ABSTRACT Title of Dissertation: METHANE VALORIZATION OVER NOVEL CATALYST SYSTEMS VIA DIRECT PATHWAYS Su Cheun Oh, Doctor of Philosophy, 2018 Dissertation directed by: Professor Dongxia Liu Department of Chemical and Biomolecular Engineering Methane, when converted to higher hydrocarbons, promises a great future as the substituent for liquid petroleum in petrochemical and fine chemical industries. Methane conversion via direct pathways such as oxidative coupling of methane (OCM) to ethylene and direct non-oxidative methane conversion (DNMC) to C2 (acetylene, ethylene and ethane) and aromatics have attracted much attention given their unique capability in circumventing the intermediate energy intensive steps found in indirect processes. In the OCM process, the more reactive nature of C2 products leads to the sequential oxidation of C2 to COx (CO or CO2). Selective catalysts that favor C2 formation are desired. The DNMC is challenged by low equilibrium conversion, high endothermicity and high coke selectivity. Catalysts or reaction systems that concurrently solve these challenges are required. This dissertation aims to develop novel catalyst systems to conquer limitations in OCM and DNMC to realize efficient and effective C2 production. For OCM reaction, hydroxyapatite (HAP), a bioceramic material with capability of cation and/or anion substitutions, was innovatively employed as a catalyst. The effects of cation and/or anion substitutions in HAP on OCM reaction were studied. Rigorous description of the reaction kinetics of OCM in HAP-based catalysts was conducted. Finally, the selective control of exposed crystalline plane of HAP was realized to further understand the catalytic behaviors of HAP-based catalysts in OCM reactions. It is shown that cation and/or anion substitution can change the physicochemical properties of the HAP catalysts, and as consequences, the OCM catalytic performances. The c-surface (i.e., (002) crystalline plane) of HAP-based catalysts exhibited significant enhancement in areal rate in OCM reaction. The single iron sites confined in the lattice of silica matrix (Fe/SiO2) is an emerging type of methane activation catalyst in DNMC. We innovated a millisecond catalytic wall reactor made of Fe/SiO2 catalyst to enabling stable and high methane conversion, C2+ selectivity, low coke yield and long-term durability. These effects originate from initiation of DNMC by surface catalysis on reactor wall, and maintenance of the reaction by gas phase chemistry in reactor compartment. Autothermal operation of the catalytic wall reactor is potentially feasible by coupling and periodical swapping of endothermic DNMC and exothermic oxidative coke removal on opposite side of the reactor. High carbon and thermal efficiencies and low cost in reactor materials are realized for the technoeconomic process viability of the DNMC technology. In addition, we created a process of tailoring product selectivity towards to C2 hydrocarbons by employing a mixture of Fe/SiO2 catalyst and mixed ionic-electronic conductive perovskite (SrCe0.8Zr0.2O3−δ) oxide in the presence of hydrogen co-feed in methane stream. Unprecedentedly high C2 yield was realized in DNMC reaction to maximize its potential as a feedstock for ethylene production in chemical industries. METHANE VALORIZATION OVER NOVEL CATALYST SYSTEMS IN DIREACT PATHWAYS by Su Cheun Oh Dissertation submitted to the Faculty of the Graduate School of the University of Maryland, College Park, in partial fulfillment of the requirements for the degree of Doctor of Philosophy 2018 Advisory Committee: Professor Dongxia Liu, Chair Professor Kyu Yong Choi Professor Taylor Woehl Professor Ashwani Gupta Dr. Dat T. Tran © Copyright by Su Cheun Oh 2018 Acknowledgements I would like to express my deepest gratitude to my PhD advisor, Professor Dongxia Liu, for giving me the opportunity and believing in me who came to the lab without any Master’s experience, giving me the financial support for a technically challenging yet meaningful project, providing me intellectual advice, feedback, and encouragement, being available every time I need more attention and guidance, and finally providing strong support for my future career pursuit. She has given me opportunities after opportunities and I could not say enough thank you. Her zeal for passion and perfection is something I looked up to during my PhD and is something I hope to hold up in the future. I would like to thank everyone at Army Research Lab (ARL), including Dr. Dat T. Tran and Dr. Ivan C. Lee. The team had been able to work closely with us and provided us the characterization facilities to perform XRD, Raman, SEM and TEM. Despite his busy schedule, Dr. Tran always performed the measurements in a timely manner. I would also like to thank Professor Eric Wachsman from University of Maryland Energy Research Center for letting me use his lab facilities, including chemicals, high temperature furnace, ball-milling and tape casting instruments, mass spectrometer, sonicator etc. Without these materials and infrastructures, my research project would not be possible. I deeply appreciate Professor Yu Lei’s (University of Alabama, Huntsville) generous help for carrying out XAS measurements at Argonne National Lab and performing data analysis for me. I also would like to thank Dr. Huiyong Chen from Northwest University of China for elemental analysis measurement. I am very grateful to my PhD committee members, Professor Kyu Yong Choi, Professor Taylor Woehl, Professor Ashwani Gupta, and Dr. Dat Tran for their commitment, time, and very constructive advice given during my defense ii presentation. I would like to thank National Science Foundation (NSF) for sponsoring the OCM and DNMC projects under the Award Number NSF-CBET 1642405 and 1351384. I would like to thank many staffs at University of Maryland, College Park who provided a helping hand throughout the years. They are the technical and administration staffs in the AIM Lab, Surface Analysis Center and business office of UMCP Department of Chemical Engineering. I would like to dedicate special appreciation to three wonderful staffs that I had the pleasure to work with, Sz-Chian Liou, Kay N. Morris, and Jenna Bishop. It has been an incredibly gratifying experience working in a lab (Liu Research Group at University of Maryland, College) full of undergraduate and graduate students who have become more like friends than colleagues to me, including Mann Sakbodin, Wei Wu, Dr. Laleh Emdadi, Junyan Zhang, Guanghui Zhu, Dongchang Qing, Jingyu Fang, Emily Schulman, Ricardo Morales and Dr. Yiqing Wu. I really appreciate their helps and guidance throughout my research work. Mann Sakbodin and Wei Wu are two close friends who have shared ideas, advice, assistance, and encouragement throughout the 5.5 years during which our times in the lab have overlapped. I would like to express my heartfelt gratitude to my parents, especially my mother, who has been showing constant faith in me and encouraging me to pursue what I desired. Even though she barely understood my research area, she was willing to support any decision I made. Despite not being able to spend much time with her during this time, she has been a Facetime away and she is always be my best listener. Her selfless love and support mean a lot to me. I would like to thank my siblings for being who they are and supportive of every one of my decision. Simply, without my family, I will not be where I am today. iii Last but not least, I would like to make a special mention to my husband, Shing Shin Cheng, who continues to believe in me long after I had lost belief in myself. I thank him for his continued and unflagging love, patience and support during my PhD journey; this work is simply impossible without him. He is always showing how proud he is of me. He has seen me through the ups and downs of the entire PhD process, encouraging me to tackle the seemingly unsolvable problems and overcome the insurmountable pressure. He has always been there to be my pillar and is always the first person that I want to share my best and worst moments. To my beloved family, this dissertation is dedicated. iv Table of Contents Acknowledgements………………………………………………………………………. ii Table of contents.….……………………………………………………………………... v List of Tables…………………………………………………………………………...... ix List of Figures…………………………………………………………………………….. x List of Abbreviations and Acronyms……………………………………………………. xv Chapter 1: Introduction………………………………………………………………… 1 1.1 The abundance of methane………………………………………………………... 1 1.2 Current catalytic methane conversion methods…………………………………… 2 1.2.1 Indirect methane conversion pathway……………………………………… 3 1.3 Direct oxidative coupling of methane (OCM) to C2 hydrocarbons (ethylene and ethane) ………………………………………………………………………………... 5 1.3.1 Challenges in OCM reaction………………………………………………... 7 1.3.2 Catalysts for OCM reaction………………………………………………… 8 1.3.2.1 Li/MgO catalysts……………………………………………………… 9 1.3.2.2 Mn/Na2WO4/SiO2 catalysts………………………………………….. 11 1.3.2.3 La2O3 catalysts……………………………………………………….. 13 1.3.2.4 Commercialized catalysts……………………………………………. 15 1.3.2.5 Hydroxyapatite (HAP)……………………………………………….. 16 1.3.3 OCM reaction kinetics…………………………………………………….. 16 1.4 Direct non-oxidative methane conversion (DNMC) to C2+ hydrocarbons……….. 18 1.4.1 Challenges in DNMC reaction…………………………………………….. 20 1.4.2 Catalysts for DNMC reaction……………………………………………… 22 1.4.2.1 Metal/zeolite-based catalysts……………………………………. 22 1.4.2.2 Iron/silica catalyst……………………………………………..… 25 1.4.2.3 Gallium nitride catalyst………………………………………….. 27 1.4.2.4 Nickel on ceria-zirconia oxide (Ni/CZ) …………………………. 28 1.4.2.5 Silica-supported tantalum hydride ((≡SiO)2Ta-H) ……………… 28 1.4.2.6 Cu/Zn/Al2O3…………………………………………………….. 29 1.4.3 Membrane reactor for DNMC reaction…………………………………… 29 1.5 Thesis overview……………………………………………………………………... 30 Chapter 2: Influences of Cation and Anion Substitutions on Oxidative Coupling of Methane over Hydroxyapatite Catalysts……………………………………………... 32 2.1 Introduction……………………………………………………………………… 32 2.2 Experiments……………………………………………………………………… 33 2.2.1 Materials…………………………………………………………………... 33 2.2.2 HAP-based catalysts preparation………………………………………….. 34 2.2.3 Catalysts characterization…………………………………………………. 35 2.2.4 Determination of acidity of catalysts………………………………………. 36 2.2.5 OCM catalytic test………………………………………………………… 36 2.3 Results and Discussion…………………………………………………………… 37 2.3.1 Textural and structural properties of catalysts……………………………... 37 2.3.2 Composition analysis of HAP-based catalysts…………………………….. 43 2.3.3 Surface acidity of HAP-based catalysts…………………………………… 45 v 2.3.4 Performance of HAP-based catalysts in OCM reactions…………………………... 47 2.4 Conclusion of Chapter 2……………………………………………………………... 53 Chapter 3: Catalytic Consequences of Cation and Anion Substitutions on Rate and Mechanism of Oxidative Coupling of Methane over Hydroxyapatite Catalysts…… 55 3.1 Introduction………………………………………………………………………. 55 3.2 Experiments………………………………………………………………………. 57 3.2.1 Materials…………………………………………………………………... 57 3.2.2 HAP-based catalysts preparation………………………………………….. 57 3.2.3 Catalyst characterization…………………………………………………... 59 3.2.4 OCM catalytic test………………………………………………………… 60 3.3 Results and discussion……………………………………………………………. 61 3.3.1 Structural analysis of HAP-based catalysts………………………………... 61 3.3.2 Composition analysis of HAP-based catalysts…………………………….. 64 3.3.3 O2-TPD and CH4-TPD profiles of HAP-based catalysts…………………... 66 3.3.4 Reaction pathways of OCM reactions over HAP-based catalysts………… 68 3.3.4.1 Derivation of reaction rate equations…………………………………. 70 3.3.4.2 Methane and oxygen consumption rates over HAP-based catalysts based on Langmuir-Hinshelwood mechanism………………………………………. 73 3.3.4.3 Methane and oxygen consumption rates over HAP-based catalysts based on Eley-Rideal mechanism…………………………………………………… 76 3.3.4.4 Product formation and selectivity over HAP-based catalysts…………. 78 3.3.4.5 Effects of Pb2+ and/or F- substitution in HAP on OCM reactions…….. 83 3.4 Conclusion of Chapter 3…………………………………………………………... 85 Chapter 4: Effects of Controlled Crystalline Surface of Hydroxyapatite on Methane Oxidation Reactions…………………………………………………………………… 87 4.1 Introduction………………………………………………………………………. 87 4.2 Experiments………………………………………………………………………. 89 4.2.1 Materials…………………………………………………………………... 89 4.2.2 HAP-based catalysts preparation………………………………………….. 90 4.2.3 Catalyst characterization…………………………………………………... 92 4.2.4 Catalytic methane oxidation reaction……………………………………… 93 4.2.4.1 Methane combustion catalytic test……………………………………. 93 4.2.4.2 Oxidative coupling of methane (OCM) catalytic test…………………. 94 4.2.5 Density functional theory (DFT) calculation…………………………… 94 4.3 Results and discussion……………………………………………………………. 96 4.3.1 Chemical structure and property of HAP crystal…………………………... 96 4.3.2 SEM and XRD characterizations for oriented and unoriented HAP-based catalysts…………………………………………………………………………. 99 4.3.3 Composition and surface area of oriented and unoriented HAP-based catalysts………………………………………………………………………... 101 4.3.4 Effects of controlled crystalline plane of HAP on methane oxidation reactions……………………………………………………………………….. 103 4.3.4.1 Performance of oriented HAP in methane combustion reaction……... 103 4.3.4.2 Performance of oriented HAP and Pb-HAP in OCM reactions……… 107 vi 4.3.4.3 Functionality of crystal plane of HAP-based catalysts………………. 109 4.3.4.4 Kinetics of c- and a-surfaces of Pb-HAP in OCM reaction………….. 113 4.4 DFT calculations on oriented and unoriented HAP catalysts……………………. 117 4.5 Conclusion of Chapter 4…………………………………………………………. 121 Chapter 5: Direct Non-oxidative Methane Conversion in a Millisecond Catalytic Wall Reactor………………………………………………………………………………... 123 5.1 Introduction…………………………………………………………………….. 123 5.2 Experiments…………………………………………………………………….. 124 5.2.1 Materials…………………………………………………………………. 124 5.2.2 Synthesis of the Fe/SiO2 catalyst…………………………………………. 124 5.2.3 Manufacturing of catalytic wall reactor………………………………….. 126 5.2.4 Catalyst characterization…………………………………………………. 127 5.2.5 DNMC reaction………………………………………………………….. 128 5.2.5.1 DNMC reaction in fixed-bed reactor………………………………... 128 5.2.5.2 DNMC reaction in catalytic wall reactor……………………………. 128 5.2.6 Model simulations for DNMC-coke combustion process concept……….. 129 5.3 Results and Discussion…………………………………………………………. 131 5.3.1 Structural analysis of Fe/SiO2 catalyst…………………………………… 131 5.3.2 Performance of catalytic wall reactor in DNMC reaction………………... 131 5.3.3 Quantification of amount of coke in catalytic wall reactor……………….. 136 5.3.4 Energy balance analysis in autothermal catalytic wall reactor…………… 140 5.3.5 Model simulations for autothermal DNMC-coke combustion process concept………………………………………………………………………… 143 5.4 Conclusion of Chapter 5………………………………………………………… 147 Chapter 6: Tailoring the Selectivity in Direct Non-Oxidative Methane Conversion: Effects of H2 Co-feed in the Presence of SrCe0.8Zr0.2O3−δ Perovskite in Fe/SiO2 Catalyst………………………………………………………………………………... 148 6.1 Introduction…………………………………………………………………….. 148 6.2. Experimental…………………………………………………………………... 151 6.2.1 Materials…………………………………………………………………. 151 6.2.2 Synthesis of Fe/SiO2 catalyst…………………………………………….. 151 6.2.3 Synthesis of SrCe0.8Zr0.2O3−δ perovskite oxide…………………………... 151 6.2.4 Catalyst characterization…………………………………………………. 152 6.2.5 Catalytic DNMC reactions……………………………………………….. 153 6.3. Results and discussion…………………………………………………………. 154 6.3.1 Physicochemical properties of SCZO perovskite and Fe/SiO2 catalyst…... 154 6.3.2 DNMC in the absence of H2 co-feed in methane stream…………………. 156 6.3.2.1 DNMC over SCZO perovskite and Fe/SiO2 catalyst………………... 156 6.3.2.2 DNMC over a mixture of SCZO and Fe/SiO2 materials…………….. 158 6.3.3 DNMC in the presence of H2-cofeed in methane stream…………………. 160 6.3.3.1 Effects of H2-cofeed on DNMC over Fe/SiO2 and SCZO materials… 160 6.3.3.2 Effects of H2-cofeed on DNMC over Fe/SiO2 and SCZO mixtures… 162 6.3.4 Coke behaviors-catalytic performances correlations of SCZO/Fe/SiO2…. 163 6.3.4.1 Amount of coke deposition………………….......………………….. 163 vii 6.3.4.2 Types of coke deposition………………...…………………………. 165 6.5. Conclusion……………………………………………………………………... 170 Chapter 7: Conclusion (Major Contributions) and Future Work……………….… 172 7.1 Conclusions (Major Contributions)…………………………………………….. 172 7.2 Future work……………………………………………………………………... 176 Appendix A: Derivation of Rate Equation for OCM Reactions over HAP-based Catalysts…..................................................................................................................... 178 Appendix B: Aspen Plus (V10) Simulation Details and Results…………………… 192 Appendix C: Flow Diagram of Reactor Unit for Heterogeneous Gas-Solid or Liquid- Solid Catalytic Reaction Tests……………………………………………………….. 198 Bibliography………………………………………………………………………….. 199 List of Publications…………………………………………………………………… 222 viii List of Tables Table 2.1. Chemical compositions and surface areas of HAP-based catalysts for OCM reactions………………………………………………………………………………… 38 Table 3.1. Chemical compositions and surface area of HAP-based catalysts for OCM reactions…………….…………………………………………………………………... 63 Table 3.2. EXAFS fit parameters for Pb oxides and Pb catalysts (k2: ∆k = 3 – 12 Å-1 and ∆r = 1.3 – 3.5 Å).……………………………………… ………………………………. 66 Table 3.3. Equilibrium constant of O2 adsorption (K1), equilibrium constant of CH4 adsorption (K2) and rate constant of CH4 activation (ka) based on Langmuir-Hinshelwood mechanism at 973K of OCM reactions over HAP-based catalysts.……………………... 76 Table 3.4. Equilibrium constant of O2 adsorption (K1) and rate constant of CH4 activation (k3∙n1) based on Eley-Rideal mechanism with associative O2 adsorption at 973K of OCM reactions over HAP-based catalysts.…………………………………………………….. 78 Table 3.5. Rate constant for formation of C2H6 (k4), CO (k5) and CO2 (k6∙n1) at 973K of OCM reactions over HAP-based catalysts.……………………………………………… 80 Table 4.1 Comparison of calculated lattice parameters with previous DFT and experimental results……………………………………………………………………... 95 Table 4.2. Chemical composition and surface area of HAP-based catalysts…………… 102 Table 4.3. Equilibrium constant of O2 adsorption (K1) and the rate constant of CH4 activation (k3∙n1) based on Eley-Rideal mechanism with associative O2 adsorption at 973 K of OCM reactions over c-surface and a-surface, respectively, of Pb-HAP catalysts… 116 Table 4.4. Vacancy formation energy on different facets……………………………… 120 Table 5.1. Amount of O2 consumed and products formed during TPO process……….. 138 Table 5.2. Summary of DNMC reaction and the corresponding coke combustion at different reaction conditions…………………………………………………………… 143 Table 5.3. Heating and cooling duties for heat exchangers with and without using heat integration and their hourly costs, respectively………………………………………… 145 Table 5.4 Hourly duties and costs for system operation with and without heat integration……………………………………………………………………………... 145 Table 5.5 Current costing and hourly production rates for feed and product species in the DNMC reaction………………………………………………………………………... 146 Table 5.6. Annual plant operational utility and raw material costs and sales credit for heat- integrated process Aspen model……………………………………………………….. 146 Table 6.1. Ratio of D band to G band determined from Raman spectroscopy analysis for Fe/SiO2 catalyst, SCZO material and 5 wt% SCZO material in Fe/SiO2 catalyst after 3.5 hours DNMC reaction at 1273 K and at different H2-cofeed concentrations…………... 167 ix List of Figures Figure 1.1. Overview of the selected direct and indirect methane conversion pathways………………………………………………………………………………….. 3 Figure 1.2. Reaction scheme showing steps involved in OCM reaction………………… 6 Figure 1.3. Gibbs free energy of reaction (ΔrG973 K) for the formation of ethane, ethylene, carbon monoxide and carbon dioxide, respectively, from methane in OCM…………….. 7 Figure 1.4. Elemental compositions of OCM catalysts with yield (C2) ≥ 25% reported in literature. All the catalysts were tested in a fixed-bed reactor in the cofeed mode under atmospheric pressure at temperatures from 943 to 1223 K, p(CH4)/p(O2) = 1.7 - 9.0, and contact times from 0.2 to 5.5 s. Reproduced with permission from reference [19]. Copyright 2011 Wiley-VCH………………………………………………………………………… 8 Figure 1.5. Products and coke formation mechanism in DNMC………………………... 20 Figure 1.6. Equilibrium conversion of methane for DNMC. (calculated using HSC Chemistry 6.0 Software) ………………………………………………………………... 21 Figure 1.7. A) STEM-HAADF image of spent Fe/SiO2 catalys, (B) in-situ XANES of the Fe/SiO2 catalyst upon activation and (C) Fourier transformed (FT) k3-weighted χ(k)- function of the EXAFS spectra for Fe/SiO2 catalyst. (Reproduced with permission from reference [99]. Copyright 2014 the American Association for the Advancement of Science………………………………………………………………………………….. 26 Figure 1.8. Schematic diagram for methane C−H bond polarization on the surface of the GaN m-plane. Adapted with permission from reference [104]. Copyright 2014 American Chemical Society………………………………………………………………………... 27 Figure 2.1. SEM images showing morphologies of HAP-based catalysts: (A) HAP, (B) HAP-CO3, (C) Pb-HAP-CO3, and (D) Pb-HAP, respectively. .…………………………. 39 Figure 2.2. XRD patterns (A), FT-IR spectra (B), Raman spectra (C), and TGA curves (D) of HAP-based catalysts, respectively, used for OCM reactions. .……………………….. 40 Figure 2.3. XPS spectra (A), Pb 4f spectra (B), Ca 2p (C) and XANES spectra (D) of HAP- based catalysts, respectively, used for OCM reactions. .………………………………… 44 Figure 2.4. NH3-TPD profiles of HAP, HAP-CO3, Pb-HAP and Pb-HAP-CO3 catalysts…………………………………………………………………………………. 46 Figure 2.5. (A) Methane conversion with time-on-stream in OCM reactions and (B) Product selectivity for OCM reactions over HAP-based catalysts at 23% conversion under 973 K and 101 kPa pressure conditions and a space velocity of 8800 mL gcat-1.hr-1……. 48 Figure 2.6. Methane conversion (A) and product selectivity (C2 (B); CO2 (C); CO (D)) of OCM reactions over HAP-based catalysts at different temperatures. (PCH4 = 27.1 kPa, PO2 = 11.0 kPa, total flow = 46 ml min-1, He was used as the balance gas)…………………. 49 Figure 2.7. Effect of methane-to-oxygen (CH4/O2) ratio on C2 selectivity and CH4 conversion at (A) 923 K, (B) 943 K and (C) 973 K at a constant CH4 partial pressure of 25 kPa. (Total flow rate = 46 mL min-1) (Filled symbol represents C2 selectivity, unfilled symbol represents CH4 conversion) .……………………………………………………. 50 x Figure 2.8. XRD pattern (A), FTIR (B) and (C) Raman of Pb-HAP-CO3 after 10-hour of OCM reaction. For comparison purpose, XRD patterns of HAP, HAP-CO3 and Pb-HAP samples after 10-hour OCM reaction have also been presented in (A)………………….. 52 Figure 3.1. SEM images showing morphologies of HAP-based catalysts: (A) HAP, (B) HAP-F, (C) Pb-HAP-F, (D) Pb-HAP, respectively...………….………………………… 62 Figure 3.2. XRD patterns (A), FT-IR spectra (B), Raman spectra (C), and XANES spectra (D) of HAP-based catalysts, respectively, used for OCM reactions. k2-weighted magnitude and imaginary component of Fourier transform EXAFS of (E) Pb-HAP and (F) Pb-HAP- F. (k2: ∆k = 3 – 12 Å-1. Blue, Fourier transform magnitude. Red, imaginary component. Dash line, experimental data. Solid line, fitted data)..………………………………….... 66 Figure 3.3. O2-TPD (A) and CH4-TPD (B) profiles of HAP, HAP-F, Pb-HAP and Pb- HAP-F catalysts, respectively. .…………………………………………………………. 67 Figure 3.4. Methane consumption rate as a function of CH4 pressure at P$% = 7.0 kPa (A) and O2 pressure at P&'(= 25 kPa (B), respectively. (973 K, 101 kPa total pressure, total flow rate = 46 mL min-1, He as balance gas.) The curves in (A) and (B) denote the fitting results using optimized rate constants in Table 3.3...……………………………………. 74 Figure 3.5. Product of CH4 pressure and the inverse rate of CH4 consumption as a function of CH4 pressure at fixed P$% =7.0 kPa (A) and product of O2 pressure and the inverse rate of CH4 consumption as a function of O2 pressure at fixed P&'(= 25 kPa (B) over HAP- based catalysts. (973 K, 101 kPa total pressure, total flow rate = 46 mL min-1, He as balance gas.) .………………………………………….………………………………………… 75 Figure 3.6. Product of CH4 pressure and the inverse rate of CH4 consumption as a function of O2 pressure at fixed P&')= 25 kPa (A), methane consumption rate as a function of O2 pressure at P&'(= 25 kPa (B) and CH4 pressure at P$% = 7.0 kPa (C), respectively. (973 K, 101 kPa total pressure, total flow rate = 46 mL min-1, He as balance gas.) The curves in (B) and (C) denote the fitting results using optimized rate.…………………………………. 77 Figure 3.7. C2H6 (A), CO (B) and CO2 (C) formation rates as a function of CH4 pressure at P$%= 7.0 kPa over HAP-based catalysts. (973 K, 101 kPa total pressure, total flow rate = 46 mL min-1, He as balance gas.) The curves denote the fitting results using optimized rate constants.……………………………………….…………………………………... 79 Figure 3.8. C2H6 selectivity as a function of CH4 pressure at P$*= 7.0 kPa (A) and as a function of O2 pressure at P&'( =25 kPa (B) over HAP-based catalysts. (973 K, 101 kPa total pressure, total flow rate = 46 mL min-1, He as balance gas.) The curves denote the fitting results using optimized rate constants...………………………………………….. 81 Figure 3.9. CO2 selectivity in oxidation reactions as a function of CH4 pressure at P$*= 7.0 kPa (A) and as a function of O2 pressure at P&'( =25 kPa (B) over HAP-based catalysts. (973 K, 101 kPa total pressure, total flow rate = 46 mL min-1, He as balance gas.) The curves denote the fitting results using optimized rate constants…………………………. 83 Figure 4.1. Illustration of unit-cell structure of hexagonal HAP. (A) side view, (B) top view. Color code: blue-Ca, violet-P, red-O, brown-C, and white-H, respectively……..... 96 xi Figure 4.2. Schematic representation of the HAP structure viewed normal to c-axis (A) and along with c-axis (B) direction, respectively. (C) shows the schematic representation of HAP crystal grown into prism-like crystal along the c-axis direction. (Ca[1] atoms: grey, Ca[2] atoms: green, P atoms: orange; O atoms belonging to PO4 tetrahedra: red; O atoms from hydroxyl groups: cyan.)…………………………………………………………… 97 Figure 4.3. SEM images showing morphologies of HAP-based catalysts prepared by electrochemical deposition of HAP seeds follow by hydrothermal growth: (A) oriented HAP (top view), (B) oriented HAP (side view), (C) oriented Pb-HAP (top view), (D) oriented Pb-HAP (side view), (E) unoriented HAP and (F) unoriented Pb-HAP, respectively……………………………………………………………………………... 99 Figure 4.4. XRD patterns of HAP-based catalysts synthesized by electrochemical deposition of HAP seeds follow by hydrothermal growth……………………………… 101 Figure 4.5. (A) Methane conversion and product selectivity in methane combustion reaction in the absence of any catalyst and in the presence of oriented and unoriented HAP catalysts, respectively. (B) shows areal rate of methane conversion and product selectivity over c-surface and a-surface of HAP catalyst. (Temperature = 973 K and total pressure = 101 kPa, space velocity = 91700 mL gcat-1.hr-1, N2 was used as internal standard) ……. 104 Figure 4.6. Long-term stability test of oriented and unoriented HAP catalysts in methane combustion reaction. (Temperature = 973 K and total pressure = 101 kPa, space velocity = 91700 mL gcat-1.hr-1, N2 was used as internal standard) ………………………………... 105 Figure 4.7. Long-term stability test of oriented and unorietend HAP and Pb-HAP in OCM reactions. (Temperature = 973 K, total pressure = 101 kPa, space velocity = 34300 mL gcat- 1.hr-1, N2 was used as internal standard) ……………………………………………….. 107 Figure 4.8. (A) Methane conversion and product selectivity in OCM reactions over oriented and unorietend HAP and Pb-HAP, respectively. (B) Areal rate of methane conversion and product selectivity in OCM reactions over c-surface and a-surface of HAP and Pb-HAP, respectively. (Temperature = 973 K, total pressure = 101 kPa, space velocity = 34300 mL gcat-1.hr-1, N2 was used as internal standard)………………………………. 109 Figure 4.9. NH3-TPD (A) and CO2-TPD (B) profiles of oriented and unoriented HAP- based catalysts…………………………………………………………………………. 111 Figure 4.10. Methane consumption rate as a function of partial pressure of methane at P$% = 4.0 kPa (A) and partial pressure of oxygen at P&') = 32 kPa (B) over c-surface and a- surface of Pb-HAP, respectively. (Temperature = 973 K, total pressure = 101 kPa, total flow rate = 46 mL min-1, N2 as internal standard and He as balance gas) The curves in (A) and (B) denote the fitting results using kinetic parameters shown in Table 4.3………... 115 Figure 4.11. Product of CH4 pressure and the inverse areal rate of CH4 consumption as a function of inverse O2 partial pressure at fixed P&'( = 32 kPa (A), areal rate of methane consumption rate as a function of CH4 pressure at P$* = 4.0 kPa (B) over c-surface and a- surface of Pb-HAP, respectively. (Temperature = 973 K, total pressure = 101 kPa, total flow rate = 46 mLmin-1, He as balance gas) The curves in (B) denote the fitting results using kinetic parameters shown in Table 4.3…………………………………………... 116 xii Figure 4.12. Optimized structures used for potential energy surface construction: (a) clean surface; (b) surface with OH vacancy; (c) adsorbed O2; (d) CH3*; (e) CH2*; (f) CH*; (g) CO*; and (h) H*.Color scheme: blue, violet, red, purple,, brown, and white represent Ca, P, O (lattice), O (gas phase), C, and H species, respectively…………………………… 118 Figure 4.13. Black solid path represents surface OH vacancy formation and oxidation of adsorbed methyl (CH3*) to CO and CO2, long dashed path represents C2H6 formation via combination of gas phase methyl groups, i.e., CH3(g), short dashed path represents C2H4 formation via combination of gas phase methylene groups, i.e., CH2(g), and dotted path represents CHx (x =1 –3) oxidation pathway on stoichiometric (002) facet. Asterisk ‘*’ indicates adsorbed species……………………………………………………………... 119 Figure 4.14. Vacancy structures, after removal of one OH group, on HAP (211) (in purple) and (112) (in yellow) facets on a-surfaces. Dashed boxes indicate the locations of OH vacancies………………………………………………………………………………. 121 Figure 5.1. (A) Aspen Plus (V10) simulation of a DNMC scale-up reaction without incorporating heat integration. (B) Aspen Plus (V10) simulation of a DNMC scale-up reaction incorporating heat integration based on HEN design and pinch analysis……... 130 Figure 5.2. (A) XRD pattern and (B) N2 adsorption/desorption isotherm of Fe/SiO2 catalyst used for catalytic wall reactor in DNMC reaction……………………………………… 131 Figure 5.3. (A) Methane conversion and product selectivity versus catalyst surface area (or mass) in a fixed-bed reactor with fixed catalyst bed length by quartz balance particles when Fe/SiO2 catalyst quantity was varied; (B) Schematic of DNMC and coke combustion on opposite sides of catalytic wall reactor for autothermal operation of DNMC; (C) Methane conversion and product selectivity in different catalyst/reactor settings: (i) blank quartz reactor, (ii) Fe/SiO2 catalyst packed in quartz reactor, (iii) catalytic wall reactor coated with Fe/SiO2 catalyst and (iv) Fe/SiO2 catalyst packed in catalytic wall reactor; (D) Long-term stability test of DNMC reaction in catalytic wall reactor. (Reaction temperature = 1273 K, total gas flow rate = 20 mL min-1, CH4:N2 = 9:1, 1 atm pressure, Fe concentration in Fe/SiO2 = 0.075wt%)………………………………………………………………... 134 Figure 5.4. Methane conversion and product selectivity of DNMC reaction in non-active quartz tube as a function of total flow rate at 1273 K (CH4:N2 = 9:1, 1 atm pressure). Methane conversion was kept below 2% when there was no active species deposited onto the wall of the reactor…………………………………………………………………... 135 Figure 5.5. (A) Methane conversion; (B) C2+ selectivity; (C) C2+ yields; (D) coke yield, respectively, as a function of reaction temperature and feed gas flow rate. (C2+ selectivity and yield and coke yield are calculated from the carbon-atom basis.)………………….. 136 Figure 5.6. (A) TGA curve of coke formed on Fe/SiO2 powder after TOS = 4 hours of DNMC reaction. (B) TPO profile of catalytic wall reactor after TOS = 4 hours of DNMC reaction (Reaction temperature = 1273 K, total gas flow rate = 20 mL min-1, CH4:N2 = 9:1, 1 atm pressure). The coke formation rate determined from TPO method (7.60 ×10-6 mole min-1) is similar to the coke formation rate determined from weight-difference method (7.81 ×10-6 mole min-1)………………………………………………………………... 138 Figure 5.7. (A) Methane conversion and product selectivity and (B) coke formation rate as a function of time on stream at 1273 K (Total flow rate = 20 mL min-1, CH4:N2 = 9:1, 1 xiii atm pressure). The coke formation rate increased significantly from TOS = 0 hour to TOS = 0.25 hours. After 0.25 hours, the coke formation rate started to decrease until it reached a plateau starting from TOS = 1.0 hour onward………………………………………… 139 Figure 5.8. Energy input for DNMC and energy output by coke combustion at their corresponding methane conversion and coke yield. The dashed line represents the energy input and output balanced from both reactions. The shaded circle indicates the operation window of autothermal DNMC in catalytic wall reactor……………………………….. 142 Figure 5.9. Process flowsheet of autothermal DNMC in millisecond catalytic wall reactor coupling both endothermic DNMC and exothermic coke combustion on opposite sides of the reactor……………………………………………………………………………… 144 Figure 6.1. SEM images showing the morphologies of Fe/SiO2 catalyst (A) and SCZO perovskite (B), respectively; XRD patterns of Fe/SiO2 catalyst and SCZO perovskite (C); and N2 adsorption/desorption isotherm of Fe/SiO2 catalyst and SCZO perovskite used in DNMC reaction (D)……………………………………………………………………. 