significantly reduces the energy of reactants by 8.00 and 11.92 kcal/mol, respectively, in Figure  5.24. Then, the two complexes evolve into transition states (TS1, TS2) with the energy of 10.43 and 19.21 kcal/mol, respectively. And the cleavage of C1–I2 and C1–Zn leads to the formation of C1─C3 and C1─C2 bonds. As RC1 (RC2) goes to TS1 (TS2), the π electrons shift from C2=C3 π orbital to C1–I2 σ* orbital, pushing electrons in C1–I2 σ orbital to I2 atom. In approaching the TS, the electrons of C1–I2 σ bond facilitate Zn─I2 bond formation. Besides, this electron transfer contributes to the partial formation of C1–C2 and C1–C3 and the partial breaking of C1–I2. The length of C1─C2 bond is slightly different from that of C1─C3 bond in both TSs, reflecting asynchronous addition of methylene to the C═C bond. This result is in agreement with the reports that monomeric Ti, Zn, Al, Sm, and Li carbenoids with low symmetry react with ethylene in asynchronous manner (Ke, Zhao, and Phillips 2007, p. 848; Li et al. 2006, p. 3735; Zhao, Wang, and Phillips 2002, p. 12903; Zhuang and Zhang 2009, p. 14). Besides, the planar ethylene moiety undergoes significant pyramidalization (5.57° and 21.25° for C2, 4.71° and 19.85° for C3 in TS1 and TS2, respectively), which demonstrates that the sp2 → sp3 rehybridization of C2 and C3 is required for the formation of cyclopropane ring. The reactions of CP C═C bond are similar to that NB C═C bond (see Figure 5.24). The energy of reactant complexes (RC3, RC4) is 7.79 and 11.75 kcal/mol lower than that of SM2. In TS3 and TS4, C1─C2 bond is longer than the corresponding C1─C3 bond, which is also a significant evidence for the asynchronous addition on CP C═C bond. The calculated barrier for R1 and R3 (18.43, 16.17 kcal/mol) is much lower than that for R2 and R4 (31.13, 22.25 kcal/mol), respectively (see Figure 5.24), demonstrating that IZnCH2I tends to approach the double bond from the exo‐ face. After carefully checking the geometries of TSs, it is found that the access from the endo‐face (R2, R4) leads the C═C bond to undergo considerable pyramidalization (21.25° and 7.33° for C2, 19.85° and 10.18° for C3, respectively). Meanwhile the pyramidalization with exo‐access (R1, R3) is much smaller (5.57° and 2.4° for C2, 4.71° and 6.32° for C3, respectively). As a result, the

195

196

5  Design and Synthesis of High‐Energy Strained Fuels

Table 5.9  The energy gaps (a.u.) between the LUMOs and HOMOs of P1–P7 computed at M06/BS level. In gas phase LUMO

P1

−0.028

In diethyl ether

HOMO

−0.252

Eg

0.224

LUMO

−0.027

HOMO

−0.253

Eg

0.226

P2

−0.030

−0.248

0.218

−0.030

−0.248

0.218

P3

−0.028

−0.253

0.225

−0.027

−0.254

0.227

P4

−0.030

−0.247

0.217

−0.029

−0.247

0.218

P5

−0.028

−0.267

0.239

−0.027

−0.267

0.240

P6

−0.029

−0.263

0.234

−0.028

−0.263

0.235

P7

−0.030

−0.266

0.236

−0.029

−0.267

0.238

a

 Eg = LUMO − HOMO. Source: Feng et al. (2012). Reproduced with permission of American Chemical Society.

RC → TS evolution with the access from the endo‐face has to overcome higher energy barrier. In addition, the energy gap between the frontier orbitals (LUMO and HOMO; see Table 5.9) of P1 and P3 are, respectively, 0.006 and 0.008 a.u. higher than that of P2 and P4, which suggests that P1 and P3 are more stable. The thermodynamically favorable products (P1 and P3) predict the stereospecificity of cyclopropanation. Actually, P1 and P3 appear in equal amount as the primary compounds in experiment (Simmons, Blanchard, and Smith 1964, p. 1347). Thus, only the further cyclopropanation of P1 and P3 is considered in the following section. Complete Cyclopropanation of P1 and P3

As shown in Figure 5.25, when monomeric IZnCH2I approaches the remained double bond in P1 and P3, π‐type complexes (RC5–RC8) are formed with the lower energy of reaction systems. The reactions overcoming the energy barrier enables the transition from RCs to TS. The discrepancy in bond length of C1–C3 and C1–C2 in TS5–TS8 also reveals the asynchronous addition to double bond. It is worth noting that the change of structure from RC to corresponding TS can reflect the order of reaction barrier. The C1─I2 bond in TS6 is 0.115 Å longer than that in TS5, so the barrier of R6 is higher (4.48 kcal/ mol) than that of R5. Similarly, the C1─I2 bond in TS8 is 0.295 Å longer than that in TS7, so R8 with a barrier 3.87 kcal/mol is higher than R7 (see Figure 5.25). The calculated barriers in Figure 5.25 show that R5 and R7 are more facile to produce while R6 and R8 are not, which implies that the monomeric IZnCH2I still tends to attack the remaining C═C bond from the exo‐face. The two channels both lead to the stereospecific final product P5, in agreement with experimental result (Simmons, Blanchard, and Smith 1964, p. 1347). In fact, P5 (0.239 a.u.) has the highest LUMO–HOMO energy gap among all possible complete cyclopropanated products (see Table 5.9).

5.3  Cyclopropane Fuel

C2

C2 2.568 2.777 I2 1 6 C 2.44 C3

6 2.897

2.32 4

2.57

C3

C1

I1 TS6(12.57)

E (kcal/mol)

TS5(8.59)

IZnCH2I+P1

I2

SM3(0.00)

21.80

R6

RC5(–8.73)

RC6(–9.23)

I1

C

PC6(–38.09) I2

2.905 2. 77 1

C3 1

C1 C22.688 9 2.75

C3

C3

I1 C2 3 C

I2

5 2.18 1 C

C1

I2

I1 PC5(–42.81)

I1 Reaction coordination

(a)

C

C3

2.265 2.3 69

I1

I2 2.852

TS8(18.25)

2

3.1 45 E (kcal/mol)

I2

IZnCH2I+P3

22.27

R8

C2

2

I

5 30 C1

3.

2.1 79

PC8(–36.58)

3. 3

I1

C3

12

C2

I2 2.1 81

R7

18.40

SM4(0.00) RC8(–4.02)

I2

2.505 C2 C12.49 3 3 C

I1 TS7(10.56)

C1

C H Zn I

RC7(–7.84) C1 2 2.635C C3 2.8 22

I2

I1

C3 1

C2 3 C

I1

C

I1

(b)

C H Zn I

80 2 2.1 I

C2

C2

I1

R5

17.32

C1

PC7(–46.98) Reaction coordination

Figure 5.25  The reaction profiles of the complete cyclopropanation of (a) P1 (R5, R6) and (b) P3 (R7, R8) with the monomeric IZnCH2I in gas phase (M06/BS level). The bond lengths are given in Å. Energies relative to the starting materials are shown in the parentheses. Source: Feng et al. (2012). Reproduced with permission of American Chemical Society.

5.3.4.2  Cyclopropanation of endo‐DCPD with Monomeric IZnCH2I in Diethyl Ether Solvent

The bulk solvent effect of the cyclopropanation with monomeric IZnCH2I was investigated by using the polarizable continuum model (PCM) solvation. Diethyl

197

5  Design and Synthesis of High‐Energy Strained Fuels Dihedral angle I2–C1–Zn–I1:0.530° 109.5° 113 .6° 105.4° 50 1.9 C1 78 I1 2.4 Zn 108.8° I2

7

18

83 2.1

105.3°

2.

Dihedral angle I2–C1–Zn–I1:–0.246° 11 4.3 109.7° ° 1.932 I1 2.429 106.1° C1 Zn I2

Gas-phase monomeric IZnCH2I

Diethyl ether solvated monomeric IZnCH2I

Dihedral angle I2–C1–Zn–I1:110.4° I2

3°

8.7 ° 1.947

91

C2

2.1

104.

10

C1 115.0°

Zn

108.7° I1

6°

91

0° 115. 1.947

.9°

104.

2.1

108

108.

9°

Diethyl ether solvated (ICH2)2Zn

Figure 5.26  The geometries of zinc carbenoids involved in this study computed at M06/BS level. The bond lengths are given in Å. Source: Feng et al. (2012). Reproduced with permission of American Chemical Society.

1

C

2.525 C2 2.5 29 C3

C2 2.551 2.764 I2 1 C3 53 C 2.4

C2

I1

C3

2.3 2.3 87 55

I2 2.81 3

2.55 4

C2

I2

TS4′

I1

I2 2.802 1 C 2.552 C2 2.5 11 C3

TS7′

C3

C2

C1

TS5′ I1

I1

TS6′

I2

C3

C2

1

C

3.068

I1

C22 .582 C1 2.767 I2 C3 9 7 4 . 2

8 2.87

C1

C3 3 41 2.

I1

TS3′

TS2′

86 2.3 2.554

I2

2.329 2.399

TS1′

3. 11 5

C1 I1

2.8 48

198

I1

C H Zn I

TS8′ I2

Figure 5.27  The geometries of transition state of the cyclopropanation with monomeric IZnCH2I in the diethyl ether solvent computed at M06/BS level. The bond lengths are given in Å. Source: Feng et al. (2012). Reproduced with permission of American Chemical Society.

