Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Hydrogenation and dehydrogenation kinetics and catalysts for new hydrogen storage liquids Sotoodeh, Farnaz 2011

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2011_fall_sotoodeh_farnaz.pdf [ 4.21MB ]
Metadata
JSON: 24-1.0059178.json
JSON-LD: 24-1.0059178-ld.json
RDF/XML (Pretty): 24-1.0059178-rdf.xml
RDF/JSON: 24-1.0059178-rdf.json
Turtle: 24-1.0059178-turtle.txt
N-Triples: 24-1.0059178-rdf-ntriples.txt
Original Record: 24-1.0059178-source.json
Full Text
24-1.0059178-fulltext.txt
Citation
24-1.0059178.ris

Full Text

Hydrogenation and Dehydrogenation Kinetics and Catalysts for New Hydrogen Storage Liquids  by Farnaz Sotoodeh M.Sc. in Chemical Engineering, Iran University of Science and Technology, 2006 B.Sc. in Chemical Engineering, Iran University of Science and Technology, 2003  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  Doctor of Philosophy in THE FACULTY OF GRADUATE STUDIES (Chemical and Biological Engineering)  The University of British Columbia (Vancouver) June 2011 c Farnaz Sotoodeh, 2011 ⃝  Abstract Due to the very low density of H2 , practical storage and recovery of H2 has been a challenge in utilizing H2 as an alternative fuel. Organic heteroaromatics have attracted interest because of their thermal stability and high storage capacity. In this study, H2 storage and recovery from these compounds were investigated. The kinetics of the hydrogenation/dehydrogenation reactions was studied and DFT calculations were used to understand the dehydrogenation product distribution. The hydrogenation of N-ethylcarbazole and carbazole at 403-423 K on a supported Ru catalyst was well described by first-order kinetics. The hydrogenation of N-ethylcarbazole was significantly faster than the hydrogenation of carbazole, and >95 % selectivity to dodecahydro-N-ethylcarbazole and dodecahydrocarbazole was achieved, respectively. The dehydrogenation kinetics of dodecahydro-N-ethylcarbazole was studied at 101 kPa and 423-443 K over a Pd catalyst prepared by wet impregnation and calcination in air. The reactions followed first-order kinetics with 100 % conversion but only 69 % recovery of H2 was achieved at 443 K, due to minimal selectivity to Nethylcarbazole. The complete recovery of H2 from dodecahydro-N-ethylcarbazole was achieved at 443 K and 101 kPa using Pd/SiO2 catalysts prepared by incipient wetness impregnation with calcination in He. The dehydrogenation TOF and selectivity to N-ethylcarbazole were dependent upon the Pd particle size. The effect of the N heteroatom on the dehydrogenation of polyaromatics was studied by comparison of dodecahydro-N-ethylcarbazole, dodecahydrocarbazole ii  and dodecahydrofluorene dehydrogenation over Pd catalysts. The dehydrogenation of dodecahydro-N-ethylcarbazole and dodecahydrocarbazole were structure sensitive. The dehydrogenation rate of dodecahydrocarbazole was slower than dodecahydro-N-ethylcarbazole. Despite catalyst poisoning through the N atom in dodecahydrocarbazole, the N heteroatom was found to favor dehydrogenation, making heteroaromatics better candidates for H2 storage than aromatics. The structure sensitivity of the reactions and the observed product distribution are explained in view of DFT calculations that showed that the adsorption of dodecahydro-N-ethylcarbazole on Pd required multiple catalytic sites and the heat of adsorption was dependent upon the surface structure. The effect of the ethyl group and the N heteroatom on the dehydrogenation rate of dodecahydroN-ethylcarbazole was also investigated by comparing the adsorption energies of dodecahydro-N-ethylcarbazole with dodecahydrocarbazole and dodecahydrofluorene.  iii  Preface This dissertation consists of six manuscripts five of which are included in Chapters 2 to 6, and one of which is included in Appendix H. This PhD work has been done by Farnaz Sotoodeh under the direct supervision of Professor Kevin J. Smith in the Department of Chemical and Biological Engineering at the University of British Columbia. Preparation of the dissertation, literature review, experimental design and set−up, data collection and analysis, reaction modeling and kinetic study, theoretical molecular modeling of the dehydrogenation reactions, as well as the analysis and interpretation of the results in this thesis have been performed by Farnaz Sotoodeh under the supervision of Professor Kevin J. Smith. The manuscripts included in this dissertation are listed below. For the manuscripts with co-authors, the contributions of Farnaz Sotoodeh have been described in detail. 1. F. Sotoodeh and K.J. Smith (2010) Kinetics of hydrogen uptake and release from heteroaromatic compounds for hydrogen storage, Industrial and Engineering Chemistry Research, 49, 1018-1026. A version of this manuscript is included in Chapter 2. The catalyst preparation and characterization, hydrogenation reactor set−up and performing the hydrogenation reactions, as well as the dehydrogenation reactor set−up, carrying out the dehydrogenation reactions, hydrogenation and dehydrogenation reactions sample collection and analysis of data, hydrogenation and dehydrogenation network mathematical modeling and coding, as well as the kinetic analysis of the reactions, have been done by Farnaz Soiv  toodeh. The preparation and writing of the manuscript was done by Farnaz Sotoodeh under direct supervision and final approval of Professor Kevin J. Smith. 2. F. Sotoodeh, L. Zhao and K.J. Smith (2009) Kinetics of H2 recovery from dodecahydro-N-ethylcarbazole over a supported Pd catalyst, Applied Catalysis A: General 362, 155-162. A version of this manuscript is included in Chapter 3. The catalyst preparation, characterization, design and experimental set−up for the dehydrogenation reactions, sample collection and data analysis, kinetic study of the reactions, mathematical modeling of the reaction network and coding have been done by Farnaz Sotoodeh under the direct supervision of Professor Kevin J. Smith. The manuscript was prepared and written by Farnaz Sotoodeh with final approval of Professor Kevin J. Smith. Some portions of the liquid sample analysis of the dehydrogenation reactions were done by Liang Zhao. 3. F. Sotoodeh and K.J. Smith (2011) Structure sensitivity of dodecahydro-Nethylcarbazole dehydrogenation over Pd catalysts, Journal of Catalysis 279, 36-47. A version of this manuscript is included in Chapter 4. The catalyst preparation and characterization, catalyst testing including the experimental reaction design and set−up, performing the reactions and data collection by sampling, data analysis of the experiments and results interpretation, kinetic studies of the dehydrogenation reactions, theoretical work including molecular design of the dehydrogenation reaction, DFT calculations using the Material Studio software and the theoretical data analysis, as well as interpretation of the results, have been done by Farnaz Sotoodeh under the direct supervision of Professor Kevin J. Smith. This manuscript was prepared and written by Farnaz Sotoodeh with final approval of Professor Kevin J. Smith. 4. F. Sotoodeh, B.J.M. Huber and K.J. Smith, The effect of the N-heteroatom v  on the dehydrogenation of polyaromatics used for hydrogen storage, to be submitted. A version of this manuscript is included in Chapter 5. The catalyst preparation and characterization, the hydrogenation and dehydrogenation reaction experimental set−up and design, performing the hydrogenation reactions of carbazole and N−ethylcarbazole, performing the dehydrogenation reactions of dodecahydro-N-ethylcarbazole, the kinetic study of the dehydrogenation reactions and the data analysis of the dehydrogenation reactions of dodecahydro-N-ethylcarbazole as well as interpretation of the results have been done by Farnaz Sotoodeh under the direct supervision of Professor Kevin J. Smith. The preparation and writing of the manuscript were done by Farnaz Sotoodeh with final approval of Professor Kevin J. Smith. Some part of the dehydrogenation reactions of dodecahydrocarbazole and dodecahydrofluorene and liquid sample analysis were done by Benjamin J.M. Huber. 5. F. Sotoodeh and K.J. Smith, Density functional theory study of dodecahydroN-ethylcarbazole dehydrogenation over a Pd catalyst, to be submitted. A version of this manuscript is included in Chapter 6. Molecular design of the dehydrogenation reaction including building the chemical structure of the reactants, products and catalyst surface, performing the geometry optimization of the designed systems using Material Studio software, performing potential energy surface calculations by doing transition state search through each catalytic step, theoretical analysis of the obtained results, as well as interpretation of the results through comparison with the experimental results, have been done by Farnaz Sotoodeh under direct supervision of Professor Kevin J. Smith. The preparation and writing of the manuscript was done by Farnaz Sotoodeh with final approval of Professor Kevin J. Smith. The following manuscript is included in the “Additional publications” section, Appendix H.  vi  L. Zhao, F. Sotoodeh and K.J. Smith (2010) Increased surface area of unsupported Mo2 C catalyst by alkali-treatment, Catalysis Communications 11, 391-395. Preparation of the 12 h alkali-treated Mo2 C catalyst, catalyst characterizations including X-ray diffraction (XRD), BET surface area and pore size distributions measurements of the 6 h and 12 h alkali-treated Mo2 C catalysts, as well as dehydrogenation reactions of tetrahydrocarbazole over a 5 wt % Pd/SiO2 were done by Farnaz Sotoodeh under direct supervision of Professor Kevin J. Smith. Writing the manuscript, some portions of the catalyst preparation and characterization, as well as CO hydrogenation reactions were done by Liang Zhao.  vii  Table of Contents Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  ii  Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  iv  Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  viii  List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  xiii  List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiv Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxviii Dedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxx 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  1  1.1  Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  2  1.2  Objective of this thesis . . . . . . . . . . . . . . . . . . . . . . .  3  1.3  Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  3  1.4  Hydrogen storage . . . . . . . . . . . . . . . . . . . . . . . . . .  4  1.5  Compressed hydrogen . . . . . . . . . . . . . . . . . . . . . . .  6  1.6  Liquified hydrogen . . . . . . . . . . . . . . . . . . . . . . . . .  7  1.7  Off-board regenerable storage . . . . . . . . . . . . . . . . . . .  7  1.7.1  Ammonia borane-based H2 storage materials . . . . . . .  8  1.7.2  Chemical hydrides . . . . . . . . . . . . . . . . . . . . .  9  1.7.3  Organic materials . . . . . . . . . . . . . . . . . . . . . .  10  viii  1.8  Hydrogenation/dehydrogenation reactions and catalysts . . . . . .  11  1.9  Selection of organic materials for hydrogen storage . . . . . . . .  11  1.10 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  15  1.11 Molecular modeling of heterogeneous catalysis . . . . . . . . . .  15  1.11.1 Schr¨odinger equation . . . . . . . . . . . . . . . . . . . .  16  1.11.2 Density functional theory (DFT) . . . . . . . . . . . . . .  18  1.12 Molecular modeling in heterogeneous catalysis . . . . . . . . . .  19  1.12.1 Cluster models . . . . . . . . . . . . . . . . . . . . . . .  20  1.12.2 Embedded-cluster model . . . . . . . . . . . . . . . . . .  20  1.12.3 Periodic quantum−mechanical calculations . . . . . . . .  20  1.13 Outline of the dissertation . . . . . . . . . . . . . . . . . . . . . .  21  2 Hydrogenation kinetics: N-ethylcarbazole and carbazole . . . . . .  23  2.1  Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  24  2.2  Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . .  25  2.2.1  Catalyst preparation and characterization . . . . . . . . .  25  2.2.2  Catalyst activity . . . . . . . . . . . . . . . . . . . . . .  25  Results and discussion . . . . . . . . . . . . . . . . . . . . . . .  27  2.3.1  Catalyst properties . . . . . . . . . . . . . . . . . . . . .  27  2.3.2  Catalyst activity . . . . . . . . . . . . . . . . . . . . . .  28  Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  38  2.3  2.4  3 Kinetics of H2 recovery from dodecahydro-N-ethylcarbazole over a supported Pd catalyst . . . . . . . . . . . . . . . . . . . . . . . . . .  39  3.1  Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  40  3.2  Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . .  40  3.2.1  Catalyst preparation . . . . . . . . . . . . . . . . . . . .  40  3.2.2  Catalyst characterization . . . . . . . . . . . . . . . . . .  41  3.2.3  Catalyst activity . . . . . . . . . . . . . . . . . . . . . .  41  Results and discussions . . . . . . . . . . . . . . . . . . . . . . .  43  3.3.1  Catalyst properties . . . . . . . . . . . . . . . . . . . . .  43  3.3.2  Catalyst activity . . . . . . . . . . . . . . . . . . . . . .  43  Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  47  3.3  3.4  ix  4 Structure sensitivity of dodecahydro-N-ethylcarbazole dehydrogenation over Pd catalysts . . . . . . . . . . . . . . . . . . . . . . . . . .  50  4.1  Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  51  4.2  Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . .  53  4.2.1  Catalyst preparation . . . . . . . . . . . . . . . . . . . .  53  4.2.2  Catalyst characterization . . . . . . . . . . . . . . . . . .  53  4.2.3  Catalyst activity . . . . . . . . . . . . . . . . . . . . . .  55  Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  55  4.3.1  Catalyst properties . . . . . . . . . . . . . . . . . . . . .  55  4.3.2  Catalyst activities . . . . . . . . . . . . . . . . . . . . . .  62  Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  71  4.4.1  Catalyst properties . . . . . . . . . . . . . . . . . . . . .  71  4.4.2  Catalyst activity and selectivity . . . . . . . . . . . . . .  75  Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  77  4.3  4.4  4.5  5 The effect of the N-heteroatom on the dehydrogenation of polyaromatics used for hydrogen storage  . . . . . . . . . . . . . . . . . . .  78  5.1  Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  79  5.2  Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . .  79  5.2.1  Catalyst preparation . . . . . . . . . . . . . . . . . . . .  79  5.2.2  Catalyst characterization . . . . . . . . . . . . . . . . . .  79  5.2.3  Catalyst activity . . . . . . . . . . . . . . . . . . . . . .  81  Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  81  5.3.1  Catalyst properties . . . . . . . . . . . . . . . . . . . . .  81  5.3.2  Catalyst activities . . . . . . . . . . . . . . . . . . . . . .  82  5.3.3  Dehydrogenation reaction comparison of dodecahydro-N-  5.3  ethylcarbazole, dodecahydrocarbazole, dodecahydrofluorene 90 5.4  5.5  Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  92  5.4.1  Catalyst properties . . . . . . . . . . . . . . . . . . . . .  92  5.4.2  Catalyst activities . . . . . . . . . . . . . . . . . . . . . .  93  Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  94  x  6 Density functional theory study of dodecahydro-N-ethylcarbazole dehydrogenation over a Pd catalyst . . . . . . . . . . . . . . . . . . . .  96  6.1  Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  97  6.2  Computational methods . . . . . . . . . . . . . . . . . . . . . . .  98  6.3  Results and discussion . . . . . . . . . . . . . . . . . . . . . . .  99  6.3.1  Molecular structures . . . . . . . . . . . . . . . . . . . .  99  6.3.2  Dodecahydro-N-ethylcarbazole adsorption on the Pd(111) surface . . . . . . . . . . . . . . . . . . . . . . . . . . . 106  6.3.3  Dodecahydro-N-ethylcarbazole dehydrogenation steps . . 112  6.3.4  Dodecahydro-N-ethylcarbazole adsorption on Pd(110) surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123  6.3.5 6.4  Heteroatom effect on the catalyst reactivity . . . . . . . . 124  Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131  7 Conclusions and recommendations . . . . . . . . . . . . . . . . . . . 132 7.1  Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133  7.2  Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . 134  Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 A Catalyst preparation: calculations of required chemicals . . . . . . 154 B Catalyst characterization techniques . . . . . . . . . . . . . . . . . . 155 B.1 Total BET surface area . . . . . . . . . . . . . . . . . . . . . . . 155 B.2 Temperature-programmed reduction (TPR) . . . . . . . . . . . . 155 B.3 Temperature-programmed desorption (TPD) . . . . . . . . . . . . 156 B.3.1  Injection loop calibration . . . . . . . . . . . . . . . . . . 156  B.4 High resolution transmission electron microscopy (HRTEM) . . . 158 C Mass spectrometery . . . . . . . . . . . . . . . . . . . . . . . . . . . 159 C.1 Mass flow controllers (MFCs) calibration . . . . . . . . . . . . . 159 C.2 Mass spectrometer (MS) calibration for H2  xi  . . . . . . . . . . . . 160  D Gas chromatography . . . . . . . . . . . . . . . . . . . . . . . . . . 163 D.1 Temperature program . . . . . . . . . . . . . . . . . . . . . . . . 163 D.2 Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 E Sample calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . 165 E.1 N-ethylcarbazole hydrogenation reaction at 413 K . . . . . . . . . 165 E.1.1  Experiment procedure . . . . . . . . . . . . . . . . . . . 165  E.2 Dodecahydro-N-ethylcarbazole dehydrogenation reaction . . . . . 168 E.2.1  Experiment procedure . . . . . . . . . . . . . . . . . . . 168  E.3 Rate constant sample calculation reported in Table 4.3 . . . . . . . 171 F Internal diffusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 F.1  Effective diffusivity . . . . . . . . . . . . . . . . . . . . . . . . . 173  F.2  Thiele modulus, effectiveness factor and Weisz-Prater criterion . . 173  G Reaction modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 G.1 Standard error of the parameter estimation . . . . . . . . . . . . . 175 G.2 Codes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176 G.2.1 Main body . . . . . . . . . . . . . . . . . . . . . . . . . 176 G.2.2 Objective function . . . . . . . . . . . . . . . . . . . . . 179 G.2.3 Ordinary differential equations . . . . . . . . . . . . . . . 180 H Additional publications . . . . . . . . . . . . . . . . . . . . . . . . . 181 H.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 H.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183 H.2.1 Catalyst preparation . . . . . . . . . . . . . . . . . . . . 183 H.2.2 Characterization . . . . . . . . . . . . . . . . . . . . . . 183 H.2.3 Catalyst activity . . . . . . . . . . . . . . . . . . . . . . 184 H.3 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . 185 H.3.1 Mo2 C properties . . . . . . . . . . . . . . . . . . . . . . 185 H.3.2 THCZ dehydrogenation . . . . . . . . . . . . . . . . . . 187 H.3.3 CO hydrogenation reaction . . . . . . . . . . . . . . . . . 190 H.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192  xii  List of Tables Table 1.1  U.S. Department of energy (D.O.E.) revised targets for hydrogen storage systems . . . . . . . . . . . . . . . . . . . . . . . . . . . .  5  Table 1.2  Different categories of hydrogen storage materials . . . . . . . . .  8  Table 1.3  Examples of different components of a heterogeneous catalyst  . . .  11  Table 1.4  Active catalytic phases for hydrogenation and/or dehydrogenation reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  Table 1.5  12  Examples of substituted carbazole structures with high hydrogen capacity and low heat of hydrogenation. . . . . . . . . . . . . . . .  13  Table 2.1  Physical properties of hydrogenation reaction materials . . . . . . .  27  Table 2.2  Catalyst properties of the pre-reduced Ru/alumina . . . . . . . . .  27  Table 2.3  1st -order  rate constants and activation energies for the hydrogenation  of N-ethylcarbazole over Ru/alumina catalyst . . . . . . . . . . . .  Table 2.4  Rate constants for N-ethylcarbazole hydrogenation with different initial concentrations at 413 K. . . . . . . . . . . . . . . . . . . . .  Table 2.5  33 34  Initial Reaction Rates, rate constants, and TOFs for hydrogenation of  .  37  Table 3.1  Pd/SiO2 catalyst properties . . . . . . . . . . . . . . . . . . . . .  43  Table 3.2  Initial reaction rates, rate constants and hydrogen recovery for dehy-  N-ethylcarbazole and carbazole at 423 K over Ru/alumina catalyst  drogenation of dodecahydro-N-ethylcarbazole . . . . . . . . . . .  Table 3.3  45  Comparison between dehydrogenation of 1,2,3,4-tetrahydrocarbazole and dodecahydro-N-ethylcarbazole over a 5 wt % Pd/SiO2 catalyst .  48  Table 4.1  Pd/SiO2 catalyst properties . . . . . . . . . . . . . . . . . . . . .  60  Table 4.2  EDX elemental analysis of the 5 wt % Pd/SiO2 catalysts . . . . . .  62  xiii  Table 4.3  First order rate constants and turn-over frequencies (TOFs) for dehydrogenation of dodecahydro-N-ethylcarbazole at 443 K and 101 kPa over Pd/SiO2 catalysts with different Pd loadings. . . . . . . . . . .  Table 4.4  68  Comparison of the selectivity to N-ethylcarbazole and the conversion of dodecahydro-N-ethylcarbazole at 443 K and 101 kPa over Pd/SiO2 catalysts calcined in He. . . . . . . . . . . . . . . . . . . . . . .  Table 4.5  70  H2 recovery for the dehydrogenation of dodecahydro-N-ethylcarbazole at 443 K and 101 kPa over Pd/SiO2 with different Pd loadings. . . .  72  Table 5.1  Catalyst properties . . . . . . . . . . . . . . . . . . . . . . . . .  82  Table 5.2  Kinetic data obtained from liquid phase analysis for dodecahydrocarbazole dehydrogenation in decalin using a 5 wt % Pd/C catalyst at various temperatures and 101 kPa. . . . . . . . . . . . . . . . . .  Table 5.3  H2 recovery for the dehydrogenation of dodecahydrocarbazole at various temperatures and 101 kPa over a 5 wt % Pd/C catalyst. . . . . .  Table 5.4  84 87  Comparison of kinetic data obtained from liquid phase analysis for dodecahydrofluorene dehydrogenation with and without a solvent at 443 K and 101 kPa over 5 wt % Pd/C catalyst. . . . . . . . . . . .  Table 5.5  90  Dehydrogenation reaction comparison of dodecahydro-N-ethylcarbazole, dodecahydrocarbazole and dodecahydrofluorene at 443 K and 101 kPa. 91  Table 6.1  Optimized geometries and energies of the most probable isomers of dodecahydro-N-ethylcarbazole in vacuum . . . . . . . . . . . . . 101  Table 6.2  Chemical structures and energies of decahydro-N-ethylcarbazole isomers in vacuum; the structures with symmetric double bond on the left 6-membered rings are not shown due to having the same energy. . 102  Table 6.3  Chemical structures and energies of octahydro-N-ethylcarbazole isomers in vacuum. . . . . . . . . . . . . . . . . . . . . . . . . . . 103  Table 6.4  Chemical structures and energies of hexahydro-N-ethylcarbazole isomers in vacuum. . . . . . . . . . . . . . . . . . . . . . . . . . . 105  Table 6.5  Chemical structures and energies of tetrahydro-N-ethylcarbazole isomers in vacuum. . . . . . . . . . . . . . . . . . . . . . . . . . . 106  xiv  Table 6.6  Bond lengths change for dodecahydro-N-ethylcarbazole’s isomers atop adsorption on Pd(111). . . . . . . . . . . . . . . . . . . . . . . . 108  Table 6.7  Hirshfield charges on dodecahydro-N-ethylcarbazole. . . . . . . . . 111  Table 6.8  Reaction barriers corresponding to the transition states (TSs) for each dehydrogenation step along the two dehydrogenation reaction pathways of dodecahydro-N-ethylcarbazole shown in Figure 6.6. . . . . 114  Table 6.9  Selected parameters for the gas phase and adsorbed structure of octahydroN-ethylcarbazole. C-Pd distances refer to corresponding Pd atom shown in Figure 6.9. . . . . . . . . . . . . . . . . . . . . . . . . 118  Table 6.10 Reaction barriers corresponding to the transition states (TSs) for conversion of octahydro-N-ethylcarbazole to tetrahydro-N-ethylcarbazole. 118  Table 6.11 Selected parameters for the gas phase and adsorbed structure of tetrahydroN-ethylcarbazole. C-Pd distances refer to corresponding Pd atom shown in Figure 6.12. . . . . . . . . . . . . . . . . . . . . . . . 123  Table 6.12 Comparison of adsorption energies. . . . . . . . . . . . . . . . . 126 Table 6.13 Bond lengths change for dodecahydrofluorene and dodecahydrocarbazole upon adsorption to Pd(111) surface. . . . . . . . . . . . . . 127  Table 6.14 Hirshfield charges for dodecahydrocarbazole in vacuum and adsorbed structure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128  Table 6.15 Hirshfield charges for dodecahydrofluorene in vacuum and adsorbed structure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129  Table C.1  He 200 SCCM MFC calibration using a bubble flow meter and stopwatch for 79.9 % of the total flow. . . . . . . . . . . . . . . . . . 159  Table C.2  H2 20 SCCM MFC calibration using a bubble flow meter and stopwatch. 160  Table C.3  MS calibration data for H2 using a constant He flow of 171.19 ml/min as the carrier gas. . . . . . . . . . . . . . . . . . . . . . . . . . 161  Table D.1  Tenperature program used in the Shimadzu QP-2010S GC/MS. . . . 163  Table E.1  Experiment information . . . . . . . . . . . . . . . . . . . . . . 166  Table E.2  Hydrogen consumption . . . . . . . . . . . . . . . . . . . . . . 166  Table E.3  N-ethylcarbazole conversion obtained by GC analysis of the samples  Table E.4  Concentrations of reactants and products in mol/ml. . . . . . . . . . 169  xv  167  Table E.5  Dehydrogenation reaction data analysis. . . . . . . . . . . . . . . 170  Table F.1  THCZ, H2 internal diffusion in mesitylene on Pd/silica at T=140 ◦ C, P=1 atm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174  Table H.1  Properties of the unsupported Mo2 C catalysts. . . . . . . . . . . . 186  Table H.2  THCZ dehydrogenation activity measured in a batch reactor at 101 kPa and 423 K over unsupported Mo2 C catalysts. . . . . . . . . . . 190  Table H.3  Syngas conversion on Mo2 C and alkali-treated Mo2 C. Reaction conditions: pressure = 8.2 MPa, temperature = 573 K, CO:H2 = 1 and GHSV = 3960 h−1 . . . . . . . . . . . . . . . . . . . . . . . . . 192  xvi  List of Figures Figure 2.1  Schematic diagram for high pressure hydrogenation system . .  26  Figure 2.2  TPR results for the reduced Ru/Al2 O3 . . . . . . . . . . . . .  28  Figure 2.3  Hydrogenation reaction pathway of N-ethylcarbazole; (a) Nethylcarbazole, (b) tetrahydro-N-ethylcarbazole, (c) octahydroN-ethylcarbazole, (d) hexahydro-N-ethylcarbazole, (e) decahydroN-ethylcarbazole, (f) dodecahydro-N-ethylcarbazole. . . . . .  Figure 2.4  Product distribution over time for N-ethylcarbazole hydrogenation at 423 K. Note the different scale of each plot. . . . . . .  Figure 2.5  31  Product distribution over time for N-ethylcarbazole hydrogenation at 403 K. Note the different scale of each plot. . . . . . .  Figure 2.7  30  Product distribution over time for N-ethylcarbazole hydrogenation at 413 K. Note the different scale of each plot. . . . . . .  Figure 2.6  29  32  Comparison between 6 wt % N-ethylcarbazole conversions at different temperatures: (a) 423 K, (b) 413 K, (c) 403 K. Data points correspond to the experimental data and the solid lines are fitted first-order kinetic models. . . . . . . . . . . . . . .  Figure 2.8  35  Comparison between measured and calculated conversion profile for N-ethylcarbazole hydrogenation at 413 K at three different initial concentrations of N-ethylcarbazole. First-order rate constants are reported in Table 2.4. (◦) experimental data, (solid line) fitted first-order kinetic model. . . . . . . . . . . .  36  Figure 3.1  Schematic diagram for atmospheric dehydrogenation system .  42  Figure 3.2  TPR profile for the calcined Pd/SiO2 . . . . . . . . . . . . . .  44  xvii  Figure 3.3  Product distribution for dehydrogenation of dodecahydro-Nethylcarbazole obtained by GC/MS: (a) 423 K, (b) 433 K, (c) 443 K. (◦) experimental data, (solid line) fitted model. The rate constants are reported in Table 3.2. . . . . . . . . . . . . . . .  Figure 3.4  46  Comparison of 1,2,3,4-tetrahydrocarbazole conversion at different temperatures obtained from liquid samples analysis . .  48  Figure 4.1  TPR patterns of SiO2 supported Pd catalysts. . . . . . . . . .  56  Figure 4.2  Profile of effluent gases during thermal treatment of catalyst samples in He as measured by mass spectrometer using 0.4 g of sample. (a) SiO2 gel calcination in He at 773 K, (b) 10 wt % Pd/SiO2 calcination in He at 773 K, (c) 10 wt % Pd/SiO2 flush in 10 % H2 in He at room temperature, (d) 10 wt % Pd/SiO2 reduction in 10 % H2 in He at 673 K. . . . . . . . . . . . . .  Figure 4.3  58  Profile of effluent gases during thermal treatment of catalyst samples in air as measured by mass spectrometer using 0.4 g of sample. (a) SiO2 gel calcination in air at 773 K, (b) 10 wt % Pd/SiO2 calcination in air at 773 K, (c) 10 wt % Pd/SiO2 flush in 10 % H2 in He at room temperature, (d) 10 wt % Pd/SiO2 reduction in 10 % H2 in He at 673 K. . . . . . . . . . . . . .  Figure 4.4  Temperature programmed desorption (TPD) of CO from Pd/SiO2 catalysts with different Pd loadings at a heating rate 10 K/min.  Figure 4.5  59 61  Selected HRTEM images of reduced Pd/SiO2 samples: (a) 10 wt % Pd, (b) 4 wt % Pd, (c) 0.5 wt % Pd. Samples were calcined in He flow at 773 K for 6 h prior to reduction in H2 at 673 K for 1 h. . . . . . . . . . . . . . . . . . . . . . . . . . .  Figure 4.6  63  Product distribution from dodecahydro-N-ethylcarbazole dehydrogenation at 443 K and 101 kPa over Pd/SiO2 with Pd loading of 0.5, 0.7 and 1 wt %; ( ) dodecahydro-N-ethylcarbazole, (△) octahydro-N-ethylcarbazole, ( ) tetrahydro- N-ethylcarbazole, ( ) N-ethylcarbazole. . . . . . . . . . . . . . . . . . . . . .  xviii  64  Figure 4.7  Product distribution from dodecahydro-N-ethylcarbazole dehydrogenation at 443 K and 101 kPa over Pd/SiO2 with Pd loading of 2, 3 and 4 wt %; ( ) dodecahydro-N-ethylcarbazole, (△) octahydro-N-ethylcarbazole, ( ) tetrahydro- N-ethylcarbazole, ( ) N-ethylcarbazole. . . . . . . . . . . . . . . . . . . . . .  Figure 4.8  65  Product distribution from dodecahydro-N-ethylcarbazole dehydrogenation at 443 K and 101 kPa over Pd/SiO2 with Pd loading of 5, 10 and 5 wt % (air calcined); ( ) dodecahydro-Nethylcarbazole, (△) octahydro-N-ethylcarbazole, ( ) tetrahydroN-ethylcarbazole, ( ) N-ethylcarbazole. . . . . . . . . . . . .  Figure 4.9  66  TOFs for dodecahydro-N-ethylcarbazole dehydrogenation at 443 K and 101 kPa over Pd/SiO2 catalyst with different Pd metal particle size. The open data point represents the TOF over the 5 wt % Pd/SiO2 reported by Sotoodeh et al.. The solid line represents the data fit using Murzin’s model. The fitted parameter values were: P1 = 1.394±1.189, P2 = 0.0020±0.0016,  γ = 81.60±12.01 and α = 0.124±0.030, with R2 = 97.7 %. . . Figure 4.10 H2 recovery comparison for dodecahydro-N-ethylcarbazole dehydrogenation at 443 K and 101 kPa (a) over 4 wt % Pd/SiO2 prepared by calcination in He (b) over 5 wt % Pd/SiO2 calcined in air and reported previously. . . . . . . . . . . . . . . . . . Figure 5.1  69  71  TPR patterns of (a) 2 wt % reduced Pd/SiO2 catalyst prepared by calcination in He; (b) 5 wt % commercial pre-reduced Pd/C catalyst. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  Figure 5.2  83  Product distributions over time for dehydrogenation of dodecahydrocarbazole in decalin at 101 kPa and (a) 443 K, (b) 433 K, (c) 423 K over the 5 wt % Pd/C catalyst. ( ) carbazole, ( ) tetrahydrocarbazole, (◦) hexahydrocarbazole, (△) octahydrocarbazole, (⋄) dodecahydrocarbazole. . . . . . . . . . . . . .  Figure 5.3  85  Arrhenius plot for dehydrogenation of dodecahydrocarbazole at 423 - 443 K and 101 kPa over a 5 wt % Pd/C. . . . . . . . .  xix  86  Figure 5.4  Product distribution over time for dodecahydrofluorene dehydrogenation in decalin at 443 K and 101 kPa over the 5 wt % Pd/C catalyst. Reactant concentrations (closed data points) are shown on the left Y-axis and products concentrations (open data points) on the right Y-axis. . . . . . . . . . . . . . . . .  Figure 5.5  88  Product distribution over time for dehydrogenation of pure dodecahydrofluorene at 443 K and 101 kPa over the 5 wt % Pd/C catalyst. Reactant concentrations (closed data points) are shown on the left Y-axis and products concentrations (open data points) on the right Y-axis. . . . . . . . . . . . . . . . .  Figure 6.1  Chemical structure of dodecahydro-N-ethylcarbazole with labeled H atoms . . . . . . . . . . . . . . . . . . . . . . . . . .  Figure 6.2  89  99  Chemical structure of dodecahydro-N-ethylcarbazole with labeled C and N atoms. For dodecahydrofluorene, N is replaced by C9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100  Figure 6.3  Hollow/atop adsorption of dodecahydro-N-ethylcarbazole isomer No. 1 on Pd(111); hollow with respect to the 5-membered ring; atop with respect to the axial C-H bonds. The dots indicate the position of the two stretched axial C-H bonds. . . . . 107  Figure 6.4  Optimized geometry of dodecahydro-N-ethylcarbazole isomer No. 1 hollow/atop adsorption on Pd(111) surface; (a) front view, (b) side view with N-Pd distance of 3.84 A˚ and the 5membered ring at an angle of 10◦ to the surface. . . . . . . . . 109  Figure 6.5  Densities of states for; (a) dodecahydro-N-ethylcarbazole in vacuum; (b) adsorbed dodecahydro-N-ethylcarbazole molecule; (c) dodecahydro-N-ethylcarbazole on Pd(111); (d) Pd(111) dstates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112  Figure 6.6  Two possible dehydrogenation pathways for dodecahydro-Nethylcarbazole conversion to octahydro-N-ethylcarbazole; for both pathways dehydrogenation begins with dissociation of H9a . 