155 Figure 6.2. CH4 conversion for Fe/SiO2 catalyst and SCZO perovskite (A) and product selectivity of Fe/SiO2 catalyst (B) and SCZO perovskite (C) in long-term stability test of DNMC reaction (temperature = 1273 K, space velocity= 3200 mLg-1h-1)……………... 157 Figure 6.3. CH4 conversion and product selectivity in DNMC over Fe/SiO2 and SCZO mixture with different packing modes under the same reaction conditions (0.1875 g SCZO, 0.1875 g Fe/SiO2, temperature = 1273 K, space velocity= 3200 mLg-1h-1). (B) Product selectivity for SCZO packed on top of Fe/SiO2, (B) Product selectivity for SCZO packed below Fe/SiO2, and (C) Product selectivity for SCZO and Fe/SiO2 well mixed……….. 159 Figure 6.4. CH4 conversion and product selectivity over Fe/SiO2 catalyst (A) and SCZO powder (B) in a fixed-bed reactor at different hydrogen co-feed concentrations (temperature=1273 K, space velocity= 3200 mLg-1h-1, TOS = 1 hour)………………… 161 Figure 6.5. H2-TPD profiles of Fe/SiO2, SCZO and CeO2 (controlled) samples………. 162 Figure 6.6. CH4 conversion and product selectivity in DNMC reaction over Fe/SiO2 catalyst and SCZO powder mixture with different catalyst arrangement (temperature=1273 K, space velocity= 3200 mLg-1h-1, TOS = 1 hour)……………………………………... 163 Figure 6.7. TGA curves of coke formed in spent SCZO/Fe/SiO2 catalysts with different SCZO amounts in a fixed-bed reactor at different hydrogen co-feed concentrations after TOS of 3.5 h in DNMC reactions………………………………………………………. 164 Figure 6.8. Raman spectra of coke formed in spent SCZO/Fe/SiO2 catalysts with different surface area ratio in a fixed-bed reactor at different hydrogen co-feed concentrations after TOS of 3.5 h in DNMC reactions………………………………………………………. 165 Figure 6.9. TPO of spent FeSiO2 catalyst, SCZO sample and Fe/SiO2/SCZO mixture at different hydrogen co-feed concentrations after TOS of 3.5 h in DNMC reactions……. 170 xiv List of Abbreviations and Acronyms OCM – Oxidative Coupling of Methane DNMC – Direct Non-Oxidative Coupling of Methane HAP – Hydroxyapatite XPS – X-Ray Photoelectron Spectroscopy EDS – Energy-Dispersive X-ray Spectroscopy BET – Brunauer–Emmett–Teller ICP-OES – Inductively coupled plasma optical emission spectroscopy XRD – X-Ray Diffraction SEM – Scanning Electron Microscopy FTIR – Fourier Transform Infrared Spectroscopy TPD – Temperature Programmed Desorption TPO – Temperature Programmed Oxidation DFT – Density Functional Theory XAS – X-Ray Absorption Spectrometry XANES – X-Ray Absorption Near Edge Structure EXAFS – Extended X-Ray Absorption Fine Structure TOS – Time-on-Stream SCZO - SrCe0.7Zr0.3O3-𝛿 xv Chapter 1: Introduction 1.1 The abundance of methane The depletion and wildly fluctuating prices in crude oil are shifting the market attention to natural gas that is still of great abundance[1]. Methane, which is the main constituent of natural gas and methane clathrates, is currently being used for power generation and home/industrial heating[2]. However, due to its abundant availability, low price, and low environmental impact, methane is deemed to be an alternative source to replace crude oil in the foreseeable future[3]. It is the raw materials to produce light olefins and benzene, which are the important building blocks for wide range of commodities such as polymers, cosmetics and lubricants. With hydrogen/carbon (H/C) ratio of 4, methane is considered a cleaner source of fossil energy since it produces less carbon dioxide (greenhouse gas) than does the combustion of oil and coal. Even though large reserves of natural gas have been discovered recently and the mature development of hydraulic fracturing technology has made natural gas production possible from impermeable rocks of shale[4], about 1/3 of these proven reserves are situated in remote areas. These stranded natural gas resources, including those offshore, are lacking gas distribution infrastructure, making the transportation of methane over a long distance through pipelines an issue since this process is not economically feasible. Therefore, vast quantities of natural gas are flared because it is relatively cheaper to do it than to capture and use it. In addition, flaring process is also more environmentally friendly compared to letting it escape to air since methane is a much stronger greenhouse gas than CO2[5]. However, methane flaring causes a huge waste in world’s fossil resources. To properly take advantage of this valuable resource, new technologies and chemical processes need to be implemented to convert methane into 1 liquid fuels and chemical feedstocks such as olefin, aromatics and hydrogen. Once liquified fuels and chemical products are formed, they can be transported via the existing crude pipelines and infrastructures, making the process more economically viable. Methane is a very stable compound of tetrahedral structure, with a melting point of 91 K and a boiling point of 108 K. It has rather strong C-H bond, negligibly small electron affinity, large ionization energy, extremely high pKa value and perfectly symmetric zero dipolar and magnetic moments. These thermochemical properties render methane with high activation barriers. Approximately 439 kJ mol-1 of energy is required to cleave the C- H bond in methane homo- and heterolytically. Therefore, elevated temperature or introduction of oxidants is usually crucial to activate methane in the gas phase[6]. Such reaction condition often leads to mostly radical reactions with intrinsic low selectivity[7]. Furthermore, the possible products formed through methane conversion reactions are more reactive than methane since the strength of C–H bond in methane is stronger than in the possible products. Acceptable product selectivity is therefore only achieved at low/moderate conversion levels, which result in extensive recycling of unreacted methane and product separation. Even though catalysts play major role in lowering the activation energy of the chemical processes, extensive fundamental research and commercially viable technology remains elusive. 1.2 Current catalytic methane conversion methods The chemical utilization of natural gas to produce basic chemicals is one of the desirable goals in the current chemical industry. However, the conversion of methane into higher hydrocarbons remains a challenge. Extensive efforts have been put to activate 2 methane more efficiently[8, 9]. Figure 1.1 shows the indirect and direct conversion processes for methane valorization. Generally, methane can be converted to useful chemicals and fuels via three approaches: (i) indirect route involving multiple steps, via synthesis gas (a mixture of CO and H2, known as syngas), (ii) direct oxidative coupling of methane (OCM) to C2 such as ethylene (C2H4) and ethane (C2H6), and (iii) direct non- oxidative methane conversion (DNMC) into C2 (i.e. acetylene (C2H2), ethylene, ethane) and aromatics (e.g., benzene (C6H6) and naphthalene (C10H8)), when combined are referred to as C2+ hydrocarbons. Figure 1.1. Overview of the selected direct and indirect methane conversion pathways. 1.2.1 Indirect methane conversion pathway The well-established Fischer-Tropsch (FT) synthesis is the state-of-the-art technology to convert methane to useful hydrocarbon via syngas intermediates, and has been practiced commercially for decades[10]. Typically, there are three possible reforming processes to convert methane to syngas: (i) steam reforming (Eq. (1.1)), (ii) partial oxidation of methane (Eq. (1.2)) and (iii) carbon dioxide (CO2) reforming (dry reforming) 3 (Eq. (1.3)). Steam reforming gives a very H2 rich synthesis while partial oxidation generates relatively lower H2. CH4 + H2O ↔ CO + 3H2 (∆ 1/H = + 226 kJ/mol, ∆ G1/ = - 71 kJ/mol) (1.1) CH4 + 1/2 O 12 ↔ CO + 2H2 (∆/H = - 44 kJ/mol, ∆ G1/ = - 254 kJ/mol) (1.2) CH4 + CO2 ↔ 2CO + 2H (∆ H12 / = + 261 kJ/mol, ∆/G1= - 73 kJ/mol) (1.3) Production of syngas by steam reforming of methane (1) over Ni-based catalysts to generate H2:CO ratio of 3 is the main process used for methane processing. However, an ideal H2:CO ratio of 2 is often desired in the later Fischer-Tropsch or methanol synthesis. Therefore, the H2:CO ratio of syngas obtained from steam reforming must be adjusted through the reverse shift reaction. In contrast, syngas obtained through direct partial oxidation reaction (Eq. 1.2) over Co, Fe or Ru based catalysts gives H2:CO ratio of 2. However, this reaction usually involves complete oxidation of methane to CO2 and H2O. Furthermore, direct partial oxidation reaction is often associated with safety concern because the mixture of methane an oxygen is extremely dangerous in industrial applications. As for CO2 reforming of methane, the carbon dioxide reforming of methane produces syngas with H2:CO composition of 1:1. The ratio is required for the synthesis of formaldehyde and polycarbonates, as well as hydroformylation process. The selection of technology for syngas production strongly depends on the downstream application. The obtained syngas gas is further processed to produce a wide range of chemicals such as olefins, gasoline, diesel, paraffin, as well as oxygenates. Other alternative way of generating hydrocarbons is first converting syngas to methanol, and then from methanol to chemicals using technologies such as methanol-to-olefins (MTO), methanol-to-hydrocarbon (MTH) and methanol-to-aromatics (MTA). A considerable 4 amount of syngas is also used for ammonia synthesis, a chemical feedstock used to produce nitric acid and chemical fertilizers. Either approach mentioned above requires the utilization of methane at high temperature and high pressure, and therefore is very energy intensive and capital costing due to the use of large reactors-furnaces and technical complexity. In addition, oxygen from CO has to be removed during hydrocarbon production, lowering the carbon atom utilization efficiency. Nonetheless, these processes are still the dominant industrial practices for methane conversion. 1.3 Direct oxidative coupling of methane (OCM) to C2 hydrocarbons (ethylene and ethane) OCM reaction is a methane valorization approach that holds great prospect in converting methane directly into higher hydrocarbons especially ethylene in the presence of an oxidant. Ethylene is known as an important chemical feedstock for the synthesis of most petrochemical products[11]. The exothermic, single-step OCM reaction was first introduced by Keller and Bhasin in 1982[12], followed by Hinsen and Baerns in 1983[13] and Lunsford in 1985[14, 15]. Following these pioneering works, OCM has become one of the most widely studied topics of research in methane activation. OCM process could provide an alternative to traditional ethylene synthesis processes based on petroleum and syngas such as naphtha pyrolysis and ethane dehydrogenation reaction. 2CH4 + O2 → C2H4 + 2H2O (∆/H1= - 281 kJ/mol) (1.4) OCM reaction usually takes place at temperature range of 953 K-1200 K under atmospheric pressure. It is generally accepted that OCM reaction follows both heterogeneous and homogeneous pathways (Figure 1.2). The reaction occurs via four steps: (i) oxygen 5 adsorption on catalyst surface to form surface adsorbed oxygen species, (ii) methane activation on the surface adsorbed oxygen species to form methyl radicals through a C-H bond breaking and a hydrogen abstraction, (iii) homogeneous coupling of two methyl radicals in the gas phase to form ethane, and (iv) oxidative dehydrogenation of ethane to ethylene or ethane activation on the catalyst surface to form C2 radicals and subsequently to ethylene. Figure 1.2. Reaction scheme showing steps involved in OCM reaction. When the OCM reaction temperature reaches beyond 873 K, ethylene and ethane can be produced homogeneously even without catalyst, even though the reaction is generally understood to be a catalytic process. The C2 formation rate in OCM reaction strongly depends on reaction conditions, such as temperature, pressure and CH4/O2 ratio. For example, methane conversion is strongly limited by oxygen availability. Increasing partial pressure of oxygen in the feed stream increases methane conversion, but it also favors total oxidation reaction. Another example is that the OCM process is a molar neutral reaction. Therefore, increasing reaction pressure does not contribute to higher methane conversion at a comparable residence time. In fact, since C2 formation steps compete with total oxidation reaction, increasing pressure increases methane consumption rate toward total oxidation products and decreases C2 yield. Even though increasing reaction 6 temperature facilitates the formation of C2, catalyst stability is another issue that is yet to be overcome. 1.3.1 Challenges in OCM reaction CH4 + 0.25O2 ® 0.5C2H6 + 0.5H2O CH4 + 0.25O2 ® 0.5C2H4 + H2O CH4 + 1.5O2 ® CO + 2H2O CH4 + 2O2 ® CO2 + 2H2O 0 -200 -400 -600 -800 Gibbs free energ y (D G973Kr , kJ/mol) Figure 1.3. Gibbs free energy of reaction (ΔrG973 K) for the formation of ethane, ethylene, carbon monoxide and carbon dioxide, respectively, from methane in OCM. The main challenge of OCM reaction lies on the low C2 selectivity. Even though ethylene is the desired product in this reaction, the presence of oxygen promotes the oxidation of methane to undesired CO, CO2 (COx) and H2O products (Figure 1.3). In addition, the more reactive C2 products also tend to undergo secondary oxidation reaction to form COx and H2O. The unfavorable side reactions lower ethylene selectivity especially at high methane conversion, limiting attainable single-pass C2 yield to 28% under fixed- bed, continuous-feed reaction conditions[16]. In addition, the formation of CO2 also significantly increases the heat of reaction, leading to heat management problem especially for industry application purpose. The nature of oxidizing agent used strongly influences 7 Methane reaction Oxidation OCM the OCM performance. Sulfur (S2)[17] and nitrous oxide (N2O)[18] have been studied as “softer” oxidants in OCM reaction to alleviate complete oxidation of hydrocarbons to COx. Investigation of various catalyst systems with S2 and N2O showed that these oxidants offered superior catalytic performance than O2 in terms of C2 yield. However, major safety concern arises when S2 is used as oxidant since H2S is formed as by-product while N2O is too expensive to be used in industry. 1.3.2 Catalysts for OCM reaction Figure 1.4. Elemental compositions of OCM catalysts with yield (C2) ≥ 25% reported in literature. All the catalysts were tested in a fixed-bed reactor in the cofeed mode under atmospheric pressure at temperatures from 943 to 1223 K, p(CH4)/p(O2) = 1.7 - 9.0, and contact times from 0.2 to 5.5 s. Reproduced with permission from reference [19]. Copyright 2011 Wiley-VCH. Hundreds of catalytic materials have been synthesized and tested for OCM reaction. Typically, a high-performance OCM catalyst should meet two criteria: initiate the formation of CH3 radicals at lower temperatures and suppress undesired total oxidation of methane and hydrocarbon products to COx. Zavyalova et al. summarized the catalysts 8 reported for OCM reaction over 400 publications since 1982 and the results are plotted in Figure 1.4[19]. Generally, the OCM catalyst materials can be categorized into four groups: (i) reducible metal oxides, (ii) nonreducible metal oxides, (iii) halogen-containing oxide materials and (iv) solid electrolytes[20, 21]. Even though some general criteria for designing an active, selective and stable OCM catalyst were proposed, an efficient catalyst that fully meets practical industrial and economic requirements has not been achieved, likely due to the complex OCM reaction network that involves both heterogenous and homogenous pathways. 1.3.2.1 Li/MgO catalysts Li/MgO catalysts was first reported by Ito et al. in 1985[22]. A lot of researches have been carried out since then to understand the active sites of this catalyst. Even though Li/MgO catalyst exhibited ethylene yield of ~12%, the catalyst faces serious deactivation (drop in 50% methane conversion) after time on stream of 5 hours due to the loss of Li in the reaction[23-25]. Myrach et al. reported that Li in Li/MgO started to agglomerate at temperature above 700 K but readily desorbed from the MgO at temperature above 1050 K[26]. The loss of Li ions in turn deactivated the catalyst and led to low methane conversion and poor C2 selectivity in OCM. Later, Huang et al. further explored the distribution and role of Li in Li-doped MgO catalysts for OCM[27]. Significant loss of Li was observed in the Li/MgO catalysts during the preparation, activation and reaction steps due to the migration of Li from the bulk to the surface and the subsequent desorption. However, such migration and desorption processes generated MgO with rough structures and exposed MgO (110) and (111) facets that are active in the OCM. Therefore, large initial Li concentration in Li/MgO catalyst in the synthesis procedures was 9 required to prevent the complete loss of Li ions but at the same time facilitate the migration process to form (110) and (111) facets. Considerable amount of research efforts has been conducted to determine the exact structure of the Li/MgO catalyst as well as its correlation to OCM activity. Early findings by Lunsford and group[14, 15] and Peng and colleagues[28] suggested that the catalytically active center of Li/MgO was Li+O-. Their findings were supported by different surface characterization tools such as X-ray photoelectron (XPS) and electron paramagnetic resonance (EPR)[14, 15, 28]. However, Kwapien et al. disproved their theories using density functional theory (DFT) calculations[29]. The calculations did not reveal any sign of Li+O- formation. Instead, the active sites of Li/MgO catalyst originated from oxygen- centered radicals on metal oxides, i.e. Mg2+O2-. Other than Li doped MgO catalyst, other types of dopant on MgO have also been studied to improve the activity and stability of the catalyst. For example, Schwach et al.[30] created highly active sites by co-doping the MgO catalyst with iron and gold (Au-Fe-MgO) in ppm quantities. The OCM catalytic activity of the Au-Fe-MgO catalyst showed exceptional OCM stability and enhance C2 selectivity. The lasting activity of the catalyst was attributed to the presence of terrace sites. The co-doping facilitated the formation of terrace sites but decreased the smaller steps found on the catalyst surface. In addition, the electronic doping by Fe also increased the C2 formation rate by over an order of magnitude. Li/MgO has also been studied at lower reaction temperature range. Roos et al. showed that when 5.3 wt% of Li was doped into the MgO catalyst to form Li20/MgO(II), C2 yield of 12.4 % was achieved at 803 K[31]. 10 1.3.2.2 Mn/Na2WO4/SiO2 catalysts Mn/Na2WO4/SiO2 catalyst was first reported by Fang et al. in 1992[32]. OCM catalytic tests based on this catalyst usually requires operating temperature ≥ 1073 K. The OCM catalytic performance over this catalyst showed that 23.9% of C2 yield was achieved at 1073 K in a packed-bed reactor, higher than the reported Li/MgO catalyst. In addition, this catalyst also exhibited better stability compared to Li/MgO catalyst. It is generally believed that the synergic effects between the metals (Mn, Na and W) and the SiO2 support lead to superior OCM catalytic performance. The thorough understanding on the structural and chemical properties of the complex Mn/Na2WO4/SiO2 catalyst, however, remains a challenge. Past researches have shown that parameters including metal loadings and catalyst preparation methods play critical roles in tuning catalytic performance[33-37]. The catalyst synthesis procedure usually produces Mn/Na2WO4/SiO2 catalyst with mixed bulk oxide composition that contains MnO2, Mn2O3, MnWO4, Na2WO4 and SiO2. During OCM reaction, the catalyst undergoes further chemical and physical changes. For example, amorphous SiO2 which promotes total oxidation to COx tend to undergo phase transformation to inert α-cristobalite phase during OCM reaction, suppressing over oxidation reaction. Two theories have been proposed to accommodate the active center of Mn/Na2WO4/SiO2 catalyst. Wu and group suggested the surface clusters including W=O and W-O-Si species to be the active sites of the Mn/Na2WO4/SiO2 catalyst in OCM reaction Catalyst surface reconstruction was observed through XRD and Raman spectroscopy analysis[38, 39]. They also suggested that bulk metal oxides were physically unstable at OCM reaction conditions as their melting points were < 973 K. On the other hand, Lunsford et al. regarded the Na-O-Mn species to be the active sites of this catalyst. Mn was proposed 11 to activate oxygen in the gas phase, Na was suggested to prevent total oxidation of methane, and W was expected to enhance the catalyst stability[40]. The theory was later proved by Xiao and group using several characterization tools such as XRD diffraction, XPS and Raman spectroscopy[41, 42]. Several methods have been attempted to improve the OCM activity and C2 yield over Mn/Na2WO4/SiO2 catalyst. For example, Mahmodi et al. studied the effect of promoter in OCM over synthesized Mn/SiO2 nanocatalysts[43]. The 12-92 nm M-Na- Mn/SiO2 nanaocatalysts (M = W, Mo, Nb, V, Cr) were synthesized using incipient wetness impregnation method. The OCM catalytic performance results showed that a maximum C2 selectivity of 31.6% and methane conversion of 12.6 % were achieved at 1 atm and 1048 K in a packed-bed reactor over these nanocatalysts. However, the catalysts started to deactivate after 9-hour time-on-stream due to the loss of catalytic surface area by sintering. Zarrinpashne patented a Nb doped Mn/Na2WO4/SiO2 catalyst that significantly increased C2 yield to 27.24% at 1123 K in a packed-bed reactor in 2011[44]. The catalyst which contained 1.9 wt% Nb, 2 wt% Mn 1.8 wt% Na, 2.8 wt% W with a SiO2 support was prepared by either co-precipitation or impregnation method. The C2 yield improvement was ascribed to the promotional effect of Nb after its substitution into the Mn/Na2WO4/SiO2 catalyst structure. In 2013, Ha et al. in his patent, on the other hand, reported a Mn/Na2WO4/SiO2 catalyst with 3.5 wt% NaW, 1.3 wt% MnO2 and SiO2 support that showed 18.4% C2 yield at 1073 K[45]. The catalyst also did not show any deactivation even after 40-hour time-on-stream. In 2015, Ikeda et al. studied the promoting effect of molten alkali chloride such as LiCl, NaCl, KCl and CsC on Mn/Na2WO4/SiO2 catalyst to improve OCM activity. Ethylene yield was enhanced to 31% at 1023 K when equimolar 12 mixture of NaCl and KCl was doped to the catalyst[34]. More recently, Lu and group reported a TiO2-doped Mn2O3/Na2WO4/SiO2 catalyst by using Ti-MWW zeolites as TiO2 dopant and SiO2 support[46]. Methane conversion of 26% and C2-C3 selectivity of 76% were reported at 993 K. MnTiO3 was formed in the synthesis process and the redox cycle between Mn2+ and Mn3+ at low temperature activated the O2 while working synergistically with Na2WO4 to convert methane to hydrocarbons. 1.3.2.3 La2O3 catalysts La2O3 is another catalyst commonly studied for OCM due to its potential viability in industry application. Alkaline-earth metals (Sr, Ca, Mg) are usually doped into La2O3 to further improve the catalytic performance since doping increases the basicity of the catalysts. Among all the types of alkaline-earth metals studied, Sr/La2O3 showed highest OCM activity. C2 yield as high as 17% was reported at 1073 K in a continuous-flow packed reactor[47]. As for CaO doped La2O3 catalyst (Ca/La2O3), when the La2O3/CaO ratio was optimized to 60%-80%, 9% of C2 yield was achieved at 1013 K[48]. In addition to alkaline-earth metals doped La2O3 catalyst, Fe-doped La2O3 catalyst has also been studied in OCM reaction[49]. The La/Fe ratio of the Fe-doped La2O3 catalyst determined the OCM activity. For example, when the La/Fe ratio was tuned to > 5, Fe-doped La2O3 catalyst showed higher OCM activity than undoped La2O3 catalyst. The effects of Bi compounds as well as Na2O and Na4P2O7 promoters addition on Fe-doped La2O3 catalyst were also studied and the results showed that C2 selectivity as high as 83% was achieved. Different catalyst preparation methods have been studied to improve the OCM catalytic performance over La2O3 catalyst. For example, Rane et al. examined the effects of synthesis methods, including physical mixing of La2O3 and CaO and co-preparation, as 13 well as source of precursors such as acetates, carbonates, nitrates ad hydroxides on OCM activity[50]. The results showed that C2 yield of 12% was achieved when the catalysts were prepared using co-precipitation method with carbonates as source precursors. However, Ca/La2O3 exhibited poor catalyst stability especially in the presence of CO2 at temperatures above 923 K since La2O3 tends to react with CO2 to form dioxymonocarbonate structure (La2O2CO3)[51, 52]. Dedov et al. developed a series of catalysts containing light group of rare earth oxidxe (REO) consisting of lanthanum, praseodymium, neodymium and cerium oxides with 5.5 wt.% of ceria and studied their OCM catalytic performance[53]. The group found that addition of 10 wt% of ceria to lanthana (La2O3-CeO2) significantly enhance the efficiency of the catalyst. The difference between La2O3 catalyst and Li/MgO and Mn/Na2WO4/SiO2-based catalysts is that La2O3 catalyst tends to produce more C2H6 than C2H4 product. However, the C2H6/C2H4 ratio can be tuned by doping different types of metal or support structure. For example, Baiya et al. showed that by switching La2O3 support to Al2O3 support and by varying Sr contents in Al2O3, OCM performance comparable to Mn/Na2WO4/SiO2-based catalyst (i.e. 18% overall C2 yield) was achieved[54]. The catalyst stability was also improved, and no deactivation was observed for up to 48-hour time-on-stream. Similar to Li/MgO catalyst, La2O3-based catalysts have also been studied at lower temperature range when the catalysts were made to different particle sizes and morphologies. Following the work by Dedov et al. in 2003, Noon et al. fabricated La2O3- CeO2 nanofibers (as opposed to La2O3-CeO2 powder) and demonstrated that methane in OCM reaction over this catalyst could be activated at temperature as low as 743 K[55]. A maximum C2 yield of 18% was reported at 793 K in a packed-bed reactor. Huang et al., on 14 the other hand, studied the effects of nano-structured La2O3 (no doping) catalysts under OCM condition[56]. The OCM reaction was performed in a packed-bed reactor at temperature range of 673 K to 1073 K. La2O3 nanorods was able to initiate the reaction at temperature as low as 673 K. They also reported that La2O3 nanorods showed higher activity and C2 selectivity towards OCM than La2O3 nanoparticles. 10% C2 yield was achieved at 723 K. The better OCM performance at low temperature over La2O3 nanorods originated from the higher surface area, strong surface basic sites, electron deficient surface oxygen species and defined surface structure of this catalyst compared to La2O3 nanoparticles. In 2015, Hou et al. studied the effects of morphologies, particles sizes and calcination temperatures on La2O2CO3 catalysts in OCM reaction[57]. La2O2CO3 catalysts with rod- and plate-shapes at nanometer scale were synthesized. The OCM catalytic performance showed that the rod-shaped La2O2CO3 catalysts exhibited higher C2 yield than plate-shaped La2O2CO3 catalysts at a temperature range of 693 K to 773 K. Furthermore, the lanthanum oxycarbonate phase also rendered this catalyst with 20 times higher specific activity than any of the other rod-shaped samples. In the same year, Song et al. also reported monodisperse Sr–La2O3 hybrid nanofibers for OCM reaction[58]. They demonstrated that such nanofiber catalysts significantly improved methane conversion and C2 selectivity in OCM, compared to conventional Sr doped La2O3 spherical catalyst. The catalyst was also able to catalyze the reaction at temperature as low as 773 K with methane conversion of ~35% and C2 selectivity of ~47 %. The enhanced OCM activity over Sr/La2O3 hybrid nanofibers could be attributed to the synergistic effects from each component’s functionality in the catalyst. 1.3.2.4 Commercialized catalysts OCM reaction has been commercialized in pilot scale by Siluria Technologies to produce ethylene. The company has filed several patents 15 for a series of nanowire catalysts that can operate at low temperature. In general, there are two categories of catalyst that were reported. The first group being the nanowire lanthanum oxide catalysts composed of two different lanthanide-group metals (Y-La, Zr-La, Pr-La and Ce-La)[59] while the second group being the nanowire catalysts composed of 20 wt% Mg, 5wt% Na and 75 wt% La2O3[60]. C2 yield > 22.5% was reported for the nanowire lanthanum oxide catalysts while C2 yield of 12 % was reported for the second group of catalysts. These catalysts not only showed excellent OCM activity but also exhibited superior long-term stability. The low operating temperature of these catalysts also significantly reduces the overall operating costs, making the process even more viable. 1.3.2.5 Hydroxyapatite (HAP) In addition to the metal oxide-based catalysts mentioned above, HAP is another type of catalyst studied for OCM reaction. Hydroxyapatite (Ca10(PO4)6(OH)2) is a type of calcium phosphate crystal, more commonly studied in biomaterials since it is the major component in bone and teeth[61, 62]. Details on HAP will be discussed in Chapter 2. Lead cation (Pb2+) substituted HAP (Pb-HAP) has been studied as a catalyst for OCM reactions. It is shown that Pb-HAP was able to increase C2 selectivity by a factor of ~ 5 compared to bare HAP[63-65]. The enhancement in OCM performance of Pb-HAP was ascribed to the stabilization of methyl radicals by Pb2+ sites for pairwise reactions against oxidation reactions[66]. 1.3.3 OCM reaction kinetics OCM reaction follows both heterogeneous and homogeneous pathways and these reaction mechanisms have been studied for different catalysts. Selectivities and yields of C2 products in OCM reactions depend on the identity and dynamics of specific elementary 16 steps involved in the primary and secondary reaction steps on the catalysts. Takanabe and Iglesia proposed a reaction pathway for methane activation over Mn/Na2WO4/SiO2 catalyst that involved OH-mediated C-H bond activation[67]. They suggested that OH radicals were generated either via reaction between H2O and surface adsorbed oxygen species or via reaction between H2O and gas-phase O2, even though the former pathway was more favorable. Methane in the gas phase then reacted with OH radicals subsequently to form methyl radicals. The more reactive nature of the OH radical led to a weaker influence of the C-H bond energies on the relative rates of H abstraction from methane, ethane and ethylene. The proposed mechanism was supported by a series of experiments, including the effects of contact time and H2O addition/removal on methane conversion and C2 yield. The results showed that methane conversion and C2 yield were significantly enhanced if 4% H2O was added into the OCM reaction at 1073 K. Lunsford and colleagues also studied OCM reaction kinetics for a variety of catalysts, including Li/MgO[68], lanthanide oxide[69, 70] and Sr/La2O3[71]. The group were able to identify the active sites of these catalysts and deduced methyl-radical (∙CH3) formation pathways through matrix isolation electron spin resonance (MIESR). In the proposed mechanism, O2 in the gas-phase first adsorbed onto the catalyst surface. Methane could either react with the surface adsorbed oxygen species to form ∙CH3 radicals or react with gas-phase O2 to form HO2 and ∙CH3 radicals. Next, ∙CH3 radicals either coupled with each other to form C2H6 or reacted with HO2 in the gas phase to form CO, CO2 and H2O. Further C2H6 dehydrogenation produced C2H4. However, Sun et al. proposed different mechanism than Lunsford and group for COx formation over Li/MgO, Sn/Li/MgO and Mn/Na2WO4/SiO2 catalysts [72]. Their results showed that COx and H2O were formed via 17 the combination of homogeneous and heterogeneous pathways. The proposed mechanism followed Eley-Rideal pathway which involved adsorption of gas-phase species, reaction of gas-phase species with surface adsorbate via radicals or molecules and surface-surface interactions. Reyes et al. also developed an OCM reaction mechanism involving both homogenous and heterogeneous pathways[73]. They proposed that methane could react with O2 in the gas phase as CH4 + O2 à ∙CH3 + HO2 (1.5) or on the surface as CH4 + O(s) à ∙CH3 + OH(s) (1.6) Stansch et al. developed a comprehensive 10-step kinetic model of the OCM reaction over a La2O3/CaO catalyst through kinetic measurements in a microcatalytic fixed-bed reactor[74]. The kinetic parameters were validated in the temperature range of 973 K≤ T≤ 1228 K. The reaction mechanism suggested that methane was converted to C2H6, CO2 and CO simultaneously. The consecutive pathways include oxidative and dehydrogenation of C2H6, oxidation and steam reforming of ethylene, oxidation of CO2 and water gas shift reaction. In the subsequent mechanism developed by Daneshpayeh at al., kinetic parameters presented in Stansch’s model were used to represent the OCM reaction network for Mn/Na2WO4/SiO2 catalyst[75]. Daneshpayeh at al. confirmed that the reaction network by Stansch et al. had the best accuracy for Mn/Na2WO4/SiO2 compare to other models. 