5.3  Cyclopropane Fuel

ether solvent does not significantly change the structures of solvated monomeric IZnCH2I (see Figure  5.26) and transition states (TS1′–TS8′; see Figure  5.27), indicating that the solvent molecules are not directly involved in the reaction. The C1─Zn, C1─I2, and Zn─I1 bond lengths of solvated IZnCH2I are just slightly elongated by 0.018, 0.004, and 0.049 Å, and the I2–C1–H, Zn–Cl–H, and H– C1–H angle is 0.7°, 0.7°, and 0.2° smaller than that in gas phase, respectively. I1, Zn, C1, I2 are still planar because the I2–C1–Zn–I1 dihedral angle is 0.530°. For TS1′–TS8′, the difference between the length of C1─C2 and C1─C3 bond confirms the asynchronous behavior of solvated monomeric IZnCH2I. Although the absolute energy of RCs and TSs is lower than of in gas phase, they are “destabilized” in diethyl ether solvent because their energies relative to the SMs are raised to some degree (see Table  5.10). Besides, the barriers of RCs decrease compared with the cases in gas phase due to the more obvious “destabilization.” For example, the total energy of RC1′ is 3.81 kcal/mol higher than that of RC1, while TS1′ is just 2.52 kcal/mol higher than TS1; therefore the barrier of R1′ decreases from 18.43 kcal/mol in gas phase to 17.14 kcal/mol in solvent. Additionally, NBO analysis reveals that the C1 and Zn natural charges of RC1′–RC8′ are more positive compared with those of RC1–RC8 in Table 5.11. For example, the natural charges of C1 increase from −1.179 in RC1 to −1.168 in Table 5.10  The relative energies (kcal/mol) of reactant complexes, transition states, and barriers of the reaction channels in diethyl ether solvent computed at M06/BS level. E (TS)

ΔE (barrier)

Reaction channels

E (RC)

R1′

−4.19

12.95

17.14

R2′

−2.06

21.30

23.35

R3′

−4.07

11.60

15.68

R4′

−5.76

14.00

19.77

R5′

−3.84

11.78

15.62

R6′

−3.94

15.73

19.67

R7′

−4.10

13.00

17.10

R8′

−2.53

20.32

22.85

R1ʺ

−5.20

10.26

15.47

R2ʺ

−5.54

18.39

23.93

R3ʺ

−4.10

9.27

13.37

R4ʺ

−6.96

12.95

19.91

R5ʺ

−3.83

10.09

13.92

R6ʺ

−4.92

14.57

19.49

R7ʺ

−5.60

9.93

15.53

R8ʺ

−6.09

17.42

23.51

R1′–R8′ represent the channels with monomeric IZnCH2I in diethyl ether solvent. R1ʺ–R8ʺ represent the channels with (ICH2)2Zn in diethyl ether solvent. Source: Feng et al. (2012). Reproduced with permission of American Chemical Society.

199

200

5  Design and Synthesis of High‐Energy Strained Fuels

Table 5.11  The natural charges of C1 and Zn in reactant complexes computed at M06/BS level.

In gas phase (RC) Reaction channels

1

In diethyl ether (RC′)

(ICH2)2Zn in diethyl ether (RCʺ)

C1

Zn

C1

Zn

C1

Zn

−1.179

0.854

−1.168

0.949

−1.160

1.194

2

−1.168

0.815

−1.160

1.011

−1.161

1.217

3

−1.182

0.856

−1.172

0.964

−1.158

1.238

4

−1.176

0.819

−1.167

0.895

−1.163

1.218

5

−1.176

0.854

−1.168

0.974

−1.157

1.241

6

−1.168

0.810

−1.169

0.868

−1.165

1.224

7

−1.179

0.857

−1.169

0.960

−1.158

1.192

8

−1.169

0.844

−1.168

1.023

−1.165

1.224

Source: Feng et al. (2012). Reproduced with permission of American Chemical Society.

RC1′, and the Zn natural charges rise from 0.854 to 0.949, which suggests that the enhanced electrophilicity of IZnCH2I in solvated reactant complexes, to improve reactivity in cyclopropanation. As illustrated in Table 5.10, the attack barriers of IZnCH2I from the exo‐ face (R1′, R3′, R5′, R7′) are much lower than that of their competitive channels from the endo‐face (R2′, R4′, R6′, R8′), predicting its reaction in solvent was still stereospecificity. The energy gap between the frontier orbitals (see Table 5.9) of P1, P3, and P5 is, respectively, 0.008, 0.009, and 0.005 a.u. higher than that of P2, P4, and P6 in solvent. As a result, P1 and P3 are the primary cyclopropanated compounds, and P5 is the sole final product, in agreement with experimental result (Simmons, Blanchard, and Smith 1964, p. 1347). The direct involvement of diethyl ether solvent was also assessed by considering the monomeric IZnCH2I coordinated with one diethyl ether molecule. Two reaction channels (R1C, R3C) with coordinated IZnCH2I attacking NB and CP C═C bonds from the exo‐face were probed. As shown in Figure 5.28, in the TSs (TS1C, TS3C), Zn atoms are coordinated by O atom of diethyl ether with Zn─O bond of 2.143 and 2.142 Å, respectively. Although the lengths of C1–I2, C1–C2, and C1–C3 in TS1C and TS3C are almost the same to those in TS1′ and TS3′, the C1–I2–Zn–C1 dihedral angle in TS1C and TS3C (−140.7° and −137.8°, respectively) is much different from that in TS1′ and TS3′ (both 180°). This indicates that the planar monomeric IZnCH2I is transformed into pyramid geometry when one diethyl ether molecule coordinates with Zn atoms, due to the strong electronic repulsion between O and the two iodine atoms. Compared with PCM solvation (see Table 5.11), the electrophilicity of coordinated monomeric IZnCH2I in the π complexes (RC1C and RC3C) is not significantly increased. The natural charge of C1 is −1.182 and −1.189 in RC1C and RC3C, and the Zn charge is 0.887 and 0.897, respectively. As a result, the barrier

5.3  Cyclopropane Fuel Dihedral angle C1–I2–Zn–I1:–137.8° I2

2.843 C1

2 2.445 C 2.5 41 C3

Zn I1

C3

2

3

I2

14

14

2.450 Zn

2.780 C1 2.

2.

O

C2 2.551

O

I1 Dihedral angle C1–I2–Zn–I1:140.7°

TS1c

TS3c

Figure 5.28  The geometries of transition state of cyclopropanation with monomeric IZnCH2I coordinated with one diethyl ether molecule computed at M06/BS level. The bond lengths are given in Å. Source: Feng et al. (2012). Reproduced with permission of American Chemical Society.

of R1C and R3C (17.90 and 15.82 kcal/mol) is slightly higher than that of R1′ and R3′ (17.14 and 15.68 kcal/mol), respectively, which indicates that the direct involvement of solvent molecule is less possible than the PCM solvation. 5.3.4.3  Cyclopropanation of endo‐DCPD with (ICH2)2Zn in Diethyl Ether Solvent

Theoretical and experimental researches have shown that (ICH2)2Zn will be formed in diethyl ether solvent via Schlenk‐type equilibrium (Blanchard and Simmons 1964, p. 1337; Fabisch and Mitchell 1984, p. 219; Staroscik and Rickborn 1972, p. 738). But it is not clear whether this carbenoid plays a role in the cyclopropanation reaction. As displayed in Figure  5.26, the dihedral angle I2–C1–Zn–I1 in solvated (ICH2)2Zn is 110.4°, predicting that the I2 and I1 are not planar and different with the monomeric IZnCH2I in gas phase. In this structure, the distance of two iodine atoms is maximized, so the electrostatic repulsion and steric hindrance are the lowest, which is most stable among all possible conformations. To the contrary, the Zn–C1(2)–H, H–C1(2)–H, I2–C1(2)–H, and Zn–C1(2)–I2 angles in (ICH2)2Zn are almost identical to those in solvated monomeric IZnCH2I, which suggests that C1 and C2 in (ICH2)2Zn are also sp3 hybridization. Besides, the lengths of C2–Zn and C2–I1 are identical to that of C1–Zn and C1–I2, respectively, so both the two methylene moieties in (ICH2)2Zn can react with C═C bond of endo‐DCPD. The difference between the C1─C3 bond and C1–C2 in TS1′–TS8─ (see Figure 5.29) means that (ICH2)2Zn also accesses the NB and CP C═C bonds in asynchronous manner. Similar to the cyclopropanation with monomeric IZnCH2I, the reaction with (ICH2)2Zn also proceeds via butterfly‐type TS (see Figure 5.29). The C1 atom of methylene moieties partially bonds with C3 and C4, forming the rudimental cyclopropane ring. C1─I2 bond is simultaneously elongated and partially broken and finally broken completely when the electron‐rich I2 atom combined with Zn atom. Unlike the cyclopropanation with IZnCH2I, the by‐product of (ICH2)2Zn is IZnCH2I instead of ZnI2, and it can participate in the formation of (ICH2)2Zn again.

201

5  Design and Synthesis of High‐Energy Strained Fuels I1 2.520 2.2 84

I

C1

C4

2.49 2.3 08

7

C3

I2

C

8 80 I2

1 . C32.515C 2

I1

95

.3 C4 2

C2

I2

3

2.772 C1 2.588C 2.3 77 2 C4

C

TS7″

C4

2.471 2.3 05

C3

I1

C1

I2

C3 C4

C1 I2

TS5″

2.974

I1

I2

C2

I1

TS4″

I2

TS3″

TS2″

8

C1

52

2.5

C4

2

786

. C1 2

2.639 24 7

2.98

TS1″

3 C2. 383

1

2. 87 0

2.557 C3 C1 2.392 C4

9

I2 2.772

C2

I1

C4

C2

2.

C3

C2

2.89

202

C2

I1

TS6″

C H Zn I

TS8″

Figure 5.29  The geometries of transition state of cyclopropanation with (ICH2)2Zn in the diethyl ether solvent computed at M06/BS level. The bond lengths are given in Å. Source: Feng et al. (2012). Reproduced with permission of American Chemical Society.

Notably, compared with monomeric IZnCH2I in solvent, the involvement of (ICH2)2Zn can decrease the barriers (0.18–2.30 kcal/mol) of most reaction channels (see Table  5.10). The barriers of R1ʺ, R3ʺ, R5ʺ, and R7ʺ are much lower than their competitive reactions (R2ʺ, R4ʺ, R6ʺ, and R8ʺ). That is, (ICH2)2Zn facilitates the reactions and retains the stereospecificity. It is noted that compared with the corresponding RC′, the natural charges of C1 and Zn atom in most RCʺ increase (see Table  5.11), which indicates the further enhanced electrophilicity of (ICH2)2Zn in RCʺ (Zhao, Wang, and Phillips 2003, p. 15200). Taking reaction channel 1 as an example, the natural charges of C1 increase from −1.168 in RC1′ to −1.160 in RC1ʺ, and the Zn natural charges rise from 0.949 to 1.194. The increase in electrophilic character of (ICH2)2Zn in RCʺ is believed to contribute to the high reactivity. Therefore, the reaction channels with (ICH2)2Zn are more preferred than those with monomeric IZnCH2I. And it is possible that the active carbenoids might be a mixture of monomeric IZnCH2I and (ICH2)2Zn in real solvated system. However, at present, it is difficult to determine which one contributes more to the reaction.