113  xx  Figure 6.7  Schematic of pathway 1 for dehydrogenation of dodecahydroN-ethylcarbazole; (a) adsorption geometry for dodecahydroN-ethylcarbazole on Pd(111) showing the interaction between ˚ H9a and H8aa and the nearest Pd atoms at 1.99 and 2.05 A, respectively, (b) the optimized geometry corresponding to the dissociation of C9a -H9a bond. The bond distance of C4a -H4aa ˚ respectively, (c) and C8a -H8aa increased to 1.13 and 1.12 A, removal step of H4aa with C8a -H8aa bond distance increased ˚ (d) removal step of H8aa with C4b -H4bb bond disto 1.16 A, ˚ and (e) production of octahydro-Ntance increased to 1.18 A, ethylcarbazole. . . . . . . . . . . . . . . . . . . . . . . . . . 115  Figure 6.8  Comparison of the calculated reaction barriers for the progressive dehydrogenation of dodecahydro-N-ethylcarbazole over Pd(111) surface following the reaction pathways 1 ( ) and 2 (◦) illustrated in Figure 6.6. . . . . . . . . . . . . . . . . . . 116  Figure 6.9  Octahydro-N-ethylcarbazole (a) aromatic hollow di-σ adsorption on Pd(111); (b) front view of the optimized geometry with N-Pd distance of 2.89 A˚ and the 5-membered ring at an angle of 5◦ to the surface. . . . . . . . . . . . . . . . . . . . . . . . 117  Figure 6.10 Densities of states (DOS) of octahydro-N-ethylcarbazole for; (a) C9a in vacuum; (b) C9a in the adsorbed molecule; (c) C4a in vacuum; (d) C4a in the adsorbed molecule. Dash lines represent the s-orbital and (solid line) represents the p-orbitals. . 119 Figure 6.11 Dehydrogenation steps for octahydro-N-ethylcarbazole; (a) adsorbed geometry octahydro-N-ethylcarbazole showing the in˚ (b) teraction between H1b and the nearest Pd atom at 2.53 A, the optimized geometry corresponding to the dissociation of C9a -H9a bond and removal of H1b . Interaction between H2b and the nearest Pd atom at 2.24 A˚ increases, (c) increasing the ˚ (d) optiinteraction of H4a with the nearest Pd atom at 2.16 A, mized geometry corresponding to the dissociation of H4a and increasing the interaction of H3a with the nearest Pd atom at ˚ (e) production of tetrahydro-N-ethylcarbazole. . . . . . 120 2.5 A, xxi  Figure 6.12 Tetrahydro-N-ethylcarbazole (a) aromatic bridge 1,2-di-σ /3,4-  π adsorption on Pd(111); (b) front view of the optimized geometry with N-Pd distance of 3.28 A˚ and the 5-membered ring at an angle of 25◦ to the surface. . . . . . . . . . . . . . . . . 122 Figure 6.13 Optimized geometry of dodecahydro-N-ethylcarbazole adsorbed over Pd(110) surface; (a) side view showing the interaction of the two H atoms with the first layer Pd surface, (b) front view showing the C-H bond distances increase from 1.10 A˚ in vacuum to 1.16 A˚ upon approaching the surface , (c) top view. The molecule moves so that the two C-H bonds adjacent to N heteroatom are at atop positions over the nearest Pd atoms. Blue: N, white: H, gray: C, light gray: first layer Pd atoms, black: second layer Pd atoms. . . . . . . . . . . . . . . . . . . . . . 124 Figure 6.14 Comparison between the adsorbed structures of (a) dodecahydroN-ethylcarbazole, (b) dodecahydrocarbazole, and (c) dodecahydrofluorene over Pd(111). (left) front view, (right) side view. 125 Figure 6.15 Comparison between the adsorbed structures of; (a) tetrahydroN-ethylcarbazole, (b) tetrahydrocarbazole, and (c) tetrahydrofluorene, over Pd(111). . . . . . . . . . . . . . . . . . . . . . . 130 Figure C.1  H2 20 SCCM MFC calibration curve obtained from the data shown in Table C.2. . . . . . . . . . . . . . . . . . . . . . . . 160  Figure C.2  MS calibration curve for H2 . . . . . . . . . . . . . . . . . . . 162  Figure E.1  First-order rate constant calculations for the hydrogenation reaction at 413 K. . . . . . . . . . . . . . . . . . . . . . . . . . 167  Figure E.2  H2 profile obtained from the gas phase analysis using MS. . . 169  Figure E.3  First-order reaction rate constant calculated using the reactant concentration. . . . . . . . . . . . . . . . . . . . . . . . . . . 170  Figure H.1  N2 adsorptiondesorption isotherms measured at 77 K for Mo2 C and alkali-treated Mo2 C catalysts. . . . . . . . . . . . . . . . 187  Figure H.2  Pore size distribution of Mo2 C and alkali-treated Mo2 C catalysts.188  Figure H.3  X-ray diffractograms of Mo2 C and alkali-treated Mo2 C catalysts.189 xxii  Figure H.4  SEM micrographs of (A) Mo2 C and (B) 2 h alkali-treated Mo2 C.190  Figure H.5  Temperature-programmed oxidation of Mo2 C and 2 h alkalitreated Mo2 C catalysts. . . . . . . . . . . . . . . . . . . . . . 191  Figure H.6  CO consumption rate and product selectivities obtained over the un-treated Mo2 C (open symbols) and the 2 h alkali-treated Mo2 C (solid symbols) measured at 8.2 MPa, 573 K, CO:H2 = 1 and GHSV = 3960 h−1 . . . . . . . . . . . . . . . . . . . . . 193  xxiii  Glossary a  =  Lattice parameter (nm)  AB  =  Ammonia borane  amu  =  Atomic mass unit  AT-2  =  Alkali-treated for 2 h  AT-6  =  Alkali-treated for 6 h  AT-12  =  Alkali-treated for 12 h  a.u.  =  Atomic unit  AWPd  =  Atomic mass of Palladium (g/mol)  BET  =  Brunauer Emmett Teller surface area analysis  c  =  Lattice parameter (nm)  CA0  =  Reactant initial concentration (mol/cm3 )  C-H  =  Carbon hydrogen bond  DAB  =  Bulk or Knudsen diffusivity (m2 /sec)  dcluster  =  Cluster diameter (nm)  De  =  Effective diffusivity (m2 /sec)  DFT  =  Density Functional Theory  DMol3  =  A unique density functional theory (DFT) quantum mechanical code  D.O.E.  =  Department of Energy  D.O.L.  =  Dehydrogenated Organic Liquid  DOS  =  Density of state  dp  =  Average pore size (nm)  DTG  =  Differential thermal analysis  xxiv  Ei  =  Numerical value of the energy of state i  EDX  =  Energy Dispersive X-ray spectroscopy  EXC  =  Exchange correlation energy  GC/MS  =  Gas Chromatograph/Mass Spectrometer  GGA  =  generalized gradient approximation  GHSV Hˆ  =  Gas hourly space velocity (h−1 )  =  Hamiltonian operator of the total energy for the system  Ha  =  Atomic energy unit (1 Hartree= 627.5 kcal/mol)  HC  =  Hydrocarbon  HF  =  Hartree-Fock  HRTEM  =  High Resolution Transmission Electron Microscopy  ∆H◦R  =  Heat of hydrogenation (kJ/mol)  i.d.  =  Inner diameter (nm)  kB  =  Boltzmann constant  LCAO  =  Linear Combination of Atomic Orbitals  LDA  =  Local density approximation  LYP  =  Gradient−corrected correlation functional of Lee, Yang, and Parr  µ MFC mMmin−1 gcat −1 MPd MS nm o.d. P1 P2 PBE PES PGMs PW91  =  Solvent viscosity (kg/(sec.m))  =  Mass flow controller  =  Millimolar per minute per gram of catalyst  =  Palladium loading (wt %)  =  Mass spectrometer  =  Nanometer (= 10−9 m)  =  Outer diameter (nm)  =  Frequency of reaction step  =  Frequency of reaction step  =  Perdew−Burke−Ernzerhof functional  =  Potential energy surface  =  Platinum group metals  =  Perdew and Wang generalized gradient approximation  xxv  R0  =  Solute radius (m)  rpm  =  Revolutions per minute  SBET  =  BET surface area (m2 /g)  St−micro  =  t-plot micropore area (m2 /g)  St−ext  =  t-plot external surface area (m2 /g)  SCCM  =  Standard cubic centimeter per minute  SCF  =  Self-consistent-field  SEM  =  Scanning Electron Microscopy  STP  =  Standard temperature and pressure  T  =  Temperature (K)  Tb  =  Boiling point (K)  TCD  =  Thermal conductivity detector  TGA  =  Thermal gravimetric analysis  THCZ  =  1,2,3,4-Tetrahydrocarbazole  Tm  =  melting point (K)  TOF  =  Turn-over frequency (s−1 )  TPD  =  Temperature Programmed Desorption  TPR  =  Temperature Programmed Reduction  UHP  =  Ultra high purity  Vm  =  Adsorbed volume at monolayer coverage (cm3 /gcat )  Vp  =  Volume of pores between 1.7 nm and 300 nm width (cm3 )  Zα  =  Charge of the nuclei  ∇2  =  Laplacian operator  ρ (r) Ψ ϕ α γ σ π  =  Electronic density  =  Wave function  =  Weisz-Prater criterion for internal mass transfer  =  Polanyi parameter (Chapter 4), degree (Chapter 6)  =  Surface free energy  =  Sigma bond  =  Pi bond (or electrons)  xxvi  ϕP σC τ˜ λ β θ  =  Pellet porosity  =  Constriction factor  =  Tortuosity  =  Wave length of radiation  =  Peak width in radian  =  Angle of diffraction  xxvii  Acknowledgments First and foremost, I wish to express my deepest and sincerest gratitude to my supervisor, Professor Kevin J. Smith whose great personality, wide knowledge, encouragement, patience, guidance and important support during my work enabled me to develop an understanding of the project. It is an honor to know Professor Smith. He is a great teacher and I have learnt a lot from him. I hope to utilize his advices in my future career and my life. I would like to sincerely thank Professor Jim C. Lim, Professor Elod Gyenge and Professor Walter Merida for being part of my examination committee. I am grateful to the Chemical and Biological Engineering Department at the University of British Columbia for giving me the opportunity to pursue this research and NSERC for providing financial support for my research. I would like to thank all of those who supported me during the completion of my thesis, the friendly catalysis group members (Xuebin Liu, Liang Zhao, Benjamin J.M. Huber, Zaman Fakhruz Sharif, Shahin Goodarznia, Hooman Rezaei, Shahrzad Jooya Ardakani, Ross Kukard, Victoria Whiffen and Rui Wang), CHBE office secretaries Helsa Leong, Lori Tanaka and Amber Lee, CHBE store personnel Ricard Ryoo and Horace Lam, CHBE IT manager Richard Zhang, CHBE workshop technicians Doug Yuen, Alex Thng, Graham Liebelt, Gordon Cheng, David Roberts and Charles Cheung, the Bioimaging facility in the Department of Botany at the University of British Columbia for the high resolution transmission electron microscopy (HRTEM) measurements specially Bradford Ross and Derrick xxviii  Horne, Marry Fletcher in the department of Material Engineering at the University of British Columbia for helping me in the SEM, TEM and X-ray measurements, Dr. Philip Wong at the advanced materials and process engineering laboratory (AMPEL) at the University of British Columbia for helping me on the set-up of the mass spectrometer, and Professor Philip G. Jessop’s group at the Chemistry department of Queen’s University, Kingston, Canada. And last but not the least, I wish to thank my beloved parents, Niloofar and Fazel, for their love and support.  xxix  Dedication To my beloved parents and to the loving memory of my grandfather  xxx  Chapter 1  Introduction  1  1.1  Introduction  Hydrogen, an alternative fuel, can be produced from different sources such as fossil fuels and renewables such as water with the aid of solar or wind energy [1–7], and biomass [8–10]. Furthermore, hydrogen is non-toxic and does not have any contaminating impact on the environment, because the only products of burning hydrogen are water and energy [1]. Although using hydrogen has many advantages, there is a gap in the technology of hydrogen storage as well as in hydrogen transport infrastructure [1]. Because of the low density of hydrogen, storage and transport of hydrogen requires the use of large scale compressors [11]. Also, current gas piping systems can not be used for pure hydrogen transport due to the embrittlement effect of hydrogen on steel [1]. Although alternative methods, including compressed or liquified hydrogen [12], or using hydrogen carriers such as methanol and organic liquids, have been considered, the hydrogen storage capacity and safety of the storage system are key factors that will determine their efficacy [1]. Due to the above mentioned reasons, the storage and transportation of hydrogen is considered one of the toughest challenges in the hydrogen economy [11]. A fuel efficient automobile requires ∼ 4 to 8 kg of H2 stored on-board the vehicle to match current fuel consumption needs [13]. Cryogenic tanks used for storage of liquid hydrogen or high pressure hydrogen tanks, are normally heavy and expensive and therefore often impractical [13]. Numerous other storage methods have been examined [13–17]. These methods can be summarized as: physical hydrogen storage, including cryogenic methods and high pressure tanks, chemical storage methods, including reversible absorption in metal or metal hydrides and reversible absorption on solids [13], and hydrogen storage on organic heteroaromatic compounds [15–17]. Among these alternatives, organic heteroaromatic compounds have received increasing attention. They have high volumetric and/or gravimetric hydrogen storage capacity, which minimizes the need for changing the existing transport infrastructure. Also, the stored hydrogen can be liberated from the storage medium through a chemical reaction without decomposition of the storage medium.  2  1.2  Objective of this thesis  The objective of this thesis was to study the kinetics of hydrogenation and dehydrogenation reactions used for hydrogen storage and recovery from organic heteroaromatic compounds. The focus was to improve the dehydrogenation catalytic reactions by synthesizing and/or modifying the catalysts in order to obtain high hydrogen recovery applicable for transportation applications. The objectives of the dehydrogenation reactions were to obtain 100 % selectivity to the completely dehydrogenated product and to avoid hydrogenolysis reactions so that the stored H2 could be completely recovered.  1.3  Approach  The approach to reach the objective was to use both experimental techniques and theoretical methods. N-ethylcarbazole was chosen as a potential candidate for hydrogen storage and recovery. Hydrogenation reactions were carried out over a prereduced commercial catalyst to obtain dodecahydro-N-ethylcarbazole, the required feed for the dehydrogenation experiments. Various catalysts with different particle sizes and preparation methods were synthesized and used for the dehydrogenation reactions. Density functional theory (DFT) calculations were used to understand the mechanism of the dehydrogenation reaction of dodecahydro-N-ethylcarbazole over Pd and to investigate the role of the N heteroatom on the reaction, by comparing the adsorption energies of dodecahydro-N-ethylcarbazole and its dehydrogenated products with dodecahydrocarbazole and dodecahydrofluorene. The goal of the experimental work was to achieve a significant improvement in the selectivity to N-ethylcarbazole by modifying the catalyst properties. In the experimental study, the Pd/SiO2 catalysts were prepared by impregnation of a silica support with different loadings of Pd (0.5 to 10 wt %). Two different environments, air and He, were used for calcination of the catalysts. Temperature programmed reduction was used to reduce the oxide form of the precursor from PdO to Pd for the former, and PdCl2 to Pd for the latter. The prepared catalysts were analyzed using different characterization techniques, including BET, TPR, CO pulse chemisorptions, TPD, SEM, EDX and HRTEM. A 300 cm3 autoclave batch reactor with continuous monitoring and control of stirrer speed, temperature, 3  and pressure was used for the hydrogenation reactions. Hydrogenation reactions were carried out at different temperatures over a commercial pre-reduced Ru catalyst and liquid samples were withdrawn from the reactor at certain time intervals in order to study the kinetics of the reactions. For the dehydrogenation reactions, a 50 cm3 glass flask reactor with flow of He as a carrier gas for continuous removal of the produced hydrogen, was used. The exit gas composition was continuously monitored using a quadrupole mass spectrometer. Liquid samples were also withdrawn from the reactor periodically. The analysis of the liquid samples from the dehydrogenation reactions was done by a GC/MS. Power law kinetic models were applied to the hydrogenation and dehydrogenation reactions for the produced intermediates and products. A Pd(111) surface was used to study the dehydrogenation of dodecahydroN-ethylcarbazole by DFT. The surface model was used to investigate the potential energy surface of the formation of N-ethylcarbazole from dodecahydro-Nethylcarbazole. For this purpose, the dodecahydro-N-ethylcarbazole molecule was first adsorbed on the surface by considering different adsorption modes. Then, stepwise removal of a H atom was investigated. In order to determine the energy barrier corresponding to each step, a transition state search was performed between two stable surface intermediates. A Pd(110) surface was created to study the adsorption geometry of the dodecahydro-N-ethylcarbazole molecule and to study the structure sensitivity of the dehydrogenation reaction. To investigate the effect of the N heteroatom as well as the ethyl group in the dehydrogenation reaction of dodecahydro-N-ethylcarbazole, adsorption geometries and energies of similar compounds, dodecahydrocarbazole and dodecahydrofluorene on the Pd(111) surface were also studied.  1.4  Hydrogen storage  To achieve global introduction of hydrogen fuel cells, a safe and practical hydrogen storage system is needed [18]. In order to achieve a hydrogen economy, developing a reliable hydrogen storage system that meets the cost and safety requirements for both on-board (vehicular applications) and off-board (such as power generation stations) uses is crucial [19]. The challenge of the past decade has been to store  4  enough hydrogen on-board a vehicle for an average driving distance of 300 miles, within reasonable cost limitations and comparable to current gasoline vehicles [19]. For transportation applications, there are many criteria that are important for hydrogen storage. Among these, hydrogen storage density (gravimetric and volumetric densities), hydrogen release (recovery) rate as well as cost are very critical. The technical targets for hydrogen storage for the years 2010 and 2015 (see Table 1.1) have been determined by the U.S. Department of Energy (D.O.E.) [20, 21], by comparing to the current gasoline systems in terms of weight, volume, operating conditions at which hydrogen is released, system cost, etc. Table 1.1: U.S. Department of energy (D.O.E.) revised targets for hydrogen storage systems [20, 21] Storage parameter Gravimetric capacity (kWh/kg) [21] H2 (wt%) [21] Volumetric capacity (kWh/L) [21] H2 (g H2 /L system) [21] Durability/Operability [20, 21]: • Operating ambient temperature • Min/max delivery temperature • Cycle life (1/4 tank to full)a • Max delivery pressure from tank [21] System fill time for 5 kg H2 (min) [21] a Equivalent to  Target 2010 1.5 4.5 0.9 28  Target 2015 1.8 5.5 1.3 40  • -30/50 ◦ C • -40/85 ◦ C • 1000 • 12 bar (abs) 4.2  • -40/60 ◦ C • -40/85 ◦ C 1500 12 bar (abs) 3.3  200,000 and 300,000 miles, respectively, based on current gasoline tank specifica-  tions [21].  There are different approaches to on-board hydrogen storage that are currently under study. These approaches include compressed hydrogen, liquified hydrogen, hydrogen storage on metal hydrides, ammonia borane based compounds, boron based compounds like metal borohydrides, organic materials for example carbazole and its substituted forms, and adsorbents with high surface area such as activated carbon [22, 23]. Those materials that can be recharged with hydrogen gas on-board the vehicle are considered as “reversible” hydrogen storage systems, for example, high pressure or liquified hydrogen, and hydrogen adsorbents [23]. On the other hand, there are methods in which hydrogen needs to be stored and recovered from the storage material through a chemical reaction [24], so called “off-board regen5  erable” systems [23]. In these systems, the dehydrogenated medium is removed from the vehicle and transported to an off-board station for re-hydrogenation or in other words “re-generation” of the fuel [23]. There are certain advantages and drawbacks using each of the indicated methods for hydrogen storage. For instance, compressed or liquified hydrogen does not have satisfactory storage capacities [25, 26]. Furthermore, cost of the cryogenic and high pressure equipment is prohibitive [23]. A very important parameter for chemical storage of hydrogen is the heat of the dehydrogenation reaction (reverse reaction of hydrogenation used for hydrogen storage). The desirable value of heat of dehydrogenation is reported to be 40 kJ/mol [27]. It has been indicated by Cooper et al. [28] that heat of dehydrogenation of ∼ 42 to 54 kJ/mol H2 will enable the catalytic dehydrogenation reaction at temperatures <473 K. The other important factor is the rate of hydrogen discharge (recovery) and the kinetics of the dehydrogenation reactions on-board the vehicle. For hydrogen storage on adsorbents such as carbon-based compounds, slow kinetics is reported [26] and normally cryogenic temperatures are required due to the weak bonding between the hydrogen and the adsorbent [23]. In this case, gravimetric capacities as well as cost of the cryogenic equipment need to be addressed. Therefore, the big challenge in transportation applications has been to develop a hydrogen storage system with sufficient storage capacity to enable a vehicle to drive an average distance of 300 miles within the limitations imposed by safety and economic considerations. This will require a minimum storage of 5 kg hydrogen on-board the vehicle [14]. The following sections discuss different approaches to hydrogen storage.  1.5  Compressed hydrogen  It has been shown that the volumetric density of hydrogen does not have a linear relationship with pressure. In other words, increasing the pressure does not lead to a linear increase in the hydrogen density [25]. For example, the hydrogen density is 20 kg/m3 at 300 bar [25]. This volumetric density can be increased to 40 kg/m3 at 700 bar and 70 kg/m3 at 2000 bar [25]. The latter is impractical [25] and can not be used on-board the vehicle. The volume of 5 kg hydrogen is 60 m3 , 0.25 m3  6  and 0.125 m3 at pressures of 1 bar, 300 bar and 700 bar, respectively [25]. The observed decrease in the hydrogen volume by increasing the pressure from 300 to 700 bar is not as significant as the volume decrease by increasing the pressure from 1 bar to 300 bar. The hydrogen pressure that is typically used is 350 bar and 700 bar [25, 29, 30]. However, the energy and the system requirements for compression of hydrogen is the main issue [12] that limits the use of compressed hydrogen and research is needed to improve the compressed hydrogen storage system [25].  1.6  Liquified hydrogen  Liquified hydrogen can be obtained by decreasing the hydrogen gas temperature at a constant pressure [25]. A liquid hydrogen container can be filled on-board the vehicle in 3 min [30]. Liquid hydrogen has a density of approximately 80 kg/m3 at 22 K and a pressure of 4 bar [25]. Compared to compressed hydrogen, the volume required for storage of liquified hydrogen is lower [25]. However, due to the physical properties of hydrogen gas, there are many limitations and constraints to the use of liquified hydrogen as a storage system [25]. The boiling point of hydrogen at atmospheric pressure is 20 K, a temperature at which other gases are solid [25]. In order to keep hydrogen liquid, the weight and volume of the cooling equipment will need to be added to the storage system [31] and strong thermal insulators are required to keep hydrogen from being vaporized [25]. The 2015 targets of the department of energy (D.O.E.) of 40 g H2 /L system include the hydrogen storage material as well as the total storage system such as high pressure tanks for compressed hydrogen [14], which means that the theoretical density of the storage material needs to be higher in order to achieve the required storage capacity for the whole system.  1.7  Off-board regenerable storage  One category of materials that has the potential to meet the H2 storage density requirements for transportation applications is off-board regenerable systems. Many of the compounds that have high volumetric/gravimetric hydrogen storage capacities have been studied as summarized in Table 1.2 [14], and are described in the following sections. 7  Table 1.2: Different categories of hydrogen storage materials [14] Hydrogen storage category  Example  Ammonia Borane-based Boron-based  Ammonia borane complexes (ammonia borane NH3 BH3 ) Metal borohydride such as NaBH4  Metal hydrides  Metal hydride such as NaAlH4  Nitrogen-Based  Metal amides and metal nitrides  Organic materials  Substituted carbazole such as N-ethylcarbazole (hydrogenated forms)  1.7.1 Ammonia borane-based H2 storage materials A chemical reaction with controlled temperature and pressure is needed in this group of materials for “charge” (or uptake) and “discharge” (or release) of hydrogen [31]. Ammonia borane (AB, NH3 BH3 ) is one example. It is solid and stable at room temperature [31] and soluble in relatively polar solvents [32]. The melting point of NH3 BH3 varies from 383 to 387 K [32]. The hydrogen content of ammonia borane is 190 g H2 /kg material, equal to 100-140 g H2 /L [31], or 19.6 wt % hydrogen [26, 33]. The high hydrogen storage capacity as well as its stability has made ammonia borane a potential candidate for hydrogen storage and recovery applications. Hydrogen can be released from ammonia borane on-board the vehicle. However, there are technical issues that need to be addressed before the commercial use of ammonia borane, one of which is the rate at which hydrogen is released. Also, since hydrogen is released from the compound, a cost effective pathway should be developed to regenerate ammonia borane [31]. Currently, no energy efficient approach is available for regenerating ammonia borane from the products [34]. In work by Stephens et al. [34], dehydropolymerization of NH3 BH3 was reported to have practical reaction rates over an Ir catalyst and very high hydrogen recovery over a Ni catalyst. However, no catalyst was reported to have both high reaction rate and high hydrogen recovery [34]. Other options, such as adding water or ionic liquids in order to improve the catalytic hydrogen recovery from ammonia borane, also decrease the storage capacity of the system [34] and  8  hence, would not be practical. Therefore, the challenge is to discover and develop a system with high storage capacity, as well as fast and complete hydrogen recovery.  1.7.2 Chemical hydrides Chemical hydrides are another group of materials which have the potential for hydrogen storage. Complex metal borohydrides with the general formula of M(BH4 )n have been of interest because of their high hydrogen storage capacity [27]. One example is sodium borohydride (NaBH4 ) that has a theoretical hydrogen content of 10.6 wt%. The hydrolysis reaction of sodium borohydride, shown in reaction 1.1, is an exothermic reaction that produces hydrogen and energy in the presence of a catalyst [35, 36], NaBH4 (aq) + 2H2 O → 4H2 + NaBO2 + ∼ 300kJ  (1.1)  Pt−LiCoO2 catalyst was reported by Kojima et al. [37] to work very well for this reaction. The produced H2 has high purity and does not poison the fuel cell catalyst [35]. A 30 wt % solution of NaBH4 has the volumetric storage capacity of about 63 g H2 /L, which is comparable to the volumetric storage of 71 g H2 /L for cryogenic liquid H2 and 23 g H2 /L for compressed hydrogen at 5000 psi [35]. Therefore, NaBH4 is a potential candidate for hydrogen storage in terms of storage capacity [36]. However, hydrogen generation through reaction 1.1 is not reversible [37]. Hence new methodologies for recycling the hydrolysis byproduct, NaBO2 , into NaBH4 are needed [37]. “Lack of reversibility” in the hydrogen generation reactions from metal borohydrides is also reported by Li et al. [38] to limit their use in transportation applications. Metal hydrides exist in different forms and categories among which hydrides in the form of A, AB, A2 B, AB2 and AB5 have been investigated extensively [25]. In these categories, A stands for the element that makes a strong hydride and B the element that forms a weak hydride [25]. An example is MgH2 with high theoretical hydrogen storage capacity of 7.5 wt % [25] which has been of interest for hydrogen storage applications [39]. However, the main obstacle in using MgH2 for hydrogen storage and recovery is its high decomposition temperature [39] and thermodynamic stability [25, 40]. Methods such as doping Mg with transition metals 9  like Ni in order to decrease the dehydrogenation temperature of the metal hydride have been studied [39], however, these will result in reducing the storage capacity of the system [40]. Therefore, using light elements such as Li, B, Na and Al in the storage systems have been considered [25]. Other approaches, such as coating sodium hydride with plastic in the form of pellets that can be sliced open to react with water, as proposed by PowerBall Technologies reported by DiPietro [41] are also very costly and not practical.  1.7.3 Organic materials Storage and release of hydrogen from organic molecules is done through hydrogenation and dehydrogenation reactions. Since the hydrogen is stored in the organic molecule covalently, system compression or liquifaction will not be needed [11]. A liquid organic compound on which hydrogen is stored through a hydrogenation reaction is used as a fuel on-board the vehicle. This fuel can be a hydrocarbon produced from crude oil or alcohols [11]. The advantage of using a liquid fuel is that the need for changing the fuel transport and storage infrastructure will be minimized [11]. The fuel undergoes a catalytic dehydrogenation reaction onboard the vehicle to release hydrogen. The liberated hydrogen is then directed to the fuel cell or the engine in the vehicle and the dehydrogenated fuel is stored in a tank for off-board regeneration through hydrogenation reactions [11, 42]. The following list describes some of the criteria for a liquid hydrogen carrier: • Absolute value of heat of hydrogenation between 42 to 54 kJ/mole H2 to enable the dehydrogenation reaction at temperatures below 473 K [43]. • High boiling point (Tb >573 K) to decrease the vapor exposure on-board the vehicle [43]. • Low melting point (Tm <233 K) for the system to stay liquid at low temperatures [44]. • Low contaminating effect on the environmental [43]. • Low cost of production.  10  • Selective hydrogenation and dehydrogenation reactions, with no decomposition of the organic compound [43].  1.8  Hydrogenation/dehydrogenation reactions and catalysts  Hydrogenation and dehydrogenation reactions are well studied in heterogeneous catalysis. A heterogeneous catalyst typically consists of an active phase which catalyzes the reaction, a support which is normally a high surface area material and a promoter that improves catalyst activity and/or selectivity [45]. Table 1.3 shows some examples of different catalyst components [45]. Table 1.3: Examples of different components of a heterogeneous catalyst [45] Catalyst component Active catalytic phase  Promoter Support  Compound example metals such as Pt, Pd, Ni metal oxides such as MoO2 , CuO metal sulfides such as MoS2 metal oxides such as K2 O, MgO activated carbon, metal oxides such as SiO2  Transition metals and their carbides [45–48], sulfides [46] and nitrides [45, 47, 48] are well known to be very active in catalytic reactions. Pt-group (precious) metals such as Pd, Ru and Pt, as well as transition metals such as Ni and Cu have been shown to be active in hydrogenation reactions [45]. However, precious metals are much more expensive than transition metals and more active [45]. Dehydrogenation is the reverse of hydrogenation and therefore, the catalysts used for hydrogenation are typically used for dehydrogenation reactions as well [45]. Dehydrogenation is an endothermic reaction [45] as opposed to hydrogenation which is exothermic. Representative active catalytic phases for hydrogenation and dehydrogenation reactions are listed in Table 1.4.  1.9  Selection of organic materials for hydrogen storage  Extensive literature is available addressing a wide variety of hydrogenation/dehydrogenation reactions relevant in H2 storage including alkenes/alkanes, benzenes/11  Table 1.4: Active catalytic phases for hydrogenation and/or dehydrogenation reactions Active phase Transition metals Transition metal sulfides Transition metal carbides Transition metal nitrides  Example Fe, Ni, Ru, Pd, Pt sulfides of Mo, Ni carbides of Fe, Mo nitrides of Mo  Reaction hydrogenation, dehydrogenation [45] hydrogenation [45] hydrogenation [45], dehydrogenation [47] dehydrogenation [48]  cyclohexanes, aldehydes, ketones/alcohols and their appropriate catalysts as well as reaction conditions [49–52]. For example, hydrogenation of benzene has been reported over Ni at 323 K, hydrogen pressure of 80 kPa and benzene pressure of 9 kPa. The reaction was first-order in hydrogen pressure with activation energy of 38 kJ/mol [52]. However, many of these organics do not meet the hydrogen storage capacity and other criteria (such as high boiling point) required to be used for H2 storage. Recent work on hydrogen storage [15, 16, 43] is available showing that organic aromatic compounds such as carbazole and its substituted forms have good potential for hydrogen storage. These compounds are mainly in the form of crystalline powders at ambient conditions with high melting and boiling points. Some of these materials are listed in Table 1.5. However, there is little information on the hydrogenation/dehydrogenation reaction mechanism, thermodynamics, kinetics and catalysts for these compounds. In work by Cooper et al. [43], fused multi-ring aromatic systems with high boiling and melting points were also reported to have high hydrogen storage capacity as well as a low absolute value of the standard heat of hydrogenation, -∆HR◦ , which was also reported through the experimental and theoretical (density functional theory) work of Pez et al. [15, 16] (See Table 1.5 for hydrogen capacity and -∆H◦R of some of these compounds). As mentioned before, substrates having a lower absolute value of -∆HR◦ are desired as they are thermodynamically more easily dehydrogenated. However, despite high hydrogen capacity of these heavy compounds and low volatility due to their high boiling points, they usually have high viscosity and are often in the form of a crystalline solid, and thus can not be used by themselves as a liquid fuel. The dehydrogenation mechanism of 1,2,3,4-tetrahydrocarbazole (THCZ), a po-  12  Table 1.5: Examples of substituted carbazole structures with high hydrogen capacity and low heat of hydrogenation. wt % H2 a  -∆H◦R , kJ/molb  carbazole [11, 16, 23]  6.7  60.4 [16]  9-Ethylcarbazole [11, 15]  5.7  50.6 [15]  Hexaazatriphenylene [16]  7.1  53.9 [16]  Pyrrole oligomer [15]  5.2  52.2 [15]  Dibenzofuran [16]  6.7  64.6 [16]  Borafluorene [15]  6.8  42.6 [15]  Compound  Structure  a Storage b Due  capacity of the hydrogenated compound. to negligible amount of entropy change[15], ∆H◦R ∼∆G◦R .  