1.4 Direct non-oxidative methane conversion (DNMC) to C2+ hydrocarbons 18 In comparison with OCM approach, DNMC is a more selective way of converting methane given its unique capability in forming C2 (acetylene, ethylene and ethane) and aromatics (when combined are referred to as C2+ hydrocarbons) as well as hydrogen while circumventing the intermediate energy intensive steps. The DNMC has been studied under both catalytic and non-catalytic conditions[76, 77] . The non-catalytic pyrolysis usually takes place at temperature above 1400 K, with acetylene as the dominant product and is accompanied by high coke formation[78-82]. However, this homogeneous reaction is often limited by the slow methane initiation and chain transfer steps. Therefore, in the past few decades, catalysts have been introduced to DNMC process to overcome the kinetic barriers. The catalytic DNMC process was first discovered by Wang et al. in 1993 using bifunctional molybdenum loaded Zeolite Socony Mobil-5 (Mo/ZSM-5) catalyst[83]. The DNMC reaction often takes place at temperature < 950 K and 1 atm with benzene as the major product since the pore size of ZSM-5 zeolite approximates that of the benzene molecules (0.59 nm). Benzene is an important chemical feedstock in petrochemical industry because it is the precursor for styrene, which can be used to make polystyrene and plastics. Nearly 80% benzene selectivity has been reported at near-equilibrium conversion on Mo/ZSM5 catalysts. The mechanism of methane activation over Mo/ZSM-5 has been an ongoing debate[1, 84, 85]. The generally accepted reaction mechanism of catalytic DNMC involves a bifunctional mechanism with participation of both Mo sites and Brønsted acid sites. In the beginning of the DNMC reaction, induction period takes place during which molybdenum is gradually transforming into molybdenum carbide or oxycarbide species from Mo6+ species upon reduction by methane. The transition between the oxide and the carbide phases can be expressed notionally as: 19 MoO3 → MoOxC → Mo2C (1.7) The C-H bond in methane is believed to be activated on the Mo2C site to form ethylene and hydrogen. The Brønsted acid sites on the zeolite are responsible for the oligomerization of ethylene to benzene. Mo/zeolite catalyst undergoes substantial deactivation easily during on-stream operation. Figure 1.5 shows the coke formation mechanism in DNMC reaction. Coke is formed via two pathways: (i) dehydrogenation of C2Hy intermediates in series with aromatics formation step and (ii) dehydrogenation of methane (CHx) in parallel with C-C bond formation step. Further oligomerization of aromatic products could lead to the deposition of heavy polyaromatics-type carbon deposits (hard coke) on the Brønsted acid sites whereas continuous, deep dehydrogenation of CHx and aromatic compounds could yield the formation of amorphous coke (soft coke). In addition, activated methane species could re-adsorb on the catalyst surface and anchor on the active Mo-C site to form coke. The coke formed on the Brønsted acid sites is irreversible and is responsible for the deactivation of Mo/ZSM-5 catalyst whereas the molybdenum-associated coke is reversible. Furthermore, the concentration of the carbonaceous deposits on zeolites grows with time- on-stream and reaction temperature. Figure 1.5. Products and coke formation mechanism in DNMC. 1.4.1 Challenges in DNMC reaction 20 Figure 1.6 shows the equilibrium conversion of methane upgrading to ethylene or benzene via the DNMC process. Clearly, the DNMC process faces several significant challenges, including (i) thermodynamically limited conversion, (ii) catalyst deactivation due to carbon deposits and (iii) high temperature endothermicity. For example, methane conversion is limited to ~12% at 973 K and 1 atm (Figure 1.6(A)). The formation of carbon in DNMC is also much more favorable than the production of ethylene or benzene, as evidenced by the ΔrG differences of these three products (Figure 1.6(B)). However, the formation of carbonaceous deposits is detrimental to the catalysts because the active sites on the catalysts will be blocked and this will eventually lead to catalyst deactivation. The product selectivity and yield to desirable C2+ hydrocarbons will also be affected by coke formation. (A) (B) 100 100 CH4 --> C + 2H2 80 80 60 60 CH4 = (1/6)C6H6 + (3/2)H2 40 DrG C2H4 20 40 0 DrG C6H6 20 -20 CH = (1/2)C H + H -40 DrG C4 2 4 2 0 -60 500 750 1000 1250 1500 500 750 1000 1250 1500 Temperature (K) Temper ature (K) Figure 1.6. Equilibrium conversion of methane for DNMC. (calculated using HSC Chemistry 6.0 Software) The high temperature endothermicity nature of DNMC suggests that higher reaction temperature is required to activate the reaction. However, for practical industrial application, such high temperature indicates that high energy input and robust reactor infrastructure are needed, making the process not economically viable. Therefore, methane conversion at mild temperature would be ideal to mitigate issue with high capital cost[86]. However, at mild temperature condition (typically < 873 K), the equilibrium methane 21 CH4 conversion (%) DrG (kJ mol -1) conversion into C2+ hydrocarbons is kept quite low, as indicated in Figure 1.6. Therefore, designing new catalysts and improved reactor technology for improved DNMC performance are necessary. 1.4.2 Catalysts for DNMC reaction 1.4.2.1 Metal/zeolite-based catalysts Zeolites are crystalline aluminosilicates of various structure types, pore connectivity, and framework compositions with typical pore sizes less than 2 nm[87-89]. They are widely used in adsorption, separation, and catalysis in chemical and petrochemical industry[90-94]. A great variety of zeolite have been investigated for DNMC reaction. ZSM-5 and Mobil Composition of Matter No.22 (MCM- 22) zeolites are by far the most widely studied catalyst for this reaction since the pioneering work by Wang et al. in 1993[83]. ZSM-5 catalyst belongs to the pentasil family of zeolite and is made up of two types of pore system, mainly the straight channel with an elliptical cross section along b-axis and the zig-zag channel consisting circular pores and straight pores along a- and c-directions. MCM-22 has larger pore sizes than ZSM-5 catalyst. It consists of two independent pore systems in which the first system consists of a two- dimensional 10 MR pore channel (0.41 nm x 0.51 nm) while the second one comprises of a 12 MR supercage (0.71 nm x 0.71 nm x 0.91 nm). Different metal ions can be dispersed into the zeolite matrix to render the catalyst with higher DNMC activity. Even though a number of metal ions dispersed on ZSM-5 zeolite to introduce bifunctionalities to the catalysts have been tested for the DNMC reaction (including Cr, Fe, Mo, W, Re, Zn, Cu, Ga, etc.), Mo remains the most active and selective catalyst for DNMC reaction, and this is followed by Zn and W. A second metal can be introduced as additives to the metal/zeolite 22 to enhance the catalyst performance and stability in DNMC reaction. For example, the effects of noble metals (Pt, Pd, Rh, Ru, Ir and Re) and transition metals (Fe, Ni, Zn, Co, Cr, W) addition have been evaluated. Pt- and Re-doped Mo/zeolite catalysts especially showed increased catalytic activity and stability. The zeolite synthesis method can significantly influence the activity and selectivity in DNMC reaction. Different synthesis methods could affect the stabilization forms, location and dispersity of metal ions in zeolite matrix. Synthesis methods such as impregnation, solid-phase synthesis and sublimation are used to incorporate Mo metal into zeolite catalysts. Among all the mentioned methods, impregnation is the most common way to introduce Mo ions into zeolite structures. Typically, ammonium heptamolybdate (NH4)6Mo7O24.4H2O solution is used as the precursor in the impregnation method. The concentration of the ammonium heptamolybdate solutions can be varied to obtain Mo/zeolite catalyst with different Mo loading, usually in the range of 1 to 20%. The metal impregnation on zeolite is carried out at either room temperature or at higher temperature (e.g. 358 K) while the impregnation time varies from 30 min to 24 hours. After the impregnation process, the zeolite will be dried at 363 K-393 K for a few hours before undergoing subsequent calcination treatment in the presence of air at temperature range of 823 K-993 K. As for the solid phase synthesis, molybdenum salt (MoCl3) or molybdenum oxide (MoO3) are generally used. The solid precursor is mixed physically with zeolite using mortar and pestle before heating the mixture at high temperature (823 K). The synthesis procedures for Mo/ZSM-5 and Mo/MCM-22 catalysts are essentially the same. In addition to different synthesis methods and impregnated metal species, modification of the structural feature of zeolite supports as well as zeolite acidity have also 23 been practiced to provide the catalyst with enhanced catalytic activity, selectivity and stability. For example, meso-/microporous multilamellar MFI zeolite impregnated with Mo ions was synthesized to systematically investigate the textural and acidity properties of the catalysts[95]. The DNMC catalytic results demonstrated that Mo loaded multilamellar supported zeolite catalysts showed higher methane conversion and aromatic production. These superior performances could be attributed to the presence of mesoporosity in the zeolite support that aids in the reactant accessibility to active sites and faster diffusion for the products. In another example, a novel Mo/ZSM-5 catalyst with capsule structure was developed and evaluated. The capsule-like zeolite catalyst was prepared through hard templating method using activated carbon[96]. Benzene production as well as catalyst stability were significantly enhanced since the hollow structure of the capsule-like zeolite facilitated the mass transfer of the zeolite. The catalyst deactivation due to carbon deposition is an inevitable process in DNMC reaction. Therefore, several strategies have been undertaken to regulate coke formation. The addition of co-feed such as CO or CO2, O2, H2, and NO have been practiced to slow down the deactivation process. For example, the addition of O2 to the methane feed can help prolong the catalyst lifetime since coke formed on the catalyst surface will react with O2 to produce CO and CO2 products. However, oxygen not only removes the deposited carbon but also tend to oxidize the active Mo2C species to form less active or inactive MoOx. Furthermore, the exothermicity nature of coke oxidation reaction could lead to catalyst degradation associated with local hot spots. Substituting O2 with NO can help to mitigate the local hot spot issue because NO reacts more vigorously with coke deposits than oxygen. Therefore, NO or a mixture of O2-NO can reduce coke oxidation 24 temperatures to the range of 603 K ≤ T ≤ 723 K. Low level of CO2 in the co-feed plays a beneficial role to DNMC reaction too. CO2 will react with surface carbon to clean the catalyst surface and produce CO. However, a study by Liu et al. showed that co-feeding of only 1.6% CO2 in the DNMC reaction increased methane conversion but decreased benzene formation rate[97]. Increasing CO2 feed to above 10% further suppressed aromatic yield in the reaction but produced higher amount of CO and H2. Higher CO2 concentration leads to the more thermodynamically favorable methane dry reforming pathway over the desired DNMC pathway. A recent approach to deal with coke formation issue in DNMC reaction over Mo/ZSM-5 catalyst relies on periodic pulsing of oxygen into methane feed[98]. This study allowed coke formed during the reaction to be burnt off and achieved significant stabilization of DNMC at 973 K. In terms of benzene yield, two times higher of yield was achieved compare to reaction without any oxygen pulsing. In addition, only CO and H2 (syngas), but no CO2 were detected as side products in the reaction. CO2 that was supposed to form from oxygen pulsing reacted further with coke (C+CO2 = 2CO) and slowed down deactivation. 1.4.2.2 Iron/silica catalyst The iron/silica (Fe/SiO2, lattice-confined single Fe site embedded in SiO2) is a recently reported catalyst by Bao’s group[99] for DNMC reaction. The catalyst exhibited exemplary conversion, product selectivity and stability. Methane conversion up to 48% at 1363 K and product selectivity as high as > 99%, with negligible coke formation, were achieved during the reaction. No obvious deactivation was observed throughout the course of 60 hours, with ethylene, benzene and naphthalene being the major products. The outstanding catalytic performance of this reaction system originates from the 25 Figure 1.7. A) STEM-HAADF image of spent Fe/SiO2 catalys, (B) in-situ XANES of the Fe/SiO2 catalyst upon activation and (C) Fourier transformed (FT) k3-weighted χ(k)- function of the EXAFS spectra for Fe/SiO2 catalyst. (Reproduced with permission from reference [99]. Copyright 2014 the American Association for the Advancement of Science) isolation of monoatomic iron sites in silica matrix. The single-atom nature is also important in circumventing carbonaceous deposition. Characterization technique such as transmission electron microscopy (TEM) has shown that the iron oxide nanoparticles in the size ranges from 3 nm to 4 nm distributed homogeneously throughout the fresh SiO2 matrix in the fresh Fe/SiO2 catalyst. Figure 1.7 shows the embedment of iron oxide species within the silica matrix through bonding of C and Si atoms. X-ray absorption near edge structure (XANES) and extended x-ray absorption fine structure (EXAFS) data also verify that the iron species is confined in silica lattices in which they are coordinated to a Si atom and two C atoms (Figures 1.7(B) and 1.7(C)). The reaction mechanism over Fe/SiO2 catalyst has been proposed to start with methane activation on the Fe sites to generate ×CH3 and ×H radicals. These radicals subsequently release from the catalyst surface and undergo a series of gas-phase reactions to form C2 and aromatics products. 26 1.4.2.3 Gallium nitride catalyst Ga species has been shown to have remarkable ability in activating C-H bond in light alkane[100-102]. The application of commercially available GaN for DMNC, however, was first reported by Li et al. in 2014 under the temperature range of 673 K-823 K[103]. The GaN catalyst was made up of wurtzite crystal structure (Figure 1.8). The catalyst not only exhibited excellent stability during the thermal- Figure 1.8. Schematic diagram for methane C−H bond polarization on the surface of the GaN m-plane. Adapted with permission from reference [104]. Copyright 2014 American Chemical Society. driven reaction of methane conversion to benzene, but also during light alkanes including propane, n-hexane and cyclohexane conversions to benzene. Benzene selectivity as high as 90% was reported. The catalyst also demonstrated good recyclability as no significant deactivation was observed when recycled catalyst was tested. The superior product selectivity could be attributed to the exposed m-plane of the wurtzite structure. The surface lattice of GaN m-plane consisted of alternating Ga3+ and N3- ions that generated strong local electrostatic polarization along the c-direction. This electrostatic polarization in turn stretched and weakened the surface adsorbed methane. GaN catalyst has also been studied under photocatalytic condition for DNMC[104]. In the photocatalytic reaction, the temperature was kept low at 278 K under ultra-violet (UV) light irradiation with a 27 wavelength of 360 nm. Even though the catalytic performance results showed that methane conversion was limited to < 1%, benzene selectivity as high as 96.5 % was achieved. 1.4.2.4 Nickel on ceria-zirconia oxide (Ni/CZ) Okolie et al. (2018) recently reported DNMC reaction to ethane, ethylene, aromatics and hydrogen over cheap and efficient nickel on ceria-zirconia oxide (Ni/CZ) catalyst at low temperatures[105]. The catalyst was able to catalyze the reaction at temperature range of 623 K to 773 K, with nickel being the active site responsible for methane activation. The activated methane species subsequently undergo a series of coupling reactions to form different products. Two types of Ni were formed on the ceria-zirconia support, as evidenced by XANES data. The larger Ni particles promoted both aromatics and coke formations while the smaller Ni particles aided in C2 formation. In addition, the small nickel sites also helped in preventing aromatics assembly, thus maintaining catalyst stability during DNMC reaction. 1.4.2.5 Silica-supported tantalum hydride ((≡SiO)2Ta-H) Soulivong et al.[106] reported the selective catalytic coupling of CH4 into C2H6 and H2 over silica-supported tantalum hydride (≡SiO)2Ta-H catalyst at temperature range of 523 K to 748 K and at PCH4 of 5 MPa in a flow reactor. The motivation behind this work originated from the well- known capability of (≡SiO)2Ta-H catalyst in hydrogenolysis of C2H6 into CH4[107]. The catalytic performance results showed that the thermodynamic equilibrium conversion was reached with ethane selectivity of > 98%. Olefins and trace amount of C3H8 were also produced at 748 K. During the reaction, tantalum-methy-methylidene intermediate was formed. The methyl ligand then moved onto the tantalum-methylidene to form tantalum- ethyl, similar to elementary steps proposed in FT-synthesis. 28 1.4.2.6 Cu/Zn/Al2O3 Other than thermal and photocatalytic methane conversion, Górska et al.[108] also reported the DNMC reaction over Cu/Zn/Al2O3 catalyst in plasma systems under dielectric-barrier discharge conditions. The reactions were performed at the frequency of about 6 kHz at 513 K, and at the pressure of 120 kPa under plasma only and plasma-catalytic condition. A wide range of hydrocarbons from C2 to C5 were detected in the experiment. C2 selectivity was improved by approximately 20% with the use of catalyst in the discharge zone even though overall methane conversion decreased. The long-term stability test was also performed for over 80 hours in the plasma-catalytic system. Methane conversion of > 20% was observed in the first 16-hour, and then leveled off to < 20% methane conversion after 60-hour time-on-stream. 1.4.3 Membrane reactor for DNMC reaction The reaction stoichiometry (CH4 = 3/52 C6H6 + 5/104 C10H8 + 7/104 C2H4 + 2/104 C2H2 +19/13 H2) in DNMC over Fe/SiO2 catalyst[99] suggests the formation of H2 in large quantities: 7.6 moles of H2 per mole of C2+ hydrocarbon. The DNMC pathway also favors low pressure and requires high temperature. Based on the Le Chatelier’s principle, the removal of the large quantity of hydrogen produced in the reaction stoichiometry of DNMC reaction can shift the thermodynamic equilibrium to higher methane conversion. This can be achieved by utilizing a membrane reactor that allows in-situ hydrogen removal, thus leading to an intensive research effort on determining a hydrogen permeable material that is thermally and chemically stable at high temperature (~1273 K), and ionically and electronically conductive[109, 110]. Metal-based membranes such as Pd and Pd alloy- based membranes have been used in DNMC reactions over Mo/ZSM-5 catalysts. For example, Larachi and group have developed a Pd-Ag/porous stainless steel membranes for 29 direct methane aromatization reaction over Ru-Mo/HZSM-5 at temperatures up to 973 K[111]. Their catalytic performance tests demonstrated that the Pd alloy-based membrane improve hydrogen permeation and result in a significant increase in methane conversion at 973 K. On the other hand, Morreale and co-workers have fabricated Pd membranes packed with Mo/HZSM-5 catalysts for non-oxidative methane aromatization reaction[112]. A significant improvement in methane conversion and total aromatic yield were achieved through in-situ H2 removal by the membrane. Nevertheless, dense mixed ionic-electronic conducting (MIEC) ceramic membranes stand out to be the best candidates for DNMC reaction compared to metal-based membranes, given the reaction temperature of DNMC. In addition, MIEC ceramic membranes are also relatively stable thermally and chemically at temperature above 873 K. Perovskites-based membranes, have been used in DNMC reactions over Mo/ZSM-5 catalysts. For instance, Iglesia and group have manufactured SrCe9.;99.5 purity) was purchased from Alfa- Aesar. 33 2.2.2 HAP-based catalysts preparation. The synthesis of bare HAP was carried out by first preparing a solution of (NH4)2HPO4 (0.25 M, 480 mL) in one flask and a solution of Ca(NO3)2 (0.37 M, 320 mL) in a second flask. After pH of each solution was raised to ~10 with NH4OH, the two source solutions were mixed together by adding (NH4)2HPO4 solution via a syringe pump (4 mL min-1) to Ca(NO3)2 solution that was preheated to 363 K in an oil bath and was equipped with a reflux condenser. After addition of the (NH4)2HPO4 solution, a milky white suspension was obtained, and the suspension was kept at 363 K under magnetic stirring overnight followed by aging for 24 hours at room temperature. Finally, the product was collected by centrifugation at 6000 rpm for 5 min and washed by dispersing in deionized (DI) water. The water washing and centrifugation steps were repeated 5 times. A vacuum oven was utilized to dry the wet product at 343 K overnight. In the synthesis of HAP-CO3 catalyst, the same procedure as that for HAP was employed except that a solution consisting of (NH4)2HPO4 (0.25 M) and NaHCO3 (0.17 M) was prepared to replace 0.25M (NH4)2HPO4 solution in HAP synthesis. For synthesis of Pb-HAP-CO3 catalyst, the Ca(NO3)2 solution in the second flask was replaced with a solution of Ca(NO3)2 (0.4 M) and Pb(NO3)2 (0.1 M). The rest of procedure was the same as that for synthesis of HAP-CO3 catalyst. The synthesis of Pb-HAP catalyst followed a reported procedure [16], which was similar to synthesis of bare HAP above except that an aqueous solution of Pb(NO3)2 (0.1 M), Ca(NO3)2 (0.4 M) and NH4Cl (1.3 M) was prepared to replace Ca(NO3)2 (0.25 M) in HAP synthesis. Finally, the vacuum dried HAP-based samples were calcined in flowing air (150 mL min-1, ultrapure, Airgas) at 823 K for 5 hours at a ramp rate of 17.5 K min-1 from 34 ambient temperature. All the samples were pelleted, crushed, and sieved to retain particle sizes between 180 and 425 μm (40-80 mesh) for the following characterization and catalysis experiments. 2.2.3 Catalysts characterization. The morphologies of the HAP-based catalysts were observed by scanning electron microscopy (SEM) on a Hitachi SU-70 electron microscope. The crystallinity of the catalysts was examined by powder X-Ray diffraction (XRD) patterns using a Bruker D8 Advance Lynx Powder Diffractometer (LynxEye PSD detector, sealed tube, Cu Kα radiation with Ni β-filter). N2 adsorption-desorption isotherms of the samples were measured using an Autosorb-iQ analyzer (Quantachrome Instruments) at 77 K. The specific surface areas of the samples were determined using (Brunauer, Emmett and Teller) (BET) method. Thermogravimetric analysis (TGA) of the catalyst samples was performed in a TGA instrument (2950, TA Instruments, Inc.) under a mixed air and N2 flow of 100 mL min-1 (40% air, 60% N2) with a heating rate of 10 K min-1 from 308 K to 1273 K. Elemental composition of the catalysts was determined by inductively coupled plasma optical emission spectroscopy (ICP-OES, Optima 4300DV Instrument, Perkin- Elmer). The FTIR spectra of the samples were recorded with a spectrometer (Nicolet Magna-IR 560) in the range of 400-4000 cm-1. The Raman spectra of the catalysts were collected with a Raman spectrometer (LabRAM Aramis, Horiba Scientific) in the range of 200-2000 cm-1. XPS data was measured over a Kratos AXIS 165 spectrometer equipped with 165 mm radius hemispherical analyzer and eight channesltron detection system coupled with monochromatic Al radiation. The X-ray absorption fine structure spectroscopy (XAFS) measurements at the Pb L3 edge (~ 13.036 keV) were conducted on 35 the bending-magnet beamline of the Materials Research Collaborative Access Team (MRCAT) at the Advanced Photon source in Argonne National Laboratory. XAFS data were collected in the transmission mode under ambient condition. PbO, PbO2 and Pb foil were considered as references and measured at the beamline. The Pb species fraction was calculated by conducting Pb X-ray absorption near edge spectroscopy (XANES) linear combination fittings using Athena in the IFEFFIT software package. 2.2.4 Determination of acidity of catalysts. The surface acidity of the HAP-based catalysts was evaluated by NH3-TPD using an Autosorb-iQ instrument (Quantachrome, ASIQM0000-4) equipped with a thermal conductivity detector (TCD). In the measurement, 0.1 g catalyst sample was loaded into a quartz reactor and heated at a rate of 10 K min-1 to 973 K under He (40 mL min-1) and maintained at this temperature for 2 hours in order to remove the surface impurities. After being cooled to room temperature under He stream, the catalyst sample was exposed to NH3 (30 mL min-1, ultrapure, Airgas) stream for 0.5 hours. Physisorped NH3 was then removed by flowing He gas (30 mL min-1) for 2 hours. Afterwards, the catalyst sample was ramped to 1100 K at a ramp rate of 10 K min-1, and the NH3-TPD profile was recorded during this step. 2.2.5 OCM catalytic test. The catalytic reaction was performed in the reactor system that has been described in our previous work [95]. Typically, 0.3 g of catalyst sample was loaded in a U-shape tubular quartz reactor (10 mm inner diameter) which was placed inside a temperature-controlled furnace (National Electric Furnace FA120 type). The temperature of the furnace was controlled by a Watlow Controller (96 series). A K-type thermocouple 36 was attached to the outer wall of the reactor to monitor the temperature of the catalyst environment. The sample was pretreated in He and O2 atmosphere (33 mL min-1, volume ratio: 91% He, 9% O2) at 823 K for 4 hours prior to the OCM reactions. Afterwards, the catalyst was heated to the desired reaction temperature. The CH4 (8 mL min-1, 99.999% purity, Airgas) and O2 (3 mL min-1, 99.9993% purity, Airgas) diluted in N2 (5 mL min-1, 99.95% purity, Airgas) and He (30 mL min-1, 99.9993% purity, Airgas) were sent via heated transfer lines hold at 343 K to the reactor. The reactant and product gases were analyzed using gas chromatograph (Agilent Technologies, 6890N) equipped with ShinCarbon ST packed column connected to a TCD. Dependency of CH4/O2 ratios was studied by changing the flow rates of O2 and the balance He gases while keeping constant flow rate in CH4. The blank test in the reactor without catalysts showed that the methane conversion was less than 1%, suggesting that the gas phase reaction between CH4 and O2 was not favored under studied conditions. 2.3 Results and Discussions 2.3.1 Textural and structural properties of catalysts. Figure 2.1 shows the SEM images of the HAP-based catalysts. The bare HAP catalyst (Figure 2.1(A)) consists of uniform nanoparticles with sizes between 10 nm to 20 nm. The particle sizes were determined by observing SEM images captured at multiple positions of the sample after loaded into SEM chamber. The nanoparticles tend to form irregular agglomerates. For HAP-CO3 catalyst in Figure 2.1(B), capsule-like particles are observed. Upon Pb substitution into HAP (Pb- HAP), the sample turned into rod-shaped clusters, as shown in Figure 2.1(D). These rod- shaped clusters stacked on each other to form agglomerates. In the case of both cation and 37 anion substituted HAP (Pb-HAP-CO3), a mixture of these morphologies was observed, as shown in Figure 2.1(C). The cation and anion, Table 2.1. Chemical compositions and surface areas of HAP-based catalysts for OCM reactions. Catalyst Composition a Composition b Surface area c Pb/Ca C/P (Ca+Pb)/(P+C) Pb/Ca C/P (Ca+Pb)/(P+C) (m2/g) HAP-CO3 0.00 0.48 1.64 0.00 0.29 1.56 36 HAP 0.00 0.00 1.65 0.00 0.00 1.59 39 Pb-HAP-CO3 0.17 0.44 1.73 0.33 0.34 1.64 37 Pb-HAP 0.16 0.00 1.60 0.47 0.00 1.63 49 a Determined by elemental analysis (ICP-OES). b Determined from XPS measurement. c Determined by BET method. or both cation and anion substitutions did not dramatically change the particle sizes of HAP catalysts. The surface areas of the HAP-based catalysts were determined from N2 adsorption-desorption isotherms and the results are listed in Table 2.1. Consistent with SEM results, there is no significant change in surface areas across these four HAP-based catalysts. XRD, FTIR and Raman spectra were used to study crystal polymorphs of the HAP- based catalysts, and the results are shown in Figure 2.2(A)-(C). In Figure 2.2(A), all peaks of the XRD spectrum for HAP, HAP-CO3, Pb-HAP and Pb-HAP-CO3, respectively, are consistent with the crystalline HAP phase. This result indicates that HAP crystalline structure was well-preserved after the incorporation of cation (Pb2+), anion (CO32-), or both cation and anion (Pb2+ and CO32-), respectively. A further examination on the XRD spectra shows that the peak position and width are slightly different across the HAP-based catalysts. Firstly, the diffraction peaks of Pb-HAP and Pb-HAP-CO3 are shifted to lower diffraction angles compared to HAP and HAP-CO3 samples. The left-shift of the diffraction peaks was caused by Pb2+ substitution, which has larger size than Ca2+ ions. It is reported that the incorporation of larger cations in the apatite materials expands the lattice parameters of the 38 (A) (B) 2.00 µm 1.00 µm (C) (D) 1.00 µm 1.00 µm Figure 2.1. SEM images showing morphologies of HAP-based catalysts: (A) HAP, (B) HAP-CO3, (C) Pb-HAP-CO3, and (D) Pb-HAP, respectively. hexagonal structure and thus leads to an increase in unit cell volume, which in turns causes a downward shifting in 2θ diffraction angles [16]. Secondly, the diffraction peaks broaden from HAP to HAP-CO3 and further to Pb-HAP-CO3 and Pb-HAP, suggesting that the crystallinity of HAP-based catalysts decreased with incorporation of cation, anion, or both type of ions into the crystalline HAP material. The broadening and left-shifting of the diffraction peaks in Pb-HAP and Pb-HAP-CO3 samples might also be resulted from the presence of trivial amount of other crystalline phases such as hydroxypyromorphite (HPY). It is reported that HPY can be precipitated from the synthesis mixture similar to that of 39 HAP used in the present study [128, 129]. The concurrent precipitation of HPY should be insignificant since no obvious diffraction peaks of HPY are observed in Figure 2.2(A). (A) (B) Pb-HAP Pb-HAP Pb-HAP-CO3 Pb-HAP-CO3 HAP-CO3 HAP-CO 2- 2-3 HAP CO 2- CO3 /HPO3 4 OH- HAP 3- PO 3- PO4 4 15 20 25 30 35 40 1600 1400 1200 1000 800 600 2 (degree) Wavenumber (cm-1q ) (C) (D)100 Pb-HAP-CO3 HAP 95 Pb-HAP HAP-CO3 Pb-HAP 90 Pb-HAP-CO3 3- 85 HAP-CO3 n1 PO4 2-n4 PO 3- n 4 1 CO3 n PO 3- HAP2 4 80 HAP-CO3 Pb-HAP-CO3 HAP Pb-HAP 75 300 600 900 1200 300 450 600 750 900 1050 1200 Wavenumber (cm-1) Temperature (K) Figure 2.2. XRD patterns (A), FT-IR spectra (B), Raman spectra (C), and TGA curves (D) of HAP-based catalysts, respectively, used for OCM reactions. Figure 2.2(B) illustrates the FTIR spectra of HAP-based catalysts. The spectrum of bare HAP shows all absorption bands characteristic for HAP [130]. For PO43- group, the ν1 vibration (symmetric stretching) occurred at 942 cm-1, the ν3 vibration (asymmetric stretching) centered at 1000 cm-1 and 1100 cm-1, while ν1 vibration (asymmetric bending) 40 Intensity (a.u.) Intensity (a.u.) Weight Loss (%) Transmittance (a.u.) located at 563 cm-1, 600 cm-1, 942 cm-1, are all observed. The librational mode of OH- groups appears at 631 cm-1, suggesting the presence of hydroxyl groups in the bare HAP sample. Absence of any distinct bands in the range of 1400-1550 cm-1 indicates that bare HAP does not contain detectable carbonate (CO32-) groups. The vibrational modes of PO43- groups are also observed in HAP-CO3, Pb-HAP-CO3, and Pb-HAP samples (Figure 2.2(B)). This result suggests that the HAP structure was preserved upon cation, anion, or both substitutions, in agreement with the XRD results. The presence of broad vibrational bands centered at 1450 cm-1 (υ3 symmetric stretching of CO32-), characteristic for substitution of PO43- by CO32- (B-type) [131, 132], in both HAP-CO3 and Pb-HAP-CO3 samples indicates the incorporation of CO32- into the HAP structure. The absorption band at 631 cm-1 disappears in HAP-CO3 and Pb-HAP-CO3, which resulted from the substitution of OH- by CO32- groups to form A-type substituted HAP. The FTIR spectra of HAP-CO3 and Pb- HAP-CO3 overall suggest that AB-type CO32- substituted HAP was formed. The band at 872 cm-1 in the FTIR spectra of these two samples result from both CO32- (υ2 in-plane bending of CO32-) and HPO42- groups. It is reported that HPO42- groups are created to counterbalance the ionic charge when the OH- groups are replaced by CO32- in HAP, which overlaps with the CO32- band at 872 cm-1. The absence of CO32- bands and appearance of OH- band at 631 cm-1 in Pb-HAP catalyst suggests no detectable CO32- groups in this sample. A slight shift of the vibration modes at 563 cm-1 and 600 cm-1 of PO43- groups toward lower wavenumbers in Pb-HAP and Pb-HAP-CO3 reveals the presence of Pb2+ ions [133], consistent with left-shift of 2θ angles observed from XRD spectra. The features of FTIR spectra overall indicate that successful synthesis of cation (Pb2+), anion (CO32-, AB- type), and cation and anion (Pb2+ and CO32-) substituted HAP catalysts. Raman spectra in 41 Figure 2.2(C) are further used to understand the structural properties of the HAP-based catalysts. The vibration modes of PO43- groups in HAP structure, ν1 (∼960 to 961 cm-1), ν2 (∼430 to 450 cm-1), ν3 (∼1035 to 1048 cm-1 and ∼1070 to 1075 cm-1), ν4 (∼587 to 604 cm- 1), confirm the apatite structure of all the catalysts [134]. The vibrational modes of the CO32- group are detected at 1073 cm-1 in HAP-CO3 and Pb-HAP-CO3 (ν1 mode of B-type carbonate) [134], suggesting the presence of CO32- groups in these two samples, consistent with FTIR results above. The Raman peak at 960 cm-1 undergoes a shift to lower wavenumber (928 cm-1) when the bare HAP was substituted with Pb2+ ions, which is attributed to the stronger Pb-O interactions compared with the Ca-O ones in HAP catalysts [24]. The absence of vibrational band at ~700 cm-1 in Figure 2.2(C), which is assigned to an asymmetric P = O stretching mode of PO43- groups in the stoichiometric HAP, indicates that all the as-synthesized HAP-based catalysts are non-stoichiometric [124, 135]. The thermal stability of the pre-calcined HAP-based catalysts was examined by TGA data presented in Figure 2.2(D). The TGA measurement was conducted on the as- synthesized HAP samples after the drying step in vacuum oven. The slight decrease in weight for all the samples at temperature from 300-750 K is assigned to the evaporation of water bound to the catalyst surface or decomposition of adsorbed ammonium or nitrate ions used in the catalyst synthesis. The weight loss at the temperature above 1100 K was caused by the structure decomposition such as dehydroxylation since the temperature approaches the thermal stability limit of the HAP material [136]. In the temperature range of 750-1100 K, these four HAP-based catalysts showed different thermal decomposition behaviors. The bare HAP is stable, but HAP-CO3 has two weight losses. The first weight loss at 1030 K was due to the removal of CO32- groups from A-type substitution, while the 42 second loss at 1100 K was resulted from the decomposition of B-type substituted CO32- groups [137]. Pb-HAP also has two weight losses in this temperature range. The first one at 830 K is assigned to the loss of chlorine component used in the catalyst preparation [124]. The second one at 1030 K can be attributed to the decomposition of HAP into a Pb- containing β-tricalcium phosphate. Compared to Pb-HAP and HAP-CO3, Pb-HAP-CO3 has much better thermal stability in the temperature range of 750-1100 K. 2.3.2 Composition analysis of HAP-based catalysts. The bulk and surface stoichiometry of the HAP-based catalysts were determined using ICP-OES and XPS analyses, respectively, and the results are listed in Table 2.1. The Pb/Ca molar ratios in the synthesized samples are lower than the theoretical ratios (Pb/Ca = 0.25) used in catalyst synthesis. The cation/anion (i.e., (Ca+Pb)/(P+C)) molar ratios are slightly different, but close to the stoichiometric ratio of 1.67. The substitution of Pb2+ cation or CO32- anion or both into the apatite structure did not alter the stoichiometry of HAP significantly. The surface element composition of the HAP-based catalysts in Table 2.1 shows that Pb/Ca ratios and (Ca+Pb)/(P+C) ratios are higher and lower, respectively, than those determined from ICP-OES analysis for these four catalyst samples. This result suggests that Pb content is rich on the surface of the Pb-HAP and Pb-HAP-CO3 samples. The basicity/acidity of the HAP catalyst and thus their catalytic performance in OCM reactions are known to depend strongly on the chemical state of the cationic sites. XPS data was therefore measured to identify the chemical states of the cationic species (Pb and Ca) in these post-calcined HAP-based catalysts. The presence of the desired elements 43 (Ca, Pb, P and O) in the final product (Figure 2.3(A)) indicates the successful preparation of (A) HAP (B) HAP Pb 4f XPS HAP-CO HAP-CO3 3 Pb-HAP-CO Pb-HAP-CO3 3 Pb-HAP Pb-HAP Pb 4f Pb 4f 7/25/2 Pb-HAP Pb-HAP Pb-HAP-CO3 Pb-HAP-CO3 HAP-CO3 HAP-CO3 HAP HAP 700 600 500 400 300 200 100 0 150 148 146 144 142 140 138 136 134 Binding Energy (eV) Binding Energy (eV) (C) (D) 1.4 HAP Ca 2p XPS PbO2 HAP-CO PbO Pb-HAP3 Pb-HAP-CO 1.2 Pb3 Pb-HAP Pb-HAP 1.0 Pb-HAP-CO3Ca 2p3/2 Ca 2p1/2 Pb Pb-HAP 0.8 PbO Pb-HAP-CO 0.63 PbO2 0.4 HAP-CO3 0.2 Pb-HAP-CO3 HAP 0.0 360 355 350 345 340 13000 13020 13040 13060 13080 13100 Binding Energy (eV) Photon Energy (eV) Figure 2.3. XPS spectra (A), Pb 4f spectra (B), Ca 2p (C) and XANES spectra (D) of HAP-based catalysts, respectively, used for OCM reactions. the HAP-based catalysts. The binding energies of Pb 4f7/2 and Pb 4f5/2 of Pb-HAP are lower than those of Pb-HAP-CO3 (Figure 2.3(B)), which suggests Pb2+ cations of Pb-HAP were positioned in a more electronegative environment than those of Pb-HAP-CO3. There are 44 Intensity (a.u.) Intensity (a.u.) Pb 4p O 1s Ca 2s Ca 2s Pb 4d Pb 4d Ca 2p C 1s P 2s P 2p Pb 5s P 2p Pb 4f Pb 5p Ca 3s Pb 5d Normalized Absorption Intensity (a.u.) two types of Ca sites in HAP materials: 9-coordinated Ca[1] and 7-coordinated Ca[2]. The basicity of an atom is well known to increase with increase in its coordination number [138, 139]. It is expected that Pb2+ ions substitution for Ca[1] would possess more basicity compared to Ca[2], and thus, Pb-HAP had more Pb2+ ions positioned at the Ca[1] sites and higher basicity than Pb-HAP-CO3 catalyst. The binding energies of Ca 2p3/2 and Ca 2p1/2 in the HAP-based catalysts, however, are almost identical (Figure 2.3(C)). The XPS data suggests that the surface basicity of the catalyst is mainly related to the Pb2+ ions and their chemical states. The oxidation state and possible phases of Pb ions in Pb-HAP and Pb- HAP-CO3 were studied using Pb L3 XANES spectra of Pb-HAP and Pb-HAP-CO3. Figure 2.3(D) shows that both Pb-HAP and Pb-HAP-CO3 exhibited similar adsorption energies at 13039 eV. The ~3 eV shift to higher energy than the metallic Pb foil indicates that Pb-HAP is in higher oxidation state than zero. Pb L3 edge represents mainly the 2p à 6p transition for Pb2+-containing structures. On the other hand, a pre-edge at ~ 13027 eV representing 2pà 6s transition can be observed for Pb4+ ions (e.g., PbO2). The absence of pre-edge features in Pb-HAP and Pb-HAP-CO3 XANES spectra indicates that most Pb ions in the catalysts are in the Pb2+ state. There is no overlapping between Pb-HAP and PbO reference, indicating that Pb2+ ions exist in the phase other than PbO. 2.3.3 Surface acidity of HAP-based catalysts. The surface acidity and acid strength of the HAP-based catalysts were investigated using NH3-TPD. Phosphate groups in bare HAP are responsible for the acidity of the catalyst, whereas the Ca2+ ions are responsible for the basicity [137]. An increase in the basicity and acidity is expected upon Pb2+ and CO32- substitution into the HAP structure, respectively. Figure 2.4 shows NH3-desorption peaks exist in the regions of lower temperatures (< 600 K), medium temperatures (750 – 930 K), 45 and also at higher temperatures (> 930 K), respectively. This indicates the presence of weak, medium and strong acid sites in the HAP-based catalyst samples. The broad variation in peak intensity across these four HAP-based catalyst samples reveals that the catalysts differ from each other widely in their total number of acid sites and also in the strength of the acid sites. In the low temperature region in Figure 2.4, bare HAP exhibits a broad peak centered at 440 K. This desorption peak is split into two peaks, one shifted downwards to 400 K and the other shifted upwards to 520 K in the sample of HAP-CO3. When Pb2+ ions were incorporated into the HAP-CO3 to form Pb-HAP-CO3, the acidity of the catalyst is significantly decreased, as reflected by the reduced intensity of the NH3-TPD peaks below 600 K. Pb-HAP shows similar NH3-deosrpiton peak to that of Pb-HAP-CO3. The weak acidity in the catalysts follows the order of HAP-CO3 > HAP > Pb-HAP ~ Pb-HAP-CO3. The medium acidity in these four catalysts shows the same trend to that of weak acidity. In the high temperature range (750 - 930 K), HAP-CO3 and HAP have higher peak intensity than that of Pb-HAP and Pb-HAP-CO3. The order of NH3-TPD peak intensity is HAP- CO3 > Pb-HAP-CO3 > HAP > Pb-HAP. 60 100 HAP HAP-CO3 Pb-HAP-CO3 x 0.5 80 Pb-HAP 40 HAP-CO3 60 Pb-HAP-CO3 40 20 HAP HAP Pb-HAP HAP-CO3 20 0 Pb-HAPPb-HAP-CO3 0 300 400 500 600 700 800 900 1000 1100 Temperature (K) Figure 2.4. NH3-TPD Profiles of HAP, HAP-CO3, Pb-HAP and Pb-HAP-CO3 catalysts. 46 TCD signal (mV) The libration of NH3 in the low temperature range in Figure 2.4 is mainly attributed to the PO43- groups present in the HAP catalysts [140]. The medium and high temperature NH3-desorption peaks are caused by the co-existed PO43-, CO32- and HPO42- groups. The AB-type CO32- substituted HAP, as suggested by FTIR data, contains both CO32- groups in the A and B sites and HPO42- that is resulted from substitution of OH- groups by CO32- [140]. The broad and strong high temperature desorption peak in HAP-CO3 and Pb-HAP-CO3 might also be caused by the release of CO2 from CO32- groups in the apatite structure [141]. It is generally accepted that the strength and quantity of the acid sites are directly related to desorption temperature and intensity in NH3-TPD analysis. Therefore, the NH3-TPD data confirms that the acidity HAP-based catalysts were tuned upon Pb2+, CO32-, or both substitutions. The absence of CO32- from Pb-HAP-CO3 catalyst further reduced the acidity of the catalyst since Pb-HAP only showed very weak NH3 desorption peaks. The reduction and enhancement in acidity upon Pb2+ and CO32- substitution, respectively, are consistent with composition and XPS analyses discussed above. 2.3.4 Performance of HAP-based catalysts in OCM reactions. The activity, stability and selectivity of the HAP-based catalysts in OCM reactions were firstly examined at 973 K and 101 kPa pressure. Figure 2.5(A) shows that the methane conversion followed a sequence of Pb-HAP-CO3 > HAP > Pb-HAP > HAP-CO3. In addition, HAP and Pb-HAP- CO3 were more stable than Pb-HAP and HAP-CO3. The slight decrease in activity of Pb- HAP with time on steam (TOS) and abrupt drop in activity of HAP-CO3 in the beginning of the reaction was resulted from their structural instability, as shown by the TGA data in Figure 2.2(D). It should be noted that Pb-HAP-CO3 maintained its activity and stability in 47 the OCM reaction conditions. Figure 2.5(B) shows the product selectivity of the HAP- based catalysts with TOS of 6 h and methane conversion of 23%. The C2 selectivity monotonically increases from HAP-CO3, HAP, Pb-HAP-CO3 to Pb-HAP catalysts. The activity and selectivity of these four HAP-based catalysts follows the reverse order of low (A) 45 (B) HAP C2H6 C2H4 CO2 CO 40 HAP-CO 1003 Pb-HAP-CO3 35 Pb-HAP 80 30 25 60 20 15 40 10 20 5 0 0 0 2 4 6 8 10 HAP-CO3 HAP Pb-HAP-CO3 Pb-HAP Time on Stream (Hours) Catalysts Figure 2.5. (A) Methane conversion with time-on-stream in OCM reactions and (B) Product selectivity for OCM reactions over HAP-based catalysts at 23% conversion under 973 K and 101 kPa pressure conditions and a space velocity of 8800 mL gcat-1.hr-1. (Error within ±0.2 %) and medium temperature desorption peak intensity in NH3-TPD profiles. Pb-HAP and Pb- HAP-CO3 are more active and selective in C2 formation. It is reported that Pb2+ ions form covalent bonds with carbon, which can stabilize methyl radicals and enable pairwise reaction of methyl radicals to form C2 products [63, 120, 142-144]. In HAP and HAP-CO3 catalysts, the acidic environment promotes dissociative adsorption and builds up a strong interaction between the catalyst surfaces and C2 products. C2 products tend to stay on the surface of the catalyst which then undergoes secondary oxidation process to form CO and CO2 [145]. The trend of C2 selectivity across these catalysts is consistent with their surface 48 CH4 Conversion (%) Product Selectivity (%) composition and acidity changes. This result suggests that the tunable composition of the HAP-based catalysts by ion substitutions influences their physicochemical properties and consequently the selectivity in OCM reactions. (A) 40 90 HAP (B) HAP HAP-CO HAP-CO 35 3 80 3 Pb-HAP-CO Pb-HAP-CO3 3 30 Pb-HAP 70 Pb-HAP 60 25 50 20 40 15 30 10 20 5 10 0 0 920 940 960 980 1000 920 940 960 980 1000 Temperature (K) Temperature (K) (C) 100 (D) 16 HAP 14 HAP-CO90 3 Pb-HAP-CO3 80 12 Pb-HAP 70 10 60 8 50 6 40 4 HAP HAP-CO3 30 2 Pb-HAP-CO3 Pb-HAP 20 0 920 940 960 980 1000 920 940 960 980 1000 Temperature (K) Temperature (K) Figure 2.6. Methane conversion (A) and product selectivity (C2 (B); CO2 (C); CO (D)) of OCM reactions over HAP-based catalysts at different temperatures. (PCH4 = 27.1 kPa, PO2 = 11.0 kPa, total flow = 46 ml min-1, He was used as the balance gas). (Error within ±0.4 %) Methane conversion and product selectivity in OCM reactions over the HAP-based catalysts at different temperatures were studied. As shown in Figure 2.6(A), the methane 49 CO2 Selectivity (%) CH4 Conversion (%) CO Selectivity (%) C2 Selectivity (%) conversion over Pb-HAP and Pb-HAP-CO3 increased significantly with increasing temperature, while the conversion increased only slightly over HAP-CO3 and HAP catalysts. This suggests that Pb species is the active component in the HAP-based catalyst for methane activation in OCM reactions, consistent with previous report [16]. Figure 2.6(B)-(D) shows the selectivity to C2, CO2, and CO products across these four catalysts at different reaction temperatures. C2 production was favored with increasing temperature, CO2 formation was inhibited, while CO formation was kept almost constant. It should be noted that HAP-CO3 generated more CO than other HAP-based catalysts (Figure 2.5(B) and 6(D)). The methane reforming or reverse water gas shift reactions, caused by the acidity of HAP-CO3 catalyst, can potentially contribute to the CO product. (A) 100 60 (B) 100 60 100 60 Pb-HAP-CO Pb-HAP-CO (C) 3 Pb-HAP-CO3 3 90 Pb-HAP 90 Pb-HAP 90 Pb-HAP Pb-HAP-CO 50 Pb-HAP-CO80 3 50 Pb-HAP-CO 80 3 5080 3 Pb-HAP Pb-HAP Pb-HAP 70 70 40 7040 40 60 60 60 50 30 50 30 50 30 40 40 40 20 30 2030 30 20 20 20 10 2010 10 10 10 10 0 0 0 0 0 0 0 1 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 6 CH4/O2 Ratio CH4/O2 Ratio CH4/O2 Ratio Figure 2.7. Effect of methane-to-oxygen (CH4/O2) ratio on C2 selectivity and CH4 conversion at (A) 923 K, (B) 943 K and (C) 973 K at a constant CH4 partial pressure of 25 kPa. (Total flow rate = 46 mL min-1) (Filled symbol represents C2 selectivity, unfilled symbol represents CH4 conversion) (Error within ±0.6 %) The effects of methane to oxygen ratio in OCM reactions was studied on the selected Pb-HAP and Pb-HAP-CO3 catalysts at different temperatures. Figure 2.7 shows the methane to oxygen ratio is tuned by varying oxygen partial pressure (8.1/11.3/21.0/32.0/40.7 kPa, respectively) while keeping a constant methane partial pressure of 25 kPa, and helium was used as the balance gas. CH4 conversion decreased 50 C2 Selectivity (%) CH4 Conve rsion (%) C2 Selectivity (%) CH4 Conv ersion (%) C2 Selectivity (%) CH4 Conve rsion (%) with increasing CH4/O2 ratio while C2 selectivity increased with increasing CH4/O2 ratio at all the studied temperatures for both catalysts. Higher CH4/O2 ratio indicates lower O2 concentration in the gas phase in the reaction, which in turns promotes C2 formation since most of the O2 are activated on the surface of the catalyst. On the other hand, higher CH4/O2 ratio, i.e., lower O2 concentration, leads to lower CH4 conversion. Figures 7(A) also demonstrates that CH4 conversion over Pb-HAP-CO3 catalyst was higher than Pb-HAP catalysts when the reaction temperature was low (923 K). With an increase in reaction temperature to 943 K (Figure 2.7(B)) or 973 K (Figure 2.7(C)), Pb-HAP and Pb-HAP-CO3 showed similar methane conversion. The selectivity to C2 product, however, showed opposite trend with the reaction temperature. At the temperature of 923 K (Figure 2.7(A)), both catalysts have similar C2 selectivity in OCM reactions. With increasing reaction temperature to 943 K (Figure 2.7(B)), Pb-HAP has higher C2 selectivity than that of Pb- HAP-CO3. The differences in C2 selectivity is more evident when the temperature was further increased to 973 K (Figure 2.7(C)). With the available C2 selectivity and CH4 conversion data, the C2 yield, which is product of C2 selectivity and CH4 conversion, can be accessed. At lower reaction temperature (923 K), Pb-HAP-CO3 showed higher C2 yield than that of Pb-HAP catalyst. When the reaction temperature was increased to 973 K, the C2 yield was higher in Pb-HAP than Pb-HAP-CO3. These results suggest that the C2 production over Pb-HAP-CO3 can be optimized under certain reaction conditions. Overall, the OCM reaction data indicate that the catalytic performance of the HAP-based catalysts can be tuned upon cation or anion substations in the apatite structure. Given the catalyst stability, activity, selectivity and O2 permeable properties under optimized reaction 51 condition, Pb-HAP-CO3 can be a potential catalyst to be integrated in membrane reactor to enable effective OCM reactions with side feeding of O2 through the membrane. (A) (B) Pb-HAP CO 2-3 CO 2-3 Pb-HAP-CO3 HAP-CO3 1600 1400 1200 1000 800 600 HAP (C) 00-008-0259 n1 PO43- n2 PO43- n4 PO43- 2- 01-075-9526 n1 CO 20 25 30 35 40 300 600 900 -1 1200 2q (Degree) Wavenumber (cm ) Figure 2.8. XRD pattern (A), FTIR (B) and (C) Raman of Pb-HAP-CO3 after 10-hour of OCM reaction. For comparison purpose, XRD patterns of HAP, HAP-CO3 and Pb-HAP samples after 10-hour OCM reaction have also been presented in (A). The crystallinity and structure of Pb-HAP-CO3 were studied after OCM reactions. Figure 2.8 shows the XRD, FTIR and Raman data, respectively, collected on Pb-HAP-CO3 sample after TOS of 10 hours. For comparison purpose, the XRD patterns of three other catalyst samples (HAP, Pb-HAP and Pb-HAP-CO3) were measured and included in Figure 2.8(A). In comparison with Figure 2.2(A), the sharpness of the XRD peaks in all the samples was decreased, which indicates that crystallinity of the HAP-based catalyst might be reduced after OCM reaction. The HAP, Pb-HAP and HAP-CO3 still maintain their initial diffraction peak positions. The Pb-HAP-CO3, however, formed a new diffraction peak centered at 2θ~30°. This diffraction peak can be indexed to the HPY structure. HPY has been reported to be synthesized from mixture of lead and phosphate precursor solutions [128, 129]. In the OCM reaction condition, acidic environment of Pb-HAP-CO3 may 52 Intensity (a.u.) Intensity (a.u.) facilitate the transformation of the HAP to HPY structure. It should be noted that the catalytic performance of Pb-HAP-CO3 did not decrease with time on stream of the OCM reaction (Figure 2.5). The FTIR (Figure 2.8(B)) spectra of Pb-HAP-CO3 sample after OCM reaction shows that the vibrational bands centered at 1450 cm-1, which corresponds to υ3 symmetric stretching of CO32-. The band at 872 cm-1, which results from CO32- (υ2 in-plane bending of CO32-) group, can still be observed after 10 hours of reaction, implying that CO32- group was still present in the HAP structure. In the case of Raman spectra (Figure 2.8(C)) of Pb-HAP-CO3, the vibrational modes of the CO32- group are detected at 1073 cm- 1 after 10 hour of OCM reaction, again indicating that Pb-HAP-CO3 still maintained the carbonate group inside the structure. 2.4 Conclusion of Chapter 2 The HAP apatite structure with cation (Pb2+), anion (CO32-), or both (Pb2+ and CO32-) substitutions have been synthesized. The physicochemical properties and catalytic behaviors of the HAP-based catalysts were assessed by a variety of characterizations and OCM catalysis tests. The Pb2+ and CO32- substitutions changed the composition and physicochemical properties of the HAP catalyst. The Pb-HAP-CO3 with both Pb2+ and CO32- substitutions exhibited the best thermal structural stability. The tunable composition of the HAP-based catalysts, and thus the tunable surface acidity environment, influences the catalytic behaviors of the catalysts in OCM reactions. The OCM catalysis studies showed that C2 selectivity increased with Pb2+ substitution in the catalysts whereas Pb- HAP-CO3 exhibited better stability in comparison with Pb-HAP. At the examined reaction temperatures of 923, 943, and 973 K, the CH4 conversion and C2 selectivity decreased and 53 increased, respectively, with increasing CH4/O2 ratios. The C2 yield showed a strong dependence on the composition of HAP-based catalysts and reaction conditions. The present work is for the first time a systematic examination of the effects of composition of HAP-based apatite materials on OCM reactions. The synthesis method, understating of the correlations between the composition, property, and catalytic performance of HAP- based catalysts are applicable to a range of apatite structured ceramic materials, which are potential membrane materials for OCM reaction in membrane reactors. 54 Chapter 3: Catalytic Consequences of Cation and Anion Substitutions on Rate and Mechanism of Oxidative Coupling of Methane over Hydroxyapatite Catalysts (Work published on Oh et al., Fuel 2017, 191, 472-485) 3.1 Introduction As discussed in Chapter 1, the OCM reaction involves the reaction of methane (CH4) and oxygen (O2) over a catalyst at high temperatures to form C2 (C2H6 and C2H4) hydrocarbons. The nature of one-step reaction to C2 products in OCM reactions is of great significance and has been a continuous interest in the past few decades [68, 146-148]. A variety of catalysts, including alkali [149-151], alkaline earth [152-156], rare earth [53, 157, 158] and transition metal oxides [159-161], have been broadly studied in OCM reactions. The generally accepted mechanism of OCM reactions [162-164] is a mixed homogeneous-heterogeneous reaction network, in which methyl radicals (CH3×) are formed on the catalyst surface by methane activation, desorbed as free CH3× radicals, and then recombined to form ethane (C2H6) as a primary product and ethylene (C2H4) as a secondary product from the subsequent dehydrogenation of C2H6. Simultaneously, methyl radicals undergo deep oxidation by adsorbed diatomic oxygen species and gas phase molecular oxygen (O2), respectively, to produce CO2 and CO. Lead cation (Pb2+) substituted HAP (Pb-HAP) has been studied as a catalyst for OCM reactions. It is shown that Pb-HAP was able to increase C2 selectivity by a factor of ~ 5 compared to bare HAP [63-65]. The enhancement in OCM performance of Pb-HAP was ascribed to the stabilization of methyl radicals by Pb2+ sites for pairwise reactions against oxidation reactions [66]. The anion substitution such as halide-substituted HAP has 55 been reported to enhance the acid resistance and mechanical properties of HAP bioceramics [165] and to reduce the number of oxygen species that are resulted from -OH groups in oxidative dehydrogenation of alkanes [166]. The substitution of HAP with both cation Pb2+ and fluoride anion (F-) has not been explored in literature, but has potential to concurrently improve the thermal/chemical stability and catalytic activity of HAP in OCM reactions. Selectivities and yields of C2 products in OCM reactions depend on the identity and dynamics of specific elementary steps involved in the primary and secondary reaction steps on the catalysts. Even though the performance of Pb-HAP in OCM reactions has been studied and the enhancement in C2 selectivity upon Pb2+ substitution in HAP has been justified [63, 142], no rigorous kinetic study is available on the kinetics and mechanisms of OCM reactions on this catalyst. In addition, the effects of both cation and anion substituted HAP catalyst on the elementary steps of OCM reaction is yet to be determined. A detailed understanding of the kinetic networks is essential to describe the reaction steps in terms of their rate constants and to define the specific contributions of cation and anion compositions in HAP-based catalysts. On basis of this understanding, a catalyst composition with desired catalytic performances can be developed for OCM reactions. In this chapter, we report the synthesis and characterization of HAP catalysts with Pb2+ cation (Pb-HAP), F- anion (HAP-F), and both cation and anion (Pb-HAP-F) substitutions. The influences of cation and/or anion substitutions in HAP on the rate and selectivity of OCM reactions together with identity and reaction rate constants of elementary steps in primary reactions of OCM were also examined. The results show that HAP and HAP-F followed Langmuir-Hinshelwood reaction pathway in which the reaction 56 occurred between associatively adsorbed O2 and CH4 species. The Pb-HAP and Pb-HAP- F catalysts, however, followed Eley-Rideal reaction pathway in which the reaction occurred between gaseous CH4 and associatively adsorbed O2 species. The substitution of Ca2+ by Pb2+ in HAP preserved C-H activation in CH4 and improved C2 selectivity due to the stabilization of methyl radicals by Pb2+ sites for pairwise reaction despite impaired CH4 and O2 adsorption capabilities. The substitution of -OH groups by F- in HAP weakened both O2 adsorption and C-H bond activation compared to HAP, leading to low methane conversion and diminished CO2 formation but higher C2 selectivity. The Pb-HAP-F exhibited highest C2 selectivity in the tested four HAP-based catalysts, which can be attributed to the integration of methyl radical pairwise reaction and reduced oxygen species on the catalyst in OCM reactions. The present study rigorously explored the mechanisms of OCM reactions and influences of cation and/or anion substitutions in HAP structure on the primary steps of these reactions. 3.2 Experiments 3.2.1 Materials. Ammonium phosphate dibasic ((NH4)2HPO4, ≥99.0%), ammonium chloride (NH4Cl, ≥99.5%), ammonium fluoride (NH4F, A.C.S. Reagent, ≥98.0%) and ammonia hydroxide solution (NH4OH, 28-30%) were supplied from Sigma-Aldrich. Lead nitrate (Pb(NO3)2, ACS. Reagent) was purchased from J.T. Baker while calcium nitrate tetrahydrate (Ca(NO3)2﹒4H2O, 99.0-103.0 %) was purchased from Alfa-Aesar. 3.2.2 HAP-based catalysts preparation. The synthesis of HAP-based catalysts was carried out by a co-precipitation method, as described in Chapter 2 [167]. In the synthesis of bare HAP, 0.24 M of (NH4)2HPO4 solution and 0.40 M of Ca(NO3)2 solution were prepared in 57 two flasks separately. NH4OH was added to each source solution to adjust the pH to ~10. Ca(NO3)2 solution was preheated to 363 K in an oil bath equipped with reflux condenser. After that, (NH4)2HPO4 solution was added to the preheated Ca(NO3)2 solution dropwise via a syringe pump in 2 hours. After addition of the (NH4)2HPO4 solution, a cloudy reaction suspension was obtained, and the suspension was kept at 363 K under magnetic stirring for 24 hours followed by aging for 24 hours at room temperature. Lastly, the product was collected by centrifugation at 6000 rpm for 5 min and washed with deionized (DI) water to remove any undesirable ions. The centrifugation and washing steps were repeated 5 times. The resulted wet product was dried in a vacuum oven at 343 K overnight. As for the synthesis of HAP-F catalyst, the same procedure as that for bare HAP was applied except that a solution consisting of 0.16 M of (NH4)2HPO4 and 0.08 M of NH4F was prepared to replace 0.24 M of (NH4)2HPO4 solution. In the synthesis of Pb- HAP-F catalyst, the Ca(NO3)2 solution used in bare HAP synthesis was replaced with a solution consisting of 0.32 M of Ca(NO3)2 and 0.08 M of Pb(NO3)2. The remaining procedures were the same as that for HAP-F catalyst. Pb-HAP was prepared using procedures reported previously [124], which was similar to the synthesis of bare HAP above except that an aqueous solution of Pb(NO3)2 (0.08 M), Ca(NO3)2 (0.32 M) and NH4Cl (1.3 M) was prepared to replace 0.40 M of Ca(NO3)2 in HAP synthesis. All the dried HAP-based catalysts were treated in flowing air (150 mL min-1, ultrapure, Airgas) at 973 K for 5 hours at a ramp rate of 17.5 K min-1 from room temperature. The catalyst samples were subsequently pelleted, crushed and sieved to retain particle sizes between 180 and 425 𝜇m (40-80 mesh) for characterization and catalysis experiments discussed below. 58 3.2.3 Catalyst characterization. Scanning electron microscopy (SEM) images were taken on a Hitachi SU-70 electron microscope to visualize the morphologies of the HAP-based catalysts. N2 adsorption-desorption isotherms of the samples were measured using an Autosorb-iQ analyzer (Quantachrome Instruments) at 77 K. The samples were outgassed at 523 K for 8 hours and 1 mm Hg prior to measurements. (Brunauer, Emmett and Teller) (BET) method were used to determine the specific surface areas of the samples. X-Ray diffraction (XRD) patterns of the catalyst samples were obtained on Bruker D8 Advance Lynx Powder Diffractometer (LynxEye PSD detector, sealed tube, Cu Kα radiation with Ni β-filter). The Fourier Transform Infrared (FTIR) spectra of the samples were recorded with a spectrometer (Nicolet Magna-IR 560) in the range of 400-4000cm-1 with 36 scans and resolution of 4 cm-1. The Raman spectra of the catalysts were collected with a Raman spectrometer (LabRAM Aramis, Horiba Scientific) in the range of 200-2000 cm-1. Inductively coupled plasma optical emission spectroscopy (ICP-EOS, Optima 4300DV Instrument, Perkin-Elmer) was used to determine the elemental composition of the catalysts, specifically Pb, Ca and P contents. X-ray photoelectron spectroscopy (XPS) data was measured over a Kratos AXIS 165 spectrometer equipped with 165 mm radius hemispherical analyzer and eight channeltron detection system coupled with monochromatic Al radiation. C1s peak was used to calibrate the binding energy of the products. The X-ray absorption fine structure spectroscopy (XAFS) measurements at Pb L3 edge (~ 13.036 keV) were conducted at beamline 10-BM at the Advanced Photon Source in Argonne National Laboratory. XAFS data were collected in the transmission 59 mode under ambient condition. PbO, PbO2 and Pb foil were considered as references and measured at the beamline. The XAFS data were analyzed using IFEFFIT software package. The oxygen temperature-programmed desorption (O2-TPD) analysis of the catalyst samples were carried out in an Autosorb-iQ unit (Quantachrome, ASIQM0000-4) equipped with a thermal conductivity detector (TCD). Typically, 0.1 g catalyst sample was pretreated at 973 K for 2 hours under He (40 mL min-1, ultrapure, Airgas) at a heating rate of 10 K min-1 from ambient temperature. After being cooled to 353 K under He stream, O2 stream (28 mL min-1, ultrapure, Airgas) was introduced into the catalysts for 0.5 hours. The adsorbed O2 was then removed by flowing He gas (40 mL min-1) for 2 hours. Afterwards, the catalyst sample was ramped to 1100 K at a ramp rate of 10 K min-1 and the O2-TPD profiles were recorded. The CH4-TPD profiles of these catalysts were measured using the same procedure as O2-TPD except that O2 was switched to CH4 in the measurement. 3.2.4 OCM catalytic test. The catalytic reaction was performed in a U-shape tubular quartz reactor (10 mm inner diameter). In the experiment, the catalyst sample was loaded in the quartz reactor in which the reactor was placed inside a temperature-controlled furnace (National Electric Furnace FA120 type). The temperature of the furnace was controlled by a Watlow Controller (96 series). A K-type thermocouple was attached to the outer wall of the reactor to monitor the temperature of the catalyst environment. The catalyst was pretreated in He and O2 atmosphere (33 mL min-1, volume ratio: 91% He, 9% O2) at 823 K for 5 hours prior to the OCM reactions. The total gas flow rate was controlled at 46 mL min-1, in which N2 (5 mL min-1, 99.95% purity, Airgas) was used as an internal standard and He was used as balance gas in all the experiments. Dependency of partial pressure of 60 methane (P&'() or partial pressure of oxygen (P$*) was examined by changing the flow rate of CH4 or O2 while keeping the flow rate of the other gas constant. The reactant and product gases were analyzed using gas chromatograph (Agilent Technologies, 6890N) equipped with ShinCarbon ST packed column connected to a TCD. All the kinetics data were measured under differential conditions in which the methane conversion was maintained below 5%. The contribution of secondary reaction pathways was negligible since only the primary products (C2H6, CO and CO2) were detected over the HAP-based catalysts. A blank test with the reactor without catalysts was carried out and the OCM result showed that methane conversion was less than 0.34%, implying that the gas phase reaction between CH4 and O2 was not favored under studied conditions. The conversion varied depending on reaction conditions too. 3.3 Results and discussion 3.3.1 Structural analysis of HAP-based catalysts. Morphologies of the HAP-based catalysts were examined by the SEM observation. Figure 3.1 shows that bare HAP consists of short rod-shaped nanoparticles with sizes between 30 nm to 50 nm (Figure 3.1(A)). Substitution of F- and/or Pb2+ into HAP to form HAP-F (Figure 3.1(B)), Pb-HAP-F (Figure 3.1(C)) and Pb-HAP (Figure 3.1(D)), respectively, did not change the morphologies significantly as short cylindrical-like nanoparticles can still be observed in these three samples. N2 adsorption-desorption isotherms were used to reveal the surface areas of the HAP-based catalysts and the results are listed in Table 3.1. The cation and anion, or both cation and anion substitutions did not significantly change the morphology and surface areas of the HAP-based catalysts, which were all prepared by the co-precipitation method. 