5.4 ­Spiro and Caged Fuels Generally, as the number of rings of the fuel increases, the density increases, but at the same time the hydrogen content decreases, resulting in less gravimetric net heat of fuel combustion. To overcome this deficiency, it is significant

5.4  Spiro and Caged Fuels

to synthesize three‐ to four‐membered ring spiro‐ or polycyclic caged fuels with extra strain energy (Zhang et al. 2018, p. 95). In addition, spiro‐fuels also have excellent cryogenic properties, while caged fuels have high density and gravimetric heat of combustion, owning to cyclic strain energy. These kinds of fuels can also be used as fuel additives blended with aviation kerosene to increase volumetric NHOC. 5.4.1 Spiro‐Fuels Xie et al. have synthesized spiro[4,5]decane and spiro[5,6]dodecane shown in Scheme 5.18 (Xie et al. 2017, p. 10303). First, cycloheptanone was used as feedstock to produce pinacols by reductive coupling reaction with the catalyst of TiCl4 and Zn. TiCl4 is reduced into low‐valence TiII (or III) by Zn, then ketones bond with TiII (or III) to produce vicinal dialkoxide intermediate, and subsequent pinacols are obtained by hydrolysis. Figure  5.30a shows that ether (Et2O) as solvent is most effective for the reductive coupling, because TiII (or III) species are easy to generate in this solvent. Meanwhile, the selectivity of pinacols is lower in THF than in Et2O, because the strongly coordination between TiCl4 with THF suppresses the reduction of TiIV. Figure 5.30b shows that the conversion of ketone increases with the increase of molar ratio of TiCl4/cycloheptanone but the selectivity remains almost unchanged. The conversion does not change any more after the ratio reaching one, because the amount of TiII (or III) is already sufficient to ensure the formation of vicinal dialkoxides. Under optimal conditions, bicyclohexyl‐1,1′‐diol and bicyclopentyl‐1,1′‐diol pinacols are synthesized in 78.5% and 76.1% yield using cyclohexanone and cyclopentanone, respectively. Next, spiro[5,6]dodecan‐7‐one and spiro[4,5]decan‐6‐one can be formed in 79.4% and 94.4% yield through pinacol rearrangement catalyzed by SnCl4. Finally, with the catalyst of t‐BuOK, spiro[5,6]dodecane and spiro[4,5] decane can be synthesized through Wolff–Kishner–Huang reduction in 83.4% and 90.1% yields. O

O

Reductive coupling

n

OH

OH

n

Pinacol rearrangement

n

Reduction

n

n

n

n

n = 1 bicyclopentyl-1,1′-diol

n = 1 spiro[4,5]decan-6-one

n = 1 spiro[4, 5]decane

n = 2 bicyclohexyl-1,1′-diol

n = 2 spiro[5,6]dodecan-7-one

n = 2 spiro[5, 6]dodecane

Scheme 5.18  The synthesis of spirocycloalkanes from cyclic ketones. Source: Xie et al. (2017). Reproduced with permission of Royal Society of Chemistry.

The freezing points of spiro[4,5]decane and spiro[5,6]dodecane are −76 and −51 °C much lower than bi‐cyclopentane and bi‐cyclohexane hydrocarbons and have 37.17 and 38.41 MJ/l volumetric NHOC also higher than the bicyclic hydrocarbons, attributed to the extra strain energy of spirocycloalkanes (see Table 5.12).

203

5  Design and Synthesis of High‐Energy Strained Fuels 100

100

80

Conversion and selectivity (%)

Conversion Selectivity

60 40 20

80 60 40 20

4l

Ti 4N

Ti C 2N

1N

C

4l

4l

C Ti

C Ti

0.

4N 0.

(b)

7N

Ti

C

4l

4l

2O

Et

TH F

H 3O

C H

3C

H C

(a)

Conversion Selectivity

0

0 N

Conversion and selectivity (%)

204

Figure 5.30  (a) Effects of solvent and (b) molar ratio TiCl4/cycloheptanone on reductive coupling reaction. Source: Xie et al. (2017). Reproduced with permission of Royal Society of Chemistry. Table 5.12  Properties of some lignocellulosic fuels.

Molecular structure

Density at 20 °C (g/ml)

0.89

0.87

0.89

0.87

Freezing point (°C)

−51

−76

1.2

−38.0

4.37 (25 °C)

2.12 (25 °C)

3.72 (25 °C)

1.62 (25 °C)

8.59 (0 °C)

3.33 (0 °C)

6.33 (5 °C)

4.6 (−35 °C)

Kinematic viscosity (mm2/s)

232.3 (−20 °C)

19.8 (−60 °C)

—

—

NHOC (MJ/kg)

43.01

42.72

42.97

42.42

Volumetric NHOC (MJ/l)

38.41

37.17

38.11

36.78

Source: Adapted from Xie et al. (2017, table 1).

Xie et al. have reported the synthesis of the spiro‐fuel through photoinduced self‐sensitized [2+2] cycloaddition process of biomass‐derived β‐pinene and isophorone (see Scheme 5.19; Xie et al. 2019, p. 5886). Biomass‐derived β‐pinene and isophorone with no extra photosensitizer are irradiated at UV–vis light for nine hours; in this reaction, isophorone works as not only a reactant but also a self‐sensitizer, attributed to the presence of carbonyl group. Researchers have Turpentine

Cellulose Fermentation

o O

Aldol Condensation

O

HDO [2+2] cycloaddition Density

0.911 g/mL

Freezing point –51 °C NHOC

42.45 MJ/kg

Scheme 5.19  The whole synthesis scheme of biomass‐derived spiro‐fuel. Source: Xie et al. (2019). Reproduced with permission of Royal Society of Chemistry.

5.4  Spiro and Caged Fuels Cisophorone

Cβ-pinene

Sco-adduct

Conversion and selectivity (%)

Conversion and selectivity (%)

100 80 60 40 20 0

1:6

(a)

1:4

1:1

1:2

2:1

Cβ-pinene

Sco-adduct

80 60 40 20 0

4:1

Isophorone/β-pinene molar ratio

Cisophorone

100

(b)

130 mW/cm2 170 mW/cm2 210 mW/cm2

Figure 5.31  Effects of (a) isophorone/β‐pinene molar ratio and (b) light intensity on the photoinduced cycloaddition. Source: Xie et al. (2019). Reproduced with permission of Royal Society of Chemistry.

found that 365 nm wavelength of irradiation achieved the highest conversion of isophorone and selectivity of target co‐adduct compared with mercury lamp at 254 nm and the wavelength larger than 360 nm. The molar ratio of isophorone and β‐pinene is also studied, and co‐adduct can be synthesized in 95.1% conversion and 91.1% yield under optimal molar ratio of 1 : 4. Further increasing molar ratio may cause self‐cycloaddition reaction and inhibit co‐cycloaddition reaction (see Figure 5.31a). Besides, light intensity has a great influence on the photoreaction, and the highest conversion is obtained at 210 mW/cm2 (see Figure 5.31b). The co‐adduct can be hydrodeoxygenated to saturated spiro‐fuel in 85.0% overall yield by bifunctional catalyst 3% Pt/HY in 6 MPa H2. Metal Pt has the effect of hydrogenation and acidic support HY can catalyze the dehydration. The saturated co‐adduct spiro‐fuel has the density of 0.911 g/ml, the volumetric NHOC of 38.67 MJ/l, the freezing point of −51 °C, and the kinematic viscosity of 176 mm2/s at −20 °C. The spiro[bicyclo[2.2.1]heptane‐2,1′‐cyclopentane] reported by Pan et al. can be synthesized through two‐step one‐pot synthesis and hydrodeoxygenation scheme (see Scheme  5.20; Pan et  al. 2019, p. 1). Two‐step one‐pot synthesis O

O H

N

O

H

H

O N

Z-130 363 K A

B

O

Pt/HY 6 MPa H2 C

D

Scheme 5.20  The synthesis of spiro[bicyclo[2.2.1]heptane‐2,1′‐cyclopentane] D. Source: Adapted from Pan et al. (2019, scheme 1).

205

206

5  Design and Synthesis of High‐Energy Strained Fuels

catalyzed by HZSM‐5 includes that the first step Mannich base is synthesized using cyclopentanone, dimethylamine, and formaldehyde through Mannich reaction and the second step Diels–Alder reaction of Mannich base and cyclopentadiene in n‐butyl alcohol solvent at 100  °C affords fuel precursor spiro[bicyclo[2.2.1]heptane‐2,1′‐cyclopentane]‐5‐ene‐2′‐one. In detail, dimethylamine and formaldehyde undergo nucleophilic addition, followed by dehydration to the iminium ion. The iminium ion is an active electrophile that will attack the enol resonance formed from cyclopentanone in acidic environment and subsequently lose a proton to obtain the Mannich base A. Then, Mannich base A will decompose to B at high temperature, which can react with cyclopentadiene (CPD) via Diels–Alder addition to produce C. In addition, B can also react with cyclopentanone or itself through aldol condensation or hetero Diels–Alder reaction to produce by‐products E and F (see Scheme 5.21). ÖH

OH H

O + H

H

H

H–

N H

H

H

N

O

O N

–Hl

N ⊕ O

∆

A

B

C

6 MPa H2 Pt/HY D

(a)

O

O

O HO

B

E

O

O

I letero Diels–Alder reaction

+ B

(b)

B

O

O

F

Scheme 5.21  Reaction pathways for Mannich–Diels–Alder reaction. (a) Reaction mechanism and (b) Side reaction. Source: Adapted from Pan et al. (2019, scheme 2).

5.4  Spiro and Caged Fuels

The different Si/Al molar ratio has an influence on the Mannich–Diels–Alder reaction. Z‐130 zeolite is the most efficient catalyst achieving 81.6% spiro[bicyclo[2.2.1]heptane‐2,1′‐cyclopentane]‐5‐ene‐2′‐one in selectivity (see Figure  5.32a). As the recycling cycle increased, the conversion of cyclopentanone keeps almost constant, while the selectivity of spiro[bicyclo[2.2.1]heptane‐2,1′‐cyclopentane]‐5‐ene‐2′‐one decreases slightly, which may be related to coke formation on the catalyst (see Figure 5.32b). In the hydrodeoxygenation process, fuel precursor can be hydrodeoxygenated into target product spiro[bicyclo[2.2.1] heptane‐2,1′‐cyclopentane] in 88.5% yield with bifunctional Pt/HY catalyst under 6 MPa H2. The volumetric net NHOC of spiro[bicyclo[2.2.1] heptane‐2,1′‐cyclopentane] is 40.18 MJ/l, which will reduce the tank volume of aircraft by 7.5% compared with spiro[4,5]decane, and the density is 0.95 g/ml at 20 °C, which is 9.2% and 6.7% improvement compared to spiro[4,5]decane and spiro[5,6]dodecane previously synthesized. Notably, spiro[cyclopentane‐1,2′‐ norbornane] has a freezing point of −53 °C and an excellent cryogenic kinematic viscosity with 61.9 mm2/s at −40 °C, satisfying the requirement for advanced jet fuels (see Table 5.13).