tential compound for hydrogen storage, was studied by Crawford et al. [53] and Hindle et al. [54] both experimentally and theoretically using density functional theory (DFT) calculations. 1,2,3,4-tetrahydrocarbazole is in the form of a solid powder with high boiling point (598 K) and melting point (391 K). Therefore, the dehydrogenation reaction was carried out in the liquid phase by dissolving the 1,2,3,4-tetrahydrocarbazole in mesitylene. Two different catalysts were used in the reaction: 5 wt % Pd/Al2 O3 with a Brunauer-Emmett-Teller (BET) surface area of 143 m2 /g and 22 % metal dispersion measured by CO pulse chemisorption [53, 54], and Pd/C with a BET surface area of 1358 m2 /g and 16.5 % metal dispersion [54]. The reaction was zero-order in the substrate (THCZ) concentration, and the cata-  13  lysts showed 100 % selectivity to the desired carbazole product. However, it was shown from the experiments that product inhibition slowed the rate of reaction significantly and 81 % conversion was obtained after about 27 h at 101 kPa and 413 K [53]. The product inhibition was confirmed by the DFT study, and was attributed to the higher adsorption energy of the product carbazole to the catalyst surface and its lower solubility in mesitylene, compared to that of 1,2,3,4-tetrahydrocarbazole [53, 54]. The DFT results in the work by Crawford et al. [53] showed that THCZ molecule in its most stable adsorption geometry adsorbs to the catalyst surface through the aromatic ring parallel to the surface and the aliphatic ring tilted away by a degree of approximately 30o . In this adsorption geometry, a maximum number of C-Pd bonds can form [53]. In work by Pez et al. [15, 16] reversible catalytic hydrogenations of extended pi-conjugated substrates, including aromatic hydrocarbons with oxygen and nitrogen heteroatoms, comprising 5 or 6 membered rings, were studied using supported Pt-group metal catalysts such as Rh/C and Ru/Al2 O3 . Because the heat of hydrogenation (-∆HR◦ ) data is only available for some small pi-conjugated molecules, this value was calculated using density functional theory (DFT) for the potential substrates for hydrogen storage [15, 16]. Based on these results, it was shown that nitrogen as a heteroatom in 5 or 6 membered rings, provided a lower -∆HR◦ in this category of compounds and are therefore effective substrates for hydrogen storage. Clot et al. [55] have also shown by DFT calculations that H2 recovery is thermodynamically favored from five membered rings compared to six membered rings in the presence of a N heteroatom in the organic materials. Wang et al. [56] reported iridium complexes to be active catalysts for the dehydrogenation of dodecahydro-N-ethyl carbazole at 473 K. However, the reaction did not proceed to the production of the completely dehydrogenated product, Nethylcarbazole, and octahydro-N-ethylcarbazole and tetrahydro-N-ethylcarbazole were reported to be the only observed products. Also, much lower catalytic activity was observed at the lower temperature of 423 K during 48 h dehydrogenation reaction. Decalin as a hydrogen storage candidate was dehydrogenated over a Pt/C catalyst. Loutfy et al. [57] performed catalytic decalin dehydrogenation by use of a batch-type membrane reactor at high temperatures of 533 K, 553 K, 573 K and 593 K and obtained conversions below 85 % after 2.2 h. Decalin conversion of 14  higher than 95 % was attained within 1.5 h at temperature of 553 K in the work by Hodoshima et al. [58]. However, the reported reaction temperature was too high and impractical for hydrogen recovery for transport applications.  1.10 Summary Research done on hydrogen storage options has led to the conclusion that the “offboard regenerable” systems are preferred to “reversible on-board”s due to several advantages including high volumetric and/or gravimetric hydrogen densities. Organic materials such as substituted carbazoles and some fused multi-ring aromatic compounds have been shown to have high storage capacity as well as low heat of hydrogenation. However, hydrogenation and dehydrogenation of these materials has been sparsely reported over platinum group metal catalysts. Furthermore, kinetics and selectivity of hydrogenation/dehydrogenation reactions with a view to maximize both hydrogenation and dehydrogenation rate and minimize the side reactions (i.e. maximize the selectivity) has not been well studied. Therefore, substituted carbazoles with high hydrogen storage capacity were chosen for further study in the present work and the hydrogenation/dehydrogenation reaction rate, kinetics and selectivity were determined over platinum group metal catalysts. Density functional theory techniques were also used to better understand the reaction network and catalyst reactivity.  1.11 Molecular modeling of heterogeneous catalysis Molecular modeling in heterogeneous catalytic systems has attracted growing attention due to the ability of first principle techniques to calculate the adsorption energies of the adsorbate to the catalyst surface, find binding energies of the substrates, and predict the potential energy surface of the chemical reactions. Catalysis is a process that occurs at the molecular level, and therefore simulating this process can provide the answer to questions regarding the catalyst behavior as well as the reaction mechanism [59]. Extensive theoretical studies are available in the literature for different heterogeneous catalytic systems including zeolites, metal or metal oxide catalysts. Quantum mechanical techniques are used to calculate the energetics of the system 15  as well as the interaction between the molecules and catalyst surface. The strength of the interaction forces between the molecules and catalyst surface determines the extent of the adsorption and desorption, compound decomposition on the catalyst surface, as well as the reaction kinetics [59].  1.11.1 Schr¨odinger equation The objective of most quantum mechanical techniques is to find the approximate solution to the Schr¨odinger equation developed by E. Schr¨odinger in 1926 [60]. The time-independent form of the Schr¨odinger equation is defined as [61],  ˆ i (⃗x1 ,⃗x2 , ...,⃗xN , ⃗R1 , ⃗R2 , ..., ⃗RM ) = Ei Ψ(⃗x1 ,⃗x2 , ...,⃗xN , ⃗R1 , ⃗R2 , ..., ⃗RM ) HΨ  (1.2)  In this equation, Hˆ is the Hamiltonian operator, M is the number of nuclei in the system, and N the number of electrons where no magnetic or electric field is present [61]. Hˆ is an operator that represents the total energy of the system and is defined as follows [61],  N N M M 1 N 1 M 1 2 N M ZA ZA ZB 1 Hˆ = − ∑ ∇2i − ∑ ∇A − ∑ ∑ +∑∑ + ∑ ∑ 2 i=1 2 A=1 MA r r i=1 A=1 iA i=1 j>i i j A=1 B>A RAB  (1.3)  The first term takes into account the kinetic energy of the electrons, the second term describes the kinetic energy of the nuclei, and the last terms show the attractive electrostatic interaction between the electrons and nuclei and repulsive forces because of electron-electron and nucleus-nucleus interactions [61]. Wave function Ψ is the solution of Eq. 1.4 that describes the “state of a system” and contains all the information about the system under study [62]. “The Schr¨odinger equation describes the wave properties of an electron in terms of its position, mass, total energy, and potential energy” [63]. The numerical value, Ei , is the energy of state “i” which is defined by Ψi [61]. The probability for an electron to be found at a given coordinate is proportional 16  to Ψ2 [63] which gives us a physical sense of the wave function Ψ, since it can not be observed [61]. According to “Pauli’s exclusion principle”, two electrons can not occupy the same state [61] and therefore, the probability of finding electrons somewhere in space equals 1 [61, 63]. This can be shown as follows [61], ∫  ∫  ...  |Ψ(⃗x1 ,⃗x2 , ...,⃗xN )|2 d⃗x1 d⃗x2 ...d⃗xN = 1  (1.4)  Finding an analytical solution to the Schr¨odinger equation is too complex for systems consisting of many atoms and therefore, approximations need to be made [64]. One approximation was proposed by Born and Oppenheimer in 1927 [65] in which the nuclei were assumed to be at a fixed position because of their much slower movement due to their heavier mass, compared to the electrons [64]. The solution obtained by this approximation resulted in a system energy which was only dependent on the nuclei position, a so called “potential energy surface” (PES) [64, 66]. The point associated with the lowest energy on the potential energy surface is the ground state energy of the system [64]. In the Hartree−Fock (HF) approximation, the N-electron wave function, which is complicated to solve, is approximated by the product of N single-electron wave functions (or spin orbitals) [61, 64, 67]. Basis functions are the functions used to define these orbitals [64] and this method is called the Linear Combination of Atomic Orbitals (LCAO) [61, 64]. In the Hartree−Fock technique, a set of orbitals are guessed and the HF equations are solved in an iterative procedure referred to as the self-consistent-field (SCF) method [61]. Quantum mechanic methods used for solving the Schr¨odinger equation can be categorized into ab initio approaches, semi empirical methods or density functional theory techniques [68]. Ab initio methods are very time consuming and costly [64] because they solve the Schr¨odinger equation [68]. Hartree−Fock method as explained earlier is one example of these approaches [68]. In semi-empirical quantum approaches however, the calculations are simplified by applying some assumptions to the equations [64] and adjusting integrations for example, by eliminating most of the two electron integrals [68]. In some semi empirical approaches experimental data are used to simplify some parts of the calculations [64]. Compared to ab  17  initio methods, semi empirical ones also require intensive calculating resources although they are less expensive than ab initio approaches [64]. Another approach in quantum mechanic calculations is density functional theory (DFT) which introduces functionals of electron density by which many electronic properties can be generated [64, 69]. DFT methods have shown to produce reasonable results at low cost and therefore are widely used in molecular simulations [68].  1.11.2 Density functional theory (DFT) Hohenberg and Kohn in 1964 proved that the ground-state energy and other electronic properties of a molecular system is a unique functional of the electron density [62, 70]. The Hohenberg and Kohn theory however did not specify exactly how to calculate the ground state electron density without solving the wave function [62]. This was developed later through a series of equations by Kohn and Sham in 1965 [62, 70, 71]. The idea of using the electron density, ρ (r), simplified the problem of solving the 3N variable Schr¨odinger equation to a function of only three spatial variables [70]. Consider the total energy functional as follows [70], E[Ψi ] = Eknown [Ψi ] + EXC [Ψi ]  (1.5)  where Eknown [Ψi ] represents all the terms which can be described in an analytical form, and EXC includes everything else [70]. For Eknown [Ψi ] we have [70],  Eknown [Ψi ] = +  h2 ∑ m i=1 e2 2  ∫  ∫ ∫  Ψ∗i ∇2 Ψi d 3 r +  ∫  V (r)ρ (r)d 3 r  ′  ρ (r)ρ (r ) 3 3 ′ d rd r + Eion. |r r′ |  (1.6)  in which the terms on the right are “the electron kinetic energies, the Coulomb interactions between the electrons and the nuclei, the Coulomb interactions between pairs of electrons, and the Coulomb interactions between pairs of nuclei”, respectively [70]. EXC [Ψi ] is referred to as the exchange−correlation functional  18  which consists of all other effects that are not considered in the Eknown [Ψi ] term [70] in Eq. 1.5. Since the exact functional for the exchange−correlation energy is not known, different models were introduced to make an approximation for this functional [70]. One of these approximations is called the local density approximation (LDA) assuming a “uniform electron gas” [61, 62, 72]. The exchange−correlation energy is obtained from [61], LDA EXC [ρ ] ∼ =  ∫  ρ (⃗r)εXC [ρ (⃗r)]d⃗r  (1.7)  where in this equation ε XC [ρ (⃗r)] is the exchange−correlation energy of an electron in a uniform electron gas [61]. Another approximation includes functionals that consider the “gradient” and “non-homogeneity” in the charge density [61, 62]. These methods are referred to as generalized gradient approximations (GGA) [61, 62] or gradient−corrected functional [62]. Although generalized gradient approximation methods contain more physical data about the charge density, they are not always more accurate than the local density approximation methods [70]. There are many GGA functionals available in the literature. Perdew and Wang (PW91) GGA for the correlation functional [73], the gradient−corrected exchange functional of Becke (B) [74], the Lee, Yang, and Parr (LYP) gradient−corrected correlation functional [75], and the Perdew−Burke−Ernzerhof functional (PBE) [76] are some examples in this category. It was found that for chemisorption problems, reasonable results can be obtained by using the nonlocal corrections to the exchange−correlation potential as well as the adsorbate−surface geometry optimization [77].  1.12 Molecular modeling in heterogeneous catalysis Techniques that are used in modeling of heterogeneous catalytic systems include cluster methods [77–79], embedded-cluster [79, 80] methods, and periodic models [79, 81]. A good and successful simulation not only benefits from an accurate method, but also the proximity of the used model to the real catalytic system [79]. These methods are briefly discussed in the following sections. 19  1.12.1 Cluster models Cluster models are built by using a limited number of atoms [79]. Cluster models are used widely in simulating the catalytic surfaces because of the limitations in the availability of the computational resources [77]. The model clusters can have different sizes and number of atoms depending on the type of the system under study as well as the desired level of accuracy [79]. In small clusters, the size of the cluster can have a significant effect on the interaction energy which are well known experimentally [82, 83]. “An important advantage of the cluster approach is that the active site is explicitly described by the interactions between the local molecular orbitals of the adsorbate and the surface” [79]. However, a cluster model is generally limited by its small size which is not capable of representing the system completely [77, 79].  1.12.2 Embedded-cluster model The embedded−cluster model tries to deal with the difficulties raised from the cluster termination [79]. In the embedded-cluster approach, the local area around the active site is modeled and the cluster is inserted in a large external environment [79]. Various systems such as metal oxides and zeolites have been modeled successfully using this methodology [79]. Although this approach simplifies the quantum mechanical calculations, accurate matching of the interface between the cluster and the external environment is one of the main difficulties of this approach [79].  1.12.3 Periodic quantum−mechanical calculations Periodic quantum−mechanical methodology can be used to model the well-defined systems such as metals surfaces [79, 84]. A unit cell is created in this approach and is repeated in one to three directions representing a linear, surface or a bulk system [79]. Therefore, the cost of calculation increases accordingly by increasing the size of the unit cell [79]. This approach provides a more complete picture of the system structure [79]. However, compared to the industrial non-ideal systems, only systems with high periodicity can be modeled currently within feasible computational time and cost [79]. 20  In this thesis, a periodic slab with four Pd layers and total number of 108 Pd atoms were used to represent the Pd(111) surface, and periodic quantum mechanical calculations were employed to investigate different reaction steps and energy barriers of dehydrogenation reactions. Material Studio and DMol3 module was used for molecular simulation of the catalyst surface, optimizing the geometry of the adsorbate/surface system, defining the adsorption energies of surface intermediates and the potential energy surface (PES) by performing the transition state search.  1.13 Outline of the dissertation This dissertation is arranged in the following order: Chapter 1: Provides an introduction to the subject of hydrogen storage and different approaches that are used for this purpose. A brief introduction to molecular heterogeneous catalysis and its application in the modeling of dehydrogenation reactions for hydrogen recovery is also described. Chapter 2: Reports the results of hydrogenation reactions of N-ethylcarbazole and carbazole, with details of a kinetic study as well as kinetic modeling of the reaction with the suggested reaction network. Chapter 3: Dehydrogenation reaction of dodecahydro-N-ethylcarbazole over a 5 wt % Pd/SiO2 catalyst is reported and compared to the data available in the literature. Kinetics of the reaction was studied and the modeling data are provided. Chapter 4: The preparation method and characterization results of a series of Pd/SiO2 catalysts with different Pd particle sizes and metal dispersions are reported. The dehydrogenation reaction of dodecahydro-N-ethylcarbazole over these catalysts is reported. Chapter 5: The effect of the N heteroatom in the dehydrogenation reactions of organic polyaromartic compounds is studied and details are reported by comparing the dehydrogenation reaction rate of dodecahydro-N-ethylcarbazole, dodecahydrocarbazole and dodecahydrofluorene. Chapter 6: Provides details of the molecular modeling and DFT calculations used to model the dehydrogenation reaction of dodecahydro-N-ethylcarbazole, as well as an investigation of the effect of N heteroatom in the dehydrogenation reaction of  21  the examined organic polyaromatic structures of dodecahydro-N-ethylcarbazole, dodecahydrocarbazole, and dodecahydrofluorene. Chapter 7: Summarizes the conclusions of the previous chapters and offers recommendations for future work.  22  Chapter 2  Hydrogenation kinetics: N-ethylcarbazole and carbazole A version of this chapter has been published: F. Sotoodeh and K.J. Smith, Kinetics of Hydrogen Uptake and Release from Heteroaromatic Compounds for Hydrogen Storage, Ind. Eng. Chem. Res. 2010, 49, 1018-1026.  23  2.1  Introduction  One approach to hydrogen storage for fuel-cell vehicles is to use organic compounds that have a high capacity to bind hydrogen covalently. Storage and release of the hydrogen is achieved by hydrogenation and dehydrogenation of the organic compounds and these reactions are done in the presence of a catalyst. For transportation applications the H2 storage (hydrogenation) would be done off-board the vehicle. Also, the candidate compounds must have high hydrogen release rates, more than 5.5 wt % H2 storage capacity (calculated as the mass of H2 liberated divided by the mass of hydrogenated compound), melting points below 313 K and boiling points above 473 K [20, 85]. Organic heteroaromatic compounds such as carbazole and its derivatives have attracted attention because they meet the storage capacity targets and the boiling point criteria. Although some results have been reported for H2 storage on heteroaromatic compounds, limited data are available on the kinetics of the reactions involved in H2 uptake and release [14, 17, 55, 86]. Carbazole and N-ethylcarbazole, when hydrogenated to dodecahydrocarbazole and dodecahydro-N-ethylcarbazole, have hydrogen storage capacities of 6.70 and 5.79 wt % [14, 16, 87], respectively. They are, therefore, potential candidates for hydrogen storage applications. However, the hydrogenation-dehydrogenation kinetics of these compounds are not well established, especially at the low temperature (< 473 K) conditions required for onboard dehydrogenation reactions. In addition, although platinum group metals are widely used for hydrogenation and dehydrogenation reactions, the influence of the heteroatom present in these compounds on the reaction kinetics needs to be understood, given that strong heteroatom adsorption can lead to a loss in catalyst activity. The N heteroatom associated with N-ethylcarbazole and dodecahydro-Nethylcarbazole is bound to an ethyl group, whereas in the corresponding carbazole compounds, the N heteroatom is free to interact with the catalyst surface through the lone pair of electrons associated with the N. In the present chapter, the complete hydrogenation cycle needed for hydrogen storage has been investigated for both carbazole and N-ethylcarbazole. The kinetics of the hydrogenation reactions needed for hydrogen storage on N-ethylcarbazole and carbazole are reported. N-ethylcarbazole was hydrogenated over a supported  24  Ru catalyst in an autoclave batch reactor at 7 MPa and 403-423 K. Kinetic models of the hydrogenation reactions, as well as the activation energies of the reactions, are reported. For comparison, carbazole was hydrogenated at 7 MPa and 443 K, using the same catalyst and experimental set-up used for the hydrogenation of Nethylcarbazole.  2.2  Experimental  2.2.1 Catalyst preparation and characterization A commercial 5 wt % Ru on alumina catalyst (Aldrich, 5 wt % Ru, reduced), received in a reduced state, was used as the hydrogenation catalyst and was rereduced prior to use. The catalyst was characterized to determine BET surface area, pore volume and pore size using a Micromeritics ASAP 2020 analyzer. A Micromeritics Autochem II 2920 was used for Temperature Programmed Reduction (TPR) analysis of the catalyst and to determine the metal dispersion by CO pulse chemisorption (see Appendices A and B for details).  2.2.2 Catalyst activity The hydrogenation of N-ethylcarbazole was carried out at 403, 413 and 423 K and a H2 pressure of 7 MPa, in a 300 cm3 autoclave batch reactor with continuous monitoring and control of stirrer speed, temperature and pressure. The schematic of the hydrogenation reactor is shown in Figure 2.1. Approximately 100 cm3 of a 6 wt % solution of N-ethylcarbazole (Aldrich, 97 %) in decalin (Sigma-Aldrich, 98 %) and 1 g of the 5 wt % reduced Ru/alumina catalyst, with a particle size < 150 µ m, were loaded into the reactor. The reactor was first purged in a 55 cm3 (STP) min−1 flow of N2 for about 1 h and then heated to the desired temperature while purging with ultra high purity (UHP) hydrogen at a 55 cm3 (STP) min−1 flow and low pressure (< 35 kPa) to ensure no reaction occurred. After the desired temperature was reached, the reactor was pressurized in hydrogen to 7 MPa, corresponding to more than 2.5 times the amount needed for complete hydrogenation of the N-ethylcarbazole reactant, while stirring at a speed of 600 rpm. The H2 pressure was monitored continuously during the experiment. Small 25  Pressure transducer H2 to reactor Stirring speed transducer  PR  Computer  Temperature transducer  Thermocouple  To vent  Sample line  Autoclave reactor H2 Tank  Figure 2.1: Schematic diagram for high pressure hydrogenation system volumes (< 0.5 cm3 ) of liquid sample were removed from the reactor periodically for analysis by gas chromatography. Hydrogenation of carbazole was carried out using the same autoclave reactor at 7 MPa and 423 K. In this case a 1 wt % solution of carbazole (Aldrich, 95 %) in decalin (Sigma-Aldrich, 98 %), and 1 g of the 5 wt % reduced Ru/alumina catalyst with a particle size < 150 µ m were loaded into the reactor. The same start-up and sampling procedure used for the hydrogenation of N-ethylcarbazole was used for the hydrogenation of carbazole. However, due to the extremely low solubility of carbazole in decalin at room temperature, liquid samples were only taken at the end of the experiment and the identity of the final product was confirmed by GC/MS analysis with a Shimadzu QP-2010S GC/MS using a Restek RTX5 30M x 0.25 mm capillary column. Table 2.1 summarizes the physical properties of the reactants, solvent and the catalyst. The absence of internal and external mass transfer effects on the measured reaction kinetics was confirmed experimentally and by theoretical analysis. For the  26  Table 2.1: Physical properties of hydrogenation reaction materials Compound Appearance MW (g/mol) Density (g/cm3 ) Boiling point (K) melting point (K) flash point (K) Purity (%) Metal (wt %)  N-ethylcarbazole white solid 195.26 1.07 621.3 343 437.4 97 —  Carbazole white solid 167.21 1.301 628 519.3 493 ≽ 95 —  Decalin clear liquid 138.25 0.896 460-469 233 330 ≽ 98 —  Ru/Al2 O3 black powder — — — — — ≽ 98.5 5  hydrogenation reactions, the gas-to-liquid mass transfer rates, estimated according to Zievernek et al. [88], were at least two orders of magnitude higher than the observed reaction rate. Furthermore, repeating the reactions at stirrer speeds of 600 and 1000 rpm showed that in both cases, the measured reaction rates were within ±10 % of the average values. Internal mass transfer effects were minimal because of the small catalyst particle size and confirmed by the calculated Weisz-Prater criterion of ϕ < 1.0 at the highest reaction temperature. Details of calculations are given in Appendix F.  2.3  Results and discussion  2.3.1 Catalyst properties Table 2.2 summarizes the characterization data for the Ru/alumina catalyst. BET surface area, pore volume and average pore size were also measured for the 5 wt % reduced Ru/alumina catalyst. Table 2.2: Catalyst properties of the pre-reduced Ru/alumina Catalyst Ru/Al2 O3  BET, m2 g−1 76.5  Pore volume, cm3 g−1 0.2  Average pore size, nm 8.4  Degree of reduction, % 3.3  Metal dispersion, % 6.2  Temperature Programmed Reduction (TPR) profile of the commercial Ru/alumina is shown in Figure 2.2. The TPR profile shows a peak with maximum at about 370 K. Another peak, smaller than the previous one, with maximum at 429 K is 27  also observed, which can be attributed to different ruthenium species. Mazzieri et al. [89] reported a sharp peak at 413 K for Ru on alumina catalyst with a smaller peak observed at about 468 K. In the present thesis, the Ru catalyst was received in a reduced state and the low degree of reduction for this catalyst reflects the stability of the reduced Ru in air. CO pulse chemisorption data were used to determine that the Ru dispersion was 6.2 % on the re-reduced Ru/alumina catalyst.  500  -2  6.0x10  T, K  480  -2  460  440  -2  4.0x10  420 -2  3.0x10  400  T, K  TCD signal (a.u.)  5.0x10  380  -2  2.0x10  360 -2  1.0x10  340  320  0.0  0  10  20  30  40  50  60  70  80  Time, min  Figure 2.2: TPR results for the reduced Ru/Al2 O3  2.3.2 Catalyst activity When N-ethylcarbazole was hydrogenated at H2 pressure of 7 MPa and temperature of 423 K, the conversion of N-ethylcarbazole proceeded to 98 % after 1 h, with an initial rate of H2 consumption of 58.4 mM min−1 gcat −1 . Analysis of the liquid samples by GC/MS confirmed the production of partially and completely hydrogenated compounds during the reaction, as represented in Figure 2.3.  28  1  k2  k1  2  (c)  (d)  3  4  k3  k4 k5  k6  5  6  7  k7  Figure 2.3: Hydrogenation reaction pathway of N-ethylcarbazole; (a) Nethylcarbazole, (b) tetrahydro-N-ethylcarbazole, (c) octahydro-Nethylcarbazole, (d) hexahydro-N-ethylcarbazole, (e) decahydro-Nethylcarbazole, (f) dodecahydro-N-ethylcarbazole. The reaction pathway in Figure 2.3 was suggested from the product distributions measured over time at three temperatures as shown in Figure 2.4-2.6. These data showed that N-ethylcarbazole hydrogenation proceeded through the stepwise saturation of double bonds, leading to the production of tetrahydro-Nethylcarbazole and octahydro-Nethylcarbazole as the primary reaction products. Rapid consumption of tetrahydro-N-ethylcarbazole and subsequent production of hexahydro-N-ethylcarbazole suggested that the reaction proceeded to the production of hexahydro-N-ethylcarbazole from tetrahydro-N-ethylcarbazole. The product distribution showed slower consumption of hexahydro-N-ethylcarbazole compared to tetrahydro-N-ethylcarbazole. Octahydro-N-ethylcarbazole, one of the main intermediates, can undergo saturation of the double bonds through two parallel  29  paths to produce decahydro-N-ethylcarbazole, and dodecahydro-N-ethylcarbazole which is the major final product. The reaction completed with more than 95 % selectivity to the desired dodecahydro-Nethylcarbazole, and less than 5 % selectivity to the product octahydro-N-ethylcarbazole. The total amount of hydrogen consumed during hydrogenation corresponded to approximately 5.3 wt % of the reactant. This is very close to the storage criterion of Department of Energy (D.O.E.) and confirms that N-ethylcarbazole is a promising compound for hydrogen storage. −3  Concentration, M  N−ethylcarbazole 0.35  3  0.3  2.5  x 10  octahydro−N−ethylcarbazole  tetrahydro−N−ethylcarbazole 0.08 0.07 0.06  0.25  2  0.05  0.2 1.5  0.04  0.15 0.02  0.05  0.5  0  0  0  100  −3  5  0.03  1  0.1  x 10  200  300  400  500  hexahydro−N−ethylcarbazole  0.01 0  100  −3  2  x 10  200  300  400  500  0  0  100  200  300  400  500  dodecahydro−N−ethylcarbazole  decahydro−N−ethylcarbazole 0.35 0.3  Concentration, M  4  1.5  0.25  3  0.2 1 0.15  2  0.1  0.5  1  0.05 0  0  100  200  300  Time, min  400  500  0  0  100  200  300  Time, min  400  500  0  0  100  200  300  400  500  Time, min  Figure 2.4: Product distribution over time for N-ethylcarbazole hydrogenation at 423 K. Note the different scale of each plot. The experimental product distribution data measured over time at three reaction temperatures were used to model the reaction kinetics. The kinetic model based on the reaction network shown in Figure 2.3 assumed that reactions were 1st -order in the concentration of the heteroaromatic organic compounds and zero-order in hydrogen. The parameters (rate constants) of the kinetic model were estimated by minimizing an objective function (see Appx. G), defined as the sum of squares of 30  −4  Concentration, M  N−ethylcarbazole 0.35  8  0.3  7  x 10  tetrahydro−N−ethylcarbazole  6  0.25  0.015  5  0.2  4 0.15 0.1  2  0.005  1 0  200  −3  1.5  0.01  3  0.05 0  octahydro−N−ethylcarbazole 0.02  x 10  400  600  0  hexahydro−N−ethylcarbazole 7  0  200  400  600  −3 x 10 decahydro−N−ethylcarbazole  0  0  200  400  600  dodecahydro−N−ethylcarbazole 0.25  6 Concentration, M  0.2 5  1  0.5  4  0.15  3  0.1  2 0.05 1  0  0  200  400  Time, min  600  0  0  200 400 Time, min  600  0  0  200 400 Time, min  600  Figure 2.5: Product distribution over time for N-ethylcarbazole hydrogenation at 413 K. Note the different scale of each plot. the difference between the experimental component concentrations and those calculated by the kinetic model, using a weighted least-squares method. The objective function was minimized using a Nelder-Mead simplex (direct search) method. At each iteration on the parameter values, the concentration of each compound as a function of time was calculated by numerical integration of the five ordinary differential equations that described the time rate-of-change of the hydrogenated product concentrations. In addition, the mole balance constraint was used to calculate the reactant concentration at each time step. Although reaction orders from 0 to 2 in increments of 0.5 were examined for all heteroaromatic organic compounds as well as hydrogen, the assumed 1st -order kinetics for reactant and hydrogenated products and zero-order kinetics for hydrogen gave the best fit of the experimental data based on a minimum magnitude of the residual sum of squares. A zero-order reaction in hydrogen likely reflects the excess hydrogen at high pressure present during 31  Concentration, M  N−ethylcarbazole  tetrahydro−N−ethylcarbazole  octahydro−N−ethylcarbazole  0.35  0.014  0.07  0.3  0.012  0.06  0.25  0.01  0.05  0.2  0.008  0.04  0.15  0.006  0.03  0.1  0.004  0.02  0.05  0.002  0.01  0  0  0  200  400  600  0 0  hexahydro−N−ethylcarbazole 0.14  200  400  600  0  decahydro−N−ethylcarbazole  200  400  600  dodecahydro−N−ethylcarbazole  0.05  0.1  0.04  0.08  0.08  0.03  0.06  0.06  0.02  0.04  0.01  0.02  Concentration, M  0.12 0.1  0.04 0.02 0  0  200 400 Time, min  600  0  0  200 400 Time, min  600  0  0  200 400 Time, min  600  Figure 2.6: Product distribution over time for N-ethylcarbazole hydrogenation at 403 K. Note the different scale of each plot. the hydrogenation reactions, and the relatively narrow range of H2 pressures investigated herein, that results in an approximate constant concentration of H2 in the liquid phase as the reaction proceeded. Therefore, although the reaction rate is influenced by a change in hydrogen pressure (i.e. atmospheric pressure will result in much lower hydrogenation reaction rate than the 7 MPa initial H2 pressure used here), the zero-order kinetic model best describes the reaction rate at the reactions condition used in the present work. The activation energies for the reactions were determined by linear regression according to the Arrhenius equation. Individual 1st -order rate constants estimated at 403, 413, and 423 K, as well as the activation energies are reported in Table 2.3. The model results calculated using the obtained rate constants are plotted against the experimental product concentrations in Figure 2.4-2.6. The standard error of the parameter estimates are reported in Table 2.3 (the method used to calculate the 32  Table 2.3: 1st -order rate constants and activation energies for the hydrogenation of N-ethylcarbazole over Ru/alumina catalyst k h−1 k1 k2 k3 k4 k5 k6 k7  403 K  413 K  423 K  0.021 ± 0.012 0.444 ± 0.018 0.840 ± 0.462 0.000 ± 0.000 0.006 ± 