61 XRD, FTIR, Raman and XANES spectra were used to determine the crystallinity and incorporation of cations and/or anions into HAP-based catalysts, and the results are shown in Figure 3.2. All peaks of the XRD spectra for HAP, HAP-F, Pb-HAP and Pb- HAP-F, respectively, in Figure 3.2(A), are consistent with the crystalline HAP phase, indicating that HAP structure was well-preserved after the incorporation of cation (Pb2+), anion (F-), or both cation and anion (Pb2+ and F2-), respectively. The downward shifting diffraction peaks of Pb-HAP and Pb-HAP-F with respect to HAP and HAP-F samples was caused by the incorporation of Pb2+, which has larger size than Ca2+ and causes an expansion of the lattice parameters [124]. The broadening in width of the diffraction peaks in Pb-HAP-F and Pb-HAP samples suggests that the crystallinity of these two catalysts slightly decreased [168, 169]. Figure 3.1. SEM images showing morphologies of HAP-based catalysts: (A) HAP, (B) HAP-F, (C) Pb-HAP-F, (D) Pb-HAP, respectively. 62 Figure 3.2(B) demonstrates the FTIR spectra of HAP-based catalysts. All absorption bands characteristic for HAP [130] are shown in the spectra. The vibrational mode of OH- group centered at 631 cm-1 indicates the presence of hydroxyl groups in the bare HAP and Pb-HAP samples. The absorption band at 631 cm-1 disappears in HAP-F and Pb-HAP-F, which could result from the substitution of OH- by F-, as evidenced by Bianco et al. [170]. Raman spectra in Figure 3.2(C) are applied to further investigate the structural properties of the HAP-based catalysts. The vibrational modes of PO43- groups, ν1 (∼960 to 961 cm-1), ν2 (∼430 to 450 cm-1), ν3 (∼1035 to 1048 cm-1 and ∼1070 to 1075 cm-1) and ν4 (∼587 to 604 cm-1), are all well-preserved, which confirms the preservation of apatite structure in all the catalysts [134]. Table 3.1. Chemical compositions and surface area of HAP-based catalysts for OCM reactions. Catalyst Composition measured by ICP Composition measured by XPS Surface a Pb/Ca Pb/(Ca+Pb) (Ca+Pb)/P Pb/Ca Pb/(Ca+Pb) (Ca+Pb)/P F/P area (m2/g) HAP 0.00 0.00 1.72 0.00 0.00 1.59 0 39 HAP-F 0.00 0.00 1.73 0.00 0.00 1.76 0.58 25 Pb-HAP-F 0.25 0.20 1.80 0.50 0.33 1.97 0.62 24 Pb-HAP 0.25 0.20 1.66 0.47 0.32 1.63 0 29 a Determined by BET method. The oxidation state and possible phases of Pb ions in Pb-HAP and Pb-HAP-F were studied using Pb L3 XANES spectra (Figure 3.2(D)). PbO2, Pb-HAP and Pb-HAP-F exhibited similar adsorption energies (E0) at 13039 eV. The ~3 eV shift to higher energy than the metallic Pb foil indicates that Pb-HAP is in higher oxidation state than zero. Pb L3 edge represents mainly the 2p à 6p transition for Pb2+-containing structures. On the other hand, a pre-edge at ~ 13027 eV representing 2pà 6s transition can be observed for Pb4+ ions (e.g., PbO2). The absence of pre-edge features in Pb-HAP and Pb-HAP-F XANES 63 spectra indicates that most of the Pb ions in the catalysts are in the Pb2+ state. Detailed extended X-ray absorption fine structure (EXAFS) analysis of the reference lead oxides as well as the Pb2+ substituted HAP is summarized in Table 3.2 and Figure 3.2(E) and (F). It is known that HAP has a hexagonal structure that is comprised of 10 Ca2+ ions located on two sets of non-equivalent sites, 4 Ca[1] on site 1 and 6 Ca[2] on site 2 in one unit cell [117, 171]. The Ca[1] is coordinated to 6 oxygen atoms belonging to different PO4 tetrahedra and also to 3 oxygen atoms at a larger distance. The Ca[2] is found in cavities in the walls of the channels formed between the Ca and O atoms. The Ca[2] and then the Ca[1] can be subsequently substituted by Pb2+ ions with increasing concentrations [117, 172]. The channels along c-axis of HAP, in which the OH- anions reside, permit substitutional solid solution of some other anions such as F- ions in the present study. In the present study, Pb partial substitutions of Ca[2] is most likely happened at low Pb loading and therefore fitted this work [44]. In Pb-HAP, Pb binds to about 4 oxygen atoms at a shorter distance (2.38 Å, 2.39 Å and 2.44 Å, respectively) and 3 oxygen atoms at a longer distance (2.51 Å and 2.72 Å). In Pb-HAP-F, Pb coordinates with 4 short distance oxygen and 3 long distance oxygen atoms. The substitution of OH- by F- sustainably also increases the Pb-O bond distance by ~ 0.1 Å compared to that in Pb-HAP. 3.3.2 Composition analysis of HAP-based catalysts. The bulk and surface compositions of the HAP-based catalysts were determined using ICP-OES and XPS analyses, and the results are listed in Table 3.1. The bulk (Ca+Pb)/P molar ratio in Pb-HAP is 1.66, nearly the same as the ratio of (Ca+Pb)/P = 1.67 in catalyst synthesis. For HAP, HAP-F and Pb- HAP-F samples, the (Ca+Pb)/P molar ratios are 1.72, 1.73 and 1.80, respectively, which 64 are slightly higher than the starting ratio in the synthesis. The concentration of Pb2+ in both Pb-HAP and Pb-HAP-F catalyst samples is the same, indicated by the same Pb/Ca or Pb/(Ca+Pb) ratio in Table 3.1. The stoichiometry of HAP-based catalysts was slightly altered after the substitution of Pb2+ cation or F- anion or both in the synthesis. The surface composition of the HAP-based catalysts varied from their bulk composition. The surfaces of Pb-HAP-F and Pb-HAP catalysts were enriched with Pb2+ species since the Pb/Ca ratios were increased to 0.50 and 0.47 from 0.33 and 0.32, respectively (Table 3.1). The F- concentrations, represented by F/P ratios, on the surfaces of HAP-F and Pb-HAP-F were similar. (A) (B) Pb-HAP Pb-HAP Pb-HAP-F Pb-HAP-F HAP-F HAP-F HAP HAP OH - PO 3- PO 3- 4 4 15 20 25 30 35 40 1600 1400 1200 1000 -1800 6002q (degree) Wavenumber (cm ) (C) (D) 1.4 PbO2 PbO 1.2 Pb Pb-HAP Pb-HAP-F Pb-HAP 1.0 0.8 Pb-HAP-F 0.6 n PO 3- n PO 3- 1 4 2 4 n PO 3- 4 4 0.4 HAP-F 0.2 HAP 300 600 900 1200 1500 0.0 -1 13000 13020 13040 13060 13080 13100Wavenumber (cm ) Photon Energy (eV) 65 Intensity (a.u.) Intensity (a.u.) Normalized Absorption Transmittance (a.u.) (E) (F) 0.3 0.3 0.2 0.2 0.1 0.1 0.0 0.0 -0.1 -0.1 -0.2 -0.2 -0.3 -0.3 1 2 3 4 5 1 2 3 4 5 R (Angstrom) R (Angstrom) Figure 3.2. XRD patterns (A), FT-IR spectra (B), Raman spectra (C), and XANES spectra (D) of HAP-based catalysts, respectively, used for OCM reactions. k2-weighted magnitude and imaginary component of Fourier transform EXAFS of (E) Pb-HAP and (F) Pb-HAP- F. (k2: ∆k = 3 – 12 Å-1. Blue, Fourier transform magnitude. Red, imaginary component. Dash line, experimental data. Solid line, fitted data). Table 3.2. EXAFS fit parameters for Pb oxides and Pb catalysts (k2: ∆k = 3 – 12 Å-1 and ∆r = 1.3 – 3.5 Å). Catalyst NPb-O R (Å) DWF (Å2) E0 (eV) PbO 4 2.23 0.009 -5.5 PbO2 2 2.15 0.004 -0.3 4 2.19 0.004 -0.3 Pb-HAP 1.8 2.39 0.030 -7.7 3.1 2.44 0.030 -7.7 3.0 2.80 0.030 -7.7 Pb-HAP-F 2.5 2.49 0.031 -0.5 2.5 2.55 0.031 -0.5 3.0 2.91 0.031 -0.5 3.3.3 O2-TPD and CH4-TPD profiles of HAP-based catalysts. Figure 3.3 shows the O2- and CH4-TPD profiles in the temperature range of 300-1000 K. Two obvious desorption peaks were observed in all the catalysts in Figure 3.3(A): one is in the lower temperature range of 350-650 K and the other one stays at higher temperatures (> 650 K). The O2-TPD peak at lower temperature is ascribed to physisorbed oxygen while the higher one is 66 FT[k2·c(k)] FT[k2·c(k)] attributed to chemisorbed oxygen species [56, 173]. Both HAP and HAP-F catalysts showed very strong desorption peaks at both temperature ranges, suggesting that both catalysts have strong physi- and chemisorption of oxygen species. A close comparison of the O2-TPD profiles between HAP and HAP-F shows that HAP-F has slightly higher physisorption of O2 but distinctly lower chemisorption than that of HAP. The Pb2+ substitution in HAP structure drastically reduced O2 adsorption, as indicated by the heavily reduced peak intensity in both temperature ranges in Figure 3.3(A). The addition of F- ions in Pb-HAP further reduced the physi- and chemisorption of O2 species in HAP structure, as evidenced by the lowest peak intensities in all the tested samples. (A) HAP (B) HAP HAP-F HAP-F HAP-F Pb-HAP-F Pb-HAP-F Pb-HAP Pb-HAP HAP HAP HAP-F Pb-HAP Pb-HAP-F Pb-HAP-F Pb-HAP 300 400 500 600 700 800 900 1000 300 400 500 600 700 800 900 1000 Temperature (K) Temperature (K) Figure 3.3. O2-TPD (A) and CH4-TPD (B) profiles of HAP, HAP-F, Pb-HAP and Pb- HAP-F catalysts, respectively. The CH4-TPD profiles in Figure 3.3(B) show similar features to those of O2-TPD profiles. Two desorption peaks (one below ~500 K and the other one above ~870 K) were observed in each catalyst. The intensity of both desorption peaks follows the order of HAP > HAP-F > Pb-HAP > Pb-HAP-F. Apparently, the substitution of Pb2+ and/or F- ions deteriorated the adsorption capability of the HAP structure for CH4 reactant. Pb2+ 67 Intensity (a.u.) Intensity (a.u.) substitution of Ca2+ ions caused more critical influence on the adsorption capability than the F- substitution of OH- groups in the HAP-based catalyst materials. The difference in O2- and CH4-TPD profiles in Figure 3.3 across these four HAP- based catalysts should be correlated to their different cation and/or anion compositions and the resultant structural and physiochemical properties. Previous attempts [174-176] on understanding of oxygen species in HAP was conducted by electron spin resonance spectra after heating HAP to 1173 K followed by exposure to O2. It was shown that O2 could be adsorbed on either Ca[2] or dehydroxylated OH- sites. The fluoroapatite (OH- substituted by F-) did not produce reactive oxygen species, which suggested no effective O2 adsorption on the HAP-F material [177]. These results are consistent with the substantial decrease in the O2-TPD peak in HAP-F and Pb-HAP-F compared to bare HAP in Figure 3.3(A). The partial replacement of Ca2+ ions by Pb2+ caused significant drop in O2-TPD peaks, suggesting that Ca2+ ions have better preference than Pb2+ in O2 adsorption. The coincidence between the decrease in CH4-TPD peaks and Pb2+ and/or F- substitutions in HAP may indicate that CH4 prefers to adsorb on Ca2+ and OH- sites. Detailed mechanism for the observed TPD profiles needs further investigation. 3.3.4 Reaction pathways of OCM reactions over HAP-based catalysts. The kinetics of OCM reactions has been intensively studied over different catalysts [150, 178-180]. A general accepted reaction network (Scheme 3.1) has been proposed to account for the observed kinetics [181, 182]. In this reaction network, activation of methane on catalysts leading to formation of methyl radicals and coupling of methyl radicals to form ethane are the primary reaction steps. The dehydrogenation of ethane to form ethylene is the 68 secondary reaction step. The COx (CO and CO2) can be formed via primary or secondary steps, as shown in Scheme 3.1. In the present study, we only focused on the primary reaction steps of OCM reactions. Scheme 3.1. General reaction pathway for oxidative coupling of methane. Only the primary reaction steps were considered in the present study. The Eley-Rideal [183-189] and Langmuir-Hinshelwood [164, 190-192] mechanisms have been commonly considered for the primary steps in OCM reactions. The Eley-Rideal mechanism states that the reaction between gaseous methane and adsorbed oxygen, either dissociative [185-187, 193] or molecularly type [188, 189], is the rate limiting step (RLS) in OCM reactions. An exemplary catalyst for this mechanism is 5%Na2WO4–2%Mn/SiO2 [40, 181, 194-196]. The Langmuir-Hinshelwood mechanism regards RLS as the reaction between adsorbed methane and oxygen species. Sm2O3 [197], Na-doped MgO [191] and perovskite [192] catalysts have been proposed to follow this mechanism. The suitability of both mechanisms for OCM reactions over the HAP-based catalysts has been examined in the present study (refer to Appendix A). The kinetic data, however, suggests that the OCM reactions for HAP and HAP-F follow the Langmuir- Hinshelwood mechanism in which the RLS is the C-H bond activation between 69 molecularly adsorbed methane and oxygen on two different active sites of these catalysts. For Pb-HAP and Pb-HAP-F, the OCM reactions follow the Eley-Rideal mechanism in which the RLS is the abstraction of hydrogen from gaseous methane by associatively adsorbed oxygen species on the catalysts. J Step (1): O I% + ∗ ↔ K O∗I% J Step (2): CH) + ∗% ↔ * CH∗%) L Step (3): CH∗% + O∗I M) % → CH@ ∙ +HO% ∙ + ∗I+∗% (RLS) L Step (4): CH (@ ∙ +CH@ ∙ → C%HO L Step (5): CH@ ∙ +O% → P CO L Step (6): CH ∙ +O∗I →Q@ % CO% Scheme 3.2. Proposed reaction steps based on Langmuir-Hinshelwood mechanism for OCM reactions over the HAP-based catalysts. 3.3.4.1 Derivation of reaction rate equations. Scheme 3.2 shows the proposed reaction steps for OCM reactions over the HAP-based catalysts following the Langmuir- Hinshelwood mechanism. The reaction involves quasi-equilibrated associative oxygen adsorption on surface site 1 (*1), leading to formation of O∗I% species (step 1). Similarly, quasi-equilibrated associative adsorption of CH4 on site 2 (*2) results in CH∗%) species (step 2). The reaction between adsorbed methane and oxygen forms methyl radicals (step 3, RLS). Under pseudo-steady state assumption for adsorbed O∗I ∗%% and CH) species, the rate law for methane consumption (r&'(, µmol gcat -1 s-1) is shown in Eq. (3.1): nI nr %&'( = k@ UKIP$* W \] UK%P&'( W \] (3.1) Y1 + KIP$*[ Y1 + K%P&'([ where k3 (µmol gcat [mol*1]-1[mol*2]-1 s-1) is methane activation rate constant in step 3, K1 (kPa-1) and K2 (kPa-1) are the equilibrium constants for the adsorptions of oxygen (step 1) and methane (step 2), respectively. P&'( (kPa) and P$* (kPa) are the partial pressures of 70 methane and oxygen in the reaction. n1 (mol*1 gcat-1) and n2 (mol*2 gcat-1) are the number of active sites for oxygen and methane adsorptions on the HAP-based catalysts. Details on the equation derivation can be referred Appendix A. J Step (1): O% + ∗ ↔ K O∗% L Step (2): CH) + O% ∗ ↔ M CH@ ∙ +HO% ∙ + ∗ (RLS) L Step (3): CH@ ∙ +CH ∙ → ( @ C%HO L Step (4): CH P@ ∙ +O% → CO L Step (5): CH@ ∙ +O∗% → Q CO% Scheme 3.3. Proposed reaction steps based on Eley-Rideal mechanism for OCM reactions over the HAP-based catalysts. Scheme 3.3 shows the proposed reaction steps for OCM reactions over the HAP- based catalysts following the Eley-Rideal mechanism. The reaction involves quasi- equilibrated associative oxygen adsorption on surface site (*) to form O∗% species (step 1). The reaction between gaseous methane and molecularly adsorbed oxygen forms methyl radicals (step 2, RLS). Under pseudo-steady state assumption for O∗% species, the rate law for methane consumption (r&'(, µmol gcat -1 s-1) is shown in Eq. (3.2): n r&'( = k I @P&'( UKIP$* W \] (3.2) Y1 + KIP$*[ where k3 (µmol [mol*1]-1 s-1 kPa-1) is methane activation rate constant in step 2, K1 (kPa-1) is the equilibrium constant for the adsorption of oxygen (step 1), P&'( (kPa) and P$* (kPa) are the partial pressures of methane and oxygen in the reaction, nI(mol*1 gcat-1) is the number of active sites for O2 adsorption in the catalyst. For consistency of rate constant notation in both Langmuir-Hinshelwood and Eley-Rideal mechanisms, k3 is used (instead 71 of k2) to represent methane activation rate constant in step 2. Section A2 of Appendix A detailed the derivation for Eq. (2). The formation of primary products (C2H6, CO and CO2) in both mechanisms are the same and is described by steps (4)-(6) in the Langmuir-Hinshelwood mechanism (Scheme 3.2) and by steps (3)-(5) in the Eley-Rideal mechanisms (Scheme 3.3). Methyl radicals couple with each other to form C2H6 (step 4 in Scheme 3.2 or step 3 in Scheme 3.3), but at the same time undergo side reactions with gaseous oxygen to form CO (step 5 in Scheme 3.2 or step 4 in Scheme 3.3) or adsorbed oxygen species to form CO2 on the catalyst surface (step 6 in Scheme 3.2 or step 5 in Scheme 3.3). Under pseudo-steady-state assumption for methyl radicals, the rate law equations for C2H6 (r&*'Q, µmol gcat -1 s-1), CO (r&$, µmol gcat-1 s-1) and CO2 (r&$*, µmol gcat -1 s-1) formation are: r %&*'Q = k)P&'M∙ (3.3) r&$ = k 0.96, positive intercept and slope in each set of kinetic data supports the proposed Langmuir-Hinshelwood mechanism for OCM reactions over the HAP-based catalysts. Table 3.3 shows that K1 follows the sequence of HAP > HAP-F > Pb-HAP ~ Pb- HAP. This result is consistent with the O2-TPD profiles of these four catalysts in Figure 3.3(A). The K2 values in Table 3.3 also follow the trend of HAP > HAP-F > Pb-HAP ~ Pb- 74 CH4 Consumption Rate (µmol g -1 -1cat s ) CH4 Consumptin Rate (µmol g -1cat s -1) (A) 8 (B) 4 HAP HAP HAP-F HAP-F Pb-HAP-F Pb-HAP-F 6 Pb-HAP 3 Pb-HAP 4 2 2 1 0 0 0 10 20 30 40 50 60 70 80 0 5 10 15 20 PCH4 (kPa) PO2 (kPa) Figure 3.5. Product of CH4 pressure and the inverse rate of CH4 consumption as a function of CH4 pressure at fixed P$% =7.0 kPa (A) and product of O2 pressure and the inverse rate of CH4 consumption as a function of O2 pressure at fixed P&'(= 25 kPa (B) over HAP- based catalysts. (973 K, 101 kPa total pressure, total flow rate = 46 mL min-1, He as balance gas.) HAP-F, consistent with the CH4-TPD profiles (Figure 3.3(B)) of these catalysts. The adsorption equilibrium constant reflects the bound nature of active species on the HAP- based catalysts. The bare HAP has the strongest oxygen and methane adsorption among these four catalysts. The F- substitution in HAP significantly weakened O2 adsorption (K1 decreased by a factor of 3 compared to bare HAP) while only slightly influenced the methane adsorption. The Pb2+ substitution in HAP weakened both the oxygen and methane adsorption. The influence of Pb2+ substitution impaired the methane adsorption strength of HAP more significantly than F-substitution (refer to K2 values in Table 3.3). Pb-HAP-F showed similar K1 and K2 values to those of Pb-HAP catalyst. The C-H bond activation in step 3 of Scheme 3.2 (indicated by ka,avg in Table 3.3) over HAP catalyst was close to that of Pb-HAP catalyst. The F- substitution in HAP apparently deteriorated the C-H activation of the HAP catalyst since HAP-F has ~4 times lower ka,avg than HAP or Pb-HAP. The co- 75 P r -1CH4 CH4 (kPa.g -1cat.s.µmol ) PO2 r -1 CH4 (kPa.g -1cat.s.µmol ) existence of Pb2+ and F- ions in HAP (Pb-HAP-F) only slightly sacrificed the C-H bond activation compared to Pb-HAP, suggesting that Pb2+ played an important role in contesting the sequestration effect of F- ions on C-H bond activation in step 3 of the OCM reaction mechanism. Table 3.3. Equilibrium constant of O2 adsorption (K1), equilibrium constant of CH4 adsorption (K2) and rate constant of CH4 activation (ka) based on Langmuir-Hinshelwood mechanism at 973K of OCM reactions over HAP-based catalysts. Catalyst K1 a K2 b ka c ka d ka, avge (kPa-1) (kPa-1) (𝜇mol gcat-1 s-1) (𝜇mol gcat-1 s-1) (𝜇mol gcat-1 s-1) HAP 0.11 0.063 116 88 102 HAP-F 0.040 0.052 43 49 47 Pb-HAP-F 0.032 0.010 68 84 76 Pb-HAP 0.032 0.010 90 100 95 aErrors are ± 0.003 kPa-1, bErrors are ± 0.004 kPa-1, cDetermined by varying partial pressure of O d2, Determined by varying partial pressure of CH4, and eka, avg=(ka,1+ka,2)/2. 3.3.4.3 Methane and oxygen consumption rates over HAP-based catalysts based on Eley-Rideal mechanism. The K2 values for Pb-HAP and Pb-HAP-F in Table 3.3 are significantly smaller (by a factor of ~ 6) compared to HAP and HAP-F, hinting that methane adsorption on the surface active sites for these two Pb-based catalysts is almost negligible. Eley-Rideal mechanism involving reaction between gaseous methane and adsorbed diatomic oxygen species was thus proposed to accommodate the OCM kinetics over these two catalysts. When P&'(is fixed, Eq. (3.2) can be linearized as: P&'(nI 1 1= + (3.8) r&'( k@KIP$* k@ where the slope is I and the intercept is I . The methane activation rate constant LMpKJK LMpK (k3∙n1) and adsorption equilibrium constant for oxygen (K1) were calculated from the intercept and slope in Eq. (8), respectively. 76 (A) 25 (B) 100 HAP HAP HAP-F HAP-F 20 Pb-HAP-F Pb-HAP-F Pb-HAP 80 Pb-HAP 15 60 10 40 5 20 0 0 0.0 0.1 0.2 0.3 0.4 0.5 0 5 10 15 20 1/PO2 (kPa -1) PO2 (kPa) (C) 100 HAP HAP-F Pb-HAP-F 80 Pb-HAP 60 40 20 0 0 10 20 30 40 50 60 70 80 PCH4 (kPa) Figure 3.6. Product of CH4 pressure and the inverse rate of CH4 consumption as a function of O2 pressure at fixed P&')= 25 kPa (A), methane consumption rate as a function of O2 pressure at P&'(= 25 kPa (B) and CH4 pressure at P$% = 7.0 kPa (C), respectively. (973 K, 101 kPa total pressure, total flow rate = 46 mL min-1, He as balance gas.) The curves in (B) and (C) denote the fitting results using optimized rate. Figure 3.6 shows the linearized correlations of eqr( versus I (Figure 3.6(A)). The /qr( ef* linear fittings with R2 > 0.97 support the proposed Eley-Rideal mechanism for OCM reactions over the HAP-based catalysts. Similar to Langmuir-Hinshelwood mechanism, Table 3.4 shows that K1 follows the sequence of HAP > HAP-F > Pb-HAP ~ Pb-HAP, 77 CH4 Consumption Rate P r -1CH CH (µmol g-1 s-1 4 4 ) (kPa.g -1cat.s.µmol ) CH4 Consumptin Rate (µmol g -1 -1cat s ) consistent with O2-TPD profiles in Figure 3.3(A). The C-H bond activation in step 3 of Scheme 3.3 (indicated by k3∙n1 in Table 3.4) over HAP catalysts follows similar trend to that of Scheme 3.2 with Langmuir-Hinshelwood pathway. HAP has the highest methane activation rate constant, follows by Pb-HAP, Pb-HAP-F and HAP-F, respectively. Figures 6(B) and (C) show fitting of the measured rates of CH4 consumption as a function of P&'(and P$*, respectively using the Eley-Rideal model. The curves on the plots denote the fitting results using optimized K1 and k3∙n1 values in Table 3.4. The good fittings shown in Pb-HAP and Pb-HAP-F again suggests that these two catalysts leaned towards Eley-Rideal pathways with reaction between gaseous methane and adsorbed diatomic oxygen species. The poor fittings on HAP and HAP-F, on the other hand, explains that these two catalysts follow Langmuir-Hinshelwood pathways. Table 3.4. Equilibrium constant of O2 adsorption (K1) and rate constant of CH4 activation (k3∙n1) based on Eley-Rideal mechanism with associative O2 adsorption at 973K of OCM reactions over HAP-based catalysts. Catalyst K1 a k ∙n b 3 1 (kPa-1) (𝜇mol g -1 s-1 cat kPa-1) HAP 0.15 2.77 HAP-F 0.07 0.41 Pb-HAP-F 0.04 0.77 Pb-HAP 0.04 1.02 aErrors are ± 0.0015 kPa-1, and bErrors are ± 0.0035 𝜇mol g -1 -1 -1cat s kPa . 3.3.4.4 Product formation and selectivity over HAP-based catalysts. Figure 3.7 shows C2H6, CO and CO2 formation rates as a function of P&') over the HAP-based catalysts. HAP exhibited slightly higher C2H6 formation rate among these four catalysts when P&'( was low, while Pb-HAP and Pb-HAP-F showed higher formation rates than bare HAP when P&'( was increased high enough (Figure 3.7(A)). HAP-F has the lowest 78 C2H6 formation rate compared to the other three catalysts. Figure 3.7(B) shows the CO formation rate followed the order of HAP > HAP-F >Pb-HAP ~Pb-HAP-F. The trend in CO2 formation rate over these catalysts (Figure 3.7(C)) is similar to that of CO formation. The kinetic data in Figure 3.7 were used to obtain the regressed values for the rate constants k4, k5 and k6 in Eqs. (3.3) – (3.5), respectively, and the fitted rate constants for each catalyst were listed in Table 3.5. For HAP and HAP-F, the values of K1, K2 and ka,avg in Table 3.3 (A) 10 (B) 25 HAP HAP HAP-F HAP-F 8 Pb-HAP-F 20 Pb-HAP-F Pb-HAP Pb-HAP 6 15 4 10 2 5 0 0 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 PCH4 (kPa) PCH4 (kPa) (C) 60 HAP 50 HAP-F Pb-HAP-F Pb-HAP 40 30 20 10 0 0 10 20 30 40 50 60 70 80 PCH4 (kPa) Figure 3.7. C2H6 (A), CO (B) and CO2 (C) formation rates as a function of CH4 pressure at P$%= 7.0 kPa over HAP-based catalysts. (973 K, 101 kPa total pressure, total flow rate = 46 mL min-1, He as balance gas.) The curves denote the fitting results using optimized rate constants. 79 CO2 formation rate C2H6 formation rate (µmol g -1 -1cat s ) (µmol g -1 cat s -1) CO formation rate (µmol g -1 -1cat s ) (from Langmuir-Hinshelwood mechanism) were used for the regression to obtain product rate constants. For Pb-HAP and Pb-HAP-F, the values of K1 and k3∙n1 in Table 3.4 (from Eley-Rideal mechanism) were applied instead. The continuous curves in Figure 3.7 denote the fitting results using the regressed rate constants for Eqs. (3) - (5). A R2 values in all cases of above 0.90 suggests that the fitting for the experimental within error range is acceptable. Table 3.5. Rate constant for formation of C2H6 (k4), CO (k5) and CO2 (k6∙n1) at 973K of OCM reactions over HAP-based catalysts. Catalyst k c k d 4 5 k6∙n e 1 (𝜇mol g -1 s-1 kPa-2) (𝜇mol g -1 s-1 kPa-2) (𝜇mol g -1 s-1 kPa-1cat cat cat ) HAPa 0.17 0.23 10.35 HAP-Fa 0.50 0.20 8.01 Pb-HAP-Fb 0.53 0.03 10.50 Pb-HAPb 0.45 0.03 9.50 aDetermined from Langmuir-Hinshelwood mechanism, bDetermined from Eley-Rideal mechanism with associative O2 adsorption, cErrors are ±0.025 𝜇mol g -1 s-1 kPa-2cat , dErrors are ±0.0013 𝜇mol g -1 s-1 cat kPa-2, and eErrors are ±0.7 𝜇mol g - -1 cat 1 s kPa-1. Figure 3.8 shows the dependence of C2H6 selectivity on P&') and P$%(represented / by q*rQ ) respectively, in the OCM reactions over the HAP-based catalysts. The C H (/qf*g/qf) 2 6 selectivity increased with increasing P&') (Figure 3.8(A)) and decreasing P$% (Figure 3.8(B)). Higher methane pressure indicates lower O2 concentration, which in turns slows down the oxidation of methane species to COx products. On the other hand, higher oxygen pressure promotes the oxidation reactions and thus leads to lower C2H6 selectivity. The dependence of C2H6 selectivity on P&') and P$%, respectively, can be explicitly expressed by Eq. (9) that was derived from the rate equations for formation of C2H6, CO and CO2 in steps (4)-(6) and steps (3)-(5) of the Langmuir-Hinshelwood and Eley-Rideal mechanisms, respectively. 80 r&*'Q k)P= &'@ . K P (3.9) r&$ + r&$% k21 times) than that in the blank reactor (Figure 4.5(A)). This indicates that HAP material is catalytically active for methane combustion reaction. A comparison across the HAP catalysts with and without (A)100 100 (B)100 160 CO CO2 2 CO CO 80 80 80 120 60 60 60 80 40 40 40 40 20 20 20 0 0 Blank Oriented HAP Unoriented HAP 0 0c-surface HAP a-surface HAP Catalyst Catalyst Figure 4.5. (A) Methane conversion and product selectivity in methane combustion reaction in the absence of any catalyst and in the presence of oriented and unoriented HAP catalysts, respectively. (B) shows areal rate of methane conversion and product selectivity over c-surface and a-surface of HAP catalyst. (Temperature = 973 K and total pressure = 101 kPa, space velocity = 91700 mL gcat-1.hr-1, N2 was used as internal standard) (Error within ±0.6 %) crystal orientation showed that unoriented HAP enabled slightly higher methane conversion than c-axis oriented HAP at 973 K. In both cases, CO and CO2 were the only carbon-containing products detected under the investigated reaction conditions over the catalysts. Oriented HAP demonstrated higher CO2 selectivity than unoriented HAP (Figure 4.5(A)), suggesting that c-surface of HAP crystals favor CO2 formation than a-surface. The stability of the oriented and unoriented HAP catalysts in methane combustion reaction was tested by running the reaction at 973 K for 10 hours and the result is shown in Figure 4.6 104 Product Selectiviy (%) Product Selectiviy (%) CH4 Conversion Rate (µmol m -2 s-1) CH4 Conversion (%) in the Supporting Information. The stability result exhibited no obvious deactivation during the test, in which CH4 conversion remained at ~83 % and ~98 % for oriented HAP and unoriented HAP, respectively. 100 80 60 40 20 Oriented HAP Unoriented HAP 0 0 2 4 6 8 10 Time on Stream (Hours) Figure 4.6. Long-term stability test of oriented and unoriented HAP catalysts in methane combustion reaction. (Temperature = 973 K and total pressure = 101 kPa, space velocity = 91700 mL gcat-1.hr-1, N2 was used as internal standard) (Error within ±0.5 %) The differences in methane conversion and product selectivity in the methane combustion reaction could result from the differences in surface areas, surface composition and exposed crystal planes between the c-axis oriented and randomly orientated HAP catalysts. As shown in Table 4.2, the surface area of unoriented HAP is 15 times higher than that of c-axis oriented HAP. The methane conversion over the unoriented HAP is only 1.2 times higher than c-axis oriented HAP catalyst. The non-consistency in enhancements between the surface area and catalytic activity of HAP catalyst hints that surface area does not contribute to the performance difference directly. It should be noted that the unoriented HAP was obtained by grinding of the c-axis oriented HAP. The increase in surface area 105 CH4 Conversion (%) mainly arose from the exposure of the side crystal surface (i.e., a-surface) of the HAP material. The non-linear increase in surface area with the increase in catalytic activity, therefore, suggests that the crystal planes of the HAP material could have different activity in the methane combustion reaction. Table 4.2 also compares the surface composition (Ca/P ratio) of the oriented and unoriented HAP materials. It shows that the unoriented HAP has higher Ca/P ratio than the c-axis oriented one. In order to verify the effect of different Ca/P ratio of HAP material on the methane combustion reaction, we purposely synthesized the unoriented HAP sample with Ca/P ratio of 1.57 and tested its performance under the same reaction condition as what we have studied above. The results showed that methane conversion of 94% and CO2 selectivity of 62% was obtained, which is comparable to that of 98% methane conversion and 67% CO2 selectivity over the unoriented HAP with Ca/P ratio of 1.71. Therefore, the crystal plane of HAP catalyst should contribute to the different catalytic activity and selectivity in the methane combustion reaction. To distinguish the differences in catalytic behaviors of crystal planes in HAP, we further analyzed the data presented in Figure 4.5(A) by evaluating the areal reaction rate and selectivity on both a- and c-surfaces of HAP catalysts. Particularly, the (002) plane, which is the major exposed surface of oriented HAP catalyst film, is defined as c-surface of HAP catalyst. Other planes of HAP catalyst are denoted as a-surface, whose area was determined by deduction of surface area of oriented HAP from that of randomly oriented HAP catalyst in Table 4.2. Apparently, the c-surface of HAP catalyst highly favors CO2 formation compared to a-surface that promotes CO formation (Figure 4.5(B)). The areal rate of methane combustion over c-surface was 47 times higher than that of a-surface of HAP catalyst. The tremendous differences in areal rate and selectivity analysis indicates 106 that the c-surface of HAP is more active in methane activation and favors complete combustion of methane to CO2 compared to other planes in HAP. 4.3.4.2 Performance of oriented HAP and Pb-HAP in OCM reactions. The catalytic activity and product selectivity of the c-axis oriented and unoriented HAP as well as Pb- HAP catalysts in OCM reactions were tested at 973 K, 101 kPa pressure, and at molar ratio of CH4/O2 ratio of 4. The stability of the oriented and unoriented HAP-based catalysts was first tested by running the reaction at 973 K for 10 hour and the results is shown in Figure 4.7. Similar to methane combustion reaction, all the catalysts maintained their activity and 20 Oriented HAP Unoriented HAP Oriented Pb-HAP Unoriented Pb-HAP 15 10 5 0 0 2 4 6 8 10 Time on Stream (Hours) Figure 4.7. Long-term stability test of oriented and unorietend HAP and Pb-HAP in OCM reactions. (Temperature = 973 K, total pressure = 101 kPa, space velocity = 34300 mL gcat- 1.hr-1, N2 was used as internal standard) (Error within ±0.6 %) stability in the OCM reaction conditions. Figure 4.8(A) shows the methane conversion and product selectivity in OCM reactions over these catalysts at time-on-stream of 6 hours. When the catalysts were ground to form irregular orientations, methane conversion was increased to ~11% for both catalysts. Similar to that in the methane combustion reaction, 107 CH4 Conersion (%) unoriented HAP-based catalysts exhibited higher catalytic activity than c-axis oriented ones. Apparently, the enhancement in methane conversion is not linearly proportional to the increase in surface areas of these catalysts. This result confirms that the surface area is not the direct cause of the different catalytic performances between oriented and unoriented HAP catalysts. Table 4.2 shows that both oriented Pb-HAP and unoriented Pb-HAP catalysts have quite similar (Ca+Pb)/P ratios, which hints that the surface composition is not directly responsible for the different performances of both types of catalysts, either. Exclusively, it seems that the different crystal planes exposed on the HAP-based catalysts should again, contribute to different methane activation and product selectivity in OCM reactions. Figure 4.8(A) shows the product selectivity of the HAP-based catalysts at time- on-stream of 6 hours in OCM reactions. Substitution of Pb into the HAP structure improved C2 selectivity in both oriented and unoriented Pb-HAP samples[65, 142, 271]. The product selectivity data shows that the c-axis oriented HAP exhibited higher CO2 selectivity but lower CO, C2H4 and C2H6 selectivity than the unoriented HAP catalyst. Similarly, the c- axis oriented Pb-HAP demonstrated higher CO2 selectivity than the unoriented Pb-HAP catalyst. Consistent with those results obtained from methane combustion reaction, the c- surface of HAP crystals favors CO2 formation than the a-surface. To access the catalytic behavior of c- and a-surface of HAP-based catalyst, we further analyzed the areal rates and selectivity for OCM reactions over these catalysts, and the results are shown in Figure 4.8(B). Clearly, the c-surface of Pb-HAP showed the highest methane conversion rate in OCM reaction, followed by c-surface HAP. Methane conversion rate on c-surface was ~13 times higher than a-surface of Pb-HAP. On the other hand, methane conversion rate on c-surface of HAP in OCM reaction was ~47 times higher 108 than a-surface of HAP. The product selectivity analysis shows that c-surface of HAP and Pb-HAP catalysts favors CO2 formation while a-surface promotes CO production. The elevated methane conversion rates in both c-axis oriented HAP and Pb-HAP catalysts again suggest that (002) plane of the HAP-based catalysts promotes methane activation and complete oxidation to CO2 product. (A) 100 15 (B)100 50 C2H4 C2H4 C2H6 C2H6 80 CO 12 80 CO2 40 CO2 CO 60 9 60 30 40 6 40 20 20 3 20 10 0 Oriented Unoriented Oriented 0 0 0Unoriented c-surface a-surface c-surface a-surface HAP HAP Pb-HAP Pb-HAP HAP HAP Pb-HAP Pb-HAP Catalyst Catalyst Figure 4.8. (A) Methane conversion and product selectivity in OCM reactions over oriented and unorietend HAP and Pb-HAP, respectively. (B) Areal rate of methane conversion and product selectivity in OCM reactions over c-surface and a-surface of HAP and Pb-HAP, respectively. (Temperature = 973 K, total pressure = 101 kPa, space velocity = 34300 mL gcat-1.hr-1, N2 was used as internal standard) (Error within ±0.4 %) 4.3.4.3 Functionality of crystal plane of HAP-based catalysts. As discussed in Section 4.3.1, the hexagonal HAP has prism-faceted a-surface and basal-faceted c-surface, which render it with anisotropic adsorption capabilities[272, 273]. In the methane combustion and OCM reactions, the higher selectivity towards CO2 on the c-surface compared to a-surface of HAP-based catalysts indicates their anisotropic catalytic characteristics. The consistency between anisotropic adsorption and catalysis suggests that different intrinsic surface characteristics of HAP-based materials could be one of the causes 109 Product Selectivity (%) Product Selectivity (%) CH4 Conversion Rate (µmol m -2 s-1) CH4 Conversion (%) for the anisotropic catalytic behaviors. The higher CO2 selectivity on the c-surface of HAP- based catalysts could be due to the preferential adsorption of O2 onto this surface, which subsequently promoted complete methane oxidation in both methane combustion and OCM reactions. Previous study[274] has also shown that CO adsorbs stronger onto the c- surface than a-surface of HAP crystals due to different exposed surface termination groups in different surface facets of the HAP crystals. As an intermediate in methane oxidation to CO2, the adsorbed CO could be easily oxidized into CO2, which leads to higher selectivity to CO2 on the c-surface of HAP catalyst. As a unique structural character, the hydroxyl ions of HAP are lined along the c- axis direction in the large channel of HAP materials. The anisotropic migration of OH- anions followed by dehydroxylation upon heating treatment could occur, as indicated previously by Wang et al. and Liu et al. [275, 276]. The oxide ion and vacancies would be formed during the dehydroxylation process, which can in turns function as active sites to accept O2 and activate O2 via the interface, i.e., c-surface, of HAP materials. Therefore, the existence and migration of these vacancy and oxide ion species along large channel of HAP and its c-surface could directly influence its performance in methane oxidation reactions. As a result, the c-surface has much higher activity that the a-surface of HAP, as well as higher CO2 selectivity. In contrast, the side planes of HAP-based materials (for example, the a-surface), disfavor oxygen migration and activation, and thus showing lower activity and making less CO2. A rigorous analysis on the kinetic behaviors of the c- and a-surfaces of the HAP-based catalysts was further carried out by employing the oriented and unoriented HAP and Pb-HAP catalysts in OCM reactions, as discussed below. 