Conversion and selectivity (%)

D

C

80 60 40 20 0

ne

8

l

Z-1

0

30

00

Z-6

Z-1

Z-2

Cyclopentanone conversion

A

B

No

(a)

Conversion and selectivity (%)

B

A

Cyclopentanone conversion 100

HC

00

Z-3 C

D

100 80 60 40 20 0

(b)

1

2

4 3 Recycle times

5

Figure 5.32  (a) Different catalysts affect the Mannich–Diels–Alder reaction. (b) Recycle performance in the Mannich–Diels–Alder reaction. Source: Pan et al. (2019). Reproduced with permission of John Wiley & Sons.

207

208

5  Design and Synthesis of High‐Energy Strained Fuels

Table 5.13  The properties of spiro[bicyclo[2.2.1]heptane‐2,1′‐cyclopentane] compared with spiro[4,5]decane and spiro[5,6]dodecane.

Molecule structure

Density at 20 °C (g/ml)

0.87

0.89

0.95

Freezing point (°C)

−76

−51

−53

2

Kinematic viscosity (mm /s) NHOC (MJ/l)

3.33

8.59

11.41

19.8 (−60 °C)

232.3 (−20 °C)

61.9 (−60 °C)

37.17

38.41

40.18

Wender and White have reported that organobis(cuprates) reagents can occur spiroannelation reaction with haloenone to synthesize spiro structure (see Scheme 5.22; Wender and White 1988, p. 2218). First, dilithiobutane is prepared through the reaction of 1,4‐dichlorobutane and sodium–lithium alloy in ether solvent in 70% yield. Next, organobis(cuprates) can be prepared by adding dilithiobutane to the preferably preformed PhSCu(Cu+) salt in THF solvent at −78 °C. As a substrate, 3‐chloroenone was added to the newly prepared organobis(cuprates) and, when the suspension was warmed to −15 °C, spiralized with organobis(cuprates) preparing product in the yield of 96% in one hour reaction time (see Scheme 5.22a). Besides, spiro product can also be obtained through spiroannelation of 3‐bromoenone and organobis(cuprates) in 87% yield at the temperature of −15 °C (see Scheme 5.22b). Although chloroenone and bromoenone are efficient to react with organobis(cuprates), iodoenone is relatively poor to get spiro product in 50% yield, and the reason is unclear. O

O

CuSPh

CuSPh

O

+ CI CuSPh 96%

(a)

O O

CuSPh

O

CuSPh

+

(b)

Br

CuSPh 87%

Scheme 5.22  The synthesis of spiro structure product with (a) 3‐chloroenone and (b) 3‐ bromoenone. Source: Adapted from Wender and White (1988, equation 1).

5.4  Spiro and Caged Fuels

To synthesize spiro‐fuels, it is essential to hydrogenate these spiroannelation fuel intermediates to saturated spirocyclic hydrocarbons. Therefore, studying the hydrodeoxygenation of spiroannelation fuel intermediates is the future research direction. Accordingly, synthesized spiro‐fuels with both bridge and spiro‐ring have high density and volumetric NHOC, as well as excellent cryogenic properties, which represent a new type of HED liquid fuels for practical applications. 5.4.2  PCU Monomer, Dimers, and Derivatives 5.4.2.1  PCU Monomer

Marchand and Allen reported the synthesis of pentacyclo[5.4.0.02,6.03,10.05,9] undecane (PCU) shown in Scheme 5.23 (Marchand and Allen 1974, p. 1596). At the first step, cyclopentadiene and p‐benzoquinone undergo Diels–Alder reaction to attain brown crystal a with the yield of 93%. At the second step, crystal a is transformed into b by photocatalytic [2+2] addition with Hanovia medium‐ pressure Hg lights at a high reaction effect. At the final step, 59% yield of colorless waxy solid PCU with the density of 1.24 g/cm3 is formed through hydrazine hydrodeoxygenation under alkaline conditions. Although PCU has a relatively high density, it is volatile, which suppresses the direct application of PCU. O

hv

+ O a

93%

O

O b

O

O N2H4–H2O

59%

Scheme 5.23  The synthesis route of PCU. Source: Adapted from Marchand and Allen (1974, scheme 1).

Singh, Raju, and Deota have synthesized pentacyclo[5.4.0.02,6.03,10.05,9]undecane‐8,11‐dione‐4‐spiro‐1‐cyclopropane 2 in 57.4% yield (see Scheme 5.24; Singh, Raju, and Deota 1986, p. 1731). Spiro[4.2]heptadiene undergoes Diels–Alder cycloaddition with p‐benzoquinone solubilized in hexadecyl trimethyl ammonium bromide (CTAB) micelle at about 57 °C for three hours to form adduct 1 in 66.0% yield. Adduct 1 solubilized in ethyl acetate can be transformed into solid pentacyclo[5.4.0.02,6.03,10.05,9]undecane‐8,11‐dione‐4‐spiro‐1‐cyclopropane 2 in 87.0% yield under UV light irradiation for four hours. The CTAB micelle makes sure the enhancement of cycloaddition due to the reactants being more ordered in the core of micellar pseudophase. Because of its cyclopropane and caged structure, the hydrodeoxygenated saturated fuel has high NHOC and density.

209

210

5  Design and Synthesis of High‐Energy Strained Fuels O In CTAB

+

hv O

O O 1

O

66%

O 2

87%

Scheme 5.24  The synthesis route of pentacyclo[5.4.0.02,6.03,10.05,9]undecane‐8,11‐dione‐4‐ spiro‐1‐cyclopropane. Source: Adapted from Singh, Raju, and Deota (1986, scheme 2).

The synthesis of methyl pentacyclo[5.4.0.02,6.03,10.05,9]undecane (MPCU) is shown in Scheme 5.25 (Jing et al. 2011, p.23; Ye et al. 2008, pp. 50–53). First, at 5 °C optimal temperature, the yield of Diels–Alder reaction that methylcyclopentadiene and p‐benzoquinone undergo can reach 87%. Next, the optimal irradiation time of intramolecular [2+2] photocyclization is 20 hours in 80% yield of methyl pentacyclo[5.4.0.02,6.03,10.05,9]undecane‐8,11‐dione. Finally, MPCU can be afforded through reduction with hydrazine in diethylene glycol solvent heated to reflux for five hours. The density of MPCU is 1.20–1.30 g/cm3 and the volumetric heat of combustion is about 50.60 MJ/l. MPCU can also work as fuel additive to increase the density and volumetric heat of combustion of other HED fuels. When the mass fraction of MPCU is 50% in RP‐3, the density and volumetric heat of combustion of RP‐3 blending fuels are increased by 13% and 10% (see Tables 5.14 and 5.15). O

CH3

H3C

H3C 5 °C CH3OH

hv 20 h O 87%

O

O 80%

O

O CH3 80% N2H4–H2O NaOH MPCU

Scheme 5.25  The synthesis route of MPCU. Source: Adapted from Jing et al. (2011, figure 6).

5.4.2.2  PCU Dimers

PCU alkene dimers have been synthesized through reductive dimerization of pentacyclo[5.4.0.02,6.03,10.05,9]undecane‐8‐one ketal 2 catalyzed by McMurry’s reagent (see Scheme  5.26; Flippen‐Anderson et  al. 1988, p. 1617). Pentacyclo[5.4.0.02,6.03,10.05,9]undecane‐8,11‐dione monoethylene ketal 1 undergoes Wolff–Kishner reaction with hydrazine and hydrolysis to form ketone 2. 2 dissolved in THF at nitrogen atmosphere undergoes reductive dimerization

5.4  Spiro and Caged Fuels

Table 5.14  The properties of HD‐01 and RP‐3 fuels mixed with MPCU. w (MPCU) (%)

Density (20 °C) (g/ml)

NHOC (MJ/l)

MPCU mixed in HD‐01 0

0.928

41.08

10

0.937

41.43

20

0.946

42.02

30

0.956

42.15

50

0.974

42.80

0

0.788

36.13

10

0.806

36.87

20

0.827

37.57

30

0.847

38.27

50

0.893

39.78

MPCU mixed in RP‐3

Source: Adapted from Jing et al. (2011, table 1).

Table 5.15  Flash point, freezing point, and viscosity of blended fuels. w (MPCU) (%)

TFa

Tf (°Cb)

η (20 °C/mPa s)c

7.78

MPCU mixed in HD‐01 0

47

360

260.5

1053

j)

16b −71.8

a) Melting point (peak point). b) Decomposition temperature (peak point). c) Density (25 °C). d) Viscosity (25 °C). e) Heat of formation. f ) Impact sensitivity. g) Friction sensitivity. h) Specific impulse (Explo5 v6.02. IL/WFNA = 24/76, w/w; isobaric conditions, equilibrium expansion, 7.0 MPa chamber pressure). i) Ignition delay time (WFNA). j) Glass transition temperature. Source: Yuan et al. (2017). Reproduced with permission of Royal Society of Chemistry.