0.006 1.140 ± 0.060 0.000 ± 0.000  0.042 ± 0.018 1.020 ± 0.018 2.034 ± 0.600 0.606 ± 0.672 0.012 ± 0.012 2.964 ± 0.528 0.048 ± 0.024  0.096 ± 0.012 1.764 ± 0.018 5.646 ± 1.188 2.538 ± 0.414 0.024 ± 0.006 3.300 ± 0.024 0.252 ± 0.084  activation energy kJ/mol 107.6 ± 7.0 (R2 = 0.99) 97.9 ± 10.3 (R2 = 0.98) 134.9 ± 7.5 (R2 = 0.99) 208.0 ± 0.0 (R2 = 1) 98.2 ± 1.4 (R2 = 0.99) 75.8 ± 33.6 (R2 = 0.83) 240.9 ± 0.0 (R2 = 1)  standard errors of the estimated parameters is described in Appendix G). The optimum parameter values of the rate constants are given in Table 2.3. Although a global search method has been used in obtaining the optimum parameters, many parameter sets were found to fit the data. The only reported parameter values are the one with the least standard error values, which shows comparable magnitude to the literature reported standard errors [90] in non-linear optimization problems. It should be noted that a simple power law model with fixed exponents has been used to fit the kinetics of a complex reaction network over a relatively wide range of reaction temperatures. The model results confirmed that the production of intermediates and the final product were well described by the simplified 1st -order kinetic model shown in Figure 2.3. The data for the concentration of tetrahydro-Nethylcarbazole showed the largest difference between the measured and calculated concentrations, but the concentration of this compound was the lowest among all compounds by approximately 1 order of magnitude. Table 2.3 summarizes the obtained rate constants for the individual reactions of the model (Figure 2.3), as well as the activation energies estimated from the rate constants. The rate constants for reaction 4 (k4 ) and reaction 7 (k7 ) of Figure 2.3, corresponding to the conversion of hexahydro-N-ethylcarbazole and decahydro- Nethylcarbazole, were zero at 403 K, confirming the higher activation barrier for the consumption of these two intermediates compared to the other products. The estimated rate constants for the conversion of tetrahydro-N-ethylcarbazole (k3 ) and hexahydro-N-ethylcarbazole (k4 ) were consistent with the observed slower consumption of hexahydro-N-ethylcarbazole compared to the more rapid consumption 33  of tetrahydro-Nethylcarbazole in the first 10 min of the reaction. Comparing the apparent activation energies of Table 2.3, it is seen that the production of octahydroN-ethylcarbazole from hexahydro-N-ethylcarbazole (reaction 4) and the production of dodecahydro-N-ethylcarbazole from decahydro-N-ethylcarbazole (reaction 7) have the highest activation energies, in agreement with the experimental observation of low consumption rates of hexahydro-N-ethylcarbazole and decahydro-Nethylcarbazole. A comparison of the N-ethylcarbazole conversion at the three reaction temperatures is shown in Fig. 2.7. The results confirmed that the reactions were 1st -order in N-ethylcarbazole concentration with an apparent activation energy for N-ethylcarbazole consumption of 99.5 kJ/mol. Hydrogenation of N-ethylcarbazole was also carried out using a range of initial reactant concentrations from 3.5 - 6 wt % at 413 K to determine the effect of initial concentration on reaction rate. The rate constants for each reactant concentration obtained from the experimental data are summarized in Table 2.4 and shown in Figure 2.8. Table 2.4: Rate constants for N-ethylcarbazole hydrogenation with different initial concentrations at 413 K.  a See  Initial concentration, wt % 3.5  k CA0 a , mol. ml−1 . min−1 2.73×10−6 ± 6.00×10−7  4.5  3.24×10−6 ± 2.50×10−7  6.0  5.39×10−6 ± 1.35×10−6  Appendix E for sample calculation  The data showed that the rate constants increased with increasing initial reactant concentration, confirming the 1st -order concentration dependence of the reaction rate. Carbazole was hydrogenated using the same autoclave reactor and catalyst to produce the reactant for the dehydrogenation reaction. At a H2 pressure of 7 MPa and a temperature of 423 K, 100 % conversion of carbazole was achieved after about 13 h. Due to the extremely low solubility of carbazole in decalin, only a 34  100  (a) 90  (b) 80  Conversion, %  70  (c)  60  50  40  30  20  10  0 0  50  100  150  200  250  300  Time, min  Figure 2.7: Comparison between 6 wt % N-ethylcarbazole conversions at different temperatures: (a) 423 K, (b) 413 K, (c) 403 K. Data points correspond to the experimental data and the solid lines are fitted first-order kinetic models. few liquid samples were withdrawn after a reaction time of 10 h. Liquid sample analysis by GC/MS confirmed the production of dodecahydrocarbazole as the main product. The reaction proceeded with more than 97 % selectivity to the completely hydrogenated product, dodecahydrocarbazole, after about 13 h. The initial rate of H2 consumption by carbazole was 18.44 mMmin−1 gcat −1 at 423 K. Since the intermediate products could not be monitored as a function of time in this case, the kinetics of carbazole consumption were described by a first-order rate equation in 35  100  90  80  Conversion, %  70  60  50  40  30  20  10  0 0  50  100  150  200  250  300  Time, min  Figure 2.8: Comparison between measured and calculated conversion profile for N-ethylcarbazole hydrogenation at 413 K at three different initial concentrations of N-ethylcarbazole. First-order rate constants are reported in Table 2.4. (◦) experimental data, (solid line) fitted first-order kinetic model.  36  carbazole concentration and zero-order in H2 concentration. The same mathematical procedure used to model the kinetics of N-ethylcarbazole hydrogenation was used to model carbazole hydrogenation. Different reaction orders for the organic heteroaromatic compounds as well as hydrogen were examined in the model. The chosen orders are based on the best fit of experimental data. A comparison between the initial rates, the first-order rate constants, and turn-over frequencies (TOFs) for the consumption of carbazole and N-ethylcarbazole at 423 K is presented in Table 2.5. Table 2.5: Initial Reaction Rates, rate constants, and TOFs for hydrogenation of N-ethylcarbazole and carbazole at 423 K over Ru/alumina catalyst reactant rate constant,a k, h−1 catalyst/reactant, g g−1 initial rateb , mM min−1 gcat −1 TOFc , s−1 reaction timed , h  N-ethylcarbazole 2.952 ± 0.122 0.17 58.36 0.223 1.16  carbazole 0.318 ± 0.034 1.00 18.44 0.108 13.33  a with  respect to N-ethylcarbazole and carbazole consumption. rate with respect to H2 consumption. c molecule reactant s−1 Ru surface atom−1 . d reaction time for 100 % conversion. b Reaction  N-ethylcarbazole hydrogenation showed a higher initial reaction rate and proceeded to 100 % conversion in a significantly shorter reaction time with lower metal loading of the catalyst, compared to carbazole. The slower hydrogenation rate of carbazole compared to N-ethylcarbazole can be attributed to the strong adsorption of the carbazole molecule through the N heteroatom to the metal surface. The inhibiting effects of basic N in organic compounds on precious metal catalysts, due to the unshared pair of electrons on N, is well-known [91]. The steric hindrance associated with the ethyl group of N-ethylcarbazole prevents the adsorption of the molecule on the metal surface through the N, as has been previously reported for similar geometries [92]. Therefore, although carbazole has a higher theoretical H2 storage capacity than N-ethylcarbazole, the inhibiting effect of the N atom slowed the hydrogenation reaction and complete H2 saturation of carbazole was obtained in a much longer reaction time compared to N-ethylcarbazole, making carbazole a 37  less practical candidate for H2 storage.  2.4  Conclusion  N-Ethylcarbazole hydrogenation was faster than the hydrogenation of carbazole over a supported Ru catalyst in the temperature range 403-423 K, whereas the selectivity to the completely hydrogenated products, dodecahydro-N-ethylcarbazole and dodecahydrocarbazole was 95 % and 97 %, respectively. Both N-ethylcarbazole and carbazole hydrogenation reactions followed first-order kinetics. The faster hydrogenation reaction rate of N-ethylcarbazole compared to carbazole is attributed to the strong interaction of N electron pair with the catalyst surface. In the case of N-ethylcarbazole, steric hindrance of the ethyl group likely prevents the N lone pair from interacting strongly with the surface, resulting in faster hydrogenation reaction rate.  38  Chapter 3  Kinetics of H2 recovery from dodecahydro-N-ethylcarbazole over a supported Pd catalyst A version of this chapter has been published: F. Sotoodeh, L. Zhao and K.J. Smith. Kinetics of H2 recovery from dodecahydroN-ethylcarbazole over a supported Pd catalyst, Applied Catalysis A: General 362 (2009) 155-162.  39  3.1  Introduction  Storing and releasing H2 from organic molecules is done through catalytic hydrogenation and dehydrogenation reactions. Platinum-group metal catalysts (PGMs) are widely used in heterogeneous catalysis for hydrogenation and dehydrogenation reactions [93, 94]. These catalysts are very active, but low selectivity as well as catalyst deactivation due to hydrogenolysis and coke formation, respectively, are issues that require careful control of process conditions and catalyst properties [95]. Key to success is a high surface area and stability of the catalyst to ensure rapid uptake and release of hydrogen. N-Carbazole and N-ethylcarbazole are organic heteroaromatic compounds with hydrogen storage capacity > 5.5 wt % and boiling point > 473 K when completely hydrogenated, making them possible candidates for hydrogen storage. Some studies have been reported on hydrogen storage by N-ethylcarbazole [16, 96]. However, there is little information on the hydrogenation and dehydrogenation reaction rates and kinetics of N-ethylcarbazole. Kinetics of hydrogen uptake by N-ethylcarbazole was studied in Chapter 2. In the present work, the kinetics of hydrogen recovery from dodecahydro-N-ethylcarbazole has been studied over a 5 wt % Pd/SiO2 catalyst. To compare the dehydrogenation rate and storage capacity, two important criteria for vehicular applications, dehydrogenation of 1,2,3,4-tetrahydrocarbazole has also been measured over the supported 5 wt % Pd catalyst.  3.2  Experimental  3.2.1 Catalyst preparation The supported Pd catalyst, with a Pd loading of 5 wt %, was prepared by the wet impregnation method reported by Song et al. [97]. Silica gel (Sigma-Aldrich) with a surface area of 300 m2 /g and pore volume of 1.15 ml/g was used as the support. The silica gel (9.5 g) was impregnated with 20 mL of a 0.2 M aqueous solution of Pd2+ prepared from PdCl2 (Sigma-Aldrich, 99.9 +%), and left to age for 24 h. The sample was then dried at 393 K for 24 h and calcined at 748 K for 3 h. The obtained sample was reduced in 30 ml (STP)/min of H2 at 723 K and cooled to room temperature in He at 30 ml (STP)/min. A commercial 5 wt % Ru on alumina 40  (Aldrich, 5 wt % Ru, reduced) was used as the hydrogenation catalyst and was re-reduced prior to use.  3.2.2 Catalyst characterization Temperature Programmed Reduction (TPR) of the calcined Pd on silica was performed using a Micromeritics Autochem II 2920. About 0.12 g of calcined Pd/silica was loaded in the reactor and heated to 673 K at a rate of 5 K/min in 30 ml (STP)/min flow of 10 % H2 in Ar. The final temperature was held for 30 min. The CO uptake of Pd/SiO2 was determined by pulsed chemisorption using a Micromeritics Autochem II 2920. Approximately 1 g of the catalyst was loaded in the reactor and pre-reduced in 50 ml (STP)/min of 10 % H2 in Ar while heating from 313 K to 673 K at a rate of 5 K/min, and holding the final temperature for 30 min. The reactor was then cooled to 313 K in a He flow of 50 ml (STP)/min before repeated injection of pulses of CO into the reactor from a 20 ml (STP)/min flow of 10 % CO in He. The injection was continued until the TCD detector response showed no more CO uptake by the sample. The metal dispersion was calculated using an assumed Pd:CO of 1:1 [98]. A Micromeritics ASAP 2020 Accelerated Surface Area and Porosimetry analyzer was used for measuring the total BET surface area, pore volume and average pore width of the Pd/SiO2 catalyst before and after reduction. Approximately 0.2 g of the catalyst was loaded in the sample tube for initial degassing. The sample tube was evacuated at a rate of 1.33 kPa/s and temperature rate of 10 K/min until the vacuum reached 0.04 kPa. The temperature was then held at 473 K for 240 min. After degassing, the sample was transferred to the analysis section for BET surface area analysis using N2 .  3.2.3 Catalyst activity The hydrogenation of N-ethylcarbazole was carried out at temperature range of 403 - 423 K and 7 MPa as in a 300 cm3 autoclave batch reactor with continuous monitoring and control of stirrer speed (600 rpm), temperature and pressure. Details of the hydrogenation experiments have been discussed in Chapter 2. The recovered product from hydrogenation of N-ethylcarbazole was filtered and used as reactant  41  for dehydrogenation. The dehydrogenation was carried out in a 50 ml glass flask reactor using 0.3 g of the reduced 5 wt % Pd/SiO2 (dehydrogenation catalyst) at a temperature range of 423 - 443 K and 101 kPa as shown in Figure 3.1.  Figure 3.1: Schematic diagram for atmospheric dehydrogenation system The reactor was stirred and heated to the desired temperature while purging in He flow at 170 cm3 (STP) /min. After reaching a stable temperature, the catalyst was injected into the reactor. Helium was used as a carrier gas for continuous removal of the produced hydrogen, and the exit gas composition was continuously monitored using a quadrupole mass spectrometer (Residual Gas Analyzer 200 amu). Liquid samples of < 0.1 ml were also withdrawn from the reactor, periodically. The analysis of the liquid samples form the dehydrogenation reactions was done using a 14-A Shimadzu gas chromatograph equipped with a flame ionization detector (FID) and an ATT M -5 25M x 0.53 mm capillary column. Product identities were also confirmed by a Shimadzu QP-2010S GC/MS using a Restek RTX5 30M x 0.25 mm capillary column.  42  Dehydrogenation reactions of 1,2,3,4-tetrahydrocarbazole (Aldrich, 99 %) were carried out for comparison of hydrogen production rates with dodecahydro-Nethylcarbazole, using the same experimental set-up and procedure. 3 wt % of 1,2,3,4-tetrahydrocarbazole in mesitylene (Aldrich, 98 %) and 0.2 g of the reduced 5 wt % Pd/SiO2 was loaded into the reactor with the constant flow of He as carrier gas for continuous removal of hydrogen, and the reaction was performed at a temperature range of 413 - 423 ◦ C and 101 kPa in a 50 ml glass flask reactor.  3.3  Results and discussions  3.3.1 Catalyst properties Temperature programmed reduction (TPR) profile for the calcined Pd/SiO2 is shown in Figure 3.2. A peak at 337 K was observed. The H2 profile is consistent with the release of H2 from the decomposition of a β -palladium hydride phase which is reported to form at room temperature over large Pd particles (> 2 nm) [99, 100]. CO pulse chemisorption data were used to determine that 4.2 % of the total Pd atoms were available at the catalyst surface. BET surface area, pore volume and average pore size were measured for the 5 wt % Pd/SiO2 after calcination and reduction and the results are summarized in Table 3.1. Table 3.1: Pd/SiO2 catalyst properties Catalyst status Calcined Reduced  BET, m2 /g 311.5 325.6  Pore volume, cm3 /g 0.9 1.1  Average pore size, nm 9.6 9.5  Metal dispersion, % — 4.2  3.3.2 Catalyst activity Dehydrogenation of dodecahydro-N-ethylcarbazole The products from N-ethylcarbazole hydrogenation in decalin at 423 K, containing 5.8 wt % dodecahydro-N-ethylcarbazole1 wt % and < 0.2 wt % octahydroN-ethylcarbazole, were transferred to a 50 ml stirred reactor for the dehydrogena43  0.030  TCD signal (a.u.)  0.025  0.020  0.015  0.010  0.005  0.000  300  325  350  375  400  425  450  475  Temperature, K  Figure 3.2: TPR profile for the calcined Pd/SiO2 tion test. Previous studies showed that decalin dehydrogenation occurs at T > 473 K [57, 58, 101, 102], and our own tests confirmed no decalin dehydrogenation at T < 453 K, which made it possible to use decalin as a solvent for the dehydrogenation experiments. The reduced 5 wt % Pd/SiO2 (0.3 g) was used as the dehydrogenation catalyst and the hydrogen generated was monitored continuously using a quadropole mass spectrometer attached to the gas outlet from the reactor. The reactions were carried out at temperatures of 423, 433 and 443 K. The product distribution over time suggested that tetrahydro-N-ethylcarbazole and octahydro-N-ethylcarbazole were the primary intermediates at all three temperatures, and no significant amounts of N-ethylcarbazole was produced, as confirmed by the GC/MS analysis of liquid samples. First order kinetics were fitted to the obtained experimental data in order to calculate the rate constant for the production of tetrahydro-N-ethylcarbazole and 44  octahydro-N-ethylcarbazole. Fig. 3.3 illustrates the experimental product distribution over time at three temperatures and the fitted first order model. The apparent activation energy for dodecahydro-N-ethylcarbazole consumption was found to be 126.7 kJ/mol. The initial reaction rates, rate constants, as well as the percent hydrogen recovered at each temperature are summarized in Table 3.2. The initial reaction rates were obtained from the first 15 min of the experiment. Table 3.2: Initial reaction rates, rate constants and hydrogen recovery for dehydrogenation of dodecahydro-N-ethylcarbazole T, K 423  Initial reaction rate, mM min−1 gcat −1 3  Rate constanta , k, min −1 0.0016  k1 , min −1 0.0012  k2 , min −1 0.0006  H2 recovery after 17 h, wt % 2.1  433  8.5  0.0044  0.0013  0.0027  3.3  443  12  0.0075  0.0027  0.0035  4.0  a For  dodecahydro-N-ethylcarbazole consumption.  One hundred percent of the reactant, dodecahydro-N-ethylcarbazole, was converted at the three temperatures, resulting in 2 wt % H2 recovery after 2 h and 4 wt % H2 recovery after 17 h at 443 K. However, although 100 % conversion was achieved in the present study, only 66 % of the total stored H2 was recovered at 443 K because of the dehydrogenation to tetrahydro-N-ethylcarbazole and octahydroN-ethylcarbazole, and not N-ethylcarbazole. Dehydrogenation of 1,2,3,4-tetrahydrocarbazole To compare the hydrogen release rate and storage capacity on N-ethylcarbazole, the dehydrogenation of 1,2,3,4-tetrahydrocarbazole was carried out at temperatures of 413 and 423 K using a solution of 3 wt % reactant in mesitylene. Reduced 5 wt % Pd/SiO2 (0.2 g) was used as dehydrogenation catalyst with constant flow of He as a carrier gas for continuous removal of the produced H2 . Gas and liquid phase analysis confirmed that no byproducts were generated, and that the reaction proceeded selectively to carbazole, as also previously reported [53]. The initial rates of 1,2,3,4,-tetrahydrocarbazole dehydrogenation on 5 wt % Pd/SiO2 catalyst at 413  45  Concentration, M  (a)  dodecahydro-N-ethylcarbazole 0.15  Concentration, M  100  200  0.02  0.02  0.01  300  0 0  100  200  300  0.2  0.04  0.08  0.15  0.03  0.06  0.1  0.02  0.04  0.05  0.01  0.02  0  0 0  Concentration, M  0.04  0 0  (c)  tetrahydro-N-ethylcarbazole 0.03  0.1  0.05  (b)  octahydro-N-ethylcarbazole 0.06  100  200  300  0.2  0  100  200  300  0  100  200  300  0  100  200  300  0 0  100  200  300  0.06  0.08  0.15  0.06 0.04  0.1  0.04 0.02  0.05  0.02  0  0 0  100  200  300  0 0  Time, min  100  200  Time, min  300  Time, min  Figure 3.3: Product distribution for dehydrogenation of dodecahydro-Nethylcarbazole obtained by GC/MS: (a) 423 K, (b) 433 K, (c) 443 K. (◦) experimental data, (solid line) fitted model. The rate constants are reported in Table 3.2.  46  and 423 K and 101 kPa, were 17 and 18.5 mM min−1 gcat −1 , respectively. These values are comparable to the average rate of 10 mM min−1 gcat −1 reported by Crawford et al. [53] for Pd/alumina at 413 K. The activation energy was estimated as 67.7 kJ/mol, in good agreement with the calculated value of 64.4 kJ/mol from DFT study done by Crawford et al. [53] and 67 kJ/mol obtained by the experimental work of Hindle et al. [54]. The reaction proceeded to 25 % conversion at 423 K after 400 min and 17 % conversion at 413 K after 350 min. The obtained results confirmed that the reaction followed zero order kinetics with respect to 1,2,3,4,tetrahydrocarbazole consumption, which is consistent with the reaction kinetics for the production of carbazole. The carbazole concentration increased approximately linearly as seen in Fig. 3.4 up to about 15 % 1,2,3,4,-tetrahydrocarbazole conversion at 423 K and about 12 % at 413 K. Then, a positive order relationship was observed because of product inhibition as also reported previously [54]. Only 0.6 wt % of the hydrogen (equal to 25 % of the theoretical reactant hydrogen capacity) was recovered during the experiment. Table 3.3 compares the dehydrogenation reaction of 1,2,3,4,-tetrahydrocarbazole and dodecahydro-N-ethylcarbazole. Comparing the two compounds, it can be seen that the theoretical H2 capacity is higher for dodecahydro-N-ethylcarbazole. At the same temperature of 423 K, the same recovery was obtained. However, because the dehydrogenation reaction of dodecahydro-N-ethylcarbazole has a higher activation energy (127 kJ/mol) compared to 1,2,3,4-tetrahydrocarbazole (67 kJ/mol), increasing the temperature from 423 K to 443 K, increased the amount of evolved H2 from dodecahydro-N-ethylcarbazole significantly from 25 % to 60 % after about 6 h.  3.4  Conclusion  Dehydrogenation of the hydrogenation product, dodecahydro-N-ethylcarbazole, showed that the reaction proceeded to 100 % conversion, but only 66 % of the stored hydrogen was recovered during dehydrogenation at 443 K after 17 h, because of the production of the two intermediates octahydro-N-ethylcarbazole and tetrahydro-N-ethylcarbazole, with minimal production of N-ethylcarbazole. However, this amount is comparable with the dehydrogenation reaction of 1,2,3,4tetrahydrocarbazole where only 25 % of the hydrogen was recovered at 423 K,  47  60  50  423 K  Conversion, %  40  413 K  30  20  10  0 0  50  100  150  200  250  300  350  400  Time, min  Figure 3.4: Comparison of 1,2,3,4-tetrahydrocarbazole conversion at different temperatures obtained from liquid samples analysis Table 3.3: Comparison between dehydrogenation of 1,2,3,4-tetrahydrocarbazole and dodecahydro-N-ethylcarbazole over a 5 wt % Pd/SiO2 catalyst Compound  1,2,3,4-tetrahydrocarbazole Dodecahydro-Nethylcarbazole b Amount  Theoretical H2 releasea , (wt %) 2.3  H2 recoveryb ,  5.8 5.8  Duration,  T,  (%) 25  Catalyst to reactant ratio, (g.g−1 ) 0.2  (min) 370  (K) 423  25 60  0.3 0.3  370 370  423 443  of H2 released in the experiment divided by theoretical amount of released H2 . amount of released H2 in grams divided by the weight of hydrogenated compound  a Theoretical  in grams  48  leading to the conclusion that N-ethylcarbazole is a more applicable compound for hydrogen storage.  49  Chapter 4  Structure sensitivity of dodecahydro-N-ethylcarbazole dehydrogenation over Pd catalysts A version of this chapter has been published: F. Sotoodeh and K.J. Smith (2011) Structure sensitivity of dodecahydro-N-ethylcarbazole dehydrogenation over Pd catalysts. Journal of Catalysis, 279, 36-47.  50  4.1  Introduction  As discussed in the previous chapters, hydrogen is an alternative, non-toxic, environmentally benign energy carrier with H2 O the only product when H2 is converted into thermal energy (by combustion) or electrical energy (using fuel cells). However, hydrogen has a very low density at ambient conditions making it difficult to store and transport. On-board vehicular hydrogen storage, including compressed or liquid (cryogenic) hydrogen, suffer from high energy inputs and low gravimetric storage capacities. On the other hand, organic aromatic compounds that bind hydrogen covalently, have the potential to meet the 6 wt % H2 storage density demands of automotive applications [20]. The catalytic dehydrogenation reaction must be fast, selective and have low enthalpy so that the reaction is thermodynamically favored at low temperature (< 473 K) [44]. However, a review of literature data shows very slow dehydrogenation reaction rates and low selectivity to the completely dehydrogenated product for some organic compounds. For example, the high enthalpy of dehydrogenation of cyclohexane (205.5 kJ/mol with H2 capacity 7.1 wt %) and methylcyclohexane (202.5 kJ/mol with H2 capacity 6.1 wt %) means that high temperatures (> 573 K) are required for on-board H2 recovery by dehydrogenation [87, 103, 104]. Pez et al. [16] have shown that incorporation of a heteroatom in the aromatic ring can lower the dehydrogenation enthalpy of the candidate compounds. Complete conversion of indoline, with a hydrogen capacity of 1.7 wt %, was reported over Pd/C and Rh/C catalysts after 24 h in refluxing toluene (∼383 K) [105]. However, the low hydrogen content of indoline means that this is not a practical choice for transport applications. Dehydrogenation of 4-aminopiperidine (hydrogen capacity 5.9 wt %) over a 10 wt % Pd/SiO2 catalyst resulted in 66 % conversion at 443 K after 4 h [44]. Complete conversion of dodecahydro-N-ethylcarbazole (hydrogen capacity 5.8 wt %) was obtained over a 5 wt % Pd/SiO2 at 443 K and 101 kPa after 17 h with only 4.0 wt % H2 recovery, due to a low selectivity toward the completely dehydrogenated N-ethylcarbazole product [106, 107]. Dehydrogenation of dodecahydrocarbazole (hydrogen capacity 6.7 wt %) was reported to be even slower than dodecahydro-N-ethylcarbazole, most likely due to product inhibition by carbazole, where only 53 % conversion was achieved after 17 h at 443 K and 101 kPa [106, 107]. In a similar dehydrogenation  51  study of dodecahydro-N-ethylcarbazole over homogeneous Ir-complex catalysts, no selectivity toward N-ethylcarbazole was reported at 473 K, and octahydro-Nethylcarbazole and tetrahydro-N-ethylcarbazole were the only products [56]. Limited data are available on the relationship between catalytic activity and catalyst dispersion for the hydrogenation and dehydrogenation reactions of organic aromatics, and contradictory interpretations have been reported. For example, Benedetti et al. [99] reported that the hydrogenation of 2,4-dinitrotoluene over Pd/SiO2 was favored by an increase in Pd particle size. However, Suh et al. [108] noted that hydrogenation of 2,4-dinitrotoluene over Pd/C catalysts was structure insensitive and that increases in specific activity on larger Pd particles was attributed to diffusion limitations within the pores of the microporous carbon support. In other work on the hydrogenation of 2,4-dinitrotoluene over a Pd/C catalyst, it was found that in the range of dispersions investigated (Pd particle size 3 - 10 nm) the specific activities increased as the Pd dispersion decreased [100]. In a study of pyrrole hydrogenation over Pt catalysts, with particle size from < 1 nm to 5 nm, reported by Kuhn et al. [109], it was demonstrated that ring hydrogenation was structure insensitive, while ring opening to n-butylamine was structure sensitive and found to be more facile over larger Pt nanoparticles. Adsorption geometries of aromatic organic compounds relevant to hydrogen storage over metal surfaces have been shown to involve several surface metal atoms, suggesting that the dehydrogenation reactions may be structure sensitive. Crawford et al. [53] investigated the mechanism of 1,2,3,4-tetrahydrocarbazole dehydrogenation over a Pd catalyst, and showed by experiment and density-functional theory (DFT) calculations that the reaction proceeded through a flat adsorption geometry of 1,2,3,4-tetrahydrocarbazole involving several Pd surface sites. Lu et al. [110], using periodic DFT calculations, also reported a flat adsorption of tetracene over several Ru surface atoms, with the long axis of tetracene parallel to the surface. The optimized structure showed that adsorption of tetracene required multiple Ru atoms. Tilted and parallel adsorption of quinoline on a Pt cluster investigated by Vargas et al. [92] also indicated that several surface Pt atoms were involved in the adsorption. In this chapter, the structure sensitivity of the dehydrogenation of dodecahydroN-ethylcarbazole was investigated over a series of Pd/SiO2 catalysts, prepared with 52  Pd loadings of 0.5 to 10 wt %, such that the Pd dispersion (particle size) varied from 6.5 % (15 nm) to 52 % (1.9 nm). The activity and selectivity of each catalyst for the dehydrogenation of dodecahydro-N-ethylcarbazole is reported.  4.2  Experimental  4.2.1 Catalyst preparation The SiO2 -supported Pd catalysts (0.5 wt % to 10 wt %) were prepared by incipient wetness of the support [111]. Silica gel (Sigma-Aldrich, BET surface area 400 m2 /g, pore volume 1.15 cm3 /g, particle size < 90 µ m) was used as the support. Approximately 10 grams of the silica gel was drop-wise impregnated with an aqueous solution of Pd2+ , prepared from 0.5 N HCl and the PdCl2 (Sigma-Aldrich, 99.9+ %), at a concentration sufficient to yield the required Pd loading from one impregnation. The impregnated support was left to age for 48 h before being dried at 393 K for 8 h and calcined at 773 K for 6 h in a He flow of 30 cm3 (STP)/min. The calcined catalyst precursor was subsequently reduced in a H2 flow of 30 cm3 (STP)/min while heating at a rate of 10 K/min to 673 K. The final temperature was held for 1 h before cooling the reduced catalyst to room temperature in a 30 cm3 (STP)/min flow of He. The Pd catalysts were compared to a previously reported 5 wt % Pd/SiO2 catalyst that was prepared by wet impregnation but dried at 393 K for 24 h and calcined at 748 K for 3 h in air rather than inert He [106]. Using the same procedure, a 10 wt % Pd/SiO2 catalyst calcined in air was also prepared in the present study.  4.2.2 Catalyst characterization Temperature-programmed reduction (TPR) of the calcined Pd/SiO2 catalyst precursors was performed using a Micromeritics Autochem II 2920 analyser fitted with a thermal conductivity detector (TCD). About 0.2 g of each sample was loaded into a u-tube reactor and flushed for 30 min in a 50 cm3 (STP)/min flow of 10 % H2 in Ar at room temperature, before heating to 673 K at a rate of 10 K/min with the final temperature held for 1 h. In order to clarify the TPR profiles, additional TPR studies were done using a 53  quartz u-tube reactor placed in a temperature programmable muffle furnace with a quadrupole mass spectrometer (SRC Residual Gas Analyzer, 200 amu) connected to the reactor effluent in order to continuously monitor the products generated during the temperature-programmed calcination and reduction. The calcination and reduction steps, using 0.4 grams of sample, followed the same procedure as that used in the Micromeritics Autochem II 2920 analyser. To investigate the reducibility of the catalyst at room temperature, the catalysts were flushed in a 50 cm3 (STP)/min flow of 10 % H2 in He for 3 h at room temperature after the calcination and prior to the TPR. Temperature-programmed desorption (TPD) of CO from the Pd catalysts was also performed using the Micromeritics Autochem II 2920 analyser. About 0.2 g of the dried catalyst was loaded into the reactor and calcined at 773 K for 6 h in a flow of Ar at 30 cm3 (STP)/min, followed by TPR at 10 K/min to a final temperature of 673 K in a 30 cm3 (STP)/min flow of 10 % H2 in Ar, holding the final temperature for 1 h. The catalyst was then flushed in a 30 cm3 (STP)/min flow of Ar for 1 h and cooled to room temperature, before being exposed to a 50 cm3 (STP)/min flow of 10 % CO in He at 313 K for 1 h. The catalyst was subsequently flushed in 50 cm3 (STP)/min He for 2 h. The temperature-programmed desorption was then started in the same He flow by heating the sample to 773 K at a rate of 10 K/min and holding the final temperature for 1 h. The catalysts were also characterized by high-resolution transmission electron microscopy (HRTEM), scanning electron microscopy (SEM) and energy dispersive X-ray analysis (EDX). SEM, EDX and EDX-mapping were done using a variable pressure 120 keV Hitachi S-3000N equipped with a light element EDX detector. A FEI Tecnai G2 200 keV with a LaB6 filament capable of 1.4 A˚ point-to-point resolution was used for high resolution transmission electron microscopy (HRTEM). Pd/SiO2 samples with variable Pd loadings were ground to a fine powder using an agate mortar and pestle and dispersed in ethanol ultrasonically. A droplet of the suspension was then put on a 200 mesh copper grid with formvar carbon coating. The sample was left to dry before analysis.  54  4.2.3 Catalyst activity The dodecahydro-N-ethylcarbazole reactant was prepared by hydrogenation of Nethylcarbazole at 423 K and a H2 pressure of 7 MPa using a 300 cm3 autoclave batch reactor, as previously discussed in Chapter 2. The dehydrogenation reactions were carried out in a 50 cm3 glass flask reactor at stirring speed of 550 rpm, operated in batch mode at 443 K and 101 kPa using the reduced Pd/SiO2 catalysts. For each experiment, 10 cm3 of the product recovered from N-ethylcarbazole hydrogenation was diluted in 20 cm3 of decalin (Sigma-Aldrich, 98 %), and added to the reactor as reactant of the dehydrogenation reaction. The reactor was stirred and heated to the desired temperature while purging in He flow at 171 cm3 (STP)/min. Once the reaction temperature was reached, the catalyst was injected into the reactor using a syringe. The He flow was maintained and acted as a carrier gas for continuous removal of the produced hydrogen. The reactor exit gas composition was continuously monitored using a quadrupole mass spectrometer (SRC Residual Gas Analyzer, 200 amu). Liquid samples of ∼0.1 cm3 were withdrawn from the reactor periodically and analysed using a Shimadzu QP-2010S GC/MS and a Restek RTX5 30 m 0.25 mm capillary column. Internal mass transfer effects were minimal because of the small catalyst particle size and confirmed by the calculated Weisz-Prater criterion of Φ < 1.0 at the reaction temperature.  