110 (A) (B) Oriented HAP Oriented HAP Unoriented HAP Unoriented HAP Oriented Pb-HAP unoriented HAP Oriented Pb-HAP Unoriented Pb-HAP Unoriented Pb-HAP unoriented Pb-HAP unoriented HAP unoriented Pb-HAP oriented Pb-HAP oriented HAP oriented HAP oriented Pb-HAP 300 400 500 600 700 800 900 300 400 500 600 700 800 900 Temperature (K) Temperature (K) Figure 4.9. NH3-TPD (A) and CO2-TPD (B) profiles of oriented and unoriented HAP- based catalysts. We further understand the acidity and basicity of the a- and c-surfaces of HAP and Pb-HAP from NH3-TPD and CO2-TPD measurements. As shown in Figure 4.9(A), both unoriented HAP and Pb-HAP have obvious NH3 desorption peaks in the temperature range 350 - 600 K, while the oriented HAP and Pb-HAP show very weak NH3 desorption peak in this temperature range. This data suggests the a-surfaces of HAP and Pb-HAP have stronger acidity than their c-surfaces. The comparison between NH3-TPD profiles of unoriented HAP and Pb-HAP indicates that the former has a higher NH3 desorption temperature, which means a-surface of HAP has higher acidity than a-surface of Pb-HAP. The Pb-substitution in HAP decreased its acidity, consistent with previous studies[65, 167]. CO2-TPD profiles in Figure 4.9(B) illustrates that these four HAP-based catalysts have two CO2 desorption peaks: the low temperature peak in the temperature range 350-600 K and the high temperature peak > 600 K. The unoriented HAP and Pb-HAP have similar low temperature CO2 desorption peaks that are stronger than those of oriented HAP and Pb- 111 Intensity (a.u.) Intensity (a.u.) HAP. This data hints that a-surfaces of HAP and Pb-HAP have weaker basic sites than their c-surfaces. The co-existence of acid and basic sites in HAP has been reported previously[277, 278]. On the contrary, oriented HAP and Pb-HAP have stronger high temperature desorption peak than the unoriented counterparts, which suggests that c- surfaces of HAP and Pb-HAP have stronger basic sites. The slightly higher intensity in CO2-desorption peak at temperature range 600-800 K in the CO2-TPD profile of Pb-HAP than that of HAP corresponds to the higher basicity due to Pb substitution. In summary, the TPD measurements indicate a-surfaces of HAP and Pb-HAP have weak acid-base pairs, but c-surfaces of HAP and Pb-HAP have strong basic sites. The Pb-substitution leads to a decrease in acidity and increase in basicity in both a-surface and c-surface of HAP-based catalysts. It is reported that phosphate (PO43-) groups in HAP are responsible for the acidity of the catalyst, whereas the calcium ions (Ca2+) are responsible for the basicity[279]. The CO2 evolution peak in the temperature range 600-800 K in the CO2-TPD profiles can predominantly be linked to the reaction of CO2 with basic OH- groups (CO2 + 2OH- → CO32- + H2O)[277]. As discussed in Section 4.3.1 in the chapter, the basal c-surface of HAP-based materials is rich in calcium ions, while the prismatic a-surface is rich in phosphate ions groups. The large channel of HAP confers mobility to hydroxyl ions and allows proton migration along the channels in the c-axis direction, especially under high temperature conditions. The higher acidity in a-surfaces of HAP and Pb-HAP can be attributed to the high density of phosphate groups. The higher basicity at elevated temperatures in c-surfaces of HAP and Pb-HAP, on the other hand, is due to the rich calcium ions and basic OH- groups. The TPD data suggest that the basic sites of higher 112 strength in oriented HAP and Pb-HAP might contribute to higher methane conversion rate of these two catalysts in OCM reaction while the strong acid sites in unoriented HAP and Pb-HAP might lead to low methane conversion rate in the same reaction. The selectivity to CO2 products in both methane combustion and OCM reactions can be linked to the high temperature basic sites as well, since higher basicity favors complete oxidation of CO into CO2[280]. 4.3.4.4 Kinetics of c- and a-surfaces of Pb-HAP in OCM reaction. In this study, the kinetics of OCM reactions over oriented and unoriented Pb-HAP catalysts were rigorously analyzed by measuring methane consumption rates (r&'() under various partial pressures of methane (P&'() or oxygen (P$*), respectively, at 973 K. The kinetic data for c-surface of Pb-HAP were determined from the oriented film catalyst directly; while the kinetics on the a-surface were evaluated from deduction of kinetic data of oriented Pb-HAP from those of unoriented catalysts. Eley-Rideal mechanism involving reaction between gaseous methane and adsorbed diatomic O2 species that was developed in our previous work[281] is employed to accommodate the OCM kinetics over the c-surface and a-surface of Pb-HAP catalyst. The reaction steps involve quasi-equilibrated associative oxygen J adsorption on surface site (∗) to form O∗ K ∗% species (O% +∗ ↔ O%). This is followed by the L rate limiting step (RLS,CH) + O∗ → M % CH@ ∙ +HO% ∙ + ∗) involving H abstraction from CH4 by O∗% to form HO% ∙ and CH@ ∙ radicals. Under pseudo-steady state assumption for O∗% species, the rate law for methane consumption (r&'(, µmol m A%AI AI „ƒ‡ s ) is shown in Eq. (4.3): /qr( = k P “K P I p @ &'K ( I $* s t” (4.3) IgJKef* 113 where k3 (µmol [m]-2 s-1 kPa-1) is methane activation rate constant, K1 (kPa-1) is the equilibrium constant for the adsorption of oxygen, P&'( (kPa) and P$* (kPa) are the partial pressures of methane and oxygen in the reaction, nI(m2) is the surface area of the catalyst for O2 adsorption. When the CH4 pressure is fixed in the OCM reactions, the rate law for CH4 consumption in Eq. (4.3) can be linearized as, P&'(nI 1 1= + (4.4) r&'( k@KIP$* k@ where the slope is I and the intercept is I . The methane activation rate constant (k ) LMJK L 3 M and adsorption equilibrium constant for oxygen (K1) were calculated from the intercept and slope in Eq. (4.4), respectively. Figure 4.10 shows that methane consumption rates depend on both P&'( and P$*on either c-surface or a-surface of the Pb-HAP catalysts. The rate of methane conversion increased with increasing P&') and P$%. A comparison across the c-surface and a-surface Pb-HAP catalysts showed that c-surface enabled much higher methane conversion than a- surface of Pb-HAP catalyst at the same P&') or P$%. Figure 4.11 shows the linearization treatment of Eq. (4.3) for extraction of kinetic parameters of the OCM reaction over c- and a-surfaces of Pb-HAP catalysts. Figure 4.11(A) shows the linearized correlations of n1eqr( /qr( versus I . The linear fitting with coefficient of determination (R2) >0.93, positive slope ef* and intercept in each set of kinetic data support the Eley-Rideal mechanism for OCM reaction over the c- and a-surfaces of Pb-HAP catalysts. Figure 4.11(B) demonstrates the correlation between /qr( = •k [K P ( I )]– P p @ I $* IgJ e &'( , in which the slope is K K f* k@[KIP I $*( )] and the intercept is 0. This results further confirm that the proposed IgJKef* 114 (A) 0.10 (B) 0.10 c-surface Pb-HAP c-surface Pb-HAP a-surface Pb-HAP a-surface Pb-HAP 0.08 0.08 0.06 0.06 0.04 0.04 0.02 0.02 0.00 0.00 0 10 20 30 40 50 60 70 0 3 6 9 12 15 18 PCH (kPa)4 PO2 (kPa) Figure 4.10. Methane consumption rate as a function of partial pressure of methane at P$% = 4.0 kPa (A) and partial pressure of oxygen at P&') = 32 kPa (B) over c-surface and a- surface of Pb-HAP, respectively. (Temperature = 973 K, total pressure = 101 kPa, total flow rate = 46 mL min-1, N2 as internal standard and He as balance gas) The curves in (A) and (B) denote the fitting results using kinetic parameters shown in Table 4.3. Eley-Rideal mechanism fits the OCM kinetics occurring on both c- and a-surface of Pb- HAP catalysts. The equilibrium constant for the adsorption of oxygen (K1) and the rate constant of methane activation (k3) were determined from the slope and intercept in Figure 4.11(A) and were summarized in Table 4.3. The values of the O2 equilibrium constant is higher on the c-surface than that of a-surface in Pb-HAP catalysts. The higher O2 equilibrium constant on the c-surface of Pb-HAP than that of a-surface further validates our explanation that O2 adsorbs stronger onto the c-surface than a-surface of HAP-based crystals. According to the mechanism proposed in our previous study[281], adsorbed O∗% species tend to react with methyl radicals to form CO2. Since c-surface of Pb-HAP catalyst 115 CH4 Consumption Rate (µmol s-1) CH4 Consumption Rate (µmol s-1) (A) 600 c-surface Pb-HAP (B) 8 a-surface Pb-HAP c-surface Pb-HAP a-surface Pb-HAP 500 6 400 300 30 4 c-surface Pb-HAP 200 20 10 2 100 0 0.0 0.1 0.2 0.3 0.4 0 0 0.0 0.1 0.2 0.3 0.4 0 10 20 30 40 50 60 70 1/PO (kPa) PCH (kPa)2 4 Figure 4.11. Product of CH4 pressure and the inverse areal rate of CH4 consumption as a function of inverse O2 partial pressure at fixed P&'( = 32 kPa (A), areal rate of methane consumption rate as a function of CH4 pressure at P$* = 4.0 kPa (B) over c-surface and a- surface of Pb-HAP, respectively. (Temperature = 973 K, total pressure = 101 kPa, total flow rate = 46 mLmin-1, He as balance gas) The curves in (B) denote the fitting results using kinetic parameters shown in Table 4.3. Table 4.3. Equilibrium constant of O2 adsorption (K1) and the rate constant of CH4 activation (k3∙n1) based on Eley-Rideal mechanism with associative O2 adsorption at 973 K of OCM reactions over c-surface and a-surface, respectively, of Pb-HAP catalysts. Pb-HAP catalyst K a k b1 3 (kPa-1) (µmol m-2 s-1 kPa-1) c-surface 0.17 0.20 a-surface 0.04 0.01 a Errors are ±0.003 kPa-1. b Errors are ±0.004 𝜇mol m-2 s-1 kPa-1. possesses more adsorbed O∗% species, methyl radicals could readily react with them and subsequently oxidize into CO2, leading to higher CO2 selectivity on this surface (Figures 4.5 and 4.8). Similar to O2 equilibrium constant, the rate constant of methane activation on the c-surfaces of Pb-HAP catalysts is much higher than that of a-surface of Pb-HAP. The higher methane activation rate constant on the c-surface of Pb-HAP than that of a-surface suggests that this surface is more active than the other surfaces in OCM reactions. The 116 P r -1CH4 CH4 (kPa.m2.s. mol-1µ ) CH4 Consumption Rate (µmol.m-2 s-1) curves on Figures 4.10(A) and (B) show the fitting of the measured rates of methane consumptions as a function of P$* and P&'( using Eley-Rideal mechanism. The good fitting suggests that the catalytic behaviors of the specific crystalline plane, i.e., both c- and a-surfaces, of Pb-HAP are interpreted successfully using the mechanisms developed for traditional bulk catalysts in OCM reactions. 4.4 DFT calculations on oriented and unoriented HAP catalysts. Periodic DFT calculations, based on the (002) facet in the oriented HAP c-plane (Figure 4.12(a)), were employed to validate the proposed kinetic model for OCM. Optimized structures of relevant surface site and reactions intermediates involved are illustrated in Figure 4.12. Energies of reaction paths on both stoichiometric and reduced (002) surface (with one hydroxyl vacancy in Figure 4.12(b)) are shown in Figure 4.13. Once the hydroxyl vacancy is formed (1.71 eV endothermic), the adsorption of one molecule of O2, with ∆EŠˆp‹ˆpŒ being -3.28 eV (Figure 4.9(c)), is included. One methyl (CH3) and one H atom are produced from methane activation. In such case, CH3 can exist as a gas phase radical or adsorb onto the surface at one lattice oxygen site, (Figure 4.12(d)), to initiate further oxidation. Subsequent oxidation, mainly via dehydrogenation, can produce CH2*, CH*, and eventually, CO at the Ca(1) site, (Figure 4.12(e), (f), and (g)). Similar mechanism has been reported in literature[282]. The hydrogen species formed from the first step of methane activation generate OOH with adsorbed O2 (Figure 4.12(h)) and the result is consistent with Eley-Rideal type mechanism. The proposed pathways and reaction energies for C2H6, C2H4, CO, and CO2 formations, based on DFT calculations, are summarized in Figure 4.13. Methane binds 117 weakly on HAP (002) facet, with a ∆EŠˆp‹ˆpŒ of -0.62 eV. Hence, methane activation occurs via direct interaction with adsorbed O2, producing CH3 and OOH*, following an associative Eley-Rideal mechanism. As shown in Figure 4.13, two routes are possible for Figure 4.12. Optimized structures used for potential energy surface construction: (a) clean surface; (b) surface with OH vacancy; (c) adsorbed O2; (d) CH3*; (e) CH2*; (f) CH*; (g) CO*; and (h) H*. Color scheme: blue, violet, red, purple, brown, and white represent Ca, P, O (lattice), O (gas phase), C, and H species, respectively. the CH3 group: (1) remain in the gas phase (represented by dashed lines), or (2) adsorb onto the HAP surface (represented by solid lines). Two gas phase CH3 radicals combine to form C2H6 and the reaction is exothermic by -2.05 eV. Furthermore, methyl radicals can 118 react with other O2* species, again via an Eley-Rideal step, to form CH2(g) and OOH* (endothermic by 1.96 eV), which is specifically represented by a path in short dashed lines. The methylene radicals, i.e., CH2 (g), combine to form C2H4, which is exothermic by -3.88 eV. Figure 4.13. Black solid path represents surface OH vacancy formation and oxidation of adsorbed methyl (CH3*) to CO and CO2, long dashed path represents C2H6 formation via combination of gas phase methyl groups, i.e., CH3(g), short dashed path represents C2H4 formation via combination of gas phase methylene groups, i.e., CH2(g), and dotted path represents CHx (x =1 –3) oxidation pathway on stoichiometric (002) facet. Asterisk ‘*’ indicates adsorbed species. The oxidation route is initiated by adsorbed methyl species, i.e., CH3*. The adsorption step for gas phase methyl onto the activated (002) facet is -2.62 eV. The following dehydrogenation step produces CH2*, in which the reaction is endothermic by 1.86 eV. The abstracted H atom is assumed to react with OOH* group to form a molecule of H2O. As shown in Figure 4.13, the formation of CH* from CH2* is exothermic. Furthermore, CH* forms a strong C–O bond with lattice oxygen and spontaneously becomes CHO*, as shown in Figure 9(f). This step is exothermic by -1.46 eV. CH* reacts 119 with OH* to break its last C–H bond to form CO* on the Ca(1) site, as shown in Figure 4.12(g), and release one molecule of H2O. This step is endothermic by 0.79 eV. The newly generated vacancy can adsorb another molecule of O2 from gas phase to oxidize CO* into CO2, which is highly exothermic (-2.17 eV). As shown in Figure 4.13, methane activation on stoichiometric surface forming adsorbed methyl radicals (for CO and CO2 formations, represented by dotted lines) is endothermic by 2.58 eV. This is much less energetically favorable compared to the same step at the O2 adsorbed surface. Therefore, it can be concluded that methyl groups produced from methane activation would prefer to exist as gas-phase radicals, and eventually form C2 hydrocarbons via C bond coupling. The mechanism derived from DFT calculations matches consistently with the kinetic measurements and supports the proposed Eley-Rideal mechanism applied in the accompanying kinetic model. Moreover, it can be deduced that the opposite product selectivities observed on randomly oriented HAP facets could be originated from fewer sites for O2 adsorptions on the a-surfaces. Hence, DFT calculations were performed to calculate respective hydroxyl vacancy formation energetics on (002), (112), and (211) facets, according to Eq. (4.1). The calculated vacancy formation energies on (002), (112), and (211) facets are 1.71 eV, 3.49 eV and 5.71 eV, respectively, as listed in Table 4.4. DFT-optimized vacancy structures and their relationships to HAP crystal are illustrated in Table 4.4. Vacancy formation energy on different facets. Facets (002) (112) (211) ΔEvac, eV 1.71 3.49 5.17 120 Figure 4.14. The much higher vacancy formation energies on (112) and (211) facets indicate that it is indeed energetically more challenging to form such vacancies on these surfaces than the (002) face for O2 adsorption during OCM. Figure 4.14. Vacancy structures, after removal of one OH group, on HAP (211) (in purple) and (112) (in yellow) facets on a-surfaces. Dashed boxes indicate the locations of OH vacancies. 4.5 Conclusion of Chapter 4 The anisotropic catalytic behavior was observed on the c- and a-surfaces of HAP and Pb-HAP catalysts in methane combustion and OCM reactions. By employing the seeded hydrothermal growth method, the catalyst films with c-surface exposure were created. The grinding of c-axis oriented HAP-based catalyst films formed unoriented catalysts with increased exposure of a-surfaces. The areal activity of c-surface was up to 121 47 times higher than the a-surface in HAP-based catalysts in methane oxidation and OCM reactions. The c-surface favors complete oxidation of methane into CO2 product compared to a-surface in both reactions. The Eley-Rideal mechanism describes the kinetics of OCM reaction on both c- and a-surfaces of Pb-HAP. The rigorous analysis of the kinetics data suggests that O2 adsorption on the c-surface is favored compared to the a-surface, which might be resulted from oxide ions and vacancies formed from the anisotropic migration of OH- ions upon dehydroxylation at high temperatures. Reaction energies for C2H6, C2H4, CO, and O2 formation pathways from DFT calculations supports the proposed Eley-Rideal mechanism. Facets with hydroxyl vacancies can readily adsorb molecular oxygen, which subsequently makes the adsorption and oxidation of CH3 much more energetically favorable. DFT calculations also confirmed that the hydroxyl vacancies necessary for O2 adsorption will be formed more favorably on the c-surface than the facets from a-surfaces. The present study exemplified the crystalline plane exposed on HAP were anisotropic and can be tailored to impact its activity and selectivity by orientation in catalytic reactions. 122 Chapter 5: Direct Non-oxidative Methane Conversion in a Millisecond Catalytic Wall Reactor 5.1 Introduction Past research efforts have studied non-catalytic[283-286] and catalytic DNMC[98, 99, 287-290] for production of chemicals and fuels from natural gas feedstock. The non- catalytic route focused on methane pyrolysis to achieve high yields of acetylene at short reaction time, but reaction temperature above 1973 K is required[77, 78, 283, 291]. In catalytic DNMC, the metal loaded zeolite catalysts such as Mo/ZSM-5 are used at temperature typically below 950 K, but the reaction is limited by low methane conversion and fast catalyst deactivation[83, 292-300]. As discussed in Chapter 1, the recently reported iron/quartz (Fe/SiO2) catalyst, comprised of lattice-confined single Fe sites embedded in SiO2 matrix, is effective for DNMC reaction[99]. It has negligible coking compared to metal/zeolite catalysts and has significant enhancement in methane conversion and C2+ yields. However, high reaction temperatures approaching or exceeding 1200 K and high heat supply for methane activation on Fe/SiO2 catalyst are required, and these hamper its effective industrial implementation. Technoeconomic and environmental aspects requires very efficient chemical reactor systems that manage heat supply and heat recovery for the highly endothermic DNMC reaction. In this chapter, we discuss the innovation of the DNMC technology by designing a millisecond catalytic wall reactor that manipulates methane conversion, product selectivity and coke formation as well as heat supply and recovery to realize an autothermal operation of DNMC from natural gas resources. The millisecond catalytic wall reactor has reactant flowing in the reactor channel and reaction happening on the wall, so the diffusion time of 123 reactant to the reactive wall is within millisecond (refer to Experiments section). Such short-contact-time is viable for chemical synthesis from alkanes, such as oxidative dehydrogenation of hydrocarbons for olefin[301] or hydrogen production[302-304] because highly nonequilibrium products with less or no carbon formation are achievable. Placing catalyst directly onto the inner and outer walls of reactor promotes heat transfer in thermal boundary layers so that heat released by exothermic reaction on one side of the reactor can be effectively coupled to heat supplied by endothermic reaction on the opposite side[305-307]. Therefore, the process can be run autothermally and almost adiabatically with a residence time of approximately a few milliseconds. It also guarantees a very high throughput using smaller amount of catalyst, energy and capital costs, compared to the traditional technology. However, millisecond catalytic wall reactors have never been attempted in DNMC reaction. 5.2 Experiments 5.2.1 Materials. Toluene (ACS Reagent grade) and methanol (ACS Reagent grade) were supplied from Fischer. Iron (II) chloride (FeCl2, 99.5% metal basis) were purchased from Alfa Aesar while sodium ethoxide (NaOC2H5, 96%) were purchased from Acros. Tetraethyl orthosilicate (TEOS, 98% purity) and sodium hydroxide (NaOH, 99%, Sigma- Aldrich) were supplied from Sigma-Aldrich 5.2.2 Synthesis of the Fe/SiO2 catalyst. To synthesize Fe/SiO2 catalyst, fayalite (Fe2SiO4) was first prepared as the iron source using the method reported by DeAngelis et al.[308] The synthesis setup comprised of a 1000 mL three-neck flask equipped with a magnetic stir bar, in which the right and left necks of the flask were sealed with rubber septa, while 124 the middle neck was connected to a condenser that was sealed at the top with a septum. The flask was heated in an oil bath under constant stirring condition to keep the solvent under reflux condition and to ensure even mixing in the synthesis process. The right septum was removed as needed to add reactants and solvents while a needle was inserted through the left septum to deliver argon gas for purging purpose. The reactor setup and the condenser were connected to a circulating cooling bath to condense the evaporated solvents during the synthesis process. In the synthesis of Fe2SiO4, 375 mL toluene (ACS Reagent 6 grade, Fischer) and 175 mL methanol (ACS Reagent grade, Fisher) were first added to the three-neck flask under magnetic stirring condition. The solution mixture was purged with flowing argon gas (100 mL·min-1) for 30 min at room temperature. Next, 8.7 g of iron (II) chloride (FeCl2, 99.5% metal basis, Alfa Aesar) and 9.3 g of sodium ethoxide (NaOC2H5, 96%, Acros) were added to the liquid mixture in the flask in sequence. The mixture was then heated to refluxing condition. During the ramping process, 7.9 g tetraethyl orthosilicate (TEOS, 98% purity, Sigma-Aldrich) was added to the mixture. The mixture was kept under refluxing condition for another 30 min. Lastly, 10 mL of 0.2 M NaOH (99%, Sigma-Aldrich) solution was added to the mixture via a syringe pump (New Era Pump Systems NE-1000) at a flow rate of 0.5 mL·min-1. After the addition, the mixture was continued to stir under reflux condition for 12 hours. After 12 hours, the heating plate was turned off and the mixture was cooled down to room temperature. The flask was continuously purged by Argon gas throughout the whole synthesis process. The gel-like mixture in the flask was transferred to a rotary evaporator (Heidolph Laborota 4000) to remove solvents. The powdered samples formed was then calcined in a 125 tube furnace (National Electric Furnace FA120 type) for 4 hours under flowing nitrogen gas (100 mL min-1) at 1073 K with a ramp rate of 5 K min-1. After calcination, the Fe2SiO4 was washed and centrifuged with hot (~353 K) deionized H2O to remove NaCl. The washing and centrifugation steps were repeated five times. Finally, the Fe2SiO4 sample was rinsed with methanol and dried with rotary evaporator. The as-prepared Fe2SiO4 sample was mixed with quartz particle and ball milled for 12 hours. The mixture was then heated to 1973 K for 6 hours under stagnant air in a high temperature furnace (Sentro Tech, ST- 1700C-101012) to produce Fe/SiO2 catalyst. The Fe/SiO2 catalyst was crushed and ground to fine powder for the catalysis tests and deposition on the wall of quartz tube to form catalytic wall reactor. 5.2.3 Manufacturing of catalytic wall reactor. The catalytic wall reactor was manufactured by heating the center of a 457 mm in length and 6.35 mm in outer diameter quartz tube to its melting temperature (~1973 K) with a torch (3A blow pipe) flowing hydrogen and oxygen. The softened part of the quartz tube was then curved 180 degrees to form a “U” shape. To make the catalytic wall reactor, the as-synthesized Fe/SiO2 catalyst in fine powder form was packed into the U-shape quartz tube. Both the Fe/SiO2 catalyst and quartz tube were then heated to ~ 1973 K to fuse the Fe/SiO2 catalyst to the wall of the quartz tube. This step was repeated several times to make sure the Fe/SiO2 catalyst was uniformly dispersed and fully embedded into the wall of the quartz tube. The leftover Fe/SiO2 catalyst that was not fused into the reactor wall was taken out. Eventually, the reactor with a flow channel and catalyst on inner wall was prepared for the catalysis tests. A simple dimensional calculation was performed to estimate the diffusion time of reactant 126 to the catalytic wall reactor[301]. The time for diffusion, t to the reactive wall in laminar flow is approximated as t ≅ (~/%) * ≅ ().@I šš/%) * * = 0.046 s (X.1) ™ I9 šš /† where x is the inner diameter (4.31 mm) of the catalytic wall reactor and D is the methane diffusion coefficient estimated to be 10 mm2 s-1 at reaction temperatures. Since the residence time of DNMC reaction in catalytic wall reactor is in the range of 5 s to 25 s, the diffusion time of the reactant to the catalytic wall is much shorter than the experimental residence time. This ensures that the feed gas contacts with the catalytic surface during the reaction. 5.2.4 Catalyst characterization. The crystallinity of the catalysts was examined by powder X-ray diffraction (XRD) patterns using a Bruker D8 Advance Lynx Powder Diffractometer (LynxEye PSD detector, sealed tube, Cu Kα radiation with Ni β-filter). N2 adsorption– desorption isotherms of the samples were measured using an Autosorb-iQ analyzer (Quantachrome Instruments) at 77 K. The specific surface areas of the samples were determined using (Brunauer, Emmett and Teller) (BET) method. The Fe composition of the catalyst was determined by inductively coupled plasma optical emission spectroscopy (ICP-EOS, Optima 4300DV Instrument, Perkin-Elmer). The amount of coke deposited on the Fe/SiO2 catalyst obtained from setting (ii) in Figure 5.1c of the main text was investigated on a thermo-gravimetric analyzer (TGA) (Shimadzu, TGA-50). In the TGA analysis, the temperature was increased to 1273 K under flowing air (50 mL min-1, breathing grade, Airgas) at a ramp rate of 10 K min-1. The amount of coke formed on the catalytic wall reactor was examined using temperature-programmed oxidation (TPO). 127 Typically, the coked catalytic wall reactor was placed inside a temperature-controlled furnace. The temperature of the furnace was held constant by a Eurotherm Controller (2408 series). The catalyst temperature was monitored by a K-type thermocouple attaching to the outer wall of the catalytic wall reactor. The furnace was ramped to 1123 K at a ramp rate of 10 K min-1 under flowing He (35 mL min−1, ultrapure, Airgas) and O2 (5 mL min−1, ultrapure, Airgas) atmosphere. The O2-TPD profile was recorded using a mass spectrometer (ABB Extrel) during this step. 5.2.5 DNMC reaction 5.2.5.1 DNMC reaction in fixed-bed reactor. The DNMC reaction was performed in a non-active U-shape quartz reactor under atmospheric pressure and at 973 K. Typically, 0.375 g of Fe/SiO2 was loaded into the quartz reactor in which the reactor was placed inside a temperature-controlled furnace (National Electric Furnace FA120 type). The temperature of the furnace was held constant by a Watlow Controller (96 series). The catalyst temperature was monitored by a K-type thermocouple attaching to the outer wall of the reactor. The catalyst was heated in N2 atmosphere (20 mL min-1, ultrapure, Airgas) to the desired reaction temperature prior to the DNMC reaction. CH4 (research grade, Airgas) and N2 (ultrapure, Airgas) were then introduced to the reactor at a total gas flow rate of 20 mL min-1 (10% N2 internal standard). The product effluents were analyzed on-line using gas chromatograph (Agilent Technologies, 6890N) equipped with ShinCarbon ST packed column connected to a TCD and DB-WAX column connected to a FID to determine methane conversion and product selectivity. 5.2.5.2 DNMC reaction in catalytic wall reactor. Same experimental setup as that of fixed-bed reactor was used to perform DNMC reaction in catalytic wall reactor. After 128 the catalytic wall reactor was heated in N2 atmosphere (20 mL min-1, ultrapure, Airgas) to the desired reaction temperature, CH4 (research grade, Airgas) and N2 (ultrapure, Airgas) were then introduced to the reactor. The reaction was run at a temperature range of 1223 K - 1323 K and at a total gas flow rate range of 10 to 50 mL min-1. 5.2.6 Model simulations for DNMC-coke combustion process concept A theoretical scale-up design was simulated in Aspen Plus (V10) to assess the feasibility of the DNMC reaction at an industrial scale. A yield-specific reactor was used to input conversion data based on the experimental reaction results from 1323 K and 20 mL min-1 (available in Table 5.2) using an initial feed of 1000 kmol/hr. To model the coke combustion process demonstrated in the concentric cylindrical autothermal reactor, an additional separator and reactor following the product stream from the reaction are proposed. The coke-free stream is cooled to 373 K and a hydrogen-permeable membrane is used to separate hydrogen from the product stream without energy-intensive separation units. The resulting hydrocarbon stream is then separated into light (C1 & C2) and heavy (aromatics) hydrocarbons, then the distillate is compressed and enters a demethanizer to separate and recycle methane. Figure 5.1(A) presents this system without using heat integration to obtain the overall energy requirements for the process. Six total heat exchangers using external utilities are required; three use various pressure and temperature steam for heating, two require cooling water, and one requires cryogenic liquid to reduce the temperature significantly for methane/C2 separation. In Figure. 