­substituents ­(–CH3, –C2H5, –C3H7, –C4H9) are thermally stable up to 260 °C, while the ILs with substituents (–CH2CCH and –CH2CN) become much less stable with Td as low as 208.2 and 207.5 °C; (ii) ILs with dense function groups (–CCH, –CN, –OH) exhibit higher densities as 1.19 (14a), 1.27 (15a), 1.16 (16a), 1.25 (14b), 1.31 (15b), and 1.22  g/cm3 (16b), respectively; and (iii) ILs with –CH2CCH and –CH2OH substituents have the highest and lowest heat of formation. Moreover, compared with the corresponding pyrrolidinium‐ and imidazolium‐based isomers (1.05 and 1.07 g/cm3), 1‐methyl‐1‐azoniabicyclo[2.2.2] octane and 1‐(prop‐2‐ynyl)‐4‐aza‐1‐azoniabicyclo[2.2.2]octane DCA possess higher densities as 1.20 (9a) and 1.19 g/cm3 (14b). Wang et al. (2017a, p. 12502) reported a new strategy of cationic functionalization by introducing the energetic nitrato group into the cationic units of HILs,

8.4  Hypergolic Ionic Liquids

t = –3 ms

t = 0 ms

t = 62 ms

t = 63 ms

t = 87 ms

t = 837 ms

t = 0 ms

t = 48 ms

t = 49 ms

t = 64 ms

t = 673 ms

(a) t = –3 ms

(b)

Figure 8.8  Ignition delay times recorded by a high‐speed camera (1000 fps) of 10a (a) and 13a (b). Source: Yuan et al. (2017). Reproduced with permission of Royal Society of Chemistry.

which significantly improve the combustion performance of HILs with larger flame diameters and duration times. The density specific impulses of these HILs are all above 279.0 s g/cm3, much higher than that of UDMH (215.7 s g/cm3). In addition, the densities of these HILs are in the range of 1.22–1.39 g/cm3, which is much higher than that of UDMH (0.79 g/cm3), showing their high loading capacity in propellant tank. By systematically analyzing the cationic and anionic structures of previously reported HILs (Bhosale and Kulkarni 2016, p. 1013; 2017, p. 1250; Chand, Zhang, and Shreeve 2015, p. 13297; Gao et al. 2009, p. 2792; Huang et al. 2015, p. 2725; Joo et al. 2010, p. 3282; Perez et al. 2013, p. 5693; Li, Gao, and Shreeve 2014, p. 2969; Liu et al. 2016, p. 2031; Schneider et al. 2011a, p. 5886; 2011b; 2014, p. 6909; 2015, p. 20664; 2016, p. 8978), it can be found that the typical cations of known HILs are mainly concentrated on some quaternary ammonium cations (e.g. 1‐alkyl‐3‐methylimidazolium, 1‐alkyl‐pyridinium, ­ N‐methyl‐N‐alkylpyrrolidinium, N‐alkyl‐trimethylammonium in Figure 8.9), while the anions have undergone a rapid evolution from traditional DCA anion ([N(CN)2]7 to strongly reducing borohydride‐rich anions (e.g. [BH4], [BH3CN], [Al(BH4)4], [BH2(CN)2], [BH3(CN)BH2(CN)], and [H2P(BH3)2]). As shown in Figure 8.10, except for HIL‐12, all other HILs (HILs‐1–11) can spontaneously ignite upon contact with WFNA, and their ID times are in the range of 26–61 ms (Table  8.7). Essentially, the spontaneous combustion between HILs and oxidizers is an oxidation–reduction reaction. Therefore, the reductive activity of anions in HILs usually plays a decisive effect on the ID

395

396

8  Design and Synthesis of Green Hypergolic Ionic Liquid Fuels

N

R

N

CH2CH3

CH2CH2CH2CH3

R = CH2CH CH2 CH2CH2CH2CH3

Cation 8 H B

N

N

N

R=

N Cation 9 R

B

CH2CH CH2 CH2CH2CH2CH3

CH3 CH2CH CH2

N

CH2CH3 CH2CH2CH2CH3

Cation 7

R=

N3 N3

H CH CH CH 2 2 CH2C CH

H B

R = CH3 CH2CH CH2

N

N R

N

NH2

N Cation 10 R

NH2 CH C CH 2

NH2

Cation 6 N

H CH CH CH 2 2

R

Cation 11

N

CH2CN CH2CH3 CH2CH2Cl

H

R=

N

N

R

R

N3

Cation 3

H

H

or

N

N

(a)

R=

Cation 5

R H

NH2

N

R

Cation 4 N

N

R R = CH2CH CH2 CH2CH2CH2CH3

Cation 2

R = CH2CH CH2 CH2CH2CH2CH3

N

N

CH2CH2CH CH2 CH2C CCH2CH3

R=

Cation 1

N

CH2C CH

CH2 CH2C CCH3

CH2CH N

N

Cation 12

R

Anions NC

N

CN

NC

N

NO2

H3B

(b)

NC

(c)

N

CN

ID = 15 ms

N3

N

N

NC

CN H

[BH3(CN)BH2(CN)]

N

C

NC

N

NO2

NC

C

NO NO2 2

ID = 8 ms

C

NO NO2 2

N3

NO3

BH4

BH3CN

BH2(CN)2 [Al(BH4)4]

H P

BH3

N

NH2 CH2CH

NO2

O2N

BH2(CN)2

ID = 4 ms

Schneider et al. Shreeve et al. Shreeve et al. (2008) (2010) (2011)

N N CH2

N

CH2CH

BH4

CH2

N

CH2CH

N

CH2

H

[BH3(CN)BH2(CN)]

H3B

P

CHCH

CH2

H BH3

ID = 2 ms

ID = 1.7 ms

ID = 1 ms

Shreeve and Gao (2014)

Zhang et al. (2015)

Zhang et al. (2016)

Figure 8.9  (a) The typical cations for known HILs, (b) the representative anions for known HILs, and (c) the evolution of the ignition delay times of some typical HILs with white fuming nitric acid (WFNA). Source: Wang et al. (2017a). Reproduced with permission of John Wiley & Sons.

times. Nitrato‐functionalized DCA HILs also fall within a normal range (26– 61 ms). For nitrato‐functionalized HILs (HILs‐1–4), except that HIL‐1 is a solid at room temperature exhibiting an ID time of 26 ms (see Figure  8.11), other HILs show a little longer ID times (41–58 ms) than those nonfunctional HILs (33–47 ms for HILs‐9–11), which are probably due to their relatively higher viscosities (Table 8.7). But for hydroxy‐functionalized HILs (HILs 5–8), they do not exhibit the expected shorter ID times by heat‐accumulating from the reaction between hydroxy group and WFNA. The main reason is that hydroxy‐functionalized HILs cannot have good contact with WFNA in a very short time due to their high viscosities, and consequently, the heat accumulating from their reaction is limited. From the viewpoint of practical applications, the combustion flame and combustion duration times of HILs are also very important, which have been ignored in previous studies. As shown in Figure 8.12, under the same conditions of droplet test, the violent‐burning brilliant whitish flames of nitrato‐functionalized HILs‐2 and 3 are obviously larger than those of their analogues, indicating that they can burn more violently and efficiently. Further analysis showed that the diameters of largest brilliant whitish flame of those nitrato‐functionalized liquid

8.4  Hypergolic Ionic Liquids Series I

Series II

Series III

Nitrato-functionalized HILs

Hydroxyl-functionalized HILs

Control HILs

OH

ONO2 N NC

N

N NC HIL-5

CN

HIL-1

N

N

NC

HIL-2 N NC

N

ONO2 N

N

N

NC HIL-6

N

NC

CN

N

[Cation] +Cl –(Br –) Precursors

NC

N

CN

HIL-10 N

CN

NC

N

CN

OH N

CN

CN

HIL-11

ONO2

HIL-4

N

HIL-7

N NC

N

CN

OH

N

HIL-3

N

OH N

CN

NC

HIL-9

ONO2 N

N

CN

NC HIL-8

N

AgN(CN)2 Deionized water

N

CN

NC

HIL-12

N

CN

[Cation] +N(CN)2 – HILs

Figure 8.10  Chemical structures and synthetic route of studied HILs. Source: Wang et al. (2017a). Reproduced with permission of John Wiley & Sons.

salts (except HIL‐1) are much bigger than their hydroxyl‐functionalized and nonfunctional analogues (HIL‐2 vs. HIL‐6/HIL‐10, HIL‐3 vs. HIL‐7/HIL‐11, HIL‐4 vs. HIL‐8; Table 8.8). The obvious improvements in both flame size and combustion time can be attributed to the high oxygen‐balance character of nitrato‐functionalized cations in these HILs, thereby promoting their combustion properties including larger brilliant whitish flames and longer combustion duration time. In summary, the nitrato‐functionalized HILs exhibit acceptable ID times (ranging from 26 to 58 ms) with the oxidizer WFNA and clearly improve combustion performance including larger brilliant whitish flames and longer combustion duration times. The nitrato‐functionalized HILs exhibit higher Isp than those of common HILs and UDMH. In addition, the densities of these nitrato‐functionalized HILs are significantly higher than those of nonfunctional HILs and UDMH. These promising properties and combustion performance of nitrato‐functionalized HILs make them a new class of potential candidates for the replacements of hydrazine derivatives in the liquid bipropellant formulations. 8.4.2  Hypergolic Ionic Liquids Based on Nitrocyanamide Anions In addition to DCA, dicyandiamide nitrate is also widely used in HILs. Recently, a new class of nitrocyanamide ([N(CN)(NO2)]−)‐based EILs with substituted imidazolium, guanidinium, and tetrazolium cations are synthesized, and their physicochemical properties and energetic applications as hypergolic fuels are studied. The NCA is an analogue of the [N(CN)2]− (DCA) and [N(NO2)2]− (DNA)

397

8.4  Hypergolic Ionic Liquids T = 0 ms

T = –20 ms

T = 8 ms

Droplet

7 cm T = 48 ms

7 cm

7 cm T = 56 ms

T = 53 ms

Flame 7 cm T = 74 ms

7 cm

7 cm T = 175 ms

7 cm

T = 225 ms

7 cm

7 cm

Figure 8.11  High‐speed camera photos that show a spatially resolved ignition event for a droplet of HIL‐3 falling into WFNA. Source: Wang et al. (2017a). Reproduced with permission of John Wiley & Sons. HIL-6

HIL-2

HIL-10

7 cm

7 cm HIL-7

HIL-3

7 cm

7 cm HIL-11

7 cm

7 cm

Figure 8.12  Comparisons of the violent‐burning largest brilliant whitish flames of three kinds of HILs. Source: Wang et al. (2017a). Reproduced with permission of Wiley.

399

400

8  Design and Synthesis of Green Hypergolic Ionic Liquid Fuels

Table 8.8  Hypergolic properties of 12 HILs. Entry

Diameter (cm)a

Duration (ms)b

HIL‐1

2

18

HIL‐2

21

98

HIL‐3

18

118

HIL‐4

16

68

HIL‐5

16

110

HIL‐6

7

67

HIL‐7

5

92

HIL‐8

7

63

HIL‐9

15

56

HIL‐10

7

72

HIL‐11

6

90

HIL‐12

—

—

a

 Diameter of violent‐burning largest brilliant whitish flame.  Violent combustion duration. Source: Wang et al. (2017a). Reproduced with permission of John Wiley & Sons. b

anions, which combines the virtues of these anions with a balance of stability and energy. Nine EILs (HIL‐7−HIL‐14) are synthesized by anion exchange reactions between substituted imidazolium, guanidinium, or tetrazolium halides and silver nitrocyanamide (Ag[N(CN)(NO2)]) in methanol. All of the resultant EILs are liquids at room temperature and exhibit desirable physicochemical properties, such as low melting points (e.g. mp  250 °C), low viscosities (20 ms), which significantly decrease their practical application potential as propellant fuels. It has been found that those EILs based on DNA and azide anions are not capable of igniting with WFNA, indicating that the anions may be largely responsible for the hypergolicity of EILs.