4.3  Results  4.3.1 Catalyst properties H2 TPR profiles of the 0.5 wt %, 10 wt % He-calcined, and 10 wt % air-calcined Pd/SiO2 catalysts are presented in Figure 4.1. A negative peak in all the TPR profiles, indicative of a net H2 evolution from the sample rather than H2 consumption, was observed at temperatures between 333 - 360 K. Babu et al. [112] reported very similar H2 TPR profiles for 5 wt % and 3 wt % Pd/alumina catalysts, where H2 evolution from the Pd in the region 348 - 373 K, was observed. In work by Amairia et al. [113], H2 TPR profiles from Pd/Al2 O3 -ZrO2 also showed H2 evolution at 358 K. The H2 profile is consistent 55  0.0 10 wt % air-calcined -1  -1.5x10  0.0  10 wt % He-calcined -2  -8.8x10  0.0  5 wt % -1  TCD signal (a.u.)  -1.1x10  0.0  4 wt % -2  -7.8x10 0.0  3 wt % -2  -6.6x10 0.0  2 wt % -2  -5.4x10 0.0  1 wt % -2  -3.7x10  0.0 0.7 wt % -2  -1.8x10 0.0  0.5 wt %  -2  -1.5x10  300  350  400  450  500  550  600  650  Temperature (K)  Figure 4.1: TPR patterns of SiO2 supported Pd catalysts.  56  700  with the release of H2 from the decomposition of a β -palladium hydride phase which is reported to form at room temperature over large Pd particles (> 2 nm) [99, 100, 112–115] and in the present work, was likely generated during the low temperature flush in 10 % H2 in He for 30 min that followed catalyst calcination. Figure 4.2 and Figure 4.3 compare the effluent gas profiles as measured by mass spectrometer during the thermal treatment of the silica gel support and the catalyst precursors. The silica gel and the 10 wt % Pd/SiO2 precursor, were calcined at 773 K in He (Figure 4.2-a and 4.2-b, respectively) or air (Figure 4.3-a and 4.3-b, respectively). The He-calcined and the air-calcined 10 wt % Pd/SiO2 catalyst precursors were then flushed in 10 % H2 in He at room temperature (Figure 4.2-c and Figure 4.3-c), and reduced in the same flow of gas by temperature programming to 673 K at 10 K/min (Figure 4.2-d and Figure 4.3-d). Calcination of the silica gel in He (Figure 4.2-a) resulted in the production of two water peaks at about 425 K and 773 K. The two water peaks were also observed during the calcination of the silica gel in air at approximately 500 K and 773 K (Figure 4.3-a). Comparing the calcination of the Pd/SiO2 in He or air (Figure 4.2-b and Figure 4.3-b, respectively), shows that a sharp peak between 423 - 500 K and a second broader peak at 773 K for water evolution were observed for both samples. HCl evolution was also detected for the catalyst calcined in He with a peak temperature at 773 K (Figure 4.2-b), whereas no HCl was observed for the sample calcined in air (Figure 4.3-b). Subsequent flushing of the catalysts in a flow of 10 % H2 in He at room temperature resulted in the production of water and HCl from both catalysts, with more HCl observed for the catalyst calcined in He (Figure 4.2-c) compared to the one calcined in air (Figure 4.3-c). The catalyst flush steps were followed by TPR in a flow of 10 % H2 in He at 20 K/min to 673 K which revealed a significant production of HCl (especially for the catalyst calcined in He) as well as water. For the Pd/SiO2 calcined in He, the peak in water evolution occurred at 488 K and the peak in HCl evolution occurred at 578 K (Figure 4.2-d). For the air calcined sample, the corresponding peak temperatures were 520 K and 620 K, respectively (Figure 4.3-d). Table 4.1 summarizes the properties of the catalysts prepared in the present study (i.e. 0.5 wt % - 10 wt % Pd/SiO2 prepared by calcination in He and the 10 wt % Pd/SiO2 catalyst prepared by calcination in air), and the 5 wt % Pd/SiO2 57  800  H O  T (K)  2  P/P(He)  (a)  -2  3.0x10  700 600  -2  2.0x10  500  -2  1.0x10  400  0.0  0  T (K)  -2  4.0x10  300  25  50  75  100  125  150  175  200  -2  2.5x10  800  H O  P/P(He)  2.0x10  (b)  T (K)  2  -2  1.5x10  600  -2  1.0x10  500  -3  400  5.0x10  HCl  0.0  0  300  25  50  75  100  125  150  175  -2  2.5x10  P/P(He)  700  T (K)  -2  200  (c)  -2  2.0x10  -2  1.5x10  H O 2  -2  1.0x10  HCl  -3  5.0x10  0.0  0  25  50  75  100  125  150  175  200 700  3.0x10  HCl  (d)  T (K)  600  -2  2.0x10  500  -2  H O  1.0x10  400  2  0.0  0  T (K)  P/P(He)  -2  300  10  20  30  40  50  60  70  80  90  100  Time (min)  Figure 4.2: Profile of effluent gases during thermal treatment of catalyst samples in He as measured by mass spectrometer using 0.4 g of sample. (a) SiO2 gel calcination in He at 773 K, (b) 10 wt % Pd/SiO2 calcination in He at 773 K, (c) 10 wt % Pd/SiO2 flush in 10 % H2 in He at room temperature, (d) 10 wt % Pd/SiO2 reduction in 10 % H2 in He at 673 K. 58  -2  1.0x10  800  H O 2  600  -3  5.0x10  500 -3  2.5x10  400  0.0  0  -2  2.5x10  300  25  50  75  100  125  150  175  200 800  (b)  T (K)  -2  P/P(Air)  700  2.0x10  H O 2  -2  1.5x10 1.0x10  500  -3  400  HCl  0.0  0 -2  2.5x10  700 600  -2  5.0x10  P/P(He)  T (K)  7.5x10  T (K)  P/P(Air)  (a)  T (K)  -3  300  25  50  75  100  125  150  175  H O  200  (c)  2  -2  2.0x10  -2  1.5x10  -2  1.0x10  -3  5.0x10  HCl  0.0  0  25  50  75  100  125  150  175  200 700  T (K)  3.0x10  P/P(He)  (d) 600  -2  2.0x10  HCl  500  -2  H O  1.0x10  400  2  0.0  0  T (K)  -2  300  10  20  30  40  50  60  70  80  90  100  Time (min)  Figure 4.3: Profile of effluent gases during thermal treatment of catalyst samples in air as measured by mass spectrometer using 0.4 g of sample. (a) SiO2 gel calcination in air at 773 K, (b) 10 wt % Pd/SiO2 calcination in air at 773 K, (c) 10 wt % Pd/SiO2 flush in 10 % H2 in He at room temperature, (d) 10 wt % Pd/SiO2 reduction in 10 % H2 in He at 673 K. 59  catalyst prepared by wet impregnation and calcination in air, reported by Sotoodeh et al. [106]. Table 4.1: Pd/SiO2 catalyst properties Pd (wt %)  CO desorbed (µ mol.g−1 )  Dispersion (%)  0.5 0.7 1 2 3 4 5 10 10d 5d  20.9 25.3 22.5 25.7 31.0 34.4 30.3 53.1 — —  51.9 44.7 28.3 16.0 12.9 11.0 7.9 6.5 3.6e 4.2e  Pd particle sizea (nm) 1.9 2.2 3.6 6.3 7.7 9.1 12.5 15.4 27.9e 23.8 e  Average Pd particle diameter (nm) 3.6b ± 0.1c 4.0 ± 0.2 4.6 ± 0.2 6.1 ± 0.1 7.0 ± 0.1 7.3 ± 0.4 13.2 ± 0.9 15.5 ± 0.7 23.7 ± 2.7 —  a obtained  from CO TPD analysis. Pd particle size obtained by fitting HRTEM data to a log-normal distribution. c Standard deviation of average particle size obtained from the log-normal fit. d reduced after calcination in air. e obtained by CO pulse chemisorption. b Average  Figure 4.4 shows the results obtained from the temperature programmed desorption (TPD) of CO from the Pd/SiO2 catalysts. The major desorption peak occurred at 343 - 362 K, in good agreement with the desorption peak temperature of 320 - 350 K reported by Myyrylainen et al. [116] for the TPD of CO from a Pd(100) surface. The CO uptake determined from the TPD experiments was used to estimate the catalyst dispersion at each Pd loading, as well as the Pd particle diameter, assuming a CO:Pd adsorption stoichiometry of 1:1 [98]. The results, summarized in Table 4.1, show that the Pd particle size of the Pd/SiO2 catalysts varied from 2 - 15 nm as the Pd loading increased from 0.5 wt % to 10 wt %. By comparing these results to the 10 wt % and the 5 wt % Pd/SiO2 [106, 107] catalysts calcined in air and shown in Table 4.1, it is clear that a significant increase in the dispersion of the Pd catalysts was obtained by calcination in He rather than air. The particle size decreased from 27.9 nm to 15.4 nm for the 10 wt % Pd/SiO2 by calcining the catalyst in He rather  60  5.08  10 wt%  5.06 5.04  5 wt%  5.05  5.04 5.05  4 wt%  5.03 5.05  3 wt%  3  CO (cm /min)  5.04  5.04  5.05 5.03  2 wt% 5.04 5.03 5.05  1 wt% 5.04 5.05  0.7 wt% 5.04  5.03  0.5 wt%  5.05  5.04  300  350  400  450  500  550  600  Temperature (K)  Figure 4.4: Temperature programmed desorption (TPD) of CO from Pd/SiO2 catalysts with different Pd loadings at a heating rate 10 K/min.  61  than air. The 5 wt % Pd/SiO2 calcined in He had a Pd metal dispersion of 7.9 %, almost twice the value of 4.2 % obtained for the 5 wt % Pd/SiO2 catalyst prepared by calcination in air [106, 107]. The reduced and calcined Pd/SiO2 samples were analyzed by SEM and EDX to investigate the catalyst morphology, as well as the elemental composition. Table 4.2 summarizes the EDX elemental analysis of the 5 wt % Pd/SiO2 catalysts, reported as the average of at least 10 point analyses, after various thermal treatments. Table 4.2: EDX elemental analysis of the 5 wt % Pd/SiO2 catalysts Status Element Oxygen Silicon Chlorine Palladium  Reduced after He calcination Concentration (wt %) 56.3 ± 3.1 38.6 ± 1.9 0.00 ± 0.00 5.1 ± 1.9  Reduced after air calcination  Calcined in He  Calcined in air  57.8 ± 0.2 35.9 ± 0.2 0.03 ± 0.01 6.2 ± 0.1  52.4 ± 0.4 37.9 ± 1.3 0.8 ± 0.1 8.8 ± 1.2  57.6 ± 0.3 37.7 ± 0.1 1.0 ± 0.1 3.7 ± 0.2  Calcination in He or air at 773 K for 6 h did not remove all the Cl present in the precursor and Cl was identified after calcination in He or air. However, as shown in Table 4.2, no Cl was present after TPR of the sample calcined in He, whereas a small amount of Cl was detected after the TPR of the sample calcined in air. Further insight into the size distribution of the Pd/SiO2 catalysts was obtained by high resolution transmission electron microscopy (HRTEM) of the reduced Pd/SiO2 catalysts as summarized in Table 4.1. Figure 4.5-a,b,c shows selected HRTEM images of the reduced 10 wt %, 4 wt % and 0.5 wt % Pd/SiO2 catalysts, respectively. The Pd size distribution obtained from several HRTEM images were fitted to a log-normal distribution to obtain an estimate of the average Pd particle diameter and standard deviation with Pd loading. The results, reported in Table 4.1, showed good agreement with the values calculated from the CO TPD data.  4.3.2 Catalyst activities Dehydrogenation of the dodecahydro-N-ethylcarbzole was carried out over each of the reduced Pd/SiO2 catalysts. The reactor was operated at 443 K and 101 kPa. Decalin was used as the solvent in the dehydrogenation reaction. The stability of 62  (a)  (b)  (c)  Figure 4.5: Selected HRTEM images of reduced Pd/SiO2 samples: (a) 10 wt % Pd, (b) 4 wt % Pd, (c) 0.5 wt % Pd. Samples were calcined in He flow at 773 K for 6 h prior to reduction in H2 at 673 K for 1 h. decalin at reaction temperatures below 473 K over Pd/SiO2 catalysts were confirmed in previous studies [57, 58, 106]. Figures 4.6, 4.7 and 4.8 show the product distribution as a function of reaction time obtained from the GC/MS liquid sample analysis of the dehydrogenation products. The data show that octahydro-N-ethylcarbazole and tetrahydro-N-ethylcarbazole were the primary products of the reaction, and the reactions proceeded with further conversion of these two intermediates and the production of the completely dehydrogenated product, N-ethylcarbazole, which can be shown by the following, 63  1 wt % -5  9.0x10  -5  6.0x10  -5  3  Concentration (mol/cm )  3.0x10  0.0  0.7 wt % -5  9.0x10  -5  6.0x10  -5  3.0x10  0.0  0.5 wt % -5  9.0x10  -5  6.0x10  -5  3.0x10  0.0 0  20  40  60  80  100  120  140  Time (min)  Figure 4.6: Product distribution from dodecahydro-N-ethylcarbazole dehydrogenation at 443 K and 101 kPa over Pd/SiO2 with Pd loading of 0.5, 0.7 and 1 wt %; ( ) dodecahydro-N-ethylcarbazole, (△) octahydro-N-ethylcarbazole, ( ) tetrahydro- N-ethylcarbazole, ( ) Nethylcarbazole. 64  4 wt % -5  9.0x10  -5  6.0x10  -5  3  Concentration (mol/cm )  3.0x10  0.0  3 wt % -5  9.0x10  -5  6.0x10  -5  3.0x10  0.0  2 wt % -5  9.0x10  -5  6.0x10  -5  3.0x10  0.0 0  25  50  75  100  125  150  Time (min)  Figure 4.7: Product distribution from dodecahydro-N-ethylcarbazole dehydrogenation at 443 K and 101 kPa over Pd/SiO2 with Pd loading of 2, 3 and 4 wt %; ( ) dodecahydro-N-ethylcarbazole, (△) octahydro-N-ethylcarbazole, ( ) tetrahydro- N-ethylcarbazole, ( ) Nethylcarbazole. 65  -4  1.5x10  5 wt %, air-calcined  -4  1.2x10  -5  9.0x10  -5  6.0x10  -5  3  Concentration (mol/cm )  3.0x10  0.0  10 wt % -5  9.0x10  -5  6.0x10  -5  3.0x10  0.0  5 wt % -5  9.0x10  -5  6.0x10  -5  3.0x10  0.0 0  25  50  75  100  125  150  Time (min)  Figure 4.8: Product distribution from dodecahydro-N-ethylcarbazole dehydrogenation at 443 K and 101 kPa over Pd/SiO2 with Pd loading of 5, 10 and 5 wt % (air calcined) [106]; ( ) dodecahydro-N-ethylcarbazole, (△) octahydro-N-ethylcarbazole, ( ) tetrahydro- N-ethylcarbazole, ( ) N-ethylcarbazole. 66  C14 H25 N → C14 H21 N + 2H2 → C14 H17 N + 2H2 → C14 H13 N + 2H2  (4.1)  Comparing the product distributions in Figures 4.6, 4.7 and 4.8 as a function of the catalyst Pd loading, it is clear that the same intermediates and products were identified for each catalyst. However, the product selectivity was dependent upon the Pd loading and hence the Pd particle size, and in no case was hexahydro-Nethylcarbazole nor decahydro-N-ethylcarbazole (two other possible hydrogenation reaction intermediates) observed. More importantly, the catalysts prepared by calcination in He were remarkably selective to the completely dehydrogenated compound, N-ethylcarbazole. As shown by the data of Figures 4.6 and 4.7, increased Pd loading from 0.5 - 4 wt %, corresponding to an increase in Pd particle size from 2 - 9 nm, resulted in a significant increase in the catalyst selectivity to the N-ethylcarbazole product. The consumption of octahydro-N-ethylcarbazole and tetrahydro-N-ethylcarbazole was much slower for the 0.5 wt % and 0.7 wt % catalysts. As seen in Figure 4.6, no consumption of tetrahydro-N-ethylcarbazole was observed in the first 120 min of the reaction. However, complete conversion of the two intermediates, octahydro-N-ethylcarbazole and tetrahydro-N-ethylcarbazole, to N-ethylcarbazole was obtained within about 95 min over the 4 wt % Pd/SiO2 (Figure 4.6). The selectivity to N-ethylcarbazole decreased with a further increase in the Pd loading to 5 wt % and 10 wt %, corresponding to Pd particle diameters of 13 - 15 nm (Figure 4.8). These results are compared to the low selectivity observed over the Pd/SiO2 prepared by calcination in air with a Pd particle size of 25 nm (Figure 4.8), reported previously [106]. The measured consumption of the reactant dodecahydro-N-ethylcarbazole as a function of time was fitted to a 1st -order kinetic model and the estimated rate constants are reported in Table E.5. Initial turn-over frequencies (TOFs) with respect to the consumption of dodecahydro-N-ethylcarbazole were calculated using the estimated 1st -order rate constants and the initial concentration of reactant used in the experiments, as follows: T OF = (k.AWPd .CA0 )/(MPd .D) 67  (4.2)  where k is the 1st -order rate constant for dodecahydro-N-ethylcarbazole consumption (L.min−1 .gcat −1 ), AWPd is the atomic mass of Pd (g.mol−1 ), CA0 is the reactant initial concentration (mol.cm3 −1 ), MPd is the Pd loading (wt %) and D is the Pd dispersion (%). Table 4.3: First order rate constants and turn-over frequencies (TOFs) for dehydrogenation of dodecahydro-N-ethylcarbazole at 443 K and 101 kPa over Pd/SiO2 catalysts with different Pd loadings. Pd (wt %) 0.5 0.7 1 2 3 4 5 10 5d  Rate constanta (min−1 ) 0.0457 ± 0.0033 0.0535 ± 0.0023 0.0935 ± 0.0022 0.1496 ± 0.0048 0.1668 ± 0.0047 0.2148 ± 0.0082 0.1139 ± 0.0015 0.0618 ± 0.0034 0.0075 ± 0.0002  Rate constanta,b (L.min−1 .gcat −1 ) 0.0036 0.0049 0.0134 0.0374 0.0626 0.0921 0.0683 0.0618 0.0045  Initial concentration (mol.cm3 −1 ) 9.9350×10−5 9.9350×10−5 9.9350×10−5 9.9272×10−5 9.9260×10−5 9.9350×10−5 9.8764×10−5 9.9350×10−5 1.3969×10−4  TOFc (min−1 ) 14.67 16.32 50.43 123.45 169.39 221.16 179.54 100.50 31.85  a with  respect to dodecahydro-N-ethylcarbazole consumption. sample calculation, see Appx. E. c molecule dodecahydro-N-ethylcarbazole.min−1 .Pd surface atom−1 . d Reported previously by Sotoodeh et al. [106]. b for  It is seen from Table E.5 that the reaction rate constant based on catalyst concentration increased significantly from 0.0036 L.min−1 .gcat −1 for the 0.5 wt % catalyst to 0.0921 L.min−1 .gcat −1 for the 4 wt % catalyst, and decreased to 0.0618 L.min−1 .gcat −1 with a further increase in Pd loading to 10 wt %. The same trends were observed for the TOF of the reaction, where the 4 wt % catalyst resulted in the highest TOF of 221.2 min−1 for the consumption of dodecahydroN-ethylcarbazole. The catalysts with 0.5 wt % and 0.7 wt % Pd loading resulted in the lowest values of the reaction rate constants and TOFs. Figure 4.9 plots the dodecahydro-N-ethylcarbazole dehydrogenation TOFs versus the Pd particle size for all the Pd/SiO2 catalysts of the present study. From Figure 4.9 it follows that the catalyst with a Pd particle size of about 9 nm, corresponding to the 4 wt % Pd/SiO2 catalyst, had the highest TOF. On the other hand, decreasing the particle size to about 2 nm with the 0.5 wt % Pd catalyst resulted in a significant reduction 68  in the TOF. Furthermore, an increase in the Pd particle size to about 12.5 nm with the 5 wt % and 15 nm with the 10 wt % catalysts, as well as the previously reported 24 nm with the 5 wt % catalyst prepared by calcination in air [106, 107], showed a decrease in the reactant consumption TOF.  250  200  TOF (min  -1  )  150  100  50 xx  0 0  5  10  15  20  25  30  Pd particle size (nm)  Figure 4.9: TOFs for dodecahydro-N-ethylcarbazole dehydrogenation at 443 K and 101 kPa over Pd/SiO2 catalyst with different Pd metal particle size. The open data point represents the TOF over the 5 wt % Pd/SiO2 reported by Sotoodeh et al. [106, 107]. The solid line represents the data fit using Murzin’s model [117]. The fitted parameter values were: P1 = 1.394±1.189, P2 = 0.0020±0.0016, γ = 81.60±12.01 and α = 0.124±0.030, with R2 = 97.7 %. Catalyst selectivity to N-ethylcarbazole at 50 % reactant conversion, as well as the selectivity after 1 h reaction at 443 K and 101 kPa over the different Pd/SiO2 catalysts, is summarized in Table 4.4. At 50 % conversion, a maximum selectivity of 15.3 % to N-ethylcarbazole was obtained in 4 min over the 4 wt % Pd/SiO2 . 69  Table 4.4: Comparison of the selectivity to N-ethylcarbazole and the conversion of dodecahydro-N-ethylcarbazole at 443 K and 101 kPa over Pd/SiO2 catalysts calcined in He. Pd (wt %) 0.5 0.7 1 2 3 4 5 10  Selectivity to N-ethylcarbazole at 50 % reactant conversion (%) 3.1 3.0 4.7 9.0 9.5 15.3 11.1 5.5  Reaction time to obtain 50 % conversion (min) 27.9 17.0 9.5 5.0 4.7 3.8 5.5 16.1  Selectivity to Nethylcarbazole at 1 h (%) 8.8 14.2 39.0 48.3 50.8 94.8 49.6 17.9  The selectivities at 50 % reactant conversion, shown in Table 4.4, demonstrate that increasing the Pd loading from 0.5 wt % to 4 wt % and accordingly increasing the Pd metal particle diameter from 2 nm to 9 nm, increased the observed selectivity to the completely dehydrogenated product, N-ethylcarbazole. With a further increase in Pd particle size a decrease in the selectivity to N-ethylcarbazole was observed. The selectivity to N-ethylcarbazole decreased to 11.1 % and 5.5 % for the 5 wt % and the 10 wt % catalysts, respectively (Table 4.4). Complete conversion of the reactant was achieved in 22 min over the 4 wt % catalyst, with 95 % selectivity to N-ethylcarbazole during the first hour of reaction. The H2 recovery profile of the dehydrogenation of dodecahydro-N-ethylcarbazole at 443 K and 101 kPa over the 4 wt % Pd/SiO2 is compared to the 5 wt % Pd/SiO2 reported previously [106] in Figure 4.10. The dramatic increase in the evolved H2 , observed using the 4 wt % Pd/SiO2 catalyst, is clear with the H2 recovery completed within 96 min. Complete conversion of the reactant was obtained in only 22 min. Table 4.5 summarizes the H2 recovery results obtained for all of the Pd catalysts. Complete H2 recovery, equal to the theoretical amount of 5.8 wt %, was obtained in 1.6 h using the 4 wt % Pd/SiO2 catalyst. These results are excellent compared to the 17 h required for complete conversion reported previously over  70  1.0  (a)  0.8  0.7  0.6  0.5  0.4  0.3  0.2  H  2  recovery profile (mmol/min)  0.9  (b)  0.1  0.0 0  20  40  60  80  100  120  140  160  180  200  Time (min)  Figure 4.10: H2 recovery comparison for dodecahydro-N-ethylcarbazole dehydrogenation at 443 K and 101 kPa (a) over 4 wt % Pd/SiO2 prepared by calcination in He (b) over 5 wt % Pd/SiO2 calcined in air and reported previously [106]. the 5 wt % Pd/SiO2 with 4.2 % metal dispersion [106]. The results demonstrate that the 4 wt % Pd/SiO2 catalyst with average particle diameter of 9 nm and 11 % Pd metal dispersion was the most active catalyst among those examined for the dehydrogenation of dodecahydro-N-ethylcarbazole.  4.4  Discussion  4.4.1 Catalyst properties The water peaks observed during temperature-programmed calcination of silica gel in He or air at 373 - 473 K (Figure 4.2 and 4.3) are ascribed to the removal 71  Table 4.5: H2 recovery for the dehydrogenation of dodecahydro-N-ethylcarbazole at 443 K and 101 kPa over Pd/SiO2 with different Pd loadings. Pd (wt %) 0.5 0.7 1 2 3 4 5 10 5d  H2 recoverya,b in 1 h, (wt %) 2.5 3.1 3.3 4.7 4.8 5.4 4.7 4.0 1.1  Final H2 recovery (wt %) 3.8 (6 h)c 4.4 (6 h) 5.2 (6 h) 5.8 (2.7 h) 5.8 (1.9 h) 5.8 (1.6 h) 5.8 (3 h) 5.8 (6 h) 4.0 (17 h)  a Mass  of H2 b Theoretical  released divided by the initial mass of dodecahydro-N-ethylcarbazole H2 recovery of dodecahydro-N-ethylcarbazole is 5.8 wt %. c Experimental time to achieve the H recovery reported. 2 d Reported by Sotoodeh et al. [106, 107]  of surface and micropore-absorbed water [118, 119], whereas the water loss at higher temperature (773 K) is ascribed to water loss from hydrogen-bonded silanol groups. The latter assignment follows work by Li et al. [119] who reported an infrared (IR) analysis of silica gel as temperature was increased to 773 K. The intensity of a band at 3650 cm−1 , related to hydrogen-bonded silanol groups, decreased with increased temperature, whereas a band at 3720 cm−1 , related to free silanol groups, increased slightly due to the conversion of some of the hydrogenbonded silanol groups absorbing at 3650 cm−1 into free silanol groups absorbing at 3720 cm−1 plus water. The preparation of Pd catalysts supported on Al2 O3 [120] and SiO2 [121] using PdCl2 as the precursor has been well studied. PdCl4 2− results when PdCl2 is dissolved in acidic solutions [122] and Contescu et al. [120] have shown that the PdCl4 2− is partially hydrolysed in solution to yield PdCl2 (H2 O)− and PdCl2 (H2 O)2 species. These species adsorb electrostatically onto acidified alumina. In the present work, the impregnation was conducted at low pH (< 2), below the isoelectric point of silica, so that a similar electrostatic adsorption on SiOH2 + can be expected. When the impregnated alumina was dried in air at 385 K for 3 h, Contescu et al. [120] demonstrated that PdCl4 (70 - 75 %) and PdClx Oy (25 - 30 72  %) were generated as a consequence of dehydration of the hydrolysed PdCl4 2− and reaction with atmospheric oxygen. Calcination at higher temperature (up to 925 K) in air resulted in further oxidation of the precursor, yielding PdOx (35 - 45 %) and an increased amount of PdClx Oy (55 - 65 %). The presence of Pdx Oy Clz species on Pd/Al2 O3 catalysts has been identified in other studies [112]. Vilarrasa-Garc´ıa et al. [123] detected the formation of metallic Pd and palladium oxychloride by XPS of activated carbons impregnated with PdCl2 and dried in He at 393 K for 6 h. Palladium oxychloride species were also identified by XPS of air-calcined Pd catalysts before reduction in H2 by Normand et al. [124]. The TPR data of Figure 4.3 are consistent with these same transformations of the precursor impregnated on silica. Accordingly, H2 O was released at ∼460 K and a second peak was observed at 773 K (Figure 4.3). The two H2 O peaks likely correspond to the water removal from the silica support as discussed earlier, as well as dehydration of two different hydrolysed Pd-species, as identified by Contescu et al. [120]. Of most significance, however, is the fact that no HCl (nor any other chlorinated products), were observed during the air drying and calcination of the catalyst precursors. This implies that the Pd precursors did not lose Cl during drying and calcination in air, and the catalyst precursor, following drying, must be mostly PdOx Cly species, such as the PdClO4 and PdCl2 O3 identified on Pd/alumina precursors following similar drying and calcination steps [120]. Furthermore, there was no colour change observed following calcination of the Pd/SiO2 catalyst precursor in air, indicative of PdOx Cly species rather than PdOx . In contrast, Figure 4.2 shows that when the impregnated silica was dried and calcined in inert He, a significant amount of HCl was detected, suggesting that reactions of the type PdCl2 + H2 O → PdO + 2 HCl occurred, rather than the oxidations that occurred in air to yield PdOx Cly . In this instance the samples recovered after calcination were black, indicative of PdOx species. The TPR profile also shows that only one peak was observed for the HCl evolution, and this occurred at a temperature very similar to the second peak temperature for water evolution (∼773 K). The profile is consistent with the dehydration of the hydrolysed Pd-precursors as already mentioned, and indicates that the removal of the Cl likely only occurs from PdCl2 or PdCl4 2− species, rather than hydrolysed Pd species. Following calcination, the precursors were flushed in H2 at room temperature 73  before being reduced according to the TPR protocol already described. HCl and H2 O were both evolved from the catalysts during the hydrogen flush and the TPR, confirming the reduction of the precursors to Pd0 during both processes. The reduction of PdCl2 at room temperature has been previously reported [125, 126]. Although the reduction profiles for the He and air calcined samples were very similar, for both the flush and the TPR, the sample calcined in He evolved more HCl, indicative of the easier reduction of PdCl2 and PdOx species on the He calcined sample than the reduction of PdOx Cly species present on the air calcined sample [120]. Different stoichiometry for CO adsorption on supported Pd has been reported in the literature [98]. The difference observed in Table 4.1 between the CO TPD and the HTREM results can be explained by the possible change in the stoichiometry for CO adsorption on Pd sites as the metal particle size increases. As reported in Table 4.1, catalyst dispersion was significantly increased for the catalysts prepared by calcination in He, compared to the catalyst prepared by wet impregnation and calcination in air [106, 107]. The effect of pre-treatment on Pd catalyst dispersion has been reported previously by Zou et al. [121] where an increase in Pd dispersion was observed by calcination in a noble gas instead of O2 . The lower dispersion in the latter case was attributed to the competitive interaction between Pd and the oxygen in the support versus Pd and the gas phase oxygen, that resulted in a partial weakening of the metal-support interaction and agglomeration of the Pd. In the present work, the improved dispersion is more likely a consequence of the different precursors that result from calcination in air (PdOx Cly ) versus He (PdOx ). The Pd oxychloride is known to be more difficult to reduce than the PdOx [120] and the data of Figure 4.2 and Figure 4.3 show that the peak reduction temperature was about ∼ 40 K higher for the air calcined sample than the He calcined sample. During the calcination process in He, both Cl and H2 O are lost from the Pd precursor, and this likely results in a strong interaction between the Pd and the O of the SiO2 support that anchors the Pd to the support. Consequently, during TPR the Pd remains well dispersed. In the case of the air dried sample however, because of the formation of PdOx Cly species that will reduce the interaction of Pd with the SiO2 , agglomeration of the PdOx Cly can occur, resulting in significantly lower dispersion of the air calcined sample compared to the He calcined sample. 74  Furthermore, the PdOx Cly has been identified as providing nucleation sites for Pd crystallite growth [120], and the presence of PdOx Cly species in the air calcined samples likely contributes to a reduced dispersion as well.  4.4.2 Catalyst activity and selectivity From Figure 4.9 and Table 4.4 it is seen that the 4 wt % Pd/SiO2 catalyst, with a Pd particle diameter of about 9 nm, resulted in a maximum in the dodecahydroN-ethylcarbazole dehydrogenation TOF and a maximum in selectivity to the completely dehydrogenated product N-ethylcarbazole. The N-ethylcarbazole selectivity over the 4 wt % Pd/SiO2 catalyst was 15.3 % at 50 % reactant conversion and 95 % after 1 h of reaction, significantly higher than that obtained for the other catalysts of Table 4.4. Complete conversion of the reactant was obtained in 22 min over the 4 wt % Pd/SiO2 catalyst, which is significantly shorter than the 17 h reaction time reported for dodecahydro-N-ethylcarbazole dehydrogenation at 443 K and 101 kPa over the 5 wt % Pd/SiO2 catalyst calcined in air [106]. In the present work, the complete H2 recovery from dodecahydro-N-ethyl carbazole obtained over the 4 wt % Pd/SiO2 catalyst after 1.6 h at 443 K, shows the significantly superior activity of the Pd/SiO2 catalysts prepared by calcination in He rather than air. The results reported herein are significant for hydrogen recovery applications, since the dehydrogenation reactions had previously been reported to be very slow with no selectivity to the completely dehydrogenated product, N-ethylcarbazole. Wang et al. [56] reported Ir complexes to be active for the dehydrogenation of dodecahydro-N-ethylcarbazole. Complete conversion occurred after 48 h at 473 K but dehydrogenation to the fully unsaturated product N-ethylcarbazole did not occur and octahydro-N-ethylcarbazole and tetrahydro-N-ethylcarbazole were reported to be the only observed products. Crawford et al. [53] and Hindle et al. [54] reported that the dehydrogenation of tetrahydrocarbazole at 413 K and 101 kPa proceeded to 81 % conversion over a 5 wt % Pd/Al2 O3 catalyst in 27 h, with an initial reaction rate of 10 mM. min−1 .gcat −1 , in reasonable agreement with the initial rate of 17 mM.min−1 .gcat −1 reported by the authors [106] over the 5 wt % Pd/SiO2 at 413 K and 101 kPa. Although 100 % selectivity to the completely dehydrogenated product carbazole was obtained, product inhibition was shown to slow  75  the reaction rate significantly. Decalin, another hydrogen storage candidate, was dehydrogenated over a Pt/C catalyst by Loutfy et al. [57] in a batch-type membrane reactor at temperatures of 533 - 593 K. Conversions below 85 % were reported after 2.2 h. Conversions > 95 % were obtained within 1.5 h at 553 K by Hodoshima et al. [58]. However, these reported reaction temperatures (> 535 K) are too high and impractical for hydrogen recovery for transport applications. The dependence of the dodecahydro-N-ethylcarbazole dehydrogenation TOF and selectivity to N-ethylcarbazole on Pd particle size, clearly demonstrated in the present work, suggests that the dehydrogenation reaction is structure sensitive. Pd particle size likely plays an important role in providing an adequate ensemble of catalytic sites for adsorption of the large dodecahydro-N-ethylcarbazole molecule and the other dehydrogenation intermediates. The maximum in TOF with particle size observed in the present work has also been observed in other reaction systems. For example, Binder et al. [127] reported a maximum in the ethene hydrogenation TOF over Pd/TiO2 catalysts at atmospheric pressure and 293 K, with the maximum occurring for a catalyst with a Pd particle size of 3 nm. Bezemer et al. [128] also reported a maximum in CO hydrogenation TOF over a series of Co catalysts, with the maximum occurring for a Co particle size of about 3 nm. Murzin [117, 129] has shown that the maximum in TOF can be accounted for by an analysis of the kinetic consequences of the different reactivities of edges and terraces present on the metal catalyst and the dependence of the fraction of edges on the cluster or particle size. Accordingly, for both two-step reaction mechanisms and Langmuir-Hinshelwood kinetics, the TOF can be written as [117, 129], T OF =  P1 e(1−α )γ /dcluster 1 + P2 eγ /dcluster  (4.3)  Where P1 and P2 are combinations of individual reaction frequencies, α is the Polanyi parameter, γ is the parameter accounting for the differences in Gibbs free energy of adsorption on terraces and edges and dcluster is the cluster diameter. The same equation will apply to the dehydrogenation TOF of dodecahydroN-ethylcarbazole, and Figure 4.9 shows the standard non-linear least-squares fit of 76  the data from the present study applied to equation 4.3. The estimated parameter values are reported in Figure 4.9 and the maximum in TOF at a particle size of about 9 nm is well explained by Murzin’s model. The Polanyi parameter obtained from the data fit was somewhat lower than the 0.5 often observed in heterogeneous catalytic reaction [117, 129]. Furthermore, the difference in the Gibbs free energy of adsorption on terraces versus edges is suggested to be significantly larger (300 kJ/mol) than that reported for ethene hydrogenation [117]. The present study has considered the effect of the preparation method on the Pd particle size, which is believed to be the major contributor to the significant change observed in catalyst selectivity. However, the change in the calcination environment may also generate surface defects due to chlorine and water removal from the precursors formed on the support, which can affect the activity of the catalyst. Density functional theory calculations were also used to confirm the structure sensitivity of the dehydrogenation reaction of dodecahydro-N-ethylcarbazole. The details are discussed in Chapter 6.  