5.1(B), a heat exchanger network is incorporated to introduce heat integration into the system as an energy- and cost- saving technique and to model the autothermal nature of the reaction process. The 129 feed stream needs to reach the reaction temperature of 1323 K, and the product stream following the combustion reaction provides adequate heat release from the exothermic reaction to maintain this feed stream temperature following the initial heating process. This energy-saving step eliminates the requirement for any external heating in the entire process. Furthermore, heating methane following the demethanizer with the heat released from the pre-treatment of the demethanizer feed stream is another optimization performed in the second simulation. As previously done for the simulation shown in Figure. 5.1(B), the heating and cooling duties for the remaining heat exchangers and for the overall system were determined in the case of incorporating heat integration. Figure 5.1. (A) Aspen Plus (V10) simulation of a DNMC scale-up reaction without incorporating heat integration. (B) Aspen Plus (V10) simulation of a DNMC scale-up reaction incorporating heat integration based on HEN design and pinch analysis. 130 5.3 Results and Discussion 5.3.1 Structural analysis of Fe/SiO2 catalyst The XRD pattern in Figure 5.2(A) shows the crystalline phase of SiO2 in the Fe/SiO2 sample. Neither Fe nor FeOx diffraction peaks were observed in the XRD pattern, hinting that only very small amount of Fe was present in the catalyst. Figure 5.2(B) shows the N2 adsorption/desorption isotherm of the Fe/SiO2 sample. The surface area of the catalyst was found to be 0.38 m2·g-1. (A) (B) 2.0 1.5 1.0 0.5 0.0 20 30 40 50 60 70 80 0.0 0.2 0.4 0.6 0.8 1.0 2q (Degree) P/P o Figure 5.2. (A) XRD pattern and (B) N2 adsorption/desorption isotherm of Fe/SiO2 catalyst used for catalytic wall reactor in DNMC reaction. 5.3.2 Performance of catalytic wall reactor in DNMC reaction Our study on the millisecond catalytic wall reactor originated from a critically low surface area of Fe/SiO2 catalyst to activate DNMC in the fixed-bed reactor (Figure 5.3(A)). By fixing the catalyst bed length with quartz-balanced particles, the surface area of Fe/SiO2 catalyst was varied from 0 to 0.15 m2, equivalent to 0 to 100 wt% mass percentage of Fe/SiO2 particles in the catalyst bed. At the tested condition, CH4 conversion was < 1.0% in the blank reactor, increased to 5.7% in the reactor packed with quartz particles, and then 131 Intensity (a.u.) N2 volume adsorbed (cm 3 g-1) increased to 10.3% by using 0.036 m2 (or 25 wt%) Fe/SiO2 in catalyst bed, and maintained at ~11.0% with further increase in Fe/SiO2 catalyst quantity. The coke selectivity was the highest in the fixed-bed reactor packed with quartz-balance particles but decreased with increasing Fe/SiO2 usage in the catalyst bed. Clearly, a critically small active surface area of Fe/SiO2 catalyst is sufficient to enable the DNMC reaction. The further increase in Fe/SiO2 catalyst usage did not increase methane conversion which suggests that the DNMC is not an exclusive heterogeneous surface reaction. Consistent with report by Bao et al.[99], a mixed heterogeneous-homogeneous reaction network is expected in DNMC over Fe/SiO2 catalyst in our experimental conditions. The requirement for small amount (i.e. surface area) of Fe/SiO2 catalyst in DNMC suggests the potential to develop millisecond catalytic wall reactors comprised of a catalyst coating layer on the reactor wall that offers equivalent catalyst surface area to that in the fixed-bed reactor. The exemplary calculation shows that a reactor tube with inner diameter of 4.3 mm and length of 228.6 mm offers 0.036 m2 of surface area, same as that of 25 wt% Fe/SiO2 mixed with 75wt% quartz-balance particles in the catalyst bed in Figure 5.3(A). We, therefore, explored for the first time the development of Fe/SiO2 millisecond catalytic wall reactor for DNMC. The manufacturing process includes the curving of softened straight quartz tube into “U” shape, loading of Fe/SiO2 into U-shaped quartz tube channel, heating of both Fe/SiO2 catalyst and quartz tube to melting temperature, and finally discharging of the catalyst residue that was not melted into the reactor wall. The low cost and abundance of catalyst materials imply the capability of fabricating the catalytic wall reactor using commercial and scale-up processes such as extrusion, which produces active surfaces on both sides of the reactor in an economic fashion. Figure 5.3(B) demonstrates 132 the structure and catalytic functionalities of the millisecond catalytic wall reactor. It consists of a smaller diameter catalytic wall reactor enclosed in a larger diameter housing tube. The inner wall can be firstly used for the DNMC reaction. After an optimal reaction period and coke formation quantity, the DNMC reaction is switched to the opposite side while the inner tube is supplied with air for coke combustion and catalyst regeneration. From now on, the two reactions, endothermic DNMC and exothermic coke combustion, are swapped periodically and thus, achieving autothermal operation of DNMC. a 100 30 (B) Coke 25 80 Naphthalene Toluene Benzene Ethane Ethylene 20 60 Acetylene 15 40 10 20 5 0 0 0 0 0.014 0.036 0.071 0.15 Catalyt active surface area (m2) c (100 30 * D) 60 30 Acetylene Ethylene Ethane Benzene Toluene Naphthalene Coke CH4 Conversion 25 50 25 80 Coke 20 40 20 Naphthalene 60 Toluene Benzene Ethane Ethylene 15 30 15 Acetylene 40 10 20 10 20 5 10 5 0 0 0 0 (i) (ii) (iii) (iv) 0 5 10 15 20 25 30 35 40 45 50 Catalyst/reactor setting condition Time on Stream (Hours) 133 Product selectivity (C-atom basis, %) Product selectivity (C-atom basis, %) Blank Reactor Catalytic Wall Reactor Product Selectivity (C-atom basis, %) Methane Conversion (%) Methane Conversion (%) Methane Conversion (%) Figure 5.3. (A) Methane conversion and product selectivity versus catalyst surface area (or mass) in a fixed-bed reactor with fixed catalyst bed length by quartz balance particles when Fe/SiO2 catalyst quantity was varied; (B) Schematic of DNMC and coke combustion on opposite sides of catalytic wall reactor for autothermal operation of DNMC; (C) Methane conversion and product selectivity in different catalyst/reactor settings: (i) blank quartz reactor, (ii) Fe/SiO2 catalyst packed in quartz reactor, (iii) catalytic wall reactor coated with Fe/SiO2 catalyst and (iv) Fe/SiO2 catalyst packed in catalytic wall reactor; (D) Long-term stability test of DNMC reaction in catalytic wall reactor. (Reaction temperature = 1273 K, total gas flow rate = 20 mL min-1, CH4:N2 = 9:1, 1 atm pressure, Fe concentration in Fe/SiO2 = 0.075wt%) (Error within ±0.3 %) We first tested the performance of the Fe/SiO2 catalytic wall reactor at 1273 K and at 20 mL min-1 gas flow rate (10% N2 internal standard) and compared it to the DNMC reaction in non-catalytic quartz reactor, a fixed-bed quartz reactor packed with 0.375 g Fe/SiO2 catalyst, and a catalytic wall reactor loaded with 0.375g Fe/SiO2, respectively. In sequence, these four reactor/catalyst settings presented 0.8%, 7.9%, 11.3% and 11.0% methane conversions (Figure 5.3(C)). It confirms that the Fe/SiO2 catalyst was successfully incorporated into the quartz tube wall, matching our expectation on the millisecond catalytic wall reactor. These four settings also have similar product selectivity except for blank quartz tube, which should be caused by the difference in methane conversions. Coke formation follows a steady state rate in the catalytic wall reactor (Figure 5.7). This data suggests that a homogeneous gas phase reaction might play a dominant role after DNMC initiation by the heterogenous catalyst surface. The long-term stability of the catalytic wall reactor for DNMC was tested by running the reaction at 1273 K for 50 hours. Methane conversion was kept stable at ~11.3%, C2 (30.3%), benzene (21.2%), toluene (6.6 %) and naphthalene (32.4%) selectivity remained constant, and the total selectivity to these products were kept at >91.0% with ~ 9.0% of coke selectivity (Figure 5.3(D)). The combination of high methane conversion, high product selectivity and excellent stability 134 in the catalytic wall reactor is undeniably remarkable. Control experiments were performed by running DNMC reactions in a non-active quartz reactor at 1273 K at total gas flow rate range of 10-30 mL min-1. The result showed that methane conversion was kept below 2% when there was no active species deposited onto the wall of the reactor (Figure 5.4). 5 4 3 2 1 0 10 15 20 30 Total gas flow rate (mL min-1) Figure 5.4. Methane conversion and product selectivity of DNMC reaction in non-active quartz tube as a function of total flow rate at 1273 K (CH4:N2 = 9:1, 1 atm pressure). Methane conversion was kept below 2% when there was no active species deposited onto the wall of the reactor. (Error within ±0.2 %) The methane conversion, C2+ selectivity and yields, and coke yield at various temperatures and gas flow rates were measured in the catalytic wall reactor (Figure 5.5). Methane conversion increased proportionally with temperature and inversely with gas flow rate (Figure 5.5(A)). The C2+ selectivity (Figure 5.5(B)) has opposite dependence, compared to methane conversion on temperature and gas flow rate while C2+ yield shows 135 CH4 conversion (%) (A) (B) CH Conversion (%) C2+ Selectivity (%)4 50 0.000 50 40.00 45 10.00 45 6600..400 40 20.00 40 35 30.00 35 8800..400 40.00 30 30 1100.40 50.00 25 25 20 20 15 15 10 10 1240 1260 1280 1300 1320 1340 1360 1240 1260 1280 1300 1320 1340 1360 Temperature (K) Temperature (K) (C) C2+ Yield (%) (D) Coke Yield (%) 50 0.000 50 0.000 45 5.000 45 5.000 10.00 10.00 40 15.00 40 15.00 35 20.00 35 20.00 25.00 30 25.0030 30.00 30.00 25 25 20 20 15 15 10 10 1240 1260 1280 1300 1320 1340 1360 1240 1260 1280 1300 1320 1340 1360 Temperature (K) Temperature (K) Figure 5.5. (A) Methane conversion; (B) C2+ selectivity; (C) C2+ yields; (D) coke yield, respectively, as a function of reaction temperature and feed gas flow rate. (C2+ selectivity and yield and coke yield are calculated from the carbon-atom basis.) (Error within ±0.5 %) 5.3.3 Quantification of amount of coke in catalytic wall reactor The amount of coke formed in the catalytic wall reactor was determined by three methods: (i) weight-difference, (ii) TGA and (iii) TPO. For the weight difference method, before loading the reactor to the reactor system, the weight of the clean catalytic wall reactor was measured. After the DNMC reaction at each reaction condition was done, the weight of the catalytic wall reactor with now coke deposited on it was measured again. The difference in weight of the catalytic wall reactor was regarded as the amount of coke 136 Total Flow Rate (mL/min) Total Flow Rate (mL/min) Total Flow Rate (mL/min) Total Flow Rate (mL/min) formed during the DNMC reaction. One example of coke formation rate quantified using weight difference method is shown below. Method 1: Weight-difference method Sample calculation of coke formation rate using weight-difference method: Reaction condition: 1273 K, 20 mL min-1, total reaction time = 4 hours Mass of reactor before DNMC reaction: 19.6837g Mass of reactor after DNMC reaction: 19.7002g Average coke formation rate = I;.›9O%ŒAI;.Oœ@›Œ = 7.81 ×10-6 mole min-1 %)9 šˆp‡∗I% Œ/š1ž Method 2: TGA The TGA profile of the Fe/SiO2 powder obtained from setting (ii) in Figure 5.3(C) of the main text is shown in Figure 5.6(A). The weight loss in the temperature range of 835 K to 945 K in Figure 5.6(A) reveals the burning-off of the coke deposited on Fe/SiO2 catalyst. A total weight loss of 2.86 % was observed on the catalyst after 4 hours of DNMC reaction, which is equivalent to 8.51 ×10-6 mole min-1 of coke formation rate. Method 3: TPO TPO of the catalytic wall reactor was performed to check the feasibility of weight-difference method to quantify coke. Figure 5.6(B) shows the TPO profiles of catalytic wall reactor after 4 hours of DNMC reaction while Table 5.1 presents the amount of O2 consumed and CO and CO2 generated during the TPO process. From the TPO results, the coke formation rate in the catalytic wall reactor over the course of 4 hours during DNMC reaction was ~ 7.60 ×10-6 mole min-1 (average of O2 consumption rate and CO2 and CO formation rates). Since the coke formation rate determined from TPO method is similar to the coke formation rate determined from weight-difference method (7.81 ×10-6 137 mole min-1), we employed the weight-difference method in all our catalytic performance data analysis later on. (A) (B)100 12002.86% CO2 CO CO2 CO2 90 O2 H2O 1000 Temperature 80 800 CO 70 600 60 H O 4002 50 400 600 800 1000 1200 0 50 100 150 Temperature (K) Time (min) Figure 5.6. (A) TGA curve of coke formed on Fe/SiO2 powder after TOS = 4 hours of DNMC reaction. (B) TPO profile of catalytic wall reactor after TOS = 4 hours of DNMC reaction (Reaction temperature = 1273 K, total gas flow rate = 20 mL min-1, CH4:N2 = 9:1, 1 atm pressure). The coke formation rate determined from TPO method (7.60 ×10-6 mole min-1) is similar to the coke formation rate determined from weight-difference method (7.81 ×10-6 mole min-1). Table 5.1. Amount of O2 consumed and products formed during TPO process. Gases Concentration Consumption/Formation rate over TOS = (moles) 4 hours (moles/min) O2 1.80 x 10-3 7.48 x 10-6 CO2 1.85 x 10-3 7.72 x 10-6 CO 4.70 x 10-8 1.96 x 10-10 The coke formation rate in the Fe/SiO2-based catalytic wall reactor was also studied over the course of 50 hours and the result is shown in Figure 5.7. The coke formation rate increased significantly from TOS = 0 hour to TOS = 0.25 hours. After 0.25 hours, the coke formation rate started to decrease until it reached a plateau starting from TOS = 1.0 hour onward. 138 Weight Loss (%) Intensity (a.u.) Temperature (K) (A) 100 30 (B) 20 80 25 15 20 60 15 10 40 10 5 20 5 0 0 0 0.25 0.5 1.0 1.5 2.0 4.0 10.0 48.0 0.25 0.5 1.0 1.5 2.0 4.0 10.0 48.0 Time on Stream (Hours) Time on Stream (Hours) Figure 5.7. (A) Methane conversion and product selectivity and (B) coke formation rate as a function of time on stream at 1273 K (Total flow rate = 20 mL min-1, CH4:N2 = 9:1, 1 atm pressure). The coke formation rate increased significantly from TOS = 0 hour to TOS = 0.25 hours. After 0.25 hours, the coke formation rate started to decrease until it reached a plateau starting from TOS = 1.0 hour onward. (Error within ±0.3 %) The amount of time required for the coke to completely fill up the reactive reactor at each reaction conditions was estimated, and the results are shown in Table 5.2. The coke growth rate was assumed to be uniform throughout the wall of the catalytic wall reactor. One example of time required calculation is shown below. Sample calculation of time required to completely fill up the reactor: Reaction temperature: 1303 K Gas flow rate: 30 mL min-1 Volume of catalytic wall reactor: 8.11 cm3 Density of coke: 2.267 g cm-3 Mass of coke formed in 2.5 hours: 0.0318 g Volume of coke formed in 2.5 hours: 9.9@Iœ Œ M = 0.013 cm3 %.%O› Œ/„š Therefore, the time taken to completely fill up the catalytic wall reactor is 1445 hours. 139 Product Selectivity (%) Coke formation rate (µmol min-1) Methane Conversion (%) 5.3.4 Energy balance analysis in autothermal catalytic wall reactor The energy balance analysis was performed to explore the techno feasibility of autothermal catalytic wall reactor for DNMC. The analysis was based on standard heat of reaction (∆H°) calculation for each product formation reaction equation at each reaction temperature. Standard heat-capacity (∆C°Ÿ) was assumed to be independent on temperature. Energy required for the high temperature endothermic DNMC is a function of methane conversion and product formation, and is strongly depended on reaction condition (temperature and gas flow rate). Energy released from coke combustion is a function of coke formation, which is also dependent on reaction temperature and gas flow rate. One sample calculation of heat required for DNMC and heat released by coke combustion reactions is shown below: Sample calculation Reaction temperature: 1323 K Gas flow rate: 20 mL min-1 Methane in feed: 1 mole basis Methane Acetylene Ethylene Ethane Benzene Toluene Naphthalene Coke Conversion Selectivity Selectivity Selectivity Selectivity Selectivity Selectivity Selectivity (%) (%) (%) (%) (%) (%) (%) (%) 33.9 3.8 12.7 0.77 16.9 1.05 39.7 25.1 Product formation reaction equation from DNMC: (1) CH4 à 1/2 C2H2 + 3/2 H2 (2) CH4 à 1/2 C2H4 + H2 (3) CH4 à 1/2 C2H6 + 1/2 H2 (4) CH4 à 1/6C6H6 + 3/2 H2 (5) CH4 à 1/7C7H8 + 10/7 H2 (6) CH4 à 1/10C10H8 + 8/5 H2 (7) CH4 à C + 2 H2 140 We assumed that the coke is mainly comprised of carbon, so the product formation reaction equation from coke combustion in air is: (1) C + O2 à CO2 Thermal properties of gases: Gas Heat capacity Enthalpy of formation (J mol-1 K-1)a (kJ mol-1) (298 K, 1 atm)a H2 29.1 0 O2 29.3 0 CH4 35.7 -74.5 C2H2 44.2 226.9 C2H4 43.7 52.51 C2H6 52.7 -84.7 C6H6 (g) 81.7 82.9 C7H8 (g) 104.4 50.0 C10H8 (g) 133.9 150.6 CO2 37.2 -393.52 Coke 20.8 0 a Source: National Standard of Institute and Technology (NIST) Chemistry WebBook. The heat of reaction was calculated based on the following equation: ∆H° = n ∆H1 + ∆C19 Ÿ(T − T9)¢ (5.1) where n (mole) is the number of moles of product, ∆H19 (kJ mol-1) is the standard enthalpy of formation, ∆C1Ÿ (J mol-1 K-1) is the standard heat capacity, T (K) is the reaction temperature and T9 (K) is the temperature at standard condition. For reaction (1) from DNMC, ∆H1I = n ∆H1 19 + ∆CŸ(T − T9)¢ = l@@.;n ∗ l @.œ n £¤I ∗ 226.9 + 0 − (−74.5)¥ ∗ 1000 + ¤I ∗ 44.2 + I ∗ 29.1 − I99 I99 % % % 35.7(1323 − 298)¥¦ /1000 = 3.82 kJ molAI 141 Same calculation steps were applied to the rest of the reaction equations for DNMC as well as the reaction equation for coke combustion. The results for DNMC reaction and the corresponding coke combustion is summarized in Table 5.2. CH4 conversion CH4 conve rsion (%) 140 0 10 20 30 40 50 6035 120 30 100 25 80 20 60 15 40 10 20 Autothermal 5 operation 0 0 0 10 20 30 40 50 60 Heat supplied for DNMC (kJ mol-1 ) Figure 5.8. Energy input for DNMC and energy output by coke combustion at their corresponding methane conversion and coke yield. The dashed line represents the energy input and output balanced from both reactions. The shaded circle indicates the operation window of autothermal DNMC in catalytic wall reactor. The calculations showed that the balance between heat absorbed during DNMC and heat released during coke combustion can be achieved when DNMC is run at ~33.9% methane conversion condition with 25.4% C2+ yield (Figure 5.8). A recent agreement framework analysis for DNMC by Maravelias et al.[309] suggests that the economically feasible DNMC is achievable at >25% methane conversion, <20% coke formation and low catalyst cost. The DNMC in our catalytic wall reactor sufficiently meet these targets, in addition to the autothermal operation viability. 142 Absolute heat released for coke combustion (kJ mol-1 ) Coke Yield (%) Table 5.2 | Summary of DNMC reaction and the corresponding coke combustion at different reaction conditions. Reaction Condition CH4 C2+ Coke Heat supplied Heat released Time required Temperature (K) Total gas flow conversion (%) yield yield for DNMC (kJ from coke to fill reactor rate (mL min-1) (%) (%) mol-1) combustion (kJ up by coke mol-1) (Hours) 10 1.7 1.7 0 2.7 0 ∞ 1223 15 0.8 0.8 0 1.3 0 ∞ 20 0.7 0.7 0 0.8 0 ∞ 30 0.5 0.5 0 0.7 0 ∞ 10 9.5 8.3 1.2 10.3 -4.7 6050 1253 15 7.4 6.8 0.6 8.2 -2.3 8187 20 4.2 3.9 0.3 5.3 -1.2 11393 30 0.9 0.9 0 1.7 0 ∞ 10 20.4 16.4 4.0 22.1 -16.4 1580 1273 20 11.2 10.1 1.1 12.6 -4.3 3402 30 2.5 2.3 0.1 3.4 -0.6 15885 40 1.3 1.3 0 2.4 0 0 20 26.2 22.0 4.3 28.0 -17.3 743 1303 30 16.6 15.0 1.6 18.2 -6.6 1445 40 13.0 11.9 1.0 14.5 -4.3 1782 50 5.4 5.2 0.2 6.8 -0.8 6967 20 33.9 25.4 8.5 34.4 -34.8 362 1323 30 25.1 20.4 4.8 25.7 -19.1 488 40 20.8 17.9 3.0 22.4 -13.8 517 50 18.7 17.3 1.5 19.2 -5.5 984 20 40.2 22.6 17.7 44.3 -71.9 190 1343 30 30.1 18.3 12.0 32.5 -48.8 223 40 24.8 16.2 8.8 26.6 -35.6 263 50 21.4 15.0 6.4 22.8 -26.1 355 20 49.0 19.3 30. 0 55.6 -121.9 98 1363 30 40.0 18.9 21.2 44.4 -86.5 113 40 33.2 16.6 16.7 36.9 -68.0 147 50 29.1 15.8 14.2 37.3 -57.9 198 5.3.5 Model simulations for autothermal DNMC-coke combustion process concept The process simulation using Aspen Plus tools was performed to evaluate the practical implications of the conceptual catalytic wall reactor. Figure 5.9 presents the schematic of a complete DNMC and coke combustion processes based on our autothermal catalytic wall reactor architecture. To effectively simulate this process, a yield reactor was incorporated using conversion and selectivity results obtained from the experimental data, and coke combustion was modeled in a separate reactor. Heat integration was applied throughout the reaction system as a substitute for the autothermal catalytic wall reactor, as this exact architecture was not achievable via simulation. 143 Figure 5.9. Process flowsheet of autothermal DNMC in millisecond catalytic wall reactor coupling both endothermic DNMC and exothermic coke combustion on opposite sides of the reactor. The process simulation using Aspen Plus tools was performed to evaluate the practical implications of the catalytic wall reactor. Figure 5.9 presents the flowsheet for DNMC, comprising of endo- and exothermic reactions as well as product separations. A heat exchanger was incorporated to utilize heat released from coke combustion to raise the feed stream to reaction temperature, mimicking the autothermal process. The Aspen Plus Utilities Object Manager and Economic Solver provided estimates for externally-supplied heating and cooling duties and costs (Tables 5.3 and 5.4). These calculations demonstrate a six-fold reduction in supplied energy costs from the autothermal process relative to a conventional system. The DNMC reaction produces multiple industrially-valuable chemicals and fuels – hydrogen, ethylene, benzene and naphthalene – whose production rates are converted into retrievable prices (Tables 5.5 and 5.6). These results, in combination with the low costs of methane feedstock and reactor material, demonstrate 144 that DNMC in a catalytic wall reactor is an economically feasible and transformative technology for shifting the petrochemical sector to natural gas feedstock in industry. Table 5.3. Heating and cooling duties for heat exchangers with and without using heat integration and their hourly costs, respectively. Unit Heating/Cooling duty Utility Utility Cost ($/hr) (x 106 kJ/hr) COOL1 -172.5 Cooling Water 36.44 COOL2 -35.08 Cooling Water 7.43 Without heat COOL3 -6.481 Liquid Propane 17.75 integration HEAT1 10.35 High Pressure Steam 389.59 HEAT2 2.145 High Pressure Steam 80.89 HEAT3 10.20 Low Pressure Steam 19.28 With heat COOL1 -16.5368 Cooling Water 32.83 integration COOL2 -19.4033 Liquid Propane 59.22 Table 5.4 Hourly duties and costs for system operation with and without heat integration. Without heat With heat integration integration Total heating duty (x 106 kJ/hr) 198 0 Total cooling duty (x 106 kJ/hr) 214 177 Net duty (x 106 kJ/hr) -15.5 -177 Total heating cost flow ($/hr) 489.76 0 Total cooling cost flow ($/hr) 61.62 93.72 Total cost ($/hr) 551.38 93.72 Comparing the results obtained from each simulation, the use of heat integration has a significant impact on the total cost of heating and cooling duties within the system. Due to the lack of further heating requirements in the system, we are able to sell all hydrogen gas produced in the process for a profit and avoid burning the hydrogen for an additional heat source. Finally, the potential profitability of this reaction is assessed, with some results in Table 5.5. Four products - ethylene, benzene, naphthalene and hydrogen - are valuable and can be sold for a substantial profit due to the low cost of obtaining natural gas. Additional electricity costs are also incorporated for powering some of the equipment operating under highly energy-intensive conditions. Table 5.6 provides an overall cost analysis of the annual plant operation expenses for utility and raw material use and the 145 annual credit received from selling valuable reaction products. However, this analysis does not account for the costs of equipment, plant operators, separation of the desirable products, or storage. As such, the hourly cost of plant operation would be significantly higher than that determined by utility of heat exchanger costs alone. Table 5.5 Current costing and hourly production rates for feed and product species in the DNMC reaction. Methane Price ($/cuft) .00288 Methane Cost ($/hr) -2632 Hydrogen Price ($/kg) 1.39 Hydrogen Cost ($/hr) 3658 Ethylene Price ($/kg) 1.16 Ethylene Cost ($/hr) 1707 Benzene Price ($/kg) 0.73 Benzene Cost ($/hr) 1326 Naphthalene Price ($/kg) 0.52 Naphthalene Cost ($/hr) 4205 Table 5.6. Annual plant operational utility and raw material costs and sales credit for heat- integrated process Aspen model. Operation Cost Sales Credit Utility/Material Cost [USD/yr] Product Cost [USD/yr] Liquid propane 287,591 Hydrogen 32,044,080 Cooling water 518,767 Ethylene 14,953,320 Methane (feed) 23,056,320 Benzene 11,615,760 Electricity 4,127,712 Naphthalene 36,835,800 Catalyst 438,000 Total cost 28,428,390 Total Credit 62,448,960 The price of methane was obtained from the Bloomberg energy database online as was reported on July 26, 2018 [310]. Ethylene prices were determined using 2018 projections from ICIS pricing models based on historical data through 2012 [311]. A linear correlation between crude oil and ethylene prices was observed and was used to determine the current price of ethylene based on the reported oil price of 69.65 USD/bbl in the morning on Monday, June 23, 2018 [310]. Benzene prices were determined using data from the ICIS analysis of CIF ARA prices in Europe in 2016 [312]. Naphthalene prices were 146 obtained using bulk costs for crude naphthalene on the Alibaba world trade website (7). The market value for hydrogen was determined using a correlation between the cost of natural gas and hydrogen prices using historical data and projections assembled by the US Department of Energy in 2012. The hydrogen cost presented here correlates to the natural gas cost of 3 USD/MMBtu, which is representative of current market conditions [313]. 5.4 Conclusion of Chapter 5 In summary, a millisecond catalytic wall reactor was successfully demonstrated for the first time for DNMC reaction. The reactor is based on coating of Fe/SiO2 catalyst onto the wall of a quartz tube. The DNMC reaction was optimized to reach methane conversion of > 25%, coke selectivity of < 10%, and stable long-term operation. To take advantage of coke formation, once DNMC is started for an optimal time, the DNMC reaction can be switched to the other side of the reactor accompanying with coke removal reaction by controlled combustion and catalyst regeneration. The periodical swapping between DNMC and coke combustion on opposite sides of the catalytic wall reactor can reach autothermal operation of DNMC, which is particularly attractive for stranded natural gas resources. In combination with low cost of Fe/SiO2 catalyst, the DNMC in catalytic wall reactor is an economically feasible and transformative technology for shifting the petrochemical sector to the natural gas feedstock in industry. 147 Chapter 6: Tailoring the Selectivity in Direct Non-Oxidative Methane Conversion: Effects of H2 Co-feed in the Presence of SrCe0.8Zr0.2O3−δ Perovskite in Fe/SiO2 Catalyst 6.1 Introduction One strategy to manipulate the product selectivity towards light hydrocarbons in DNMC is to co-feed hydrogen in the methane stream. For example, the increase in H2 concentration in the reaction over Mo/ZSM-5 catalyst led to reduction in coke in the catalyst and increase C2H4 selectivity due to coke hydrogenation [314-319]. However, this is at an expense of lower methane conversion, due to a shift in the thermodynamic equilibrium. An alternative strategy to tailor the product selectivity is to use the hydrogen permeable membrane reactor. In the DNMC reaction, H2 is the only by-product, whose yield reached up to ~50% in the product effluent[320] , directly influencing the kinetics and thermodynamics of DNMC chemistry. According to the Le Chatelier’s principle, the removal of hydrogen produced in the reaction can shift the thermodynamic equilibrium to higher methane conversion. This can be achieved by utilizing a membrane reactor that allows in-situ hydrogen removal, thus leading to an intensive research effort on determining a hydrogen permeable material that is thermally and chemically stable at high temperature (~1273 K), and ionically and electronically conductive[109, 110, 299]. Metal-based membranes such as Pd and Pd alloy-based membranes have been used in DNMC reactions over Mo/ZSM-5 catalysts. For example, Larachi and group have developed a Pd-Ag/porous stainless steel membranes for direct methane aromatization reaction over Ru-Mo/HZSM-5 at temperatures up to 973 K[111]. Their catalytic performance tests demonstrated that the Pd alloy-based membrane improve hydrogen permeation and result in a significant increase in methane conversion at 973 K. When the 148 DNMC was performed at 973 K, maximum methane conversion of 17% was achieved even though it was accompanied by fast deactivation. On the other hand, Morreale and co- workers have fabricated Pd membranes packed with Mo/HZSM-5 catalysts for non- oxidative methane aromatization reaction[112]. A significant improvement in methane conversion (~10%) and benzene selectivity (~80%) were achieved through in-situ H2 removal by the membrane. Perovskites-based membranes, which exhibit mixed ionic- electronic conductivity, have also been used in DNMC reactions over Mo/ZSM-5 catalysts. For instance, Iglesia and group have manufactured SrCe9.;91.0% with ~ 9.0% of coke selectivity in the Fe/SiO2 catalyst (Figure 6.2(B)). However, the product selectivity in the SCZO perovskite changed with time-on-stream. For example, coke formation increased with increasing time-on-stream while aromatics products decreased with increasing time-on- stream (Figure 6.2 (C)). Coke selectivity reached up to ~ 60% when pure SCZO material was used. The significantly higher coke selectivity in SCZO material than Fe/SiO2 catalyst hints that this material favored surface reaction over both surface and gas phase reactions, as demonstrated by Fe/SiO2 catalyst. It has been proposed that coke in DNMC can be formed via two pathways: (i) dehydrogenation of C2Hy intermediates in series with aromatics formation step and (ii) dehydrogenation of methane (CHx) in parallel with C-C bond formation step[321]. During the reaction, methane in the gas phase first adsorbs on the SCZO surface. The surface adsorbed methane species then undergo further 156 dehydrogenation to form coke (C). The carbonaceous deposits are detrimental to the SCZO because the active sites on the SCZO are blocked and this eventually leads to catalyst deactivation, as shown in Figure 6.2(C). The low aromatics selectivity on SCZO sample further confirms the proposed surface reaction. Aromatic formation steps through C2Hy intermediates in the gas-phase were restricted but dehydrogenation of methane was favored. (A) 30 (B) 100 Fe/SiO2 catalyst SCZO perovskite 25 Coke 80 Naphthalene Toluene Benzene Ethane 20 Ethylene Acetylene 60 15 40 10 20 5 0 0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 Time-on-Stream (Hours) Time-on-Stream (Hours) (C) 100 Coke Naphthalene Toluene Benzene 80 Ethane Ethylene Acetylene 60 40 20 0 0 1 2 3 4 5 6 7 8 9 10 Time-on-Stream (Hours) Figure 6.2. CH4 conversion for Fe/SiO2 catalyst and SCZO perovskite (A) and product selectivity of Fe/SiO2 catalyst (B) and SCZO perovskite (C) in long-term stability test of DNMC reaction (temperature = 1273 K, space velocity= 3200 mLg-1h-1). (Error within ±0.4 %) 157 Product Selectivity (%) Methane Conversion (%) Product Selectivity (%) 6.3.2.2 DNMC over a mixture of SCZO and Fe/SiO2 materials. The catalytic activity and product selectivity of the SCZO/Fe/SiO2 mixture at different packing modes in the reactor in DNMC reactions were examined. Figure 6.3 shows the methane conversion and product selectivity when SCZO powder was placed on top and at the bottom of the Fe/SiO2 catalyst, as well as when both SCZO and Fe/SiO2 powder were mixed. The overall methane conversion was slightly lower when SCZO was placed on top of the Fe/SiO2 layer compared to when SCZO was placed at the bottom of the Fe/SiO2 layer. Methane conversion was lower in the first case because methane reacted with the SCZO material first before reaching Fe/SiO2 powder. As discussed above, the SCZO material promoted coke formation more easily than Fe/SiO2 powder. Therefore, coke formed on the SCZO material tend to block the active sites on the Fe/SiO2 catalyst located at the bottom layer, preventing some unreacted methane in the SCZO layer from reacting further on the Fe/SiO2 catalyst. This explains the lower methane conversion in this case. In terms of product selectivity, ethylene and acetylene selectivity were higher when SCZO was arranged at the top layer. Since methane reacted on SCZO material through surface reaction, the dehydrogenation of methane on the surface formed not only coke but also C2 products. Aromatic products were proposed to form in the gas phase homogeneously through a series of cyclization reactions. The aromatic selectivity was lower in the case when SCZO was placed at the top layer because less methane was reacted on Fe/SiO2 catalyst to form reaction intermediates for gas phase cyclization reactions. In the case where SCZO was placed at the bottom of the Fe/SiO2 catalyst, the overall methane conversion and coke selectivity was higher. As shown in Figure 6.2(A), Fe/SiO2 catalyst did not show significant deactivation over the course of 10-hour reaction. 158 (A) 20 (B) 100 SCZO (Top) SCZO (Bottom) Coke SCZO & Fe/SiO2 mixture Naphthalene80 Toluene Benzene 15 Ethane Ethylene Acetylene 60 10 40 5 20 0 0 0 1 2 3 4 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Time-on-Stream (Hours) Time-on-Stream (Hours) (C) 100 (D) 100 Coke Coke Naphthalene Naphthalene 80 Toluene 80 Toluene Benzene Benzene Ethane Ethane Ethylene Ethylene Acetylene Acetylene 60 60 40 40 20 20 0 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Time-on-Stream (Hours) Time-on-Stream (Hours) Figure 6.3. CH4 conversion and product selectivity in DNMC over Fe/SiO2 and SCZO mixture with different packing modes under the same reaction conditions (0.1875 g SCZO, 0.1875 g Fe/SiO2, temperature = 1273 K, space velocity= 3200 mLg-1h-1). (B) Product selectivity for SCZO packed on top of Fe/SiO2, (B) Product selectivity for SCZO packed below Fe/SiO2, and (C) Product selectivity for SCZO and Fe/SiO2 well mixed. (Error within ±0.4 %) Therefore, when methane reacted with the Fe/SiO2 catalyst layer first, C2+ products were formed through both surface and gas-phase reactions. However, when these products (including C2 and aromatics) encountered the SCZO layer at the bottom, they underwent 159 Product Selectivity (%) Methane Conversion (%) Product Selectivity (%) Product Selectivity (%) surface reactions on the SCZO material to form coke even though methane continued to react at the top Fe/SiO2 layer. The higher coke selectivity and lower C2+ selectivity in such SCZO/ Fe/SiO2 arrangement verify the proposed surface kinetics of SCZO, and both surface and gas phase kinetics of Fe/SiO2 catalyst. In addition, comparing both catalyst arrangements in Figures 6.3, methane deactivated faster when the SCZO material was put on the top layer. As explained earlier, methane reacted with SCZO first to form coke, which in turn led to catalyst deactivation. As for the case when both SCZO and Fe/SiO2 powder were physically mixed, methane conversion showed only very slight deactivation from ~11% to ~9.5% over the course of 3.5 hours. Coke selectivity also increased at a slower rate compared to the previous two cases. Such catalyst arrangement allowed methane to react with SCZO and Fe/SiO2 powder at the same probability. The coke formed through surface reaction on SCZO material again blocked the active sites of the Fe/SiO2 catalyst, causing slight deactivation on the overall methane conversion. Overall, the C2+ selectivity was also more stable compared to the previous two cases. 6.3.3 DNMC in the presence of H2-cofeed in methane stream 6.3.3.1 Effects of H2-cofeed on DNMC over Fe/SiO2 and SCZO materials. Figure 6.4 shows the methane conversion and product selectivity in DNMC reactions over these materials at different H2 co-feed amount. Methane conversion of Fe/SiO2 sample decreased with increasing H2 co-feed concentration (Figure 6.4(A)), consistent with Le Chatelier’s principle. Product selectivity shifted from heavy aromatics to lighter hydrocarbons while coke formation decreased when more H2 was added to the reaction, consistent with previous reports[82, 314, 315, 318, 319, 322, 323]. However, H2 addition did not show 160 significant detrimental effect on methane conversion for SCZO sample than for Fe/SiO2 sample, except when the H2 co-feed amount was high enough, as shown in Figure 6.4(B). (A) 100 30 (B) 100 30 Coke Coke Naphthalene Naphthalene 25 Toluene 80 Toluene Benzene 2580 Benzene Ethane Ethane Ethylene Ethylene Acetylene Acetylene 20 20 60 60 15 15 40 40 10 10 20 5 20 5 0 0 0 5 10 15 20 0 00 5 10 15 20 Fraction of H2 in Feed (%) Fraction of H2 in Feed (%) Figure 6.4. CH4 conversion and product selectivity over Fe/SiO2 catalyst (A) and SCZO powder (B) in a fixed-bed reactor at different hydrogen co-feed concentrations (temperature=1273 K, space velocity= 3200 mLg-1h-1, TOS = 1 hour). (Error within ±0.2 %) Methane conversion was kept at ~12.5% when no H2 was added to the reaction. When H2 co-feed concentration increased from 5% to 15%, methane conversions were maintained at ~11%. Unlike Fe/SiO2 catalyst, coke selectivity remained almost the same at around 52%-58% for SCZO material, except for 20% H2 co-feed concentration where the selectivity dropped more significantly. C2 selectivity, on the other hand, increased slightly with increasing H2 co-feed concentration. The catalytic behavior verifies that SCZO favored surface reaction by cleaving the C-H bond in methane to form C2 products and coke, instead of both surface and gas-phase reaction. Any H2 addition to the system will not affect the methane conversion since gas-phase reaction does not happen. We performed H2-TPD on the Fe/SiO2 and SCZO to study the H2-adsorption on these samples. For 161 Product Selectivity (%) Product Selectivity (%) Methane Conversion (%) Methane Conversion (%) comparison, we did H2-TPD for CeO2 sample. As shown in Figure 6.5(B), CeO2 has significant H2 desorption peak in the temperature range of 820-1200 K. On the contrary, both Fe/SiO2 and SCZO samples did not have such obvious H2-TPD peak. It should be mentioned that all the samples were measured following the same procedure and the same sample mass was used in each measurement. The H2-TPD data indicates that both Fe/SiO2 and SCZO samples did not react with H2 under H2-cofeed condition. Such data further confirms that SCZO promoted surface reaction since methane conversion has less effects on H2 concentration. Fe/SiO2 SCZO CeO2 CeO2 SCZO Fe/SiO2 400 600 800 1000 1200 Temperature (K) Figure 6.5. H2-TPD profiles of Fe/SiO2, SCZO and CeO2 (controlled) samples. 6.3.3.2 Effects of H2-cofeed on DNMC over Fe/SiO2 and SCZO mixtures. The effects of H2 addition on the catalytic activity and product selectivity of the Fe/SiO2 catalyst and SCZO material at different catalyst packing modes were also studied at different H2 co-feed concentrations. Figure 6.6 shows the methane conversion and product selectivity for different catalyst arrangements. In the physically mixed sample, 2 wt% of SCZO 162 Intensity (a.u.) powder was used as it was similar to the wt% of SCZO membrane material that came into contact with Fe/SiO2 catalyst in the membrane reactor used in our previous study. Again, increasing H2 co-feed concentration did not show significant adverse effects on methane conversion, regardless of catalyst arrangement, as methane conversions only decreased slightly. Coke selectivity also did not decrease significantly with increasing H2 co-feed concentration. In contrast, when only 2 wt% of SCZO was physically mixed with the Fe/SiO2 catalyst, methane conversion dropped significantly from ~11% to ~3% when H2 co-feed concentration increased from 0% to 20%. Coke selectivity also approached zero while heavy aromatics selectivity shifted to lighter hydrocarbons when more H2 was added to the reaction. However, a closer look at the data showed that methane conversion did not decreased as seriously when H2 co-feed concentration was increased from 5% to 10%. Such reaction condition is favorable as methane conversion can be maintained while excellent C2 selectivity can be achieved. (A) 100 20 (B)100 20 (C) 100 SCZO (top layer) SCZO (bottom layer) 2 wt% SCZO 20 Coke Coke Naphthalene Naphthalene Toluene Toluene Benzene Benzene Ethane 80 Ethane 80 Ethylene 80 Ethylene 15 Acetylene 15 15 Acetylene Coke 60 60 60 Naphthalene Toluene Benzene 10 10 Ethane 10 Ethylene Acetylene 40 40 40 5 5 5 20 20 20 0 0 0 0 0 0 0 5 10 15 20 0 5 10 15 20 0 5 10 15 20 Fraction of H2 in Feed (%) Fraction of H2 in Feed (%) Fraction of H2 in Feed (%) Figure 6.6. CH4 conversion and product selectivity in DNMC reaction over Fe/SiO2 catalyst and SCZO powder mixture with different catalyst arrangement (temperature=1273 K, space velocity= 3200 mLg-1h-1, TOS = 1 hour). (Error within ±0.2 %) 6.3.4 Coke behaviors-catalytic performances correlations of SCZO/Fe/SiO2 163 Product Selectivity (%) Product Selectivity (%) Product Selectivity (%) Methane Conversion (%) Methane Conversion (%) Methane Conversion (%) 6.3.4.1 Amount of coke deposition. To investigate the influence of the coke deposition on the SCZO/Fe/SiO2 catalyst mixture and its influences on the catalytic performances, the amount of coke formed on SCZO/Fe/SiO2 catalyst mixture after TOS of 3.5 h under DNMC conditions at different H2 co-feed concentrations were studied by TGA analysis. The TGA curves, shown in Figure 6.7, reveal a weight loss in the temperature range of 820–1050 K that corresponds to the burning-off of the coke deposits due to the oxidation of coke to CO and CO2. The coke deposition in these catalysts increased with increasing SCZO concentration. The TGA curves also shows that coke formation became less significant when higher H2 co-feed concentration was used in the reactions, consistent with coke selectivity results in Figures 6.5 and 6.6. In addition, the TGA curves demonstrated that lower temperature (820 K) was required to start burning off the coke formed on the SCZO sample while higher temperature (~980 K) was necessary to start burning off the coke formed on the Fe/SiO2 catalyst. Such behaviors suggest that loose coke was formed on SCZO sample while hard coke was generated on Fe/SiO2 catalyst. (A) 0 % H2 in feed (B) 5 % H2 in feed 10 % H i (C) 2 n feed 100 100 100 96 96 96 92 92 92 88 88 88 84 84 84 Fe/SiO Fe/SiO2 Fe/SiO2 2 5 wt% SCZO in Fe/SiO2 5 wt% SCZO in Fe/SiO 5 wt% SCZO in Fe/SiO2 2 SCZO SCZO SCZO 80 80 80 400 600 800 1000 1200 400 600 800 1000 1200 400 600 800 1000 1200 Temperature (K) Temperature (K) Temperature (K) 164 Weight Loss (%) Weight Loss (%) Weight Loss (%) (D) 15 % H2 in feed (E) 20 % H in 2 feed 100 100 96 96 92 92 88 88 84 Fe/SiO 842 Fe/SiO2 5 wt% SCZO in Fe/SiO2 5 wt% SCZO in Fe/SiO2 SCZO SCZO 80 80 400 600 800 1000 1200 400 600 800 1000 1200 Temperature (K) Temperature (K) Figure 6.7. TGA curves of coke formed in spent SCZO/Fe/SiO2 catalysts with different SCZO amounts in a fixed-bed reactor at different hydrogen co-feed concentrations after TOS of 3.5 h in DNMC reactions. 6.3.4.2 Types of coke deposition. Raman spectra were measured from spent SCZO/ Fe/SiO2 catalysts (pure Fe/SiO2, pure SCZO and 5 wt% SCZO in Fe/SiO2) after TOS of 3.5 h under DNMC conditions at different H2 co-feed concentrations, and the results are shown in Figure 6.8. Nearly no fluorescence background was detected in all these spectra. The Raman analysis confirmed the existence of two types of carbon structures. The spectra of all the spent catalysts are similar with two observed narrow peaks centered at 1320 cm- 1 and 1600 cm-1, respectively. The band at 1320 cm-1 is assigned to D band while the band at 1600 cm-1 is assigned to G band [324, 325]. D band represents disordered graphitic structure, amorphous carbon or polyaromatic type species, while G band represents pure graphite which involve out-of-phase intra-layer displacement in the graphene structure [326]. There is no significant shift in the two peaks as a function of SCZO concentration 0% H in feed 5% H in feed 10% H (A) 2 (B) 2 (C) 2 in feed Fe/SiO2 Fe/SiO2 Fe/SiO2 5 wt% SCZO in Fe/SiO2 5 wt% SCZO in Fe/SiO2 5 wt% SCZO in Fe/SiO2 SCZO SCZO SCZO 800 1000 1200 1400 1600 1800 800 1000 1200 1400 1600 1800 800 1000 1200 1400 1600 1800 Raman Shift (cm-1) Raman Shift (cm-1) Raman Shift (cm-1 ) 165 Weight Loss (%) Intensity (a.u.) Weight Loss (%) Intensity (a.u.) Intensity (a.u.) (D) 15% H2 in feed (E) 20% H2 in feed Fe/SiO2 Fe/SiO2 5 wt% SCZO in Fe/SiO2 5 wt% SCZO in Fe/SiO2 SCZO SCZO 800 1000 1200 1400 1600 1800 800 1000 1200 1400 1600 1800 Raman Shift (cm-1) Raman Shift (cm -1) Figure 6.8. Raman spectra of coke formed in spent SCZO/Fe/SiO2 catalysts with different surface area ratio in a fixed-bed reactor at different hydrogen co-feed concentrations after TOS of 3.5 h in DNMC reactions. and amount of H2 co-feed. Table 6.1 shows the ratio of D band to G band for all the samples at different H2 co-feed concentration. The lower ratio in pure SCZO samples at all studied H2 co-feed concentration suggests that more coke of order graphitic structure was formed on the sample while more coke of amorphous types was formed in the pure Fe/SiO2 catalyst after 3.5 hours of DNMC reaction. In addition, the relative intensities of both the D band and G band in pure Fe/SiO2 catalyst and 5 wt% SCZO sample became smaller with increasing hydrogen co-feed concentration, hinting that coke formations on this sample became lesser when more H2 was cofed. In contrast, the relative intensity of pure SCZO sample remained almost the same regardless of H2-cofeed concentration, and these results are consistent with DNMC performance results. It has been shown that the ratio of the intensity of the G band to D band depends inversely on the size of the graphite microcrystals in the samples[327, 328]. The intensity ratios of I(1600)/I(1320) in all the samples, regardless of the amount of hydrogen co-fed to the reactions, were less than unity, except for pure SCZO sample at 20% hydrogen feed. This phenomenon indicates that big crystallite size of coke was formed on the samples. In addition, the width of the intensity broadens with decreasing amount of hydrogen co-fed during the reaction, suggesting that 166 Intensity (a.u.) Intensity (a.u.) the coke formed during DNMC reaction is of less organized structure. The Raman results suggests that disordered/amorphous structure of coke does not contribute to catalyst deactivation while ordered coke aids in deactivation since there was no significant drop in methane conversion over the course of 10 hours during the reaction for Fe/SiO2 catalyst. A comparison of G Raman band between samples at 0% H2 co-feed and 20% H2 co-feed shows that the position of this band shifted to lower wavenumber when more H2 was added to the reaction. This implies that the addition of H2 decreased the crystallite size of graphite particles[327]. Table 6.1. Ratio of D band to G band determined from Raman spectroscopy analysis for Fe/SiO2 catalyst, SCZO material and 5 wt% SCZO material in Fe/SiO2 catalyst after 3.5 hours DNMC reaction at 1273 K and at different H2-cofeed concentrations. Catalyst H2-cofeed D to G band ratio concentration (%) 0 1.25 5 1.43 Fe/SiO2 10 1.39 15 1.68 20 1.34 0 1.30 5 1.11 SCZO 10 1.14 15 1.21 20 0.92 0 1.24 5 wt% SCZO 5 1.37 in Fe/SiO2 10 1.45 15 1.46 20 1.59 TPO data were collected on all the spent SCZO and Fe/SiO2 catalysts to study the ease of removal of the coke formed by combustion and the types of coke formed on the samples after 3.5 h DNMC reactions. The combustion products in all the spent catalysts consisted of CO and CO2 with CO2 as the primary product overall (Figure 6.9), indicating 167 that the deposit on the catalyst was mainly made up of carbon, similar to what has been observed in the past[329]. Low to no H2O evolution was observed in all the samples, indicating that the coke formed were of aromatic form or highly unsaturated hydrocarbons, consistent with Raman data. Overall, the CO and CO2 formations on the pure SCZO were higher compared to pure Fe/SiO2 sample, indicating that coke formed more extensively on the SCZO in the DNMC reaction, consistent with reaction data. CO2 dominated for pure SCZO sample while CO prevailed for pure Fe/SiO2 catalyst. Such phenomenon suggests that coke formed on the Fe/SiO2 catalyst was of hard coke and the amount of oxygen fed during TPO was not sufficient to react with such coke to form CO2. In all cases, CO and CO2 evolutions decreased with increasing hydrogen feed. Only one peak was observed in the CO2 evolution profile of pure Fe/SiO2 catalyst while two peaks were observed for pure SCZO and 5 wt% SCZO samples. Additionally, the coke oxidation temperature for pure Fe/SiO2 sample was also higher than the pure SCZO and 5 wt% SCZO samples, consistent with TGA data. Such behaviors again suggest that the coke formed on the Fe/SiO2 sample was of hard type, and therefore higher temperature was required to oxidize it. Even though two peaks were observed in the pure SCZO and 5 wt% SCZO samples, the evolution peaks at higher temperature were much smaller compared to the low-temperature peaks, suggesting that the coke formed on SCZO samples mainly consisted of soft-type. It has also been proposed that the coke that appeared at low temperature was more reactive to oxygen, and has been assigned to coke deposited on metallic center[330]. Since SCZO is made up of three different metals and this material generated more coke that caused catalyst deactivation, the TPO analysis further verifies that DNMC on SCZO was catalyzed by surface reaction through the metal active centers. On the other hand, the coke that was 168 burnt off at higher temperature can be corresponded to coke deposited on the support[330]. Since the coke mainly deposited on the SiO2 support in the Fe/SiO2 sample, the Fe active centers were not blocked by carbonaceous deposit, and therefore, no deactivation was observed for this catalyst. The O2, CO and CO2 evolution peaks shifted from higher temperature to lower temperature with increasing hydrogen co-feed concentration for all samples, suggesting the ease of coke removal in the catalysts when coke formation was lesser. The TPO results demonstrate that the coke formed during DNMC reaction was mostly soft coke for SCZO samples that potentially cause deactivation to this sample while it was mostly hard coke for Fe/SiO2 sample that does not cause catalyst deactivation. (A) O 2 (B) O (C) 2 O2 CO CO CO CO2 CO2 CO2 Fe/SiO Fe/SiO2 Fe/SiO2 2 5 wt% SCZO in Fe/SiO 5 wt% SCZO in Fe/SiO2 5 wt% SCZO in Fe/SiO22 SCZO SCZO SCZO 500 600 700 800 900 1000 500 600 700 800 900 1000 500 600 700 800 900 1000 Temperature (K) Temperature (K) Temperature KC) 169 Peak Intensity (a.u.) Peak Intensity (a.u.) Peak Intensity (a.u.) Peak Intensity (a.u.) Peak Intensity (a.u.) Peak Intensity (a.u.) Peak Intensity (a.u.) Peak Intensity (a.u.) Peak Intensity (a.u.) (D) O (E) 2 O2 CO CO CO2 CO2 Fe/SiO2 Fe/SiO 5 wt% SCZO in Fe/SiO 22 SCZO 5 wt% SCZO in Fe/SiO2 SCZO 500 600 700 800 900 1000 500 600 700 800 900 1000 Temperature (K) Temperature (K) Figure 6.9. TPO of spent FeSiO2 catalyst, SCZO sample and Fe/SiO2/SCZO mixture at different hydrogen co-feed concentrations after TOS of 3.5 h in DNMC reactions. 6.4. Conclusion 170 Peak Intensity (a.u.) Peak Intensity (a.u.) Peak Intensity (a.u.) Peak Intensity (a.u.) Peak Intensity (a.u.) Peak Intensity (a.u.) The hydrogen functionality of SCZO membrane material in DNMC reaction over Fe/SiO2 catalyst was investigated via two approaches: effects of SCZO membrane material concentration and arrangement in Fe/SiO2 catalyst and effects of H2 co-feed to the SCZO/Fe/SiO2 catalyst mixture, respectively. The DNMC catalytic performance demonstrated that stable methane conversion, higher C2 (acetylene, ethylene and ethane) selectivity and negligible coke formation could be achieved with Fe/SiO2 catalyst. SCZO material, on the other hand, showed deactivation during DNMC reaction. DNMC reaction over SCZO was proposed to follow surface reaction while DNMC over Fe/SiO2 catalyst was proposed to follow both surface and gas phase mechanisms. The DNMC catalytic performance also suggests that methane conversion could be maintained while excellent C2 selectivity could be achieved when 5 wt% of SCZO material in Fe/SiO2 catalyst was employed at 5% and 10% H2-cofeed concentrations reaction condition. Such catalyst and membrane material composition match well with the tubular membrane reactor setup in our membrane reactor study. Raman analysis suggested that the coke formed on the spent Fe/SiO2 was mainly amorphous graphitic type while the coke formed on the spent SCZO was mainly ordered graphitic type. TPO analysis confirmed that hard coke formed on Fe/SiO2 catalyst did not contribute to catalyst deactivation while soft coke formed on SCZO sample led to catalyst degradation. The present study exemplified for the first time that the SCZO membrane material concentration in Fe/SiO2 catalyst, and amount of hydrogen co-feed to the reaction can be tailored to impact its activity and selectivity in DNMC reaction. 171 Chapter 7: Conclusions (Major Contributions) and Future Works 7.1 Conclusions (Major Contributions) This work was motivated by the need to improve current methane upgrading technology which focuses on indirect, energy-intensive pathways. However, in the more economical viable direct methane conversion pathways, namely OCM and DNMC, several challenges need to be overcome. OCM reaction is challenged by the dilemma between methane conversion and catalyst selectivity to desired C2 products while DNMC process faces a number of significant challenges including low thermodynamically limited methane conversion, high endothermicity and high coke selectivity. In this study, we aim to synthesize catalysts that are active for both OCM and DNMC reactions. The principle objectives of this work include: (i) developing both cation and anion substituted HAP materials to be potentially used as O2 permeable membranes for OCM in the future, (ii) assessing the reaction kinetics of HAP-based material in OCM reaction, (iii) synthesizing HAP-based materials with controlled crystalline plane to systematically tune the methane conversion and C2 selectivity to reach high C2 yield in OCM reaction, (iv) fabricating millisecond catalytic wall reactor for DNMC process to understand kinetics of metal/silica 172 catalyst, and (v) examining the hydrogen functionality of H2-permeable membrane reactor material in DNMC reaction. Lead substituted hydroxyapatite (Pb-HAP) has been an active catalyst for oxidative coupling of methane (OCM) reactions. CO32- substituted HAP (HAP-CO3) has showed enhanced oxide ion conductivity than bare HAP in high temperature solid oxide fuel cells. Substitutions for both cations and anions in HAP structure (Pb-HAP-CO3) are promising to integrate the catalytic property of Pb-HAP and oxide ion conductive property of HAP- CO3 into one apatite-based ceramic material that can be manufactured into membrane reactors for possessing CH4 activation and O2 permeation capabilities for efficient OCM reactions. In Chapter 2, the effects of substitutions for both cation (Pb2+) and anion (CO32-) in HAP structure on OCM reactions were studied. The composition and physicochemical properties of HAP catalysts were changed by the cation and anion substitutions, respectively, and as consequences, they influenced the catalytic performances of HAP structure in OCM reactions. The selectivity to C2 (ethylene and ethane) products increased in the order of HAP-CO3 < HAP < Pb-HAP-CO3 < Pb-HAP, while Pb-HAP-CO3 showed the best stability and comparable C2 yield (under optimized reaction conditions) to Pb-HAP catalyst. Under different reaction temperature and/or CH4/O2 ratio in the OCM reactions, the CH4 conversion and C2 or COx (CO and CO2) selectivity showed a strong dependence on the composition of HAP-based catalysts. The present study forms a basis for understanding of the correlations between the composition, structure, and catalytic performance of HAP and other apatite structured catalysts, which are potential membrane materials for OCM reactions in membrane reactors. 173 In Chapter 3, the identity and rate constants of elementary steps in primary reactions of oxidative coupling of methane (OCM) over Pb2+ and/or F- substituted hydroxyapatite (HAP, Pb-HAP, HAP-F, and Pb-HAP-F) catalysts have been studied. The rigorous kinetics analysis suggests that HAP and HAP-F initiated the reaction between adsorbed methane and O2 following Langmuir-Hinshelwood behavior. The Pb-HAP and Pb-HAP-F, however, enabled the reaction between gaseous methane and adsorbed O2 in the Eley-Rideal mechanism. The F- substitution of OH- weakened both O2 adsorption and C-H bond activation, leading to low methane conversion and slightly higher C2H6 selectivity. The substitution of Ca2+ by Pb2+ weakened both methane and oxygen adsorption, but maintained C-H bond activation and raised C2H6 selectivity. The present analysis explored for the first time the effects of cation and/or anion in HAP on OCM reactions in which the analysis has been detailed and quantified in the nature of the mechanism-based kinetic models. In Chapter 4, the predominant exposure of c-surface of HAP was controlled and its influences in methane oxidation reactions (combustion and oxidative coupling over HAP and lead-substituted HAP (Pb-HAP), respectively) were studied. The c-surface exposure was realized by crystal orientation in HAP-based catalyst film, which was created by an electrochemical deposition of HAP seeds on a titanium substrate, followed by hydrothermal crystallization and peeling-off of the crystalline films from the substrate. In comparison to a-surface that is prevalently exposed in unoriented HAP-based catalysts, the c-surface (i.e., (002) crystalline plane) of HAP-based catalysts exhibited up to 47-fold enhancement in areal rate in both reactions. The distinct catalytic activity between these two crystalline surfaces is attributed to the preferential formation of oxide ions and 174 vacancies on c-surfaces. The oxide ions and vacancies in turns function as actives sites for promoting methane activation and complete oxidation into CO2. Density functional theory calculations confirmed the close relationship between different catalytic activities in c- surface of oriented and a-surface of unoriented HAP through the tendency of vacancy formation. Without the presence of vacancies, the methyl or methylene group after methane activation forms ethane or ethylene via coupling. The present study explored the effects of HAP crystal orientation in methane oxidation reactions, which revealed distinct catalytic behaviors of crystal surfaces in HAP-based materials. In Chapter 5, we showed that a millisecond catalytic wall reactor made of iron/quartz enables stable methane conversion, C2+ selectivity, coke yield and long-term durability. These effects originated from initiation of DNMC by surface catalysis on reactor wall, and maintenance of the reaction by gas phase reaction in reactor compartment. We further demonstrated the concept of millisecond catalytic wall reactor operated autothermally by coupling and periodical swapping of endothermic DNMC and exothermic oxidative coke removal on opposite side of the reactor. High carbon and thermal efficiencies and low cost in reactor materials are realized for the technoeconomic process viability of the DNMC technology. In Chapter 6, we present a strategy to reach light hydrocarbon selectivity as high as ~80% and C2 ~70% at methane conversion of ~10%. This is enabled by a bifunctional catalyst, i.e., a mixture of iron/silica (Fe/SiO2) and zirconium-doped strontium-cerium perovskite oxide (SrCe0.8Zr0.2O3−δ), in the presence of hydrogen (H2) co-feed in methane stream, affording two types of active sites with complementary properties. The Fe/SiO2 175 catalyst has lattice-confined single Fe sites embedded in the SiO2 matrix that activates methane to higher hydrocarbons. The SrCe0.8Zr0.2O3−δ is a mixed ionic–electronic conductor (MIEC), capable of H2 permeation, which enriches hydrogen species in the DNMC reaction. Under co-feeding of H2 in methane stream, product selectivity is shifted to C2 hydrocarbons from aromatics such as naphthalene without the expense of lower methane conversion. The interface of DNMC catalyst and MIEC oxide under variant H2 concentrations is often experienced in the studies of DNMC in H2-permeable membrane reactors that manipulate methane conversion and product selectivity via Le Chatelier’s principle. Therefore, the present study further allows mechanistic understanding of DNMC in such reactor systems. 7.2 Future Works (1) Fabrication of O2-permeable membrane reactor for OCM reaction We have also developed methods for synthesis of oxygen permeable HAP hollow fiber membrane and support materials for OCM reaction. We have also established the procedures for fabrication and assembly of the hollow fiber membrane reactor for O2 permeation and OCM reaction tests. We will continue to fabricate Pb2+ and CO32- substituted HAP hollow fiber membrane. After that, we will conduct O2 permeation test on the membrane reactor. We will evaluate the catalytic performance of the Pb-HAP-CO3 hollow fiber membrane reactor. The activity, selectivity and stability of the membrane reactor in the reaction environment will be systematically measured. We will study the influence of O2 distribution on methane conversion rate and product selectivity. Kinetics 176 of OCM reactions over the membrane reaction under O2 permeation conditions will be examined. (2) OMC reaction using HAP catalyst obtained from nature Instead of performing OCM reaction using synthetic HAP powder, we will obtain HAP catalyst directly from natural resources such as egg shells and etc. We will investigate the effects of impurities in HAP material obtained from nature on OCM reaction. Such strategy could potentially be a huge leap in OCM reaction since HAP from nature is abundant and thus, operation cost could be significantly reduced. (3) Reaction mechanism of DNMC reaction over Fe/SiO2 catalyst The understanding of reaction mechanism and kinetics is necessary to improve the catalytic performance of DNMC process for higher activity and selectivity. We aim to deduce reaction mechanism describing DNMC pathways over Fe/SiO2 catalyst that involve a series of reaction sequences comprising of adsorption, charge transfer and chain growth. We will perform C2 and H2 cofeed experiments to understand the effects of cofeed gases on heterogeneous activation of methane and product distribution, and eventually get some hints on reaction pathways. Since our preliminary data showed that the DNMC over Fe/SiO2 consists of both heterogeneous and homogeneous pathways, we will carry out experiments to decouple these reactions to understand more about the surface kinetics. We will also characterize radical/intermediate species in the DNMC reactions using a molecular-beam mass spectroscopy. Reaction parameters such a temperature and space velocity will be changed and the resultant product in each condition will be analyzed and correlated to the reaction mechanism of DNMC over Fe/SiO2 catalyst. 177 Appendix A: Derivation of rate equation for OCM reactions over HAP-based catalysts A1. Derivation of rate equation for OCM reactions over HAP-based catalysts via the Eley-Rideal mechanism A1.1. Eley-Rideal mechanism with dissociative O2 adsorption In this mechanism, oxygen (O2) is dissociatively adsorbed on the active sites of the HAP-based catalyst to form surface adsorbed oxygen species, O* (step 1). This is followed by the H-atom abstraction from CH4 by O* to form adsorbed OH* species and gas-phase methyl radical (rate limiting step (RLS), step 2). Two adsorbed OH* species subsequently recombine to form water and regenerate surface-active species (step 3). Methyl radicals, on the other hand, couple with each other to form ethane (step 4). Simultaneously, methyl radicals react with gaseous oxygen (step 5) and adsorbed oxygen species, O* (step 6) to produce CO and CO2, respectively. Scheme A1 shows the reaction equation in each step of this mechanism. J Step (1): O% + 2 ∗ ↔ K 2O ∗ L Step (2): CH) + O ∗ → * CH@ ∙ +OH ∗ (RLS) J Step (3): 2OH ∗↔M H%O + O ∗ + ∗ L Step (4): CH (@ ∙ +CH@ ∙ → C%HO L Step (5): CH@ ∙ +O% → P CO L Step (6): CH Q@ ∙ +O ∗ → CO% Scheme A1. Elementary steps in Eley-Rideal reaction mechanism with dissociative O2 adsorption in OCM reactions. A1.1.1 Derivation of rate equation for CH4 consumption Since step 2 is the RLS, the CH4 reaction rate (r&'() is written as, r&'( = k%P&'(θ$∗ (Eq. A1.1) 178 where k2 is the rate constant of methane consumption in the RLS, P&'( is methane pressure and θ$∗ is the coverage of active sites by adsorbed oxygen species on the catalyst. Under the quasi-equilibrium approximation, step (1) can be written as, r % %I = kIP$*θ∗ − kAIθ$∗ ≈ 0 Therefore, k θ$∗ = ¬KIP$*θ∗ sKI = I t (Eq. A1.2) kAI where K1 is the oxygen adsorption equilibrium constant, P$* is oxygen pressure, θ∗ is the fraction of free active sites on the catalyst, k1 and k-1 are the forward and backward reaction rate constants in step (1). Similarly, when quasi-equilibrium approximation is applied to step (3), r %@ = k@θ$'∗ − kA@P'*$θ$∗θ∗ ≈ 0 We can solve for coverage of active site by OH* (θ$'∗) species. 1 k θ$'∗ = o P'*$θ$∗θ∗ (K @ K @ = ) @ kA@ After substitution of θ$∗ into θ$'∗, we can get θ = o ­KI P P9.<$'∗ '*$ $ θ∗ (Eq. A1.3) K@ * where K3 is OH* recombination equilibrium constant and P'*$ is water partial pressure. From the site balance equation, θ = θ + θ + θ = θ + K P θ + ¬­J‡1‡ƒž ∗ $∗ $'∗ ∗ ­ K 9.< I $* ∗ P'*$P$ θ∗ (Eq. A1.4) JM * We can solve for θ∗, θ$∗, and θ$'∗, respectively. 179 θ θ = ‡1‡ƒž∗ (Eq. A1.5) o­K1 + ­KIP I 9.< $* + K P'*$P@ $ * ­K P θ θ = I $* ‡1‡ƒž$∗ (Eq. A1.6) o­K1 + ­K P + II $* K P'*$P 9.< $ @ * o­KI K P'*$P 9.< $ θ* ‡1‡ƒž@ θ$'∗ = (Eq. A1.7) ­K 1 + ­K I 9.