8.4  Hypergolic Ionic Liquids

Therefore, for most hypergolic ILs, the anion plays a decisive role in the inducing stage of ignition process, while the cation has obvious influence on the ID time. The anions of DCA and NCA can exhibit hypergolic behavior when in contact with oxidizer such as WFNA. However, most of them only show moderate ID time (>20 ms). No doubt that the synthesis of new hypergolic ILs with ultrashort ID times ( [BImB]+, which can explain that the [BTzB]− ILs would be hypergolic while [TTzB]− ILs and ILs precursors could not be ignited upon ­contact with WFNA. Second, the viscosity has great impact on the ignition activity of [BTzB]−‐based ILs. Lower viscosity value is favorable to mass transfer between oxidizer and fuel, resulting in shorter ID time. Figure 8.18 shows that the ID variations of [BTzB]− ILs are almost consistent with that of viscosity values. Of the above HILs, 1b and 1e show short ID time of 26 and 20 ms, respectively. Based on the charge distribution analysis, the B–H groups on cation and anion all contribute to the hypergolic reaction. Even though large molecular weights can result in high viscosities, it perhaps gave a useful idea about HIL structural design to further improve their properties that the cation and anion are modified with both B–H hypergolic triggers and azole energetic groups. Weng et  al. (2018, p. 464) studied the hypergolic ignition of [BH3(CN) BH2(CN)]−‐based ILs with WFNA. Results show that the hypergolic ignition process of the present ILs is of a completely different three‐stage nature and ultrafast hypergolic ignition is observed for this family of ILs ( 1). Subsequently in the second stage, liquid spikes that are ejected into the oxidizer pool during CEDT significantly increase the reactive surface area underneath the liquid surface. As a consequence, local

409

8.4  Hypergolic Ionic Liquids t = –10 ms

t = 0 ms

t = 20 ms

t = 25 ms

t = 30 ms

t = 50 ms

t = 90 ms

t = 680 ms

Figure 8.18  High‐speed camera photos that show a droplet of ionic liquid 1e falling into WFNA. Source: Jiao et al. (2018). Reproduced with permission of John Wiley & Sons.

t(ms) = 0

0.43

0.57

0.86

0

0.43

0.57

0.86

0

0.43

0.57

0.86

(a)

(b)

(c)

Figure 8.19  Images of the very early stage upon contact of the IL droplet with WFNA pool taken by V611 camera. (a)–(c) IL‐6, 8, and 5 droplet vs. WFNA, D0 ≈ 2.25 mm, U0 ≈ 1.55 ms−1. Source: Adapted from Weng et al. (2018, figure 2).

temperature increases, and gas‐phase intermediate product and oxidizer/fuel vapor accumulate due to the continuous liquid‐phase reaction underneath the surface. When the local pressure overcomes the surface tension, disintegration of the surface in terms of larger ligaments and secondary droplet ejection was observed, together with shooting out of the vapor/smoke. Finally, in the third stage, when the local temperature and the vapor concentration increase sufficiently, further gas‐phase reaction leads to ignition. The ID time decreases with the increase of unsaturation index of heterocyclic core in the cation and

411

412

8  Design and Synthesis of Green Hypergolic Ionic Liquid Fuels V611 camera 0.86 ms

1.14

Liquid spikes Vapor smoke Ligaments/ droplets

Vapor/smoke

1.43

1.86

Ligaments/ droplets Flame kernel Solid product 2.43

13.6

44.1

220

2.43

V1 camera

Figure 8.20  Images of the hypergolic ignition and flame propagation process between the IL‐6 droplet with WFNA pool from V611 and V1. Source: Weng et al. (2018). Reproduced with permission of Elsevier.

also the decrease of CEDT. The enthalpy of formation of the ILs for different cation structures is correlated with ID time, which represents both the fuel structure chemistry and the “coulomb explosion” that enhances mixing effect on the overall reactivity of this diffusive system. Figure 8.21 shows the average ID time for the eight ILs as a function of cation structure. It is seen that for all the tested cases, the ID time is smaller than 5 ms. For given side alkyl functional group, increasing unsaturation index in the heterocyclic core decreases the ID time. It is noted that hypergolic ignition in this diffusive system involves complex coupling of the reactants chemistry, mixing, heat release, and phase change. Figure  8.21c shows that the ID time for these ILs depends on the ΔHf in Table 8.13. Generally, a larger ΔHf leads to shorter ID time. It is noted that the ΔHf depends on the chemical structures of ILs, including both the side alkyl functional group and the heterocyclic core in the cation. Two hydrophobic imidazolylidene cyanoborane complexes were prepared by the introduction of [BH2CN] into the molecular formula via treatment of imidazolium iodide and NaBH3CN (see Figure 8.22a; Li et al. 2018c, p. 939).

8.4  Hypergolic Ionic Liquids 5

3 2 1 0

R2 = 0.99

4

4

Average IDT (ms)

Average IDT (ms)

5

IL1IL4

IL3IL6IL8

(a)

3 2 1 0

IL2IL5IL7

0.4

(b)

5

0.5

0.6

0.7

0.8

0.9

CEDT (ms) R2 = 0.67

4 IDT (ms)

3 2 1 0 –150 –100 –50

(c)

0

50

100 150 200 250

Δ Hf (kJ/mol)

Figure 8.21  Average ignition delay time for ILs with different cation structures (a), average IDT as a function of CEDT (line is the linear fitting of the data) (b), and IDT of tested cases for all the ILs as a function of their Hf (line is the linear fitting with an R2 of 0.67) (c). Source: Weng et al. (2018). Reproduced with permission of Elsevier.

NHC‐2 is hypergolic with WFNA and displays the attractive properties such as water immiscibility, wide liquid range (Tg = –22 °C), short ID time (13 ms), high density (0.98 g/cm3), and high density impulse (347 s g/cm3). The calculated heats of formation of NHC‐1 and NHC‐2 are 1.053 and 0.498 kJ/g, respectively. Although the Isp value of NHC‐2 (262 s) is lower than that of TEAB (267 s), its density impulse value (347 s g/cm3) is higher than that of TEAB (338 s g/cm3); the physical properties and calculated performance of NHC‐1 and NHC‐2 are shown in Table 8.14. Noteworthy, NHC‐1 and NHC‐2 display exciting water immiscibility. [BH2CN] moiety might play a major role in determining the hydrophobicity of the molecules. NHC‐2 is shown in Figure 8.22b, which displays the ID as 13 ms, while NHC‐1 reacts vigorously with copious production of white smoke, but without ignition. Nitrogen‐rich EILs often show high heats of formation and high densities as a result of aza‐heterocyclic cations and anions such as 3‐butyl‐1‐methyl‐1H‐imidazolium 5‐aminotetrazolate (heats of formation  =  1.242  kJ/g, density  =  1.30 g/cm3). Thus, a new family of N‐rich (≥43% N) hypergolic ILs, bishydrobis‐(tetrazol‐1‐yl)borate (BTB)‐based HILs, were designed and prepared with superior energy capacity, with high positive ΔHf, high density, and high Isp,

413

414

8  Design and Synthesis of Green Hypergolic Ionic Liquid Fuels

Table 8.13  Chemical structures, physical properties of the ILs. Physical properties

IL No.

Chemical formula

IL‐1

C8H16B2N4

IL‐2

C10H20B2N4

IL‐3

C9H16B2N4

IL‐4

C9H15B2N3

ρa (g/cm3)

μb (mPa s)

ΔHfc (kJ/ mol)

Td (onset)d (°C)

0.9260

19

42.18

240

0.9131

22

15.97

230

0.9348

17

167.25

181

0.9339

22

93.88

261

+ N

0.9195

27

66.36

243

+ N

0.9412

10

219.80

267

0.8859

25

−120.32

245

0.9011

14

29.11

166

Cation structure + CH3 N

N

H3C

+ CH3 N

N

H3C H2C

+ CH3 N

N + N H3C

IL‐5

C11H19B2N3 H3C

IL‐6

C10H15B2N3

IL‐7

C9H21B2N3

IL‐8

C10H21B2N3

H2C CH3

CH3

N+

CH3 +

N

CH2

a

 Density at 25 °C.  Viscosity at 25 °C. c  Heat of formation. d  Thermal decomposition temperature. Source: Weng et al. (2018). Reproduced with permission of Elsevier. b

almost covering all the concerns for a good propellant. Some BTB‐based EILs are synthesized by a straightforward N‐hydroboration of tetrazole followed by an ionic metathesis (see Scheme 8.10; Li et al. 2017b, p. 15525). The BTB‐based ILs are hypergolic with oxidizers and show superior energy capacity with high positive ΔHf (often ≥2.0 kJ/g), high density (≥1.2 g/cm3), high density impulse (often ≥367 s g/cm3), and short ID time (220 °C), high densities (1.00–1.10 g/cm3), good hydrolytic stabilities, and short ID times (2.3– 9.7 ms) with WFNA as the oxidizer. The viscosities and specific impulses of the new HILs are shown in Table 8.16. Figure 8.25 shows a representation of the droplet test process performed on MBP‐1 and MBP‐6. Following the contact of MBP‐1 with WFNA for 1 ms, a drastic oxidation–reduction reaction occurs immediately, and a hypergolic ignition with a flame kernel is clearly observed after 2 ms (Figure 8.25a, the first test of MBP‐1). In the case of MBP‐6, as shown in Figure 8.25b (first test), under the same conditions as droplet test, the violent‐burning brilliant whitish flames of MBP‐6 are obviously larger than those of MBP‐1, indicating that the cationic structures have some influences on the combustion behaviors of the HILs. The dimethyl boranophosphate‐based HILs (MBP‐1–MBP‐6) show shorter ID times than the diethyl boranophosphate‐based ones (EBP‐1–EBP‐6; Figure  8.26, blue bars vs. orange bars). Besides, with an increase in the chain length of the alkyl substituents on the cation, the ID times became longer, which is evidenced by the increase in the ID times of MBP‐3 (4.8 ms), MBP‐4 (5.8 ms), and MBP‐5 (6.2 ms). Among them, MBP‐1 exhibits the shortest ID time of 2.3 ms, which is superior to that of UMDH (4.8 ms). EBP‐5 shows the longest ID time of 9.7 ms; however, it is still shorter than that of the most reported DCA and nitrocyanamide ion‐based HILs (e.g. 47 ms for [BMIm][DCA], 16 ms for 2,2‐ dimethyltriazanium nitrocyanamide). Table 8.16  Viscosities and specific impulses of the new HILs.

a

ηa (mPa s)

Ispb (s)

ρIspc (s g/cm3)

MBP‐1

155

243.2

331.2

2

MBP‐2

237

246.0

333.1

3

MBP‐3

235

246.9

336.3

4

MBP‐4

226

247.9

337.4

5

MBP‐5

279

248.8

336.4

6

MBP‐6

190

249.0

335.3

7

EBP‐1

216

245.6

331.3

8

EBP‐2

218

247.5

332.4

9

EBP‐3

382

248.4

336.8

10

EBP‐4

223

249.2

335.9

11

EBP‐5

397

249.8

336.2

12

EBP‐6

70

250.0

334.8

Entry

HILs

1

 Viscosity at 25 °C.  Specific impulse (Explo5 v6.02) at the optimum oxidizer (WFNA)‐to‐fuel ratio; isobaric conditions, equilibrium expansion, 7.0 MPa chamber pressure. c  Density impulse. Source: Liu et al. (2018). Reproduced with permission of John Wiley & Sons. b

8.4  Hypergolic Ionic Liquids T = –2 ms

T = 0 ms

T = 1 ms

T = 2 ms

T = 3 ms

T = 0 ms

T = 5 ms

T = 8 ms

T = 12 ms

T = 20 ms

(a)

(b)

Figure 8.25  Droplet test recorded for the new HILs with WFNA captured by a high‐speed camera at 2000 fps. (a) First droplet test of MBP‐1 and (b) first droplet test of MBP‐6. Source: Liu et al. (2018). Reproduced with permission of John Wiley & Sons.

11

EBP-

8.8

9 Ignition delay time (ms)

MBP-

9.7

9.5

10 8.2

8

7.0

7

5.8

6 5

4.5

6.2

5.8

6.3

4.8

4 3

2.3

2 1

1 2 P- P-1 P- P-2 MB EB MB EB

3 P- P-3 MB EB

5 4 P- P-5 P- P-4 MB EB MB EB

6 P- P-6 MB EB

Figure 8.26  ID times of the boranophosphate‐based HILs (MBP and EBP) with WFNA. The ID values are given as an average of three hypergolic drop tests. Source: Liu et al. (2018). Reproduced with permission of John Wiley & Sons.

8.4.4  Hypergolic Ionic Liquids Based on Other Anions Apart from the abovementioned HILs, there are many other typical ILs with novel physicochemical properties and short ID time. For example, Wang et al. (2018, p. 22819) synthesized and characterized four iodocuprate‐containing ILs (see Scheme 8.12). The performance of CuILs 1–4 as promoters of hypergolic ignition with highly concentrated H2O2 (95%) was evaluated for the

421

8  Design and Synthesis of Green Hypergolic Ionic Liquid Fuels

I

N

+ CuI

N

CH3OH

N

–

4 [Cu4I8 ]

N

4

I

N

+ CuI

N

CuIL 1

1 equiv.

3. equiv.

CH3OH

N

–

6 [Cu8I14 ]

N

6 1 equiv.

1 equiv. I

N

+ CuI

N

2 equiv.

CuIL 2

CH3OH

NH2

N N

–

2 [Cu5I7 ]n

N

2n

CuIL 3

3 equiv.

NH2

N

I

N

+ CuI CH3

1 equiv.

CH3OH

N N N

[CuI2 ]n CH3 n

CuIL 4

1 equiv.

Scheme 8.12  Synthesis of CuILs 1–4. Source: Wang et al. (2018). Reproduced with permission of Royal Society of Chemistry.

hypergolic fuels F1 and F2. In their evaluation of hypergolic ignitions of CuIL‐1‐containing fuels F1 and F2 (with highly concentrated H2O2), they observed 10‐fold shortened ID times (vs. promoter‐free ignitions) of 37 and 30 ms, respectively, and also found that the promoter CuIL‐2 is able to reduce the ID times for F1 and F2 fuels to 39 and 23 ms, respectively, while the best performing promoter was found to be CuIL‐3, exhibiting impressive ID times for F1 and F2 of 24 and 14 ms, respectively (Figure 8.27). The physicochemical –3 ms

0 ms

24 ms

25 ms

31 ms

250

(a) –3 ms

0 ms

14 ms

15 ms

21 ms

Ignition delay (ms)

422

200 150 CuIL 3 + F1 CuIL 3 + F2

100 50 0

(b)

(c)

0

2

4

6

8 10 wt%

12

14

16

Figure 8.27  Fast camera images from the hypergolic ignition tests with H2O2 (95%) and CuIL‐3 as a promoter. (a) Using F1 as a fuel; (b) using F2 as a fuel; (c) results of hypergolic ignition tests with H2O2 (95%), F1 and F2 fuels and CuIL‐3 as a promoter. Source: Wang et al. (2018). Reproduced with permission of Royal Society of Chemistry.

8.4  Hypergolic Ionic Liquids

properties of the hypergolic fuels with and without CuILs are shown in Table 8.17. The specific CuIL can be obtained in excellent yields (>92%) by controlling the molar ratios of the starting materials and by variating the cation structure. Structures of the [Cu8I146] cluster in CuIL‐2 and the [Cu5I72]n coordination polymer in CuIL‐3 are unprecedented, while examples of the [Cu4I84] cluster (found in CuIL‐1) and the [CuI2]n coordination polymer (found in CuIL‐4) are very rarely observed in hundreds of reported iodocuprate compounds. In terms of CuIL performance as promoters of hypergolic ignition with H2O2, at the concentration of 10 wt%, CuIL‐3 showed the shortest ID times of 24 and 14 ms for [EMIm][H3BCN] (F1) and [MIM][BH3] (F2) fuels, respectively. Also, at the same concentration, CuIL‐3 was able to keep the viscosity of the promoter‐in‐fuel CuIL‐3–F1 mixture at 50 mPa s and to remain in the F1 solution for at least four weeks without any signs of its degradation and the degradation of fuel, showing excellent chemical stability of this promoter‐in‐fuel mixture. Jin et  al. (2018, p. 4620) synthesized triaminocyclopropenium cation‐based ILs, which show excellent integrated properties, including high decomposition temperature (194 °C), high density (0.95 g/cm3), moderate viscosity (44 mPa s), ultrafast ID time (6 ms), and high specific impulse (301.9 s). As shown in Scheme  8.13, four TAC cations with different substituents are produced from tetrachlorocyclopropene, which subsequently undergo metathesis reactions with fuel‐rich anions to yield 14 TAC‐based HILs (1–14). These include Table 8.17  Physicochemical properties of the hypergolic fuels with and without CuILs. Fuel Promoter

a

Tda (°C)

ρb (g/cm3)

ηc (mPa s)

ΔHfd (kJ/g)

Ispe (s)

IDf (ms)

1

F1

No

247

0.980

19

−34.06

269.0

>1000

2

F1

1

220

1.018

42

−31.48

264.7

37

3

F1

2

221

1.021

48

−31.22

261.3

36

4

F1

3

219

1.025

50

−31.56

263.9

23

5

F1

4

214

1.023

55

−32.55

262.3

38

6

F2

No

263

0.930

5

−36.17

266.6

>800

7

F2

1

161

1.005

80

−32.96

262.5

30

8

F2

2

162

1.007

87

−33.86

259.1

23

9

F2

3

160

1.010

82

−33.35

261.7

14

10 F2

4

158

1.008

89

−33.76

260.1

28

 Decomposition temperature.  Density at 25 °C. c  Viscosity at 25 °C. d  Calculated heat of combustion, based on bomb calorimetry measurements. e  Isp values for promoter‐in‐fuel mixtures that were calculated from the Isp values of the pure fuels F1 and F2 and CuIL promoters. f  Ignition delay times. Source: Wang et al. (2018). Reproduced with permission of Royal Society of Chemistry. b

423

424

8  Design and Synthesis of Green Hypergolic Ionic Liquid Fuels

R1 Cl

Cl

R2

R 1-NH-R2 Cl

Cl

N

NaX/AgX

Cl

R2

R2

R1

N R1

N

N

R1

R2

X– R2

N

N

R1

R2

R1

N(Ethyl)2

N(Allyl)2 M 2 +=

M1+= (Allyl)2N

N(Allyl)2

N(Ethyl)2

(Ethyl)2N

Allyl

N(Propyl)2

N M4+=

M 3 += (Propyl)2N M1+BH3CN–

M1+N(CN)2– IL5 M 1 N3

–

M1+ BH2CN N N IL11

M2+N(CN)2– IL6

IL8

–

M2+BH3CN– IL2

IL1

+

N(Propyl)2

Allyl N M3+BH3CN– IL3

IL9

IL10

–

N

N IL12

M4+BH3CN– IL4

IL7 M 3 +N3 –

BH2CN

Allyl

M3+N(CN)2–

M2+N 3– M 2+

N

–

M3+ BH2CN N N IL13

–

M4+ BH2CN N N IL14

Scheme 8.13  Synthesis of TAC cation‐based HILs (1–14) discussed herein. Source: Jin et al. (2018). Reproduced with permission of John Wiley & Sons.

CB‐based TAC ILs 1–4, DCA‐based TAC ILs 5–7, azide‐based TAC ILs 8–9, and CIB‐based TAC ILs 11–14. In their tests, 12 of the 14 TAC‐based ILs show expected hypergolic behavior upon contact with the oxidizer WFNA, and their ID times range from 6 to 330 ms. Table 8.18 shows the physicochemical properties of as‐prepared TAC‐ based ILs. Among TAC‐based HILs, IL‐12 exhibited the shortest ID time (6 ms). Salts 11 and 14 also illustrated relatively short ID times of 7 and 6.5 ms (see Figure 8.28), respectively, which were comparable with that of UDMH (4.8 ms). In this study, two TAC salts based on the azide anion (8 and 9) show hypergolic properties with WFNA that are not observed previously for azide‐based ILs. The interaction between the acid and base is known to originate from the energy gap between the HOMO of the anion and LUMO of nitric acid. This gap can be used as a criterion for evaluating the hypergolic ability (or reactivity), and a smaller gap results in a larger interaction.

8.4  Hypergolic Ionic Liquids

Table 8.18  Properties of ILs 1–14. Tga (°C)

Tdb (°C)

ρc (g/cm3)

ΔHfd (kJ/g)

ηe (mPa s)

Ispf (s)

IDg (ms)

1

4

155

0.95

2.25

128

302.9

74.0

2

−24

211

0.93

0.49

52

304.6

14.0

IL

3

23

219

0.96

0.04

171

304.5

330.0

4

−18

165

0.99

1.93

61

303.9

37.5

5

−65

261

1.04

2.64

166

300.8

111.0

6

−5

299

1.1

1.10

56

301.9

52.0

7

14

277

1.09

0.60

178

302.2

—

8

−35

171

0.97

2.97

235

302.9

147.0

9

9

298

1.01

1.40

196

304.6

90.0

10

56

302

0.98

0.79

—

304.2

—

11

−20

170

0.99

2.00

204

300.9

7.0

12

−28

194

0.95

0.54

44

301.9

6.0

13

−53

199

0.93

0.16

241

302.2

63.0

14

−24

155

0.97

1.72

79

301.2

6.5

a

 Glass transition temperature.  Decomposition temperature (TGA, onset). c  Density at 25 °C. d  Heat of formation. e  Viscosity at 25 °C. f  Specific impulse (CEA 400, pressure 0.95 MPa; area expansion ratio of nozzle 70; oxidizer N2O4 [equivalence ratio = 1.0]). g  Ignition delay times with WFNA. Source: Adapted from Jin et al. (2018, table 1). b

Figure 8.29 shows the energy gap values for the anions used here. It is clear that the energy gap value for the azide anion is much larger than that other commonly used hypergolic anions, which demonstrates the relatively weak hypergolic ability. On the other hand, for a given anion, the cationic structure may have some effect on the energy gap of the ILs. Herein, four azide‐based ILs with different cationic structures are selected, and their energy gap values of cations are calculated (Figure 8.29b). The energy gaps of two TAC cations are found to be lower than those of imidazole cations, which may lead to the hypergolicity of TAC‐ based ILs. Cho and coworkers studied the hydrolysis of [Bmim][PF6], [Bmim][BF4], and [Bmim][SbF6] (Cho et al. 2008, p. 67). They found that hydrolysis of [SbF6]− is easier than that of [BF4]− and [PF6]− under identical conditions. Aqueous extraction has been widely used to recycle ILs from reaction mixtures. However, these studies illustrate that specific caution should be taken to avoid potential hydrolysis when handling aqueous solutions of [SbF6]−, [BF4]−, and [PF6]− ILs (Mai, Ahn, and Koo 2014, p. 872). Rachiero, Titi, and Rogers (2017, p. 7736) found that the

425

426

8  Design and Synthesis of Green Hypergolic Ionic Liquid Fuels

T =–7 ms

T= 0 ms

T =3 ms

T =6.5 ms

T = 14.5 ms

T = 30 ms

T =49 ms

T=130 ms

T = 207 ms

Figure 8.28  Droplet tests of IL‐14 with WFNA, as captured by using a high‐speed camera. Source: Jin et al. (2018). Reproduced with permission of John Wiley & Sons.

combination of an energetic imidazole bonded to a hypergolic trigger, such as certain boron hydride clusters, dramatically improved both the energy density and ID of the resulting compounds while also introducing tunability via alkyl group substitution as demonstrated here with (Rim)2B10H12 (R = H, C1, C2, and C4). The synthesis process is shown in Scheme 8.14. In control experiments, the imidazoles and n‐B10H14 are not hypergolic under these experimental conditions. In the hypergolic tests for 1–4, a green‐red flame is observed in all experiments, indicating the oxidation of B─H bonds to boron oxide. The compounds showed hypergolic ignition with exceptionally short ID times, 1(0) ms for 2–4 and 2(1) ms for 1 (Table 8.19), and high and continuous flames (45 cm) with long durations (480 ms). The observed ID times are comparable with those observed for previously synthesized [B9H14]‐based IL salts (ID times of 3 ms) and considerably lower than the values reported for [N(CN)2]‐ based HILs, e.g. [1‐methyl‐3‐aminotriazolium][N(CN)2] (ID  =  39(2) ms). The explosiveness diminishes with the lengthening of the alkyl chain. These results suggest that the introduction of electron‐donating groups, increasing the heat of combustion of the imidazole moiety, favors the explosive nature observed. In summary, the presence of the imidazole also introduces a versatile platform for structural modification of the compounds. The compounds prepared here are solids, which exhibit extremely short (≤2 ms) ID time with vigorous flames. As shown in Figure 8.30, compounds 1–4 appear to induce hypergolicity in kerosene when present as undissolved solids. The results also open the path to tuning

8.4  Hypergolic Ionic Liquids 3

2.5

Gap (eV)

2

1.5

1

0.5

(a)

0 Azide

DCA

CB

CIB

TAC-M1+

TAC-M2+

BMIm

EMIm

9

8

Gap (eV)

7

6

5

4

3

(b)

Figure 8.29  Computed energy gaps of various experimentally studied anions (a) and cations (b) EMIm = 1‐ethyl‐3‐methylimidazolium. Source: Jin et al. (2018). Reproduced with permission of John Wiley & Sons.

the melting points with the possibility of forming lower melting or perhaps liquid analogues of these compounds. The oxygen‐rich aluminum complex anion was also used for the production of oxidizer ILs for rocket propellant applications. As early as 1966, the salt [Et4N] [Al(NO3)4] was reported as the first substantiated example of the [Al(NO3)4]− anion (Addison, Boorman, and Logan 1966, p. 1434). In 2003, an oxygen‐balanced EIL, 1‐ethyl‐3‐methylimidazolium tetranitratoborate, was reported by Christe et al. They designed and synthesized another new oxygen‐balanced EIL using the [Al(NO3)4]− anion as a thermally more stable high oxygen carrier and the 1‐ethyl‐4,5‐dimethyltetrazolium cation as a more energetic counterion (Jones et al. 2006, p. 4981). The synthesis process is shown in Scheme 8.15.

427

N N N

H

H

H B HBH B H HB B B H B B H B H H H

OH NH O

N N

OH 5 rt THF/C6H6

85 °C 5h N

H N H H B B H HB HB B B H B B B H H H H

C6H6

H

B

H H H 2.1 B H N N H B B H HB HB B B H C H 6 6 B B BH H H 85 °C H 5h

H

85 °C 5h 2.1 N

N

exo-6,exo-9-(C1im)2B10H12

C6H6

3, 60%

85 °C 5h

2.1 N

OH

O

[C4Him][B5O6(OH)4]

N N

B B

C6H6 N

O

O

2, 45% N

B

O

exo-6,exo-9-(C1im)2B10H12

2.1 N

O

B

HO

B

B

NH

HN

N

H N H B HB H B H B HB B HB B B H H H H exo-6,exo-9-(C4im)2B10H12 H

N

B

4, 65%

NH H N H B HBH B H H B B BH B B BH H H H exo-6,exo-9-(Him)2B10H12 N

H

B

1, 83%

Scheme 8.14  Synthesis of 1–4 and isolation of 5. Source: Rachiero, Titi, and Rogers (2017). Reproduced with permission of Royal Society of Chemistry.

8.4  Hypergolic Ionic Liquids

Table 8.19  Hypergolicity and IDs of 1–4. Compound

Ignition delay

1, (Him)2B10H12

2, (C1im)2B10H12

3, (C2im)2B10H12

4, (C4im)2B10H12

Values in parentheses denote the standard deviation of three averages tests. Source: Rachiero, Titi, and Rogers (2017). Reproduced with permission of Royal Society of Chemistry.

1, ID = 35

2, ID = 31

3, ID = 564a

4, ID = 4

Figure 8.30  Hypergolic ignition of 1–4 in kerosene‐based mixtures. a IDs (ms) observed after contact of the 2nd drop of WFNA. Source: Rachiero, Titi, and Rogers (2017). Reproduced with permission of Royal Society of Chemistry.

1‐Ethyl‐4,5‐dimethyltetrazolium cation is chosen because of its intrinsically large positive heat of formation and its ability to form a RTIL with the tetranitratoaluminate anion. Interestingly, the ignition of this IL and self‐sustained

429

430

8  Design and Synthesis of Green Hypergolic Ionic Liquid Fuels

N

+

N

N

+

N

N N

N

AlCl3

N

CH3NO2

N

Cl–

N

+

N

+ 4N2O4

Cl

+

N

N

CH3NO2 N

Cl

O N O

ON

O

O N O O

O

+

N

N

AlCl4–

Al– Cl

Cl

Al–

O

O

O N O

Scheme 8.15  Synthesis of 1‐ethyl‐4,5‐dimethyltetrazolium tetranitratoaluminate. Source: Adapted from Jones (2006, scheme 1).

­ urning were readily achieved by either thermal heating to about 200 °C or the b use of a hot 40 gauge Ni/Cr wire wrapped around the sample container, a glass melting point capillary. An ID of a few seconds was observed. In search of green hypergolic propellants, Schneider et  al. (2011a, p. 5886; 2011b) have great interest not only in the new high‐performance HILs but also in the choice of greener oxidizers. On the basis of above considerations, two hydrogen‐rich EILs (Scheme 8.16) were designed and synthesized, and their ignition performance with 98% H2O2 as oxidizer was studied (as shown in Table 8.20). BH4

+

BH4

BH4

Al

Al BH4

H 4B

(A)

BH4

n-C14H29

n-C14H29

P

P

n-C6H13[BH4]

n-C6H13

BH4

H4 B

n-C6H13 [Al(BH4)4]

n-C6H13

n-C6H13 [THTDP][BH4]

n-C6H13 [THTDP][Al(BH4)4] H

4

P

B H

H

H

1 (a)

H H

H H

(B)

H H

4

P

H

B

(b)

2

B

H Al

H

H

H B

B H

H

H

H

H

Scheme 8.16  (A) Synthesis and structures of [Al(BH4)4]− based RTILs. (B) Borohydride ILs 1 (a) and 2 (b). Source: Adapted from Schneider et al. (2011a, scheme 2).

­  References

Table 8.20  Drop test results of 2 on four oxidizers (N2 atmosphere). 90% H2O2

Reactivity with 2 ignition delay

98% H2O2

N2O4

WFNA

Ignition

Ignition

Ignition

Explosion