4.5  Conclusion  The complete recovery of the H2 stored on dodecahydro-N-ethylcarbazole was achieved at 443 K and 101 kPa using Pd/SiO2 catalysts, prepared by incipient wetness impregnation with calcination in He rather than air. The dehydrogenation turn-over frequency (TOF) and selectivity to the completely dehydrogenated product, N-ethylcarbazole, were dependent upon Pd particle size, with a maximum in both occurring with a 4 wt % Pd/SiO2 catalyst that had an average Pd particle size of 9 nm. Over this catalyst the dehydrogenation reaction proceeded to complete conversion within 22 min and complete H2 recovery (5.8 wt %) within 1.6 h at 443 K and 101 kPa.  77  Chapter 5  The effect of the N-heteroatom on the dehydrogenation of polyaromatics used for hydrogen storage A version of this chapter is to be submitted: F. Sotoodeh, B.J.M. Huber and K.J. Smith. The effect of the N-heteroatom on the dehydrogenation of polyaromatics used for hydrogen storage, to be submitted.  78  5.1  Introduction  As indicated in the previous chapters, incorporation of a heteroatom in the aromatic ring can lower the dehydrogenation enthalpy of the candidate compounds [16]. In the present chapter, the effect of N heteroatom on the dehydrogenation reaction rate was studied by comparing the rates of hydrogen recovery from dodecahydro-Nethylcarbazole, dodecahydrocarbazole and dodecahydrofluorene over Pd catalysts. The N heteroatom of dodecahydro-N-ethylcarbazole is bound to an ethyl group that should decrease the N - catalyst interaction compared to dodecahydrocarbazole, whereas there is no N heteroatom in dodecahydrofluorene.  5.2  Experimental  5.2.1 Catalyst preparation A commercial pre-reduced 5 wt % Pd on activated carbon catalyst (Aldrich, 5 wt % Pd, reduced) was used for the dehydrogenation reactions. A 2 wt % Pd/SiO2 catalyst was prepared by incipient wetness of the support [97, 111], described previously in Chapter 4. The prepared 2 wt % Pd/SiO2 catalyst was used for a comparative activity study with a 5 wt % Pd/SiO2 prepared previously by wet impregnation, as well as the 5 wt % Pd/C with the same particle size. The reported 5 wt % Pd/SiO2 catalyst was dried at 393 K for 24 h and calcined at 748 K for 3 h in air instead of He. Details of the catalyst preparation procedure have been presented elsewhere [106] and aslo discussed in Chapter 3.  5.2.2 Catalyst characterization N2 adsorption-desorption isotherms measured at 77 K using a Micromeritics ASAP 2020 Accelerated Surface Area and Porosimetry analyzer, were used to determine the total BET surface area, pore volume and average pore diameter of the catalysts. Approximately 0.12 g of the catalyst was weighed and loaded into a quartz sample tube for initial degassing. The sample was first evacuated at a rate of 1.33 kPa/s until the vacuum reached 0.04 kPa. Simultaneously, the temperature was increased at a ramp rate of 10 K/min until it reached 363 K. The evacuation phase was followed  79  by a heating phase where the temperature was further increased at the same rate and then held at 473 K for 60 min. After initial degassing the catalyst was weighed again and transferred to the analysis section for the N2 adsorption-desorption measurement. Temperature-programmed reduction (TPR) analysis and carbon monoxide (CO) pulse chemisorption were carried out on the 5 wt % Pd/C catalyst using a Micromeritics Autochem II 2920 analyzer. Approximately 0.12 g of the catalyst was loaded into the quartz sample tube. Prior to the TPR the sample was degassed in a He flow of 30 cm3 /min (STP) while heating to 423 K at a rate of 15 K/min. After subsequent cooling to ambient temperature, the sample was flushed in a flow of 10 % H2 in He for 30 min prior to the start of the TPR. The catalyst was then reduced by heating to 673 K at a rate of 10 K/min in a 30 cm3 /min (STP) flow of 10 % H2 in Ar. The final temperature was held for 30 min. After cooling the sample to ambient temperature, the CO pulse chemisorption analysis was initiated. The sample was flushed in a 30 cm3 /min (STP) flow of He carrier gas for 1 h. This was followed by repeated injections of pulses of CO into the reactor from a 20 cm3 /min (STP) flow of 10 % CO in He. The injection was continued until the TCD showed no further CO uptake by the sample. The metal dispersion was calculated using an assumed Pd:CO stoichiometry of 1:1 [98]. Temperature-programmed desorption (TPD) of CO from the 2 wt % Pd/SiO2 catalyst was performed using a Micromeritics Autochem II 2920 analyser. About 0.2 g of the dried catalyst was loaded into the reactor and calcined at 773 K for 6 h in a 30 cm3 (STP)/min flow of Ar, followed by TPR at 10 K/min to a final temperature of 673 K in a 30 cm3 (STP)/min flow of 10 % H2 in Ar, holding the final temperature for 1 h. The catalyst was then flushed in 30 cm3 (STP)/min flow of Ar for 1 h and cooled to room temperature, before being exposed to a 50 cm3 (STP)/min flow of 10 % CO in He at 313 K for 1 h. The catalyst was subsequently flushed in 50 cm3 (STP)/min He for 2 h. The temperature-programmed desorption was then started by heating the sample to 773 K at a rate of 10 K/min and holding the final temperature for 1 h.  80  5.2.3 Catalyst activity The hydrogenation of N-ethylcarbazole (Aldrich, 97 %) and carbazole (Sigma, Approx. 95 %) was carried out at 423 K and a H2 pressure of 7 MPa to produce the reactant for the dehydrogenation reactions. A 300 cm3 autoclave batch reactor with continuous monitoring and control of stirrer speed, temperature, and pressure was used for the hydrogenation reactions. Details of the experimental procedure and reactor set-up for hydrogenation reactions have been reported in Chapter 2. The dehydrogenation of dodecahydrocarbazole as well as dodecahydrofluorene (Aldrich, 97 %) were carried out in a 50 cm3 glass flask reactor at temperature in the range 423 - 443 K and 101 kPa, over the Pd/C catalyst, using the same reactor setup and procedure discussed in Chapter 3. Liquid samples of volume of < 0.1 cm3 were withdrawn from the reactor periodically. The analysis of the liquid samples obtained from the dehydrogenation reactions was done by a Shimadzu QP-2010S GC/MS using a Restek RTX5 30 m 0.25 mm capillary column. A comparative study on dehydrogenation of dodecahydro-N-ethylcarbazole was performed over the 2 wt % Pd/SiO2 and the 5 wt % Pd/C catalyst at 443 K and 101 kPa using the same dehydrogenation experiment set-up and procedure as that of used for dehydrogenation of dodecahydrocarbazole and dodecahydrofluorene.  5.3  Results  5.3.1 Catalyst properties The BET surface area, pore volume and average pore size of the catalysts are summarized in Table 5.1. As expected, the Pd/C catalyst had a significantly larger BET surface area (1055.1 m2 .g−1 ) compared to the 325.6 m2 .g−1 of the 5 wt % Pd/SiO2 , due to the highly porous nature of the activated carbon compared to the support silica. Hindle et al. [54] also reported a BET surface area of 1358 m2 .g−1 for a 10 wt % Pd/C. The average pore size obtained for the Pd/C catalyst was 2.0 nm, about 5 times smaller than the average pore size of 9.5 nm of the Pd/SiO2 . H2 TPR profiles of the 2 wt % Pd/SiO2 and the 5 wt % Pd/C catalysts are presented in Figure 5.1-a,b, respectively. A negative peak in the TPR profiles was 81  Table 5.1: Catalyst properties Catalyst  5 wt % Pd/C 5 wt % Pd/SiO2 b 2 wt % Pd/SiO2  BET surface area (m2 .g−1 ) 1055.1 325.6 346.63  Pore volume (cm3 .g−1 ) 0.55 1.1 1.1  Average pore size (nm) 2.0 9.5 10.1  Active particle size (nm) 6 24 6  Pd metal dispersion (%) 19a 4a 16c  a Obtained  from CO pulse chemisorption. by Sotoodeh et al. [106] c Obtained from TPD of CO.  b Reported  observed at temperatures between 333 - 337 K. In the work of Babu et al. [112] very similar H2 TPR profiles were observed for 3 and 5 wt % alumina-supported Pd catalysts and a negative peak in the region of 348-373 K was detected. Amairia et al. [113] also reported a negative H2 peak at 358 K for the H2 TPR profiles of Pd/Al2 O3 -ZrO2 samples. The observed negative peaks in the H2 profiles from the TPR experiments are likely due to the release of H2 from the decomposition of a Pd-hydride phase formed on large Pd particles (> 2 nm) [100, 112, 113]. As shown in Table 5.1, CO pulse chemisorption of the commercial 5 wt % prereduced Pd/C catalyst resulted in about 19 % metal dispersion and an active particle diameter of 6 nm. The CO TPD results show a 16 % Pd metal dispersion as well as an active particle size of about 6 nm for the 2 wt % Pd/SiO2 catalyst prepared by calcination in He. The results are comparable with the metal dispersion of 4 % and the particle size of 24 nm reported previously for a 5 wt % Pd/SiO2 catalyst prepared by calcination in air [106].  5.3.2 Catalyst activities Dehydrogenation of dodecahydrocarbazole The hydrogenation of carbazole at 423 K and 7 MPa using the 300 cm3 autoclave batch reactor proceeded with more than 97 % selectivity to the completely hydrogenated product, dodecahydrocarbazole, after about 13 h. Details of the kinetics of carbazole hydrogenation have been discussed previously [107]. The hydrogenated product, containing 1 wt % dodecahydrocarbazole dissolved in decalin, was trans82  TCD signal (a.u.)  (a)  (b)  300  350  400  450  500  Temperature (K)  Figure 5.1: TPR patterns of (a) 2 wt % reduced Pd/SiO2 catalyst prepared by calcination in He; (b) 5 wt % commercial pre-reduced Pd/C catalyst.  83  ferred to the 50 cm3 dehydrogenation glass flask reactor for the dehydrogenation reaction at three different temperatures (423 K, 433 K and 443 K) using the reduced 5 wt % Pd/C catalyst. For all three experiments the catalyst to reactant mass ratio was kept at 0.28 to make it possible to compare this work with the previously reported dehydrogenation of dodecahydrocarbazole reported using a 5 wt % Pd/SiO2 with 4 % dispersion and a Pd particle diameter of 24 nm [106, 107]. From the product distribution of the dehydrogenation reaction shown in Figure 5.2 it follows that the same intermediates and products were present in each case and the reaction rate increased by increasing the temperature. Complete conversion of the dodecahydrocarbazole was achieved after 30 min at 443 K, after 2 h at 433 K and after > 6.5 h at 423 K. Previous works reported 53 % reactant conversion for the dehydrogenation of dodecahydrocarbazole after 17 h over a 5 wt % Pd/SiO2 with Pd particle diameter of 24 nm at 443 K and 101 kPa [106], and 81 % of 1,2,3,4-tetrahydrocarbazole conversion after 27 h over a 5 wt % Pd/Al2 O3 at 413 K and 101 kPa [53]. As shown in Figure 5.2, octahydrocarbazole was produced in a significant amount as a primary product that was subsequently consumed. Tetrahydrocarbazole appeared as a secondary product in an even higher concentration and reacted more slowly. Hexahydrocarbazole was only produced in very small amounts observed shortly after octahydrocarbazole and was consumed very quickly suggesting a sequential dehydrogenation reaction pathway. Table 5.2 compares the kinetic data obtained for the dehydrogenation reactions at three different temperatures. Table 5.2: Kinetic data obtained from liquid phase analysis for dodecahydrocarbazole dehydrogenation in decalin using a 5 wt % Pd/C catalyst at various temperatures and 101 kPa. T (K) 443 433 423  Rate constant, k (h−1 ) 7.96 2.14 0.55  Initial rate (mM.min−1 .gcat −1 ) 65.1 20.3 5.2  Initial concentration (mol.ml−1 ) 4.63×10−5 2.84×10−5 2.83×10−5  TOF (min−1 ) 21.2 8.1 2.1  Time for 100 % reactant conversion (h) 0.5 2.0 > 6.5  The 1st -order rate constants (k) were calculated by fitting 1st -oder kinetics to the dodecahydrocarbazole conversion data. The initial reaction rates and turn-over 84  (a) -5  5.1x10  -5  3.4x10  -5  Concentration (mol/ml)  1.7x10  0.0  (b) -5  2.5x10  -5  1.7x10  -6  8.5x10  0.0  (c) -5  2.5x10  -5  1.7x10  -6  8.5x10  0.0 0  2  4  6  20  21  22  23  24  25  Time, h  Figure 5.2: Product distributions over time for dehydrogenation of dodecahydrocarbazole in decalin at 101 kPa and (a) 443 K, (b) 433 K, (c) 423 K over the 5 wt % Pd/C catalyst. ( ) carbazole, ( ) tetrahydrocarbazole, (◦) hexahydrocarbazole, (△) octahydrocarbazole, (⋄) dodecahydrocarbazole.  85  frequencies (TOFs) were calculated from the 1st -order rate constants and the initial reactant concentrations. From Table 5.2 it is seen that increasing the temperature from 423 - 443 K increased the dodecahydrocarbazole consumption TOF from 2.1 min−1 to 21.2 min−1 , respectively. Accordingly, the 1st -order rate constant for dodecahydrocarbazole consumption also increased from 0.55 h−1 at 423 K to 7.96 h−1 at 443 K. In previous work on the dodecahydrocarbazole dehydrogenation at 443 K over a 5 wt % Pd/SiO2 catalyst, the 1st -order rate constant was determined as 0.45 h−1 [107]. An apparent activation energy of 208.9 ± 0.2 kJ/mol for dehydrogenation of dodecahydrocarbazole was obtained by linear regression of the Arrhenius equation for the calculated 1st -order reaction rate constants as shown in Figure 5.3.  -6.0  -6.5  Ln (k)  -7.0  -7.5  -8.0  -8.5  -9.0  0.270  0.272  0.274  0.276  0.278  1/RT x10  0.280  0.282  0.284  0.286  3  Figure 5.3: Arrhenius plot for dehydrogenation of dodecahydrocarbazole at 423 - 443 K and 101 kPa over a 5 wt % Pd/C. Table 5.3 summarizes the H2 recovery data for the dehydrogenation of dodec86  ahydrocarbazole at three different temperatures. Table 5.3: H2 recovery for the dehydrogenation of dodecahydrocarbazole at various temperatures and 101 kPa over a 5 wt % Pd/C catalyst. Temperature (K) 443 433 423  Reaction time (h) 6.3 21.5 21.1  H2 recoverya (wt %) 6.6 6.6 4.3  a amount  of released H2 in grams divided by the weight of hydrogenated compound in grams (theoretical H2 recovery of doecahydrocarbazole is 6.7 wt %).  The H2 recovery increased by increasing the temperature. A 6.6 wt % H2 recovery (very close to the theoretical value of 6.7 wt % H2 release capacity) was obtained after 6.3 h for the dehydrogenation reaction at 443 K, while the same recovery was obtained after 21.5 h for the reaction at 433 K. For the experiment at 423 K, only 4.3 wt % of the stored H2 was recovered after 21.1 h. Dehydrogenation of dodecahydrofluorene in decalin Dehydrogenation of dodecahydrofluorene was conducted at 443 K and 101 kPa using 0.2 g of the 5 wt % Pd/C catalyst and a 2.3 wt % solution of commercially available dodecahydrofluorene in decalin. The product distribution obtained by analyzing the liquid samples is shown in Figure 5.4. It was found that decahydrofluorene and hexahydrofluorene were two reaction intermediates produced in very small amounts, and the reaction proceeded with production of the completely dehydrogenated compound fluorene as the major product. Conversion of the reactant, dodecahydrofluorene, was found to be very slow. By fitting 1st -order kinetics to the data obtained for the consumption of dodecahydrofluorene, a rate constant of 0.012 h−1 , an initial reaction rate of 0.127 mM.min−1 .gcat −1 and a TOF of 0.043 min−1 were obtained. After 19 h of reaction, a dodecahydrofluorene conversion of only 10.1 % and a yield of 9.6 % for the completely dehydrogenated product fluorine were achieved. Although the reaction kinetics were slow, the catalyst selectivity to the final product fluorene was found 95.6 %. The product distribution in Figure 5.4 suggested that the decahydrofluorene and hexahydrofluorene were the primary intermediates and fluorene was a 87  -4  -5  1.30x10  1.4x10  Hexahydrofluorene  -5  1.2x10  Decahydrofluorene Fluorene  -4  -5  1.25x10  1.0x10  -4  -6  1.23x10  8.0x10  -4  -6  1.20x10  6.0x10  -4  -6  1.18x10  4.0x10  -4  1.15x10  -6  2.0x10  -4  1.13x10  Product concentration (mol/ml)  Reactant concentration (mol/ml)  Dodecahydrofluorene -4  1.28x10  0.0  0  5  20  Time (h)  Figure 5.4: Product distribution over time for dodecahydrofluorene dehydrogenation in decalin at 443 K and 101 kPa over the 5 wt % Pd/C catalyst. Reactant concentrations (closed data points) are shown on the left Y-axis and products concentrations (open data points) on the right Y-axis. secondary product. In the work by Cooper et al. [130] dehydrogenation of pure perhydrofluorene (dodecahydrofluorene) under 101 kPa H2 pressure at 503 K using a Pt/Al2 O3 catalyst resulted in 99 % conversion to fluorene with > 99 % selectivity. However, the reported temperature is too high for on-board hydrogen recovery and therefore impractical for transport applications. Solvent effect in the dehydrogenation of dodecahydrofluorene In order to investigate whether the solvent decalin had an inhibitory effect on the dehydrogenation reaction rate of the dodecahydrofluorene, the dehydrogenation reaction was carried out without a solvent at 443 K and 101 kPa using 15 cm3 pure 88  dodecahydrofluorene loaded into the dehydrogenation reactor with 0.5 g of the 5 wt % Pd/C catalyst. The product distribution obtained from pure dodecahydrofluorene dehydrogenation, as shown in Figure 5.5, illustrates the presence of the same products as in the experiment carried out in the solvent decalin.  -3  -4  5.1x10  3.0x10 Dodecahydrofluorene  Decahydrofluorene  -4  2.5x10  Fluorene  -4  -3  2.0x10  -3  1.5x10  -3  1.0x10  -3  5.0x10  4.9x10  -4  4.8x10  -4  4.7x10  -5  4.6x10  Product concentration (mol/ml)  Reactant concentration (mol/ml)  Hexahydrofluorene -3  5.0x10  0.0  -3  4.5x10  0  5  20  Time (h)  Figure 5.5: Product distribution over time for dehydrogenation of pure dodecahydrofluorene at 443 K and 101 kPa over the 5 wt % Pd/C catalyst. Reactant concentrations (closed data points) are shown on the left Y-axis and products concentrations (open data points) on the right Y-axis. Again, fluorene exhibited the progression of a secondary product and first appeared in the sample taken after 30 min. The primary product hexahydrofluorene exhibited the behaviour of an intermediate that was initially formed and then consumed. Table 5.4 summarizes the obtained kinetic results for the dehydrogenation of dodecahydrofluorene with and without the presence of solvent. A rate constant of 0.0122 h−1 was found by fitting 1st -order kinetics to the ex89  Table 5.4: Comparison of kinetic data obtained from liquid phase analysis for dodecahydrofluorene dehydrogenation with and without a solvent at 443 K and 101 kPa over 5 wt % Pd/C catalyst. Status Using solvent (decalin) Pure reactant  Rate constant, k (h−1 ) 0.0120 0.0122  Initial rate (mM.min−1 .gcat −1 ) 0.127 2.040  TOF (min−1 ) 0.043 0.349  perimental data obtained for the consumption of dodecahydrofluorene. The initial reaction rate and the catalyst TOF with respect to the reactant consumption were found to be 2.04 mM. min−1 .gcat −1 and 0.349 min−1 , respectively. Also, dodecahydrofluorene conversion of 12.4 % with a yield of 12.0 % and a selectivity of 96.5 % for the production of the completely dehydrogenated product fluorene was achieved. The obtained results for the 1st -order rate constant of the dehydrogenation of the pure dodecahydrofluorene compared to the reaction in decalin solvent confirms that there is no solvent effect in the reaction. In both cases, only about 1 wt % of the hydrogen was recovered from the hydrogenated compound, which is far from the theoretical capacity of 6.74 wt %. Although the previous work on the dodecahydrofluorene by Cooper et al. [130] resulted in full hydrogen recovery at 503 K, the department of energy (D.O.E.) temperature criterion of 473 K for on-board dehydrogenation reactions limits the use of dodecahydrofluorene as a candidate for hydrogen recovery.  5.3.3 Dehydrogenation reaction comparison of dodecahydro-N-ethylcarbazole, dodecahydrocarbazole, dodecahydrofluorene Table 5.5 summarizes the kinetic results as well as the H2 recovery for the dehydrogenation reactions of dodecahydro-N-ethylcarbazole, dodecahydrocarbazole and dodecahydrofluorene. Comparing the dehydrogenation of dodecahydro-Nethylcarbazole with dodecahydrocarbazole over the Pd catalyst with an active Pd particle diameter of 24 nm, it is seen that the reaction TOF of 31.8 min−1 obtained for dodecahydro-N-ethylcarbazole is significantly higher than the 0.79 min−1 for dodecahydrocarbazole. Decreasing the Pd particle size from 24 nm to 6 nm resulted in a dramatic 90  Table 5.5: Dehydrogenation  reaction comparison of dodecahydro-Nethylcarbazole, dodecahydrocarbazole and dodecahydrofluorene at 443 K and 101 kPa. H2 capacitya (wt %) 5.79  Particle size (nm) 24d 6g 6h  Initial concentration (mol.ml−1 ) 1.40×10−4 9.92×10−5 9.92×10−5  TOFb (min−1 )  Conversion (%)  31.8 97.8 59.6  100(17 h)e 100(40 min) 100(94 min)  H2 recoveryc (wt %) 4.00(17 h)f 5.79(2.7 h) 5.79(12 h)  Dodecahydrocarbazole  6.70  24d 6h  1.14×10−4 4.63×10−5  0.8 21.2  53(17 h) 100(30 min)  1.90(17 h) 6.57(6.3 h)  Dodecahydrofluorene (in decalin)  6.74  6h  2.84×10−5  0.04  10.1(19 h)  1.06(19 h)  Dodecahydrofluorene (pure)  6.74  6h  5.00×10−3  0.35  12.4(> 70 h)  0.75(70 h)  Reactant  DodecahydroN-ethylcarbazole  of reactant.min−1 .Pd surface atom−1 . H2 released (grams) divided by the weight of hydrogenated compound (grams). c H released in experiment (grams) divided by the weight of hydrogenated compound (grams). 2 d Using 5 wt % Pd/SiO . 2 e Experimental time for the obtained conversion. f Experimental time for the obtained H recovery. 2 g Using 2 wt % Pd/SiO . 2 h Using 5 wt % Pd/C.  b Molecule  a Theoretical  increase in the reaction TOF for dodecahydro-N-ethylcarbazole consistent with the structure sensitivity of this reaction reported previously over a series of Pd catalysts [131]. The data for dodecahydro-N-ethylcarbazole dehydrogenation over the 2 wt % Pd/SiO2 and the 5 wt % Pd/C, both of which have an average Pd particle size of 6 nm, also show some impact of the SiO2 and C supports, with the dehydrogenation TOF higher on the SiO2 supported Pd than the C supported Pd. The reason for the difference in TOF is not clearly understood, but it is likely a consequence of different thermal treatments during preparation between the Pd/SiO2 catalyst and the commercial Pd/C catalyst. As shown recently, the atmosphere 91  present during thermal treatment of the Pd catalyst precursor can have a profound effect on the catalyst activity [131]. Nonetheless, the data of Table 5.5 show that both dodecahydro-N-ethylcarbazole and dodecahydrocarbazole dehydrogenation are structure sensitive reactions. For both reactants, the TOF increased as the Pd particle size decreased. Note, however, that the shorter time for complete H2 recovery reported in Table 5.5 for dodecahydrocarbazole dehydrogenation compared to dodecahydro-N-ethylcarbazole dehydrogenation, is a consequence of differences in reactant molar concentration and catalyst concentration in each case.  5.4  Discussion  5.4.1 Catalyst properties As reported in Table 5.1, catalyst dispersion was significantly increased for the 2 wt % Pd/SiO2 catalyst prepared by the modified preparation method and calcination in He, compared to the catalyst prepared by wet impregnation and calcination in air [106, 107]. Zou et al. [121] also reported the effect of pre-treatment on the Pd catalyst dispersion where an increase in the catalyst dispersion was observed by calcination in noble gas instead of O2 . The lower dispersion in the latter was attributed to the competitive interaction of oxygen in the support with gas phase oxygen that caused partial weakening of surface metal bond and metal agglomeration and therefore resulting in a lower metal dispersion. More recently, the improved dispersion that follows calcination in He rather than air has been shown to be a consequence of the different precursors that result from calcinations in air (PdOx Cly ) versus He (PdOx ). In the TPR profiles, the observed negative peak for the pre-reduced 5 wt % Pd/C and the reduced 2 wt % Pd/SiO2 catalysts is attributed to the decomposition of the β -palladium hydride phase which is reported to form at room temperature over large Pd particles (> 2 nm) [100, 114, 115] during the flushing step that was done in a flow of 10 % H2 in He for 30 min prior to the start of the TPR.  92  5.4.2 Catalyst activities The sudden increase in the H2 recovery rate from dodecahydrocarbazole by increasing the temperature from 423 K to 443 K is expected as very high activation energy of 208.9 ± 0.2 kJ/mol was found for the dehydrogenation reaction of dodecahydrocarbazole. The obtained activation energy for dodecahydrocarbazole dehydrogenation is relatively high compared to the reported 126.8 kJ/mol value for the dehydrogenation of dodecahydro-N-ethylcarbazole over a 5 wt % Pd/SiO2 catalyst [107], and the 67.7 kJ/mol value for the dehydrogenation of tetrahydrocarbazole over the same 5 wt % Pd/SiO2 [106]. The latter was found in good agreement with the 64.4 kJ/mol value obtained by the DFT study of Crawford et al. [53] and confirmed by the experimental work with the value of about 67 kJ/mol reported by Hindle et al. [54]. The high activation energy for dehydrogenation of dodecahydrocarbazole can be attributed to the strong interaction of the N heteroatom with the Pd surface atoms, resulting in strong adsorption of the molecule to the surface. In dodecahydro-N-ethylcarbazole, the ethyl group hindrance prevents the N heteroatom to interact strongly with the surface and hence results in a lower dehydrogenation activation energy. The present study has compared kinetic results as well as the H2 recovery for the dehydrogenation reactions of dodecahydro-N-ethylcarbazole, dodecahydrocarbazole and dodecahydrofluorene. The data show that the dehydrogenation reactions of dodecahydro-N-ehylcarbazole and dodecahydrocarbazole are structure sensitive. The lower reaction TOF obtained for dodecahydrocarbazole dehydrogenation is attributed to the effect of the N-heteroatom lone electron pair which interacts strongly with the catalyst surface and poisons the active catalytic sites and therefore, decreases the reaction rate. Catalyst poisoning by the product carbazole was also reported by the work of Hindle et al. [54] for dehydrogenation reaction of tetrahydrocarbazole. Interaction of the N heteroatom with the catalyst surface however, is decreased for dodecahydro-N-ethylcarbazole and its dehydrogenated product most likely due to the steric hindrance of the ethyl group. The effect of the N heteroatom in catalyst poisoning was also reported for H2 storage through hydrogenation reaction of carbazole compared to N-ethylcarbazole [107]. 100 % conversion of the reactant N-ethylcarbazole was obtained in about 1 h at 423 K over  93  a Ru/Al2 O3 catalyst, whereas the complete conversion of carbazole was obtained after about 13 h at the same reaction temperature and a catalyst/reactant ratio of 6 times higher than that used for N-ethylcarbazole hydrogenation. Comparing the kinetics of the dehydrogenation reaction of dodecahydrofluorene with dodecahydrocarbazole in Table 5.5, it is seen that a much lower TOF for dodecahydrofluorene was obtained compared to dodecahydrocarbazole. Furthermore, the very small amount of H2 recovery at the reaction temperature of 443 K after a very long experimental time, suggests that the N-heteroatom favors the dehydrogenation reactions of organic heteroaromatic compounds significantly. The effect of introducing the N heteroatom to replace a carbon in polycyclic aromatic hydrocarbons was also reported to result in a lower absolute value of the standard enthalpy of hydrogenation, and therefore decrease the reversible dehydrogenation barrier [16]. As discussed earlier, complete H2 recovery from dodecahydrofluorene obtained in a work by Cooper et al. [130] at 503 K suggests a high reaction barrier for the dehydrogenation of dodecahydrofluorene at 101 kPa. However, this temperature is not within the limits of on-board H2 recovery determined by the department of energy. Noticeably, although dodecahydrocarbazole and its dehydrogenated products have been shown to poison the catalytic active sites significantly due to the strong interaction of the N-heteroatom lone electron pair with the surface, the presence of the N-heteroatom in the polycyclic aromatic structures still has a positive effect in increasing the H2 recovery rate.  5.5  Conclusion  Dodecahydro-N-ethylcarbazole dehydrogenation was compared to dodecahydrocarbazole and dodecahydrofluorene. Complete H2 recovery from both dodecahydroN-ethylcarbazole and dodecahydrocarbazole was achieved over a Pd catalyst with an active particle diameter of about 6 nm. However, the dehydrogenation reaction TOF of dodecahydro-N-ethylcarbazole was found to be about 5 times faster than that of dodecahydrocarbazole. The faster reaction rate for dodecahydro-Nethylcarbazole was attributed to the effect of the ethyl group which prevented the N lone electron pair to interact strongly with the catalyst surface. It was found that the dehydrogenation reactions of the both reactants were structure sensitive, where  94  the active particle size had a significant effect on the selectivity of the dehydrogenation reactions to the completely dehydrogenated products, N-ethylcarbazole and carbazole. It is concluded that N-ethylcarbazole is a more practical candidate for both H2 storage and release than carbazole, although its storage capacity is about 13 % lower (5.8 wt % for N-ethylcarbazole compared to 6.7 wt % capacity for carbazole). Dodecahydrofluorene dehydrogenation reaction was shown to be the slowest among those examined. Only about 12 % of the reactant was converted and a H2 recovery of less than 1 wt % was achieved in more than 20 h which is impractical for on-board H2 recovery applications. The results supported the assertion that the presence of a N-heteroatom in the polycyclic structure favors the dehydrogenation reactions significantly. It was found from the present work that despite catalyst deactivation by strong adsorption through the N-heteroatom of dodecahydrocarbazole and the other dehydrogenated intermediates, the TOF of the dodecahydrocarbazole dehydrogenation reaction was still about 80 times higher than the corresponding value for dodecahydrofluorene dehydrogenation.  95  Chapter 6  Density functional theory study of dodecahydro-N-ethylcarbazole dehydrogenation over a Pd catalyst A version of this chapter is to be submitted: F. Sotoodeh and K.J. Smith. Density Functional Theory Study of Dodecahydro-Nethylcarbazole Dehydrogenation over a Pd Catalyst, to be submitted.  96  6.1  Introduction  Organic heteroaromatic compounds such as carbazole and N-ethylcarbazole have high gravimetric storage capacity as discussed earlier, and their hydrogenated forms can release hydrogen at low temperatures of less than 473 K and therefore are potential candidates for hydrogen storage and recovery. As indicated in Chapter 4, dehydrogenation of dodecahydro-N-ethylcarbazole was found to be structure sensitive. In a DFT study by Crawford et al. [53] on the mechanism of 1,2,3,4tetrahydrocarbazole dehydrogenation over a Pd catalyst, it was shown that the reaction proceeded through a flat 1,2,3,4-tetrahydrocarbazole adsorption geometry involving several Pd surface sites, indicative of a structure sensitive reaction. Adsorption of tetracene over a Ru surface was also reported to be flat requiring multiple Ru surface atoms [110]. Investigation of the adsorptions of quinoline on a Pt cluster reported by Vargas et al. [92], also indicated that several Pt atoms on the surface were involved in the adsorption. From the experimental and theoretical studies on benzene adsorption on Pd(111) it was revealed that depending on the surface coverage, benzene is adsorbed parallel with the surface (through the interaction of π electrons) or in a tilted geometry [132–137] with multiple Pd atoms involved in the adsorption. As discussed in Chapter 4, it was found that the dehydrogenation reaction of dodecahydro-N-ethylcarbazole was structure sensitive and a modified Pd catalyst resulted in 100 % selectivity to the product N-ethylcarbazole in a significantly shorter reaction time. In the present chapter, density functional theory (DFT) calculations are used to study the dehydrogenation reaction network of dodecahydroN-ethylcarbazole over a Pd(111) surface and to better understand the observed dehydrogenation reaction pathway, the product distribution and the structure sensitivity of the reaction. Furthermore, the obtained results for the adsorption geometry of dodecahydro-N-ethylcarbazole are compared to those of dodecahydrocarbazole and dodecahydrofluorene, two other candidates for H2 recovery for transport applications, in order to understand the slow H2 recovery reaction rate observed experimentally for the two latter compounds.  97  6.2  Computational methods  The DFT calculations were performed using DMol3 [138, 139] program and the graphical molecular displays were generated with Materials Studio 4.4.0.0. (Accelrys Software Inc.). The geometry of the reactant and the products were optimized using the double-numeric quality basis set (DNP) with the gradient-corrected GGA functional of Becke [74] and PBE description of exchange and correlation effects [140, 141]. For the numerical integration, a Fermi smearing of 2×10−3 Ha (1 Ha = 27.21 eV), a real-space cut-off of 4.0 A˚ and the MEDIUM quality mesh size of the program (using about 1000 grid points for each atom in the calculation) were used. Different possible geometries of the reactant and product molecules were optimized in order to identify the most probable isomers. Adsorption geometries of the most probable isomers of the reactant and products on Pd(111) surface were examined to determine the first adsorption sites. For this purpose, the Pd(111) surface was modeled by periodic slabs of four Pd layers with a 30 A˚ vacuum region between the slabs, and a (3×3) supercell was created to provide adequate surface. For each adsorption case, the molecule and the top layer of the Pd(111) surface were allowed to relax. Density functional semi-core pseudopotentials (DSPP) were used for the Pd substrate. The tolerances of energy, gradient, and displacement conver˚ and 5×10−3 A, ˚ respectively. Adsorption gence were 2×10−5 Ha, 4×10−3 Ha/A, energies were determined by subtracting the energies of the molecule in vacuum and the Pd(111) surface from the energy of the adsorption system, as follows,  Eadsorption = Eadsorbed molecule/sur f ace − Emolecule in vacuum − Esur f ace  (6.1)  Where in this equation Eadsorbed molecule/sur f ace is the energy of the adsorption system, Emolecule in vacuum is the energy of the adsorbed molecule in vacuum, and Esur f ace is the energy of the surface.  98  6.3  Results and discussion  6.3.1 Molecular structures Dodecahydro-N-ethylcarbazole Figure 6.1 and 6.2 shows the chemical structure of dodecahydro-N-ethylcarbazole molecule with a label assigned to H and C atoms, respectively.  a) vvvv  H6a  H5b H4bb  H6b  H4aa  H3a  H4a  H3b H4b  H5a  H2b  H7a  H2a  H9a  H8aa  H7b  H1a  H8b  H1b  H8a  H10b  H11c  H10a  H11a H11b  Figure 6.1: Chemical structure of dodecahydro-N-ethylcarbazole with labeled H atoms The geometries of different isomers of the dodecahydro-N-ethylcarbazole molecule, the dehydrogenation reactant, were optimized to find the most probable isomers. Dodecahydro-N-ethylcarbazole molecule has a non-flat, distorted chemical structure. Table 6.1 compares the optimized structure and the total and binding energies  99  b) vvvv C5  C6  C4  C3  C4a  C4b  C7  C2  C9a  C8a  C1  C8 N  C10  C11  Figure 6.2: Chemical structure of dodecahydro-N-ethylcarbazole with labeled C and N atoms. For dodecahydrofluorene, N is replaced by C9 . of the most probable isomers of dodecahydro-N-ethylcarbazole in vacuum. The optimized geometries of the isomers were obtained by changing the position of the H atom in the 5-membered ring. As in Table 6.1, isomer No. 1 was found to have total and binding energies of -602.6875 Ha and -6.5514 Ha, respectively. The two 6-membered rings of isomer No. 1 are symmetric with respect to the yz plane and all four H atoms in the 5-membered ring face down. With this configuration, the ethyl group and the two 6-membered rings tilt away upwards. In isomer No. 2, with total and binding energies of -602.6886 Ha and -6.5524 Ha, H4aa (refer to Figures 6.1 and 6.2 for atoms labeling) faces up whereas the other three H atoms in the 5-membered ring face down. The absolute value of the binding energy of this isomer was found to be 3 kJ/mol higher than isomer No. 1 (1 Ha = 2625.5 kJ/mol). The higher binding energy of isomer No. 1 implies more stability of this isomer in vacuum. In isomer No. 3 however two of H atoms in the 5-membered ring are in the opposite direction with the other two. The calculated total and binding energies for this isomer are -602.6848 Ha and -6.5487 Ha, respectively. The  100  Table 6.1: Optimized geometries and energies of the most probable isomers of dodecahydro-N-ethylcarbazole in vacuum Isomer No.  Top view  Front view  Energy (a.u.a ): Total (Binding)  1  -602.6875 (-6.5514)  2  -602.6886 (-6.5524)  3  -602.6848 (-6.5487)  a Atomic  unit or Hartree (Ha), 1 a.u. = 27.211 eV = 2625.5 kJ/mol.  existence of three stable isomers for dodecahydro-N-ethylcarbazole was confirmed experimentally by observing three separate peaks from GC/MS liquid sample analysis of hydrogenation experiments of N-ethylcarbazole reported previously by the authors [107], with one dominant and two small peaks. Eblagon et al. [96] also reported three peaks for dodecahydro-N-ethylcarbazole in the hydrogenation reaction of N-ethylcarbazole. Decahydro-N-ethylcarbazole The optimized geometries of different possible isomers of decahydro-N-ethylcarbazole are shown in Table 6.2. 101  Table 6.2: Chemical structures and energies of decahydro-N-ethylcarbazole isomers in vacuum; the structures with symmetric double bond on the left 6membered rings are not shown due to having the same energy.  It was found that isomer No. 1 wherein one of the bonds in the 5-membered ring is dehydrogenated had the most stable structure with total and binding energy of -601.49 Ha and -6.35 Ha, respectively. The smallest energy difference between this isomer and the other examined ones is 26 kJ/mol, indicating that isomer No. 1 is likely the most stable isomer of decahydro-N-ethylcarbazole. As opposed to the hydrogenation reaction of N-ethylcarbazole where decahydro-N102  ethylcarbazole was reported as one of the intermediates produced in small quantities [107], this compound was not observed during the dehydrogenation reactions [106]. Octahydro-N-ethylcarbazole Table 6.3 illustrates the optimized geometries and chemical structures of different isomers of octahydro-N-ethylcarbazole. Table 6.3: Chemical structures and energies of octahydro-N-ethylcarbazole isomers in vacuum.  103  It is seen that isomer No. 1 in which the 5-membered ring is unsaturated had the highest absolute value of the total and binding energies (-600.3 Ha and -6.2 Ha, respectively). The energy difference of at least 131 kJ/mol between this isomer and the other isomers indicates that the suggested structure for isomer No. 1 most likely represents the structure of the produced octahydro-N-ethylcarbazole observed in the hydrogenation/dehydrogenation reactions [106]. Unlike dodecahydro-N-ethylcarbazole for which three different isomers were identified, only one peak in the GC/MS analysis of the liquid samples was detected for octahydro-N-ethylcarbazole implying that only one of the isomers of octahydro-N-ethylcarbazole is produced in the hydrogenation/dehydrogenation reactions. The same structure for octahydro-N-ethylcarbazole was reported in the work reported by Eblagon et al. [96] on the hydrogenation of N-ethylcarbazole by comparing the energy difference between the octahydro-N-ethylcarbazole isomers and N-ethylcarbazole molecule. Hexahydro-N-ethylcarbazole Table 6.4 shows the isomers of hexahydro-N-ethylcarbazole. This molecule was reported as one of the intermediates of the hydrogenation reaction of N-ethylcarbazole which was produced in small amounts [107]. However, this intermediate was not observed in the dehydrogenation reactions. Among the three isomers shown in Table 6.4, isomer No. 1 and 2 have very similar total and binding energies. In these two isomers, the 5-membered ring and one of the bonds in the sextets are dehydrogenated. However, the most stable structure for hexahydro-N-ethylcarbazole molecule was found to be isomer No. 3 with 53 kJ/mol energy difference, in which one of the 6-membered rings is completely dehydrogenated. The higher stability of this isomer of hexahydro-N-ethylcarbazole, compared to the other two, is expected as partial dehydrogenation of one of the 6-membered rings results in the loss of resonance of the sextet and hence increasing instability of the molecule [106, 142]. Isomer No.3 was also found to be the stable structure of hexahydro-N-ethylcarbazole in other work on the hydrogenation of N-ethylcarbazole [96].  104  Table 6.4: Chemical structures and energies of hexahydro-N-ethylcarbazole isomers in vacuum.  Tetrahydro-N-ethylcarbazole Geometries of different isomers of tetrahydro-N-ethylcarbazole were optimized as summarized in Table 6.5. It was found that isomer No. 2 in which the 5-membered ring and one of the 6-membered rings are unsaturated had the most stable structure in vacuum with at least 79 kJ/mol energy diference with the other isomers. The calculated total and binding energies were -597.9 Ha and -5.8 Ha, respectively. The same structure was reported by Crawford et al. [53] in the study of the adsorption of 1,2,3,4tetrahydrocarbazole molecule on Pd(111) surface through the 5-membered and the aromatic rings for which the dehydrogenation reaction was initiated by removing the hydrogen atoms from the aliphatic rings. In work on N-ethylcarbazole hydrogenation, the same structure was reported for tetrahydro-N-ethylcarbazole in the suggested hydrogenation reaction network [96].  105  Table 6.5: Chemical structures and energies of tetrahydro-N-ethylcarbazole isomers in vacuum.  6.3.2 Dodecahydro-N-ethylcarbazole adsorption on the Pd(111) surface Benzene adsorption structure is often used as a model system for the adsorption of aromatic compounds on transition metal surfaces [143]. Ab initio studies of the ad106  sorption of other aromatic molecules on transition metals have been limited, however, because of the large size of the aromatic molecules and thus the large number of transition metal atoms required to create an appropriate model of the metal surface. Larger molecules such as dodecahydro-N-ethylcarbazole also increase the number of possible adsorption orientations. The fcc(111) surface is an ideal substrate in that it is highly symmetric. In addition, it is also the closest-packed facet and the thermodynamically most stable [144]. Therefore, in the present work, the Pd(111) surface was chosen to study the adsorption and the dehydrogenation mechanism of dodecahydro-N-ethylcarbazole and to explain the observed dehydrogenation reaction network. Eight adsorption modes with respect to each axial C-H bond positioned above a Pd atom (referred to as atop adsorption) were considered for the adsorption of dodecahydro-N-ethylcarbazole isomer No. 1 to the Pd(111) surface. As seen in Figure 6.3, the hollow/atop adsorption mode with respect to the 5-membered ring and the axial C-H bonds were distinguished, respectively.  Figure 6.3: Hollow/atop adsorption of dodecahydro-N-ethylcarbazole isomer No. 1 on Pd(111); hollow with respect to the 5-membered ring; atop with respect to the axial C-H bonds. The dots indicate the position of the two stretched axial C-H bonds.  107  Among the four axial C-H bonds in the 5-membered ring and the two axial C-H bonds (perpendicular to the Pd(111) surface) in each of the 6-membered rings, atop adsorptions of dodecahydro-N-ethylcarbazole through H9a and H8aa atoms were found to be favorable and had very similar adsorption energies of -95.0 kJ/mol and -94.6 kJ/mol, respectively. As seen in Figure 6.3, the 5-membered ring is adsorbed on the hollow site of the Pd(111) surface and the two axial C-H bonds next to the N heteroatom are at atop position. The detailed structural properties of the adsorbed dodecahydro-N-ethylcarbazole molecule are summarized in Table 6.6. The weakening of the C-H bonds are characterized by a significant increase in the axial C9a -H9a and C8a -H8aa bond distance from 1.11 A˚ to about 1.16 A˚ as shown in Table 6.6. Table 6.6: Bond lengths change for dodecahydro-N-ethylcarbazole’s isomers atop adsorption on Pd(111). No.  Adsorbing hydrogen  1  H9a  Adsorption energy (kJ/mol) -95.0  H4aa H4bb H8aa  -88.8 -86.7 -94.6  H2b H3a H6a H7a  -93.1 -44.7 -36.8 -94.2  2  H9a  -93.0  3  H3a  -42.4  C-H distance, gas phase ˚ (A) C9a -H9a : 1.11 C8a -H8aa : 1.11 1.10 1.10 C8a -H8aa : 1.11 C9a -H9a : 1.11 1.10 1.10 1.10 C7 -H7a : 1.10 C8a -H8aa :1.11 C9a -H9a : 1.11 C9a -H9a : 1.12 C8a -H8aa : 1.11 C3 -H3a : 1.10 C6 -H6a : 1.10  C-H distance, adsorbed ˚ (A) 1.16 1.15 1.10 1.11 1.16 1.15 1.11 1.11 1.12 1.11 1.16 1.15 1.15 1.15 1.11 1.12  H-Pd distance, adsorbed ˚ (A) 1.99 2.05 2.66 2.60 2.01 2.08 2.49 2.49 2.26 2.73 2.00 2.16 2.22 2.16 2.35 2.19  This bond increase is larger than the increase from 1.11 A˚ to about 1.13 A˚ for the axial C-H bonds in cyclohexane [143] and likely explains the higher adsorption energy of -95 kJ/mol for dodecahydro-N-ethylcarbazole compared to -26.8 kJ/mol for cyclohexane adsorption on Pt(111) surface. Comparing to the 1.6 A˚ bond  108  length of atop adsorbed H on Pt [145], this C-H bond increase is significant taking ˚ Figure 6.4-a shows the into account the obtained long H-Pt bond length of 2.0 A. optimized geometry of dodecahydro-N-ethylcarbazole hollow/atop adsorption on Pd(111) surface with the stretched axial C9a -H9a and C8a -H8aa bonds.  Figure 6.4: Optimized geometry of dodecahydro-N-ethylcarbazole isomer No. 1 hollow/atop adsorption on Pd(111) surface; (a) front view, (b) side view with N-Pd distance of 3.84 A˚ and the 5-membered ring at an angle of 10◦ to the surface. The stretch in the two axial C-H bonds made the molecule tilt from the surface as seen in Figure 6.4-b, where the plane of 5-membered ring is tilted away at about 10◦ away from the Pd(111) surface. Adsorption through H7a and H2b where in each case the H atom was positioned approximately above a Pd atom, led to a similar adsorption geometry to that of H9a and H8aa with adsorption energies of -94.2 kJ/mol and -93.1 kJ/mol, respectively. The geometry optimization of both systems led to H7a and H2b atoms to tilt away from the surface with a small C-H ˚ and H9a and H8aa relocate at atop bond length increase from 1.10 A˚ to 1.11 A, 109  position again with the corresponding C-H bond length increase from 1.11 A˚ to ˚ This confirmed the stability of the adsorption mode through the axial about 1.16 A. H9a and H8aa . Positioning H4aa and H4bb on top of a Pd atom however, led to weaker adsorption energies of -88.8 kJ/mol and -86.7 kJ/mol, respectively, with little increase in C4b -H4bb bond length from 1.10 A˚ to 1.11 A˚ and no change in the C4a -H4aa bond distance. Adsorption through H3a and H6a resulted in the least favorable adsorption mode with the adsorption energies of -44.7 kJ/mol and -36.8 kJ/mol, respectively, and the C3 -H3a bond length increase from 1.10 A˚ to 1.11 A˚ ˚ An adsorption energy and C6 -H6a bond length increase from 1.10 A˚ to 1.12 A. of -93.0 kJ/mol and similar adsorbed geometry was obtained for dodecahydro-Nethylcarbazole isomer No. 2 compared to that of isomer No. 1 through H9a . The C9a -H9a and C8a -H8aa bond distances increased from their corresponding values ˚ confirming the 5-membered ring as the favorable site in the gas phase to 1.15 A, for the first C-H bond weakening in the molecule. Isomer No. 3 adsorbed weakly to the surface with adsorption energy of -42.4 kJ/mol, with C3 -H3a bond distance change from 1.10 A˚ to 1.11 A˚ and C6 -H6a bond distance change from 1.10 A˚ to ˚ As mentioned before, isomers No. 1, 2 and 3 were found to be the most 1.12 A. stable isomers of dodecahydro-N-ethylcarbazole in the present work. Hirshfield charges were obtained to have a better understanding of the reason for the observed stretch and weakening in the axial C-H bonds in the adsorption of dodecahydro-N-ethylcarbazole isomer No. 1 (shown in Table 6.7). Density of states of the adsorbed dodecahydro-N-ethylcarbazole were also calculated as illustrated in Figure 6.5. From Table 6.7 it is seen that electron density of the two axial hydrogen atoms nearest to the surface increases significantly when adsorbed. The increase for H9a and H8aa is about 0.06 a.u., larger than the 0.04 a.u. increase for the axial hydrogen atoms reported for cyclohexane upon adsorption on Pt(111) and the 0.03 a.u. increase for the carbon atoms of adsorbed benzene on Pt(111) [144]. The charge increase for the other two hydrogen atoms in the 5-membered ring, H4aa and H4bb , is about 0.03 a.u. upon adsorption. The charge increase at the four axial hydrogen atoms in the 5-membered ring is compensated by a decrease in charge of about 0.06 a.u. for the corresponding surface Pd atoms. The charge on the N heteroatom however, decreases by about 0.06 a.u., indicating some interaction of 110  Table 6.7: Hirshfield charges on dodecahydro-N-ethylcarbazole. Atom H9a H8aa H4aa H4bb N Pd9a a Pd8aa a Pd4aa -4bb a C9a C8a C4a C4b a Refers  Gas phase (a.u.) 0.0110 0.0109 0.0275 0.0273 -0.0910  Pd(111)  -0.0156 -0.0156 -0.0156 0.0138 0.0136 -0.0205 -0.0201  Hollow/atop (a.u.) -0.0470 -0.0447 -0.0026 -0.0086 -0.0338 0.0479 0.0441 0.0364 0.0235 0.0240 -0.0273 -0.0259  Charge change (a.u.) -0.0580 -0.0556 -0.0301 -0.0359 0.0572 0.0635 0.0597 0.0520 0.0097 0.0104 -0.0068 -0.0058  to Pd atoms under the corresponding H atom.  the N heteroatom to the closest Pd atom on the surface. Overall, the total charge on the dodecahydro-N-ethylcarbazole molecule increases by 0.22 a.u. upon adsorption corresponding to a net electron density flow from the Pd surface to the dodecahydro-N-ethylcarbazole molecule. This electron density flow is mostly located at the two axial H9a and H8aa atoms. The projected density of states (PDOS) shown in Figure 6.5 indicates stabilization energies for the dodecahydro-N-ethylcarbazole molecule adsorbed atop on Pd(111) through H9a and H8aa by comparing the characteristic peaks for dodecahydroN-ethylcarbazole molecule in vacuum (Figure 6.5-a) with that in the adsorbed state (Figure 6.5-b). The peaks corresponding to the energy value between 0.2-0.3 Ha for dodecahydro-N-ethylcarbazole in vacuum shown in Figure 6.5-a) are broadened and shifted to the negative values. The electronic states belonging to the adsorbed configuration of dodecahydro-N-ethylcarbazole in Figure 6.5-b, show the contributions from s and p-orbitals. The states above -0.35 Ha mainly have p characteristics, while the features below -0.45 Ha mainly come from s-orbitals. By comparing the density of states (DOS) of dodecahydro-N-ethylcarbazole in vacuum (Figure 6.5-a) with the adsorbed one in Figure 6.5-b, it is clear that the density of states of s and p-orbitals close the Fermi level, have changed when the molecule is adsorbed. Figure 6.5-c shows the change in the DOS of the Pd d-states (Figure 111  Figure 6.5: Densities of states for; (a) dodecahydro-N-ethylcarbazole in vacuum; (b) adsorbed dodecahydro-N-ethylcarbazole molecule; (c) dodecahydro-N-ethylcarbazole on Pd(111); (d) Pd(111) d-states. 6.5-d) upon adsorption of dodechydro-N-ethylcarbazole molecule.  6.3.3 Dodecahydro-N-ethylcarbazole dehydrogenation steps For the dodecahydro-N-ethylcarbazole adsorption mode shown in Figure 6.4, H9a and H8aa are in close proximity to the Pd surface atoms and can react with the surface. The H9a and H8aa atoms are 1.99 A˚ and 2.05 A˚ from the nearest Pd surface atoms, respectively. As discussed earlier, from the C-H bond changes shown in  112  Table 6.6, it is clear that the routes involving an initial cleavage of either H9a or H8aa are the most likely. In order to gain some insight into the relative reactivity of the four sequential hydrogen removal steps, two reaction pathways shown in Figure 6.6 were considered in the present work for dehydrogenation of dodecahydro-Nethylcarbazole to octahydro-N-ethylcarbazole over Pd(111).  Figure 6.6: Two possible dehydrogenation pathways for dodecahydro-Nethylcarbazole conversion to octahydro-N-ethylcarbazole; for both pathways dehydrogenation begins with dissociation of H9a . 113  Both pathways start with dissociation of H9a (intermediate I) as C9a -H9a showed the largest bond distance change upon adsorption of dodecahydro-N-ethylcarbazole (Table 6.6). In the reaction pathways illustrated in Figure 6.6, it is clear that the dissociation of H4aa in pathway 1 and H8aa in pathway 2 lead to the production of two unique intermediates denoted as intermediate II, resulting in the formation of a C=C double bond in pathway 1. This produced intermediate has the stable structure obtained for decahydro-N-ethylcarbazole (Table 6.2). Intermediate III for each pathway is produced by the third dehydrogenation step; and both species subsequently lose a further H atom to complete the dehydrogenation process. Reaction pathways 1 and 2 have been modeled over Pd(111) to determine their respective transition state energies and the reaction barriers for each of the dehydrogenation steps. As seen in Figure 6.7, removing the first hydrogen atom, H9a , resulted in C4a -H4aa and C8a -H8aa bonds to stretch from an equilibrium bond length of 1.10 A˚ ˚ respectively. to 1.13 A˚ and 1.12 A, The reaction barriers to all the dehydrogenation steps for removal of the first four hydrogen atoms are summarized in Table 6.8 and compared in Figure 6.8. Table 6.8: Reaction barriers corresponding to the transition states (TSs) for each dehydrogenation step along the two dehydrogenation reaction pathways of dodecahydro-N-ethylcarbazole shown in Figure 6.6. Pathway 1 Transition state (TS) TS1 TS2 TS3 TS4  Barrier (kJ.mol−1 ) 14.1 126.3 106.5 41.6  Pathway 2 Transition state (TS) TS1 TS2 TS3 TS4  Barrier (kJ.mol−1 ) 14.1 159.5 69.1 129.6  The calculated reaction barrier for the dissociation of the C9a -H9a bond was 14.1 kJ/mol. For both pathways, the reactivity trend showed an increase in the energy barrier from the dehydrogenation step 1 to 2. The energy barrier decreased in both pathways from the dehydrogenation step 2 to 3. However, for the dehydrogenation step 4, the reaction barrier along pathway 1 decreased significantly, whereas a large increase in the reaction barriers from step 3 to 4 was obtained in pathway 2. From Table 6.8 it follows that the dehydrogenation pathway 1 is more  114  Figure 6.7: Schematic of pathway 1 for dehydrogenation of dodecahydroN-ethylcarbazole; (a) adsorption geometry for dodecahydro-Nethylcarbazole on Pd(111) showing the interaction between H9a and ˚ respectively, (b) the H8aa and the nearest Pd atoms at 1.99 and 2.05 A, optimized geometry corresponding to the dissociation of C9a -H9a bond. The bond distance of C4a -H4aa and C8a -H8aa increased to 1.13 and 1.12 ˚ respectively, (c) removal step of H4aa with C8a -H8aa bond distance inA, ˚ (d) removal step of H8aa with C4b -H4bb bond distance creased to 1.16 A, ˚ and (e) production of octahydro-N-ethylcarbazole. increased to 1.18 A,  115  180  Reaction barrier (kJ/mol)  160 140 120 100 80 60 40 20 0  1  2  3  4  Dehydrogenation step  Figure 6.8: Comparison of the calculated reaction barriers for the progressive dehydrogenation of dodecahydro-N-ethylcarbazole over Pd(111) surface following the reaction pathways 1 ( ) and 2 (◦) illustrated in Figure 6.6. likely, as a larger bond distance increase was observed for C4a -H4aa compared to C8a -H8aa after the 1st dehydrogenation step (Figure 6.7) and the activation barrier for removal of H4aa in pathway 1 was about 33 kJ/mol smaller than that of H8aa in pathway 2. Also, the cleavage barrier of the 4th hydrogen, H4bb , leading to the formation of the 2nd C=C double bond and production of octahydro-N-ethylcarbazole was 88 kJ/mol smaller in pathway 1 compared to pathway 2. As seen in both Table 6.8 and Figure 6.8, for pathway 1, the removal of the 2nd hydrogen resulting in the production of decahydro-N-ethylcarbazole had the highest energy barrier, after which the hydrogen removal barriers decreased significantly. Furthermore, removal of the 2nd hydrogen (H4aa ) resulted in a significant stretch in the C8a -H8aa  116  bond length from 1.11 A˚ (in vacuum) to 1.16 A˚ (Figure 6.7-c) indicating that the 3rd H removal step is favored. This implies that decahydro-N-ethylcarbazole is, in fact, produced and consumed very quickly and confirms the experimental observations where no decahydro-N-ethylcarbazole was measured in the dehydrogenation of dodecahydro-N-ethylcarbazole. It was found that octahydro-N-ethylcarbazole adsorbs to the surface in a di-  σ mode through C9a and C4a atoms in the 5-membered ring with σ C−Pd bond ˚ respectively (Figure 6.9), and an adsorption energy distances of 2.25 A˚ and 2.29 A, of -56.0 kJ/mol.  Figure 6.9: Octahydro-N-ethylcarbazole (a) aromatic hollow di-σ adsorption on Pd(111); (b) front view of the optimized geometry with N-Pd distance of 2.89 A˚ and the 5-membered ring at an angle of 5◦ to the surface. The C-Pd bond distances are larger than the σ C−Pt distances of 2.19 A˚ and 2.16 A˚ obtained for di-σ adsorption of 1,3-cyclohexadiene on Pt(111) [143] and σ C−Pt 117  distances of 2.16 A˚ for adsorbed benzene [144]. Table 6.9 summarizes the structural properties of the adsorption of octahydroN-ethylcarbazole. The C9a =C4a bond distance increased from 1.39 A˚ for the gas phase octahydro-N-ethylcarbazole molecule to 1.49 A˚ upon adsorption to the Pd(111) surface. This increase in the C9a =C4a bond distance supports the di-σ adsorption of the molecule through C9a and C4a . Saeys et al. [143] reported a bond distance of 1.51 A˚ for the C=C bond in the di-σ adsorption of 1,3-cyclohexadiene on Pt(111) surface. Table 6.9: Selected parameters for the gas phase and adsorbed structure of octahydro-N-ethylcarbazole. C-Pd distances refer to corresponding Pd atom shown in Figure 6.9. Variable C9a -Pd C4a -Pd C9a -C4a  ˚ Gas phase (A) — — 1.39  ˚ Adsorbed (A) 2.25 2.29 1.49  In the optimized conformation shown in Figure 6.9, the two aliphatic rings as well as the other two C atoms in the 5-membered ring tilt away from the Pd(111) surface. Figure 6.10 compares the density of states for C9a and C4a in vacuum and upon adsorption to the surface obtained from periodic calculations. A significant shift and broadening of the energy levels of carbon p-orbitals are observed for both C9a and C4a , which likely point to the σ interaction of carbon atoms with the Pd surface. Table 6.10 summarizes the energy barriers to the dehydrogenation steps from octahydro-N-ethylcarbazole to tetrahydro-N-ethylcarbazole. Table 6.10: Reaction barriers corresponding to the transition states (TSs) for conversion of octahydro-N-ethylcarbazole to tetrahydro-N-ethylcarbazole. Transition state (TS) TS1 TS2 TS3 TS4  Removed H H1b H2b H4a H3a  Barrier (kJ.mol−1 ) 137.7 81.2 83.9 198.9  The adsorbed geometry of octahydro-N-ethylcarbazole dehydrogenation steps is also shown in Figure 6.11. Figure 6.11-a illustrates the optimized adsorbed 118  20  Density of states (a.u.)  35  a)  18  30 25  14 12  20  10 15  8 6  10  4 5  2  0  0 -0.8  -0.6  -0.4  -0.2  0.0  0.2  0.4  30  -1.0  -0.8  -0.6  -0.4  -0.2  0.0  0.2  -0.8  -0.6  -0.4  -0.2  0.0  0.2  20  35  Density of states (a.u.)  b)  16  c)  18  d)  16 25  14 12  20  10 15  8 6  10  4 5  2  0  0 -0.8  -0.6  -0.4  -0.2  0.0  0.2  0.4  -1.0  Energy (Ha)  Energy (Ha)  Figure 6.10: Densities of states (DOS) of octahydro-N-ethylcarbazole for; (a) C9a in vacuum; (b) C9a in the adsorbed molecule; (c) C4a in vacuum; (d) C4a in the adsorbed molecule. Dash lines represent the s-orbital and (solid line) represents the p-orbitals. geometry of octahydro-N-ethylcarbazole with the interaction of H1b atom with the surface. Both C1 -H1b and C2 -H2b bond distances increased to 1.11 A˚ upon the molecule adsorption, and therefore were likely to be the first bonds to dissociate. The energy barrier for C2 -H2b cleavage was found to be 259.7 kJ/mol (not shown here), while this value was about 137.7 kJ/mol for cleavage of C1 -H1b bond (Table 6.10), leading to the conclusion that dehydrogenation pathway for octahydro-Nethylcarbazole starting from C1 -H1b bond cleavage is likely more favored. 119  Figure 6.11: Dehydrogenation steps for octahydro-N-ethylcarbazole; (a) adsorbed geometry octahydro-N-ethylcarbazole showing the interaction ˚ (b) the optimized gebetween H1b and the nearest Pd atom at 2.53 A, ometry corresponding to the dissociation of C9a -H9a bond and removal of H1b . Interaction between H2b and the nearest Pd atom at 2.24 A˚ increases, (c) increasing the interaction of H4a with the nearest Pd atom ˚ (d) optimized geometry corresponding to the dissociation of at 2.16 A, H4a and increasing the interaction of H3a with the nearest Pd atom at ˚ (e) production of tetrahydro-N-ethylcarbazole. 2.5 A,  120  Comparing the obtained first step reaction barriers for production of octahydroN-ethylcarbazole (14.1 kJ/mol, Table 6.8) with that of tetrahydro-N-ethylcarbazole (137.7 kJ/mol, Table 6.10), it is concluded that hydrogen removal from the 5membered ring with a N-heteroatom is more favored than the 6-membered ring. This is in good agreement with the obtained experimental apparent activation energy of 144.3 kJ/mol for production of tetrahydro-N-ethylcarbazole [106], compared to the 67.1 kJ/mol obtained for production of octahydro-N-ethylcarbazole [106]. Figure 6.11-b illustrates the optimized geometry after removal of H1b , showing the increased interaction of H2b to the nearest Pd atom by an increase in C2 -H2b ˚ It was noticeable that upon cleavage of C1 -H1b bond, C9a bond distance to 1.12 A. and C4a detached from the surface and C1 made a σ C−Pd bond with the nearest Pd atom. The 2nd dehydrogenation step resulted in removal of H2b (Figure 6.11-c) and formation of a σ -bond between C2 and the nearest Pd on the surface with an energy barrier of 81.2 kJ/mol. The removal of H2b results in the formation of one of the hexahydro-N-ethylcarbazole isomers (Table 6.4). However, the considered dehydrogenation pathway for octahydro-N-ethylcarbazole (which is very likely to occur), does not lead to the formation of the most probable isomer of hexahydroN-ethylcarbazole (isomer No. 3 in Table 6.4). This is in agreement with the product distribution observed in dehydrogenation reactions where no hexahydroN-ethylcarbazole was produced. Further removal of H atoms resulted in cleavage of C4a -H4a and C3 -H3a . The adsorption energy of tetrahydro-N-ethylcarbazole to the Pd(111) surface was calculated as -117.5 kJ/mol, significantly smaller than the -143.8 kJ/mol value reported for the adsorption of tetrahydrocarbazole on Pd(111) [53]. The tetrahydroN-ethylcarbazole stable adsorption geometry and structural properties are shown in Figure 6.12 and Table 6.11. From Figure 6.12-a it is seen that the bridge adsorption of the aromatic ring of tetrahydro-N-ethylcarbazole is the favored geometry. Tetrahydro-N-ethylcarbazole adsorbs to the surface through C1−4 and the 5-membered ring tilts away from the ˚ The surface at an angle of 25◦ (Figure 6.12-b) and a N-Pd distance of 3.28 A. ˚ respectively. These values are C1 -Pd and C2 -Pd distances are 2.24 A˚ and 2.25 A, comparable to σ C−Pt distance of 2.16 A˚ for adsorbed benzene and 2.17 A˚ for di-  σ adsorbed ethane [143]. The C3 -Pd and C4 -Pd distances are longer at 2.28 A˚ 121  Figure 6.12: Tetrahydro-N-ethylcarbazole (a) aromatic bridge 1,2-di-σ /3,4-π adsorption on Pd(111); (b) front view of the optimized geometry with N-Pd distance of 3.28 A˚ and the 5-membered ring at an angle of 25◦ to the surface. ˚ respectively, more typical for π -adsorbed molecules, supported also and 2.29 A, by the larger increase observed in the C1 -C2 bond distance of 1.46 A˚ for their σ bonds to Pd atoms compared to 1.43 A˚ for the C3 -C4 bond for their less strong  π -adsorbed mode (Table 6.11). The obtained adsorbed geometry of tetrahydro-Nethylcarbazole supports the dehydrogenation pathway considered for octahydroN-ethylcarbazole, in which dehydrogenation starts by removal of hydrogen atoms from one of the sextets.  122  Table 6.11: Selected parameters for the gas phase and adsorbed structure of tetrahydro-N-ethylcarbazole. C-Pd distances refer to corresponding Pd atom shown in Figure 6.12. Variable C1 -Pd C2 -Pd C3 -Pd C4 -Pd C1 -C2 C3 -C4  ˚ Gas phase (A) — — — — 1.39 1.39  ˚ Adsorbed (A) 2.24 2.25 2.28 2.29 1.46 1.43  6.3.4 Dodecahydro-N-ethylcarbazole adsorption on Pd(110) surface The structure sensitivity of the dodecahydro-N-ethylcarbazole dehydrogenation reaction was also investigated by comparing the adsorption geometry of dodecahydroN-ethylcarbazole over a Pd(110) surface with that of Pd(111). Figure 6.13 illustrates the optimized geometry of the dodecahydro-N-ethylcarbazole adsorbed on top of the Pd(110) surface. The molecule locates on the surface so that the interaction of the two H atoms adjacent to N, increases (Figure 6.13-a). The calculated adsorption energy on the Pd(110) surface was -83.2 kJ/mol. The C-H bond distance increase from 1.11 A˚ to about 1.16 A˚ was observed for the adsorption of the molecule over the Pd(110) surface (Figure 6.13-b) which was the same as that obtained on the Pd(111) surface. No other C-H bond distance change was identified. Figure 6.13-c shows the top view of the adsorption geometry of dodecahydro-N-ethylcarbazole. The two H atoms adjacent to the N heteroatom are in atop positions, but note that the remaining H atoms of the 5-membered ring are located above Pd atoms in the second layer of the Pd(110) surface, making H removal on this surface less likely than on the Pd(111) surface. Because of the less packed structure of the Pd(110) surface compared to the Pd(111) surface, the interaction between Pd and the remaining H atoms in the 5-membered ring is not possible. These H atoms are located above Pd atoms of the second layer of the Pd(110) surface (shown in Figure 6.13-a in black) and the distance between the H and Pd atoms is such that there is no interaction. Hence to form octahydro-N-ethylcarbazole on the Pd (110) surface would require additional desorption and re-adsorption steps that are not required on the  123  Figure 6.13: Optimized geometry of dodecahydro-N-ethylcarbazole adsorbed over Pd(110) surface; (a) side view showing the interaction of the two H atoms with the first layer Pd surface, (b) front view showing the C-H bond distances increase from 1.10 A˚ in vacuum to 1.16 A˚ upon approaching the surface , (c) top view. The molecule moves so that the two C-H bonds adjacent to N heteroatom are at atop positions over the nearest Pd atoms. Blue: N, white: H, gray: C, light gray: first layer Pd atoms, black: second layer Pd atoms. Pd(111) surface. These observations also point to the fact that the dehydrogenation of dodecahydro-N-ethylcarbazole is structure dependent.  6.3.5 Heteroatom effect on the catalyst reactivity The effect of the N heteroatom on the dehydrogenation reaction was studied by comparing the adsorption energies and surface/adsorbate charge transfer of dodecahydroN-ethylcarbazole, dodecahydro-carbazole and dodecahydrofluorene molecules. Fig124  ure 6.14 compares the adsorption geometries of the three different molecules.  Figure 6.14: Comparison between the adsorbed structures of (a) dodecahydro-N-ethylcarbazole, (b) dodecahydrocarbazole, and (c) dodecahydrofluorene over Pd(111). (left) front view, (right) side view. As discussed earlier, dodecahydro-N-ethylcarbazole adsorption to the Pd(111) surface was characterized by a significant stretch of C9a -H9a and C8a -H8aa bonds (Figure 6.14-a) with adsorption energy of -95.0 kJ/mol. These stretches in the C-H bonds resulted in an approximate 10◦ rotation of the 5-membered ring plane with respect to the axis parallel with the surface. The N heteroatom in dodecahydroN-ethylcarbazole had a distance of 3.84 A˚ to the nearest Pd atom on the sur125  face, whereas this distance for dodecahydrocarbazole was found to be 2.22 A˚ (Figure 6.14-b). The shorter N-Pd distance obtained for dodecahydrocarbazole compared to dodecahydro-N-ethylcarbazole is indicative of a stronger interaction of the docecahydrocarbazole N heteroatom with the Pd(111) surface, whereas for dodecahydro-N-ethylcarbazole the ethyl group prevents the N heteroatom from interacting as strongly as that in dodecahydrocarbazole. This is also seen by the approximately 19◦ rotation of the plane of the 5-membered ring from the axis parallel to the Pd(111) surface (Figure 6.14-b), assisting the interaction of N atom with the surface compared to dodecahydro-N-ethylcarbazole. The adsorption energy of dodecahydrocarbazole on the Pd(111) was -109.4 kJ/mol, with an absolute value of 14 kJ/mol higher than that of dodecahydro-N-ethylcarbazole (Table 6.12). Table 6.12: Comparison of adsorption energies. Compound name DodecahydroN-ethylcarbazole Dodecahydrocarbazole Dodecahydrofluorene a Reported  Adsorption energy (kJ.mol−1 ) -95.0 -109.4 -180.8  Compound name TetrahydroN-ethylcarbazole Tetrahydrocarbazole Tetrahydrofluorene  Adsorption energy (kJ.mol−1 ) -117.5 -143.8a -218.6  by Crawford et al. [53].  No significant change in the C-H bond distances of the 5-membered ring was observed for dodecahydrocarbazole upon adsorption (Table 6.13). The stronger interaction of N heteroatom in dodecahydrocarbazole is likely responsible for the much slower dehydrogenation reaction rate observed compared to dodecahydroN-ethylcarbazole. Dodecahydrofluorene however, showed a much stronger adsorption than dodecahydrocarbazole and dodecahydro-N-ethylcarbazole. The N heteroatom in the dodecahydrocarbazole molecule is replaced by a C atom through which the adsorption of dodecahydrofluorene occurred (Figure 6.14-c) with a short ˚ The adsorption energy of dodecahydrofluorene was C-Pd distance of 2.11 A. found to be -180.8 kJ/mol, which points to a significantly stronger surface adsorption of the molecule compared to dodecahydrocarbazole or dodecahydro-Nethylcarbazole. The 5-memebred ring plane had a rotation of about 22◦ in this case ˚ shown in Figure 6.14-c, as expected by the observed short C-Pd distance of 2.11 A. 126  Compared to the large increases in the C-H bond lengths of the 5-membered ring in the adsorbed dodecahydro-N-ethylcarbazole (Table 6.6), no significant increase was observed for doecahydrofluorene, as summarized in Table 6.13. Table 6.13: Bond lengths change for dodecahydrofluorene and dodecahydrocarbazole upon adsorption to Pd(111) surface. Dodecahydrofluorene Adsorbing atom C9 -Pd ˚ (2.11 A)  Vacuum  Adsorbed  C-H ˚ (A) C9 -H9 : 1.09 C9a -H9a : 1.11 C8a -H8aa : 1.11 C4a -H4aa : 1.10 C4b -H4bb : 1.10  C-H ˚ (A) 1.10 1.12 1.13 1.11 1.10  Dodecahydrocarbazole Adsorbing atom N-Pd ˚ (2.22 A)  Vacuum  Adsorbed  C-H ˚ (A) N-H9 : 1.02 C9a -H9a : 1.11 C8a -H8aa : 1.11 C4a -H4aa : 1.10 C4b -H4bb : 1.10  C-H ˚ (A) 1.02 1.11 1.11 1.10 1.11  Hirshfield charges for dodecahydrocarbazole and dodecahydrofluorene upon adsorption to the surface were obtained from the periodic calculations in order to make a comparison with dodecahydro-N-ethylcarbazole. The charge decrease of about 0.08 a.u. on the N heteroatom in dodecahydrocarbazole upon adsorption due to the N lone pair charge donation (Table 6.14), together with the large charge decrease of 0.13 a.u. on the attaching Pd atom is responsible for the formed σ N−Pd bond. The total charge on the dodecahydrocarbazole molecule increased by 0.095 a.u., which is almost two times smaller than the 0.22 a.u. charge increase on dodecahydro-N-ethylcarbazole molecule upon adsorption. This can be explained by the larger amount of electron donation through N in dodecahydrocarbazole that is closer to the surface in this case, which compensates the net charge transfer through the Pd surface to the molecule. The smaller overall charge increase on the molecule is likely the reason for the small change in the C-H bond distances of the 5-membered ring for the adsorbed dodecahydrocarbazole and therefore, a slower dehydrogenation rate observed for dodecahydrocarbazole compared to dodecahydro-N-ethylcarbazole. For dodecahydrofluorene on the other hand, the charge on the replaced C atom increased by about 0.05 a.u.(Table 6.15) with a significant charge decrease of about 0.16 a.u. on the attaching Pd atom. The overall charge on the dodecahydrofluorene molecule was increased by about 0.31 a.u., in127  Table 6.14: Hirshfield charges for dodecahydrocarbazole in vacuum and adsorbed structure. Atom H9a H8aa H4aa H4bb Pd9 a Pd9a a Pd8aa a Pd4aa−4bb a N C9a C8a C4a C4b a Refers  Gas phase (a.u.) 0.0200 0.0236 0.0283 0.0297  Pd(111) (a.u.)  -0.0156 -0.0156 -0.0156 -0.0156 -0.1701 -0.0148 -0.0112 -0.0185 -0.0185  Adsorbed (a.u.) -0.0382 -0.0216 0.0011 -0.0243 0.1134 0.0426 0.0225 0.0359 -0.0918 0.0042 0.0037 -0.0277 -0.0278  Charge change (a.u.) -0.0582 -0.0452 -0.0272 -0.054 0.129 0.0582 0.0381 0.0515 0.0783 0.019 0.0149 -0.0092 -0.0093  to Pd atoms under the corresponding H atom.  dicative of a net flow of charge transfer from the surface to the molecule which is mostly concentrated on the σ bond formed between the C and the Pd and is responsible for the strong adsorption of the molecule to the surface. The very small change in the C-H bond distances of the molecule can explain the very slow H2 recovery from this molecule as observed experimentally (see Chapter 5). The adsorption geometries of tetrahydro-N-ethylcarbazole, tetrahydrocarbazole and tetrahydrofluorene molecules on Pd (111) are shown in Figure 6.15. TetrahydroN-ethylcarbazole adsorption occurred through four carbon atoms in the aromatic 6membered ring, with the 5-membered ring and the aliphatic 6-membered ring tilted away from the surface. The N atom distance from the nearest Pd atom was 3.28 A˚ (Figure 6.15-a). The adsorption energies are summarized in Table 6.12. Lower adsorption energy of tetrahydro-N-ethylcarbazole compared to tetrahydrocarbazole, can be attributed to the effect of the ethyl group. The steric hindrance of the ethyl group does not allow the molecule to interact strongly with the catalyst through the N atom. As can be seen in Figure 6.15-b, tetrahydrocarbazole adsorbs through ˚ from the N six carbon atoms in the aromatic ring, with a shorter distance (3.05 A) atom to the nearest surface atom. The stronger adsorption of tetrahydrocarbazole 128  Table 6.15: Hirshfield charges for dodecahydrofluorene in vacuum and adsorbed structure. Atom H9a H8aa H4aa H4bb Pd9 a Pd9a a Pd8aa a Pd4aa−4bb a C9 b C9a C8a C4a C4b a Refers  Gas phase (a.u.) 0.0273 0.0281 0.0255 0.0263  Pd(111) (a.u.)  -0.0156 -0.0156 -0.0156 -0.0156 -0.0332 -0.0266 -0.0294 -0.0187 -0.0184  Adsorbed (a.u.) -0.0326 -0.0415 -0.0203 0.0125 0.1395 0.0500 0.0563 0.0382 -0.0823 -0.0402 -0.0333 -0.0276 -0.0236  Charge change (a.u.) -0.0599 -0.0696 -0.0458 -0.0138 0.1551 0.0656 0.0719 0.0538 -0.0491 -0.0136 -0.0039 -0.0089 -0.0052  to Pd atoms under the corresponding H atom. for N.  b Replaced  compared to tetrahydro-N-ethylcarbazole is also clear from the higher adsorption energy of tetrahydrocarbazole (Table 6.12). A similar adsorption geometry for tetrahydrocarbazole was obtained by the DFT work of Crawford et al. [53], who reported an adsorption energy of -143.8 kJ/mol. The optimized geometry of the adsorbed tetrahydrofluorene on the other hand, suggested that the molecule adsorbs to the surface through the 6 carbon atoms in the aromatic ring, as well as the N replaced carbon atom with an even shorter distance of 2.89 A˚ than tetrahydrocarbazole to the attaching Pd atom. The adsorption energy of tetrahydrofluorene was found to be -218.6 kJ/mol, significantly higher than that of tetrahydro-N-ethylcarbazole and tetrahydrocarbazole, implying a much stronger adsorption of tetrahydrofluorene to the Pd(111) surface. The very slow dehydrogenation reaction rate observed for dodecahydrofluorene (as discussed in Chapter 5) is very well supported by the strong adsorption of dodecahydrofluorene and tetrahydrofluorene to the catalyst surface.  129  Figure 6.15: Comparison between the adsorbed structures of; (a) tetrahydroN-ethylcarbazole, (b) tetrahydrocarbazole, and (c) tetrahydrofluorene, over Pd(111).  130  6.4  Conclusion  The structure sensitivity of the dodecahydro-N-ethylcarbazole dehydrogenation reaction was confirmed by the DFT results indicating that dodecahydro-N-ethylcarbazole interacts with the Pd(111) surface involving several Pd atoms. The dehydrogenation reaction was found to start by removing of the H atoms from the 5-membered ring to produce octahydro-N-ethylcarbazole, followed by stepwise H removal of one of the aliphatic rings to produce tetrahydro-N-ethylcarbazole. The obtained results supported the observed experimental product distribution for dodecahydro-Nethylcarbazole dehydrogenation, with no production of hexahydro-N-ethylcarbazole and decahydro-Nethylcarbazole. The strong interaction of the N heteroatom in dodecahydrocarbazole was found to be responsible for the slower dehydrogenation reaction rate of doecahydrocarbazole compared to that of dodecahydro-Nethylcarbazole. Dodecahydrofluorene was found to dehydrogenate much slower than dodecahydrocarbazole, due to its very strong adsorption to the catalyst surface. It can be concluded that the N heteroatom favors the dehydrogenation reaction of dodecahydrocarbazole compared to dodecahydrofluorene, although it was shown that dodecahydrocarbazole dehydrogenation product inhibition decreased the reaction rate due to the strong interaction of the N with the surface.  131  Chapter 7  Conclusions and recommendations  132  7.1  Conclusion  Organic heteroaromatic compounds were chosen for the hydrogen storage study of this work, due to their thermal stability and high H2 storage capacity. H2 storage was carried out through hydrogenation reactions over a commercial Ru catalyst and the kinetics of the reactions were quantified. The hydrogenated compounds were tested for H2 recovery through a series of dehydrogenation reactions at different temperatures over Pd/SiO2 catalysts prepared over a range of Pd loadings with different particle sizes and metal dispersions. The kinetics of the dehydrogenation reactions were studied and the obtained experimental results were confirmed by theoretical DFT calculations. In the hydrogenation of N-ethylcarbazole and carbazole, two H2 storage candidates, it was found that N-ethylcarbazole hydrogenates much faster than carbazole. The slower hydrogenation reaction rate obtained for carbazole compared to N-ethylcarbazole was attributed to the strong interaction of the N heteroatom electron pair with the catalyst surface, which poisons the catalyst. In the case of N-ethylcarbazole and its hydrogenated products, the steric hindrance by the ethyl group prevents the N from interacting strongly with the surface and hence, increase the hydrogenation rate. From the obtained results, it is concluded that Nethylcarbazole is a better candidate for H2 storage applications than carbazole. The measured TOFs of the dehydrogenation of dodecahydro-N-ethylcarbazole over Pd catalysts with a range of Pd loadings and particle sizes, suggested that the dehydrogenation reaction of dodecahydro-N-ethylcarbazole is structure sensitive. The results were well described by the model proposed by Murzin [117] supporting the structure sensitivity of the reactions. It was found that modifying the catalyst preparation method by calcining the catalyst in He rather than air had a significant effect on the catalyst activity and selectivity. Complete conversion and 100 % selectivity to the completely dehydrogenated product, N-ethylcarbazole, was achieved over a 4 wt% Pd/SiO2 catalyst with a Pd particle diameter of ∼9 nm. The results were a significant improvement compared to previously reported dehydrogenation results for organic (hetero)aromatic compounds [53, 54, 56, 106].  133  The effect of the N heteroatom on the dehydrogenation reaction rates of the organic (hetero)aromatics was studied by comparing the dehydrogenation reaction rate of dodecahydro-N-ethylcarbazole, dodecahydrocarbazole and dodecahydrofluorene. Dodecahydro-N-ethylcarbazole had the highest reaction TOF compared to the other two compounds. Dodecahydrocarbazole showed a smaller TOF compared to dodecahydro-N-ethylcarbazole most likely due to catalyst poisoning through the N atom. The dehydrogenation TOF for dodecahydrofluorene was found to be even lower than dodecahydrocarbazole. From the measured TOFs of the dehydrogenation reactions it was concluded that the presence of a N heteroatom favors the dehydrogenation of organic polyaromatic compounds, although catalyst poisoning occurs due to the interaction of the N electron pair with the catalyst surface. Therefore, compounds with N heteroatoms are more favored for H2 recovery applications. Furthermore, comparing the dehydrogenation of dodecahydroN-ethylcarbazole and dodecahydrocarbazole, it is concluded that dodecahydro-Nethylcarbazole is a better H2 recovery candidate than dodecahydrocarbazole. Density functional theory calculations were used to model the dehydrogenation reaction of dodecahydro-N-ethylcarbazole, to study the obtained product distribustion observed experimentally, and to investigate the effect of the N heteroatom in the dehydrogenation reactions of organic (hetero)aromatics. The obtained results confirmed the structure sensitivity of the dehydrogenation reactions of dodecahydroN-ethylcarbazole and dodecahydrocarbazole. The DFT results supported the dehydrogenation product distribution observed experimentally for dodecahydro-Nethylcarbazle dehydrogenation. Also, the small dehydrogenation reaction TOFs obtained for dodecahydrocarbazole and dodecahydrofluorene was attributed to the strong adsorption of these two compounds and their dehydrogenated intermediates to the catalyst surface.  7.2  Recommendations  1. Heteroatom effect: In the present work, it was shown that the presence of a N heteroatom de-  134  creases the dehydrogenation barrier and improves the dehydrogenation reaction rate significantly. The effect of other heteroatoms (such as S, P and O) in the chemical structure of the organic polyaromatic compounds on the dehydrogenation reaction rates, should also be studied. Both a theoretical approach by comparing the electronic structure of the heteroatom and its effect on the adsorption structure and energies of the organic compound, and an experimental approach through dehydrogenation reactions, should be used. Furthermore, the effect of having more than one heteroatom in the chemical structure of the organic heteroaromatics on the adsorption energies and dehydrogenation reaction barrier should be investigated. 2. Effect of the chemical structure of the organic heteroaromatics: DFT calculations could be used to study the applicability of organic compounds having more fused rings or a larger branch attached to the heteroatom in their chemical structure for on-board H2 recovery by comparing their adsorption energies to the catalyst surface with the currently studied ones. 3. Use of catalyst promoters: Catalyst promoters such as Ni or K can be used in the Pd catalysts to study their effect on the dehydrogenation reaction TOF and selectivity. 4. Recyclability, cost and system modification: The catalysts used in the present work need to be tested for recyclability in order to be used for transport applications. This can be done by recovering the catalyst after dehydrogenation reaction through filtration and drying of the catalyst. The recovered catalyst can then be tested in a fresh reactant solution in decalin for recyclability. Also, Pd is known to form a Pd−hydride phase by adsorbing the hydrogen present in the system and this might lead to some deactivation of the catalyst after many recycling attempts. The possible deactivation of Pd catalyst can be further studied and other alternatives such as Pt can be investigated. Furthermore, the dehydrogenation system studied in this thesis is a batch reactor in which the solid catalyst is dispersed by continuous mixing. For an on-board dehydrogenation application, a fixed 135  catalytic bed may be required in order to be able to use the catalyst for many cycles. Material cost also needs to be addressed. As was indicated earlier, about 5 kg of H2 is required for a fuel cell vehicle to be able to travel approximately 300 miles. This will mean that ∼1.5 kg catalyst and ∼85 kg of the pure reactant will be required on board the vehicle. Considering the cost of approximately $ 1/kg [146] for the reactant, and $ 10/kg for the catalyst [146] precursor, the total cost of reactant and the catalyst will be approximately $ 100/5 kg H2 ($ 20/kg H2 ). Having the D.O.E. criterion of total system cost of $ 133/kg H2 for year 2010 [21], the presented result in this work are promising in terms of cost of H2 recovery.  136  Bibliography [1] V. Balema. Material Matters, Hydrogen Storage Materials, volume 2, chapter Introduction, page 2. Aldrich, 2007. → pages 2 [2] J.Y. Lee, M. Yoo, K. Cha, T.W. Lim, and T. Hur. Life cycle cost analysis to examine the economical feasibility of hydrogen as an alternative fuel. International Journal of Hydrogen Energy, 34:4243–4255, 2009. → pages [3] T. Pregger, D. Graf, W. Krewitt, C. Sattler, M. Roeb, and S. Moller. Prospects of solar thermal hydrogen production processes. International Journal of Hydrogen Energy, 34:4256–4267, 2009. → pages [4] A.S. Joshi, I. Dincer, and B.V. Reddy. Solar hydrogen production: A comparative performance assessment . International Journal of Hydrogen Energy, (doi:10.1016/j.ijhydene.2010.11.122), 2010. → pages [5] A. Steinfeld. Solar thermochemical production of hydrogen, a review. Solar Energy, 78(5):60315, 2005. → pages [6] R.J. Mantz and H. De Battista. Hydrogen production from idle generation capacity of wind turbines. International Journal of Hydrogen Energy, 33:4291–4300, 2008. → pages [7] M. Santarelli, M. Cali, and S. Macagno. Design and analysis of stand-alone hydrogen energy systems with different renewable sources. International Journal of Hydrogen Energy, 29(15):157186, 2004. → pages 2 [8] J. Chen, Y. Lu, L. Guo, X. Zhang, and P. Xiao. Hydrogen production by biomass gasification in supercritical water using concentrated solar energy: System development and proof of concept. International Journal of Hydrogen Energy, 35:7134–7141, 2010. → pages 2 [9] Y.J. Lu, H. Jin, L.J. Guo, X.M. Zhang, C.Q. Cao, and X. Guo. Hydrogen productionnext term by previous termbiomassnext term gasification in 137  supercritical water with a fluidized bed reactor. International Journal of Hydrogen Energy, 33:6066–6075, 2008. → pages [10] P.C. Hulteberg and H.T. Karlsson. A study of combined biomass gasification and electrolysis for hydrogen production. International Journal of Hydrogen Energy, 34:772–782, 2009. → pages 2 [11] E.D. Naeemi, D. Graham, and B.F. Norton. Material Matters, Hydrogen Storage Materials, volume 2, chapter Introducing HYDRNOL: Organic Liquid Storage for Hydrogen, page 23. Aldrich, 2007. → pages 2, 10, 13 [12] L.M. DAS. On-board hydrogen storage systems for automotive application . International Journal of Hydrogen Energy, 21(9):789–800, 1996. → pages 2, 7 [13] R.H. Crabtree. Hydrogen storage in liquid organic heterocycles. Energy and Environmental Science, 1:134–138, 2008. → pages 2 [14] Material Matters, Hydrogen Storage Materials, volume 2. Aldrich, 2007. → pages 6, 7, 8, 24 [15] G. Pez, A. Scott, A.C. Cooper, and H. Cheng H. Hydrogen Storage by Reversible Hydrogenation of Pi-conjugated Substrates. US patents, (Pub. No. 2857A1), 2005. → pages 2, 12, 13, 14 [16] G. Pez, A. Scott, A.C. Cooper, and H. Cheng H. Hydrogen Storage by Reversible Hydrogenation of Pi-conjugated Substrates. US patents, (7101530B2), 2006. → pages 12, 13, 14, 24, 40, 51, 79, 94 [17] G. Pez, A. Scott, A.C. Cooper, and H. Cheng H. Hydrogen Storage by Reversible Hydrogenation of Pi-conjugated Substrates. US patents, (7429372), 2008. → pages 2, 24 [18] A.C. Cooper, K.M. Campbell, and G.P. Pez. An Integrated Hydrogen Storage and Delivery Approach Using Organic Liquid Phase Carriers. Technical Report 16, Lyon France, 13 - 16 June 2006. → pages 4 [19] Global Greenhouse Warming.com. → pages 4, 5 [20] S. Satyapal, J. Petrovic, C. Read, G. Thomas, and G. Ordaz. The U.S. Department of Energy’s National Hydrogen Storage Project: Progress Towards Meeting Hydrogen-Powered‘ Vehicle Requirements. Catalysis Today, 120:246–256, 2007. → pages 5, 24, 51 138  [21] Targets for Onboard Hydrogen Storage Systems for Light-Duty Vehicles. US Department of Energy, Office of Energy Efficiency and Renewable Energy and The FreedomCAR and Fuel Partnership, September 2009. → pages 5, 136 [22] T. Zhang, L.T. Fan, W.P. Walawender, M. Fan, A.E. Bland, T. Zuo, and D.W. Collins. Hydrogen storage on carbon adsorbents: a review. Environanotechnology, pages 137–163, 2010. → pages 5 [23] C. Read, G. Thomas, G. Ordaz, and S. Satyapal. Material Matters, Hydrogen Storage Materials, volume 2, chapter U.S. department of energys system targets for on-board vehicular hydrogen storage, page 3. Aldrich, 2007. → pages 5, 6, 13 [24] C.L. Aardahl and S.D. Rassat. Overview of systems considerations for on-board chemical hydrogen storage. International Journal of Hydrogen Energy, 34:6676–6683, 2009. → pages 5 [25] A. Leon (Ed.). Hydrogen Technology, Mobile and Portable Applications. Green Energy and Technology. Springer, 2008. → pages 6, 7, 9, 10 [26] M. Diwan, D. Hanna, and A. Varma. Method to release hydrogen from ammonia borane for portable fuel cell applications. International Journal of Hydrogen Energy, 35:577–584, 2010. → pages 6, 8 [27] G.L. Soloveichik. Material Matters, Hydrogen Storage Materials, volume 2, chapter Metal Borohydrides as Hydrogen Storage Materials, page 11. Aldrich, 2007. → pages 6, 9 [28] A.C. Cooper, H. Cheng, and G.P. Pez. Hydrogen Storage by Reversible Hydrogenation of Liquid-phase Hydrogen Carriers. Air Products and Chemicals Inc., 2006. → pages 6 [29] T.Q. Hua, R.K. Ahluwalia, J.K. Peng, M. Kromer, S. Lasher, K. McKenney, K. Law, and J. Sinha. Technical assessment of compressed hydrogen storage tank systems for automotive applications. International Journal of Hydrogen Energy, 36:3037–3049, 2011. → pages 7 [30] R.K. Ahluwalia and J.K. Peng. Dynamics of cryogenic hydrogen storage in insulated pressure vessels for automotive applications. International Journal of Hydrogen Energy, 33:4622–4633, 2008. → pages 7  139  [31] A. Karkamkar, C. Aardahl, and T. Autrey. Material Matters, Hydrogen Storage Materials, volume 2, chapter Recent Developments on Hydrogen Release from Ammonia Borane, page 6. Aldrich, 2007. → pages 7, 8 [32] J. Yang, A. Sudik, D.J. Siegel, Sh. Hirano, A.R. Drews, and C.M. Wolverton. Hydrogen Storage Materials Containing Ammonia Borane. United States Patent Application Publication, (US 2011 0008693A1), January 2011. → pages 8 [33] H.L. Jiang and Q. Xu. Catalytic hydrolysis of ammonia borane for chemical hydrogen storage. Catalysis Today, (doi:10.1016/j.cattod.2010.09.019), 2010. → pages 8 [34] F.H. Stephens, V. Pons, and R.T .Baker. Ammoniaborane: the Hydrogen Source par Excellence? The Royal Society of Chemistry, Dalton Transactions, page 26132626, 2007. → pages 8 [35] Y. Wu and R. M. Mohring. Sodium Borohydride for Hydrogen Storage. Prepr. Pap.-Am. Chem. Soc., Div. Fuel Chem., 48(2):940, 2003. → pages 9 [36] I.E. Ture, F.O. Tabakoglu, and G. Kurtulus. Economical Aspects of Sodium Borohydride for Hydrogen Storage. In 16th World Hydrogen Energy Conference (WHEC), Lyon France, 13-16 June 2006. → pages 9 [37] Y. Kojima and T. Haga. Recycling Process of Sodium Metaborate to Sodium Borohydride. International Journal of Hydrogen Energy, 28:989 993, 2003. → pages 9 [38] S. Li, M. Willis, and P. Jena. Reaction intermediates during the dehydrogenation of metal borohydrides: A cluster perspective. The Journal of Physical Chemistry C, 114(39):1684916854, 2010. → pages 9 [39] J. Lu, Y.J. Choi, Z.Z. Fang, H.Y. Sohn, and E. Ronnebro. Hydrogen Storage Properties of Nanosized MgH2 0.1TiH2 Prepared by Ultrahigh Energy High Pressure Milling. Journal of American Chemical Society, 131:1584315852, 2009. → pages 9, 10 [40] G. Liang. Synthesis and Hydrogen Storage Properties of Mg-Based Alloys. Journal of Alloys and Compounds, 370(123), 2004. → pages 9, 10 [41] J.P. DiPietro and E.G. Skolnik. Analysis of Sodium Hydride-based Hydrogen Storage system. Prepared for US Department of Energy Office of Power Technologies Hydrogen Program, 1999. → pages 10 140  [42] T. Tsuji, Y. Shinya, T. Hiaki, and N. Itoh. Hydrogen solubility in a chemical hydrogen storage medium, aromatic hydrocarbon, cyclic hydrocarbon, and their mixture for fuel cell systems. Fluid Phase Equilibria, 228-229:499–503, 2005. → pages 10 [43] A.C. Cooper, D.E. Fowler, A.R. Scott, A.H. Abdourazak, H. Cheng, F.C. Wilhelm, B.A. Toseland, K.M. Campbell, and G.P. Pez. Hydrogen Storage and Delivery by Reversible Hydrogenation of Liquid-phase Hydrogen Carriers. Air Products and Chemicals Inc., 2005. → pages 10, 11, 12 [44] Y. Cui, S. Kwok, A. Bucholtz, B. Davis, R. Whitney, and P. Jessop. The Effect of Substitution on the Utility of Piperidines and Octahydroindoles for Reversible Hydrogen Storage. New Journal of Chemistry, 32:1027–1037, 2008. → pages 10, 51 [45] C.H. Bartholomew and R.J. Farrauto. Fundamentals of Industrial Catalytic Processes. Wiley, 2 edition, 2006. → pages 11, 12, 155 [46] G.A. Somorjai. Introduction to Surface Chemistry and Catalysis. Wiley, New York, 1994. Ch. 7. → pages 11 [47] Sh.K. Bej, Ch. A. Bennett, and l.T. Thompson. Acid and Base Characteristics of Molybdenum Carbide Catalysts. Applied. → pages 11, 12 [48] J.G. Chen. Carbide and Nitride Overlayers on Early Transition Metal Surfaces: Preparation, Characterization, and Reactivities. Chemical Reviews, 96:1477–1498, 1996. → pages 11, 12 [49] P.N. Rylander. Catalytic Hydrogenation over Platinum Metals. Academic Press NY, 1967. → pages 12 [50] P.N. Rylander. Catalytic Hydrogenation in Organic Synthesis. Academic Press NY, 1979. → pages [51] P.N. Rylander. Catalytic Processes in organic conversions, in Catalysis, Science and Technology. Springer Verlag Berlin, 1983. → pages [52] R. Mezaki and H. Inoue. Rate Equations of Solid-Catalyzed Reactions. University of Tokyo Press, 1991. → pages 12 [53] P. Crawford, R. Burch, C. Hardacre, K.T. Hindle, P. Hu, B. Kalirai, and D.W. Rooney. Understanding the Dehydrogenation Mechanism of Tetrahydrocarbazole over Palladium Using a Combined Experimental and 141  Density Functional Theory Approach. J. Phys. Chem., 111:6434–6439, 2007. → pages 13, 14, 45, 47, 52, 75, 84, 93, 97, 105, 121, 126, 129, 133, 182, 188 [54] K.T. Hindle, R. Burch, P. Crawford, C. Hardacre, P. Hu, B. Kalirai, and D.W. Rooney. Dramatic Liquid-phase Dehydrogenation Rate Enhancements Using Gas-phase Hydrogen Acceptors. Journal of Catalysis, 251:338–344, 2007. → pages 13, 14, 47, 75, 81, 93, 133, 182 [55] E. Clot, O. Eisenstein, and R.H. Crabtree. Computational structure-activity Relationships in H2 Storage: How Placement of N Atoms Affects Release Temperatures in Organic Liquid Storage Materials. Chemical Communications, Cambridge, United Kingdom, 22:2231–2233, 2007. → pages 14, 24 [56] Zh. Wang, I. Tonks, J. Belli, and C.M. Jensen. Dehydrogenation of N-ethylperhydrocarbazole Catalyzed by PCP Pincer Iridium Complexes: Evaluation of a Homogenous Hydrogen Storage System. Organomet. Chem., (doi: 10.1016/ j.jorganchem 2009.03.052), 2009. → pages 14, 52, 75, 133 [57] R.O. Loutfy and E.M. Veksler. Investigation of Hydrogen Storage in Liquid Organichydrides. In International Hydrogen Energy Forum, pages 335–340, Munich, Germany, September 11-15 2000. → pages 14, 44, 63, 76 [58] Sh. Hodoshima, H. Arai, Sh. Takaiwa, and Y. Saito. Catalytic Decalin Dehydrogenation/Naphthalene Hydrogenation Pair as a Hydrogen Source for Fuel-Cell Vehicle. Intl. J. Hydrogen Energy, 28(1255-1262), 2003. → pages 15, 44, 63, 76 [59] R.A. van Santen and M. Neurock. Molecular Heterogeneous Catalysis A Conceptual and Computational Approach. Wiley-VCH, 2006. → pages 15, 16 [60] E. Schrodinger. Ann. Phys., page 79, 1926. → pages 16 [61] W. Koch and M.C. Holthausen. A Chemist’s Guide to Density Functional Theory. Wiley-VCH, 2 edition, 2001. → pages 16, 17, 19 [62] I.N. Levine. Quantum Chemistry. Prentice Hall, New Jersey, 5 edition. → pages 16, 18, 19  142  [63] G.L. Miessler and D.A. Tarr. Inorganic Chemistry. Pearson Prentice Hall, third edition, 2004. → pages 16, 17 [64] A.B. Richon. An Introduction to Molecular Modeling. Mathematec, 83(1), 1994. → pages 17, 18 [65] M. Born and J.R. Oppenheimer. Annals of Physics, 84:457, 1927. → pages 17 [66] Accelrys Software inc., San Diego: Accelrys Software Inc. Materials Studio Manual, release 4.4 edition, 2008. → pages 17 [67] A. Szabo and N.S. Ostlund. Modern Quantum Chemistry: Introduction to Advanced Electronic Structure Theory. MacMillan Publishing Co., New York, 1982. → pages 17 [68] P.R. Westmoreland, P.A. Kollman, A.M. Chaka, P.T. Cummings, K. Morokuma, M. Neurock, E.B. Stechel, and P. Vashishta. Applications of Molecular and Materials Modeling. WTEC panel report Ch. 2: Electronic Structure, Thermochemistry and Kinetics, In Application of Molecular and Material Model, International Technology Research Institute, Baltimore, Maryland, January 2002. → pages 17, 18 [69] M. Springborg, editor. Density-Functional Methods in Chemistry and Materials Science. Research in Theoretical Chemistry. Wiley, 1997. → pages 18 [70] D.S. Sholl and J.A. Steckel. Density Functional Theory , A Practical Introduction. Wiley, 2009. → pages 18, 19 [71] W. Kohn and L.J. Sham. Self-Consistent Equations Including Exchange and Correlation Effects. Physical Review: A, 140:1133–8, 1965. → pages 18 [72] L. Hedin and B.I. Lundqvist. Explicit Local Exchange Correlation Potentials. J. Phys. C, 4:2064–2083, 1971. → pages 19 [73] J.P. Perdew and Y. Wang. Accurate and Simple Analytic Representation of the Electron-Gas Correlation Energy. Phys. Rev., B45:13244–13249, 1992. → pages 19 [74] A.D. Becke. A Multicenter Numerical Integration Scheme for Polyatomic Molecules. J. Chem. Phys., 88:2547–2553, 1988. → pages 19, 98 143  [75] C.T. Lee, W. Yang, and R.G. Parr. Development of the Colle-Salvetti Correlation-Energy Formula into a Functional of the Electron Density. Phys. Rev. B, 37:785–789, 1988. → pages 19, 181, 182 [76] J.P. Perdew, K. Burke, and M. Ernzerhof. M. Phys. Rev. Lett., 77:3865, 1996. → pages 19 [77] R.A. Van Santen and M. Neurock. Concepts in Theoretical Heterogeneous Catalytic Reactivity. Catal. Rev.-Sci. Eng., 37(4):557–698, 1995. → pages 19, 20 [78] R.A. van Santen and M. Neeurock. Theory of Surface Chemical Reactivity. VCH-Wiley, Berlin, 1998. → pages [79] M. Neurock. Catalysis In Application of Molecular and Material Modeling; Applications of Molecular and Materials Modeling, chapter 8, pages 77–105. International Technology Research Institute, January 2002. → pages 19, 20 [80] J.L. Whitten and H. Yang. Surf. Sci. Rep., 24:59–112, 1996. → pages 19 [81] B. Hammer and J.K. Norskov. Adv. in Catal., 45:71–129, 2000. → pages 19 [82] A. Kaldor and D.M. Cox. High Temp. Sci., 27:1, 1990. → pages 20 [83] D.M. Cox, M.R. Zakin, and A. Kaldor. NATO ASI Ser., in Physical Chemistry of Small Clusters, (158):741, 1987. → pages 20 [84] M.C. Payne, M.P. Teter, D.C Allan, T.A. Arias, and J.D. Joannopoulos. Reviews of Modern Physics, 64:1045– 1097, 1992. → pages 20 [85] W. Oelerch, T. Klassen, and R. Bormann. Mg-based Hydrogen Storage Materials with Improved Hydrogen Sorption. Material Transactions, 42(8), 2001. → pages 24 [86] H.Y. Zhao, S.T. Oyama, and E.D. Naeemi. Hydrogen Storage Using Heterocyclic Compounds: The Hydrogenation of 2-methylthiophene. Catalysis Today, 149(1-2):172–184, 2010. → pages 24 [87] U. Eberle, M. Felderhoff, and F. Schuth. Chemical and Physical Solutions for Hydrogen Storage. Angew. Chem. Int. Ed., 48:66086630, 2009. → pages 24, 51  144  [88] M.M.P. Zieverink, M.T. Kreutzer, F. Kapteijn, and J.A. Moulijn. Gas-Liquid Mass Transfer in Benchscale Stirred Tanks: Fluid Properties and Critical Impeller Speed for Gas Induction. Ind. Eng. Chem. Res., 45:45744581, 2006. → pages 27 [89] V. Mazzieri, F. Coloma-Pascual, A. Arcoya, P.C. LArgentiere, and N.S. Fygoli. XPS, FTIR and TPR characterization of Ru/Al2O3 catalysts. Applied Surface Science, 210:222230, 2003. → pages 28 [90] R. Grover. Estimation and use of standard errors of latent class model parameters. Journal of Marketing Research, 24(3):298–304, 1987. → pages 33 [91] L. Hegedus and T. Mathe. Hydrogenation of Pyrrole Derivatives Part V. Poisoning Effect of Nitrogen on Precious Metal on Carbon Catalysts. Applied Catalysis A: General, 226:319322, 2002. → pages 37 [92] A. Vargas, Th. Brgi, and A. Baiker. Adsorption of cinchonidine on platinum: a DFT insight in the mechanism of enantioselective hydrogenation of activated ketones. Journal of Catalysis, 226:6982, 2004. → pages 37, 52, 97 [93] P. Klusona and L. Cerveny. Selective Hydrogenation over Ruthenium Catalysts. Applied Catalysis A: General, 128(1):13–31, 1995. → pages 40 [94] P. Chou and M.A. Vannice. Benzene Hydrogenation over Supported and Unsupported Palladium : II. Reaction Model. Journal of Catalysis, 107:140–153, 1987. → pages 40 [95] A. Arnold, F. Dobert, and J. Gaube. Handbook of Heterogeneous Catalysis, volume 5 - p. 2165. VCH, Germany, 1997. → pages 40 [96] K. M. Eblagon, D. Rentsch, O. Friedrichs, A. Remhof, A. Zuettel, A. J. Ramirez-Cuesta, and Sh. C. Tsang. Hydrogenation of 9-Ethylcarbazole as a Prototype of a Liquid Hydrogen Carrier. Internayional Journal of Hydrogen Energy, 35:11609–11621, 2010. → pages 40, 101, 104, 105 [97] J.R. Song, L.X. Wen, L. Shao, and J.F. Chen. Appl. Surf. Sci., (253):26782684, 2006. → pages 40, 79 [98] R. F. HICKS, Q. J. YEN, and A. T. BELL. Effects of Metal-Support Interactions on the Chemisorption of H2 and CO on Pd/SiO2 and Pd/La203. Journal of Catalysis, 89:498–510, 1984. → pages 41, 60, 74, 80 145  [99] A. Benedetti, G. Fagherazzi, F. Pinna, G. Rampazzo, M. Selva, and G. Strukul. The Influence of a Second Metal Component (Cu, Sn, Fe) on Pd/SiO2 Activity in the Hydrogenation of 2,4-Dinitrotoluene. Catalysis Letters, 10:215–224, 1991. → pages 43, 52, 57 [100] G. Neri, M. Musolino, C. Milone, D. Pietropaolo, and S. Galvagno. Particle Size Effect in the Catalytic Hydrogenation of 2,4-Dinitrotoluene over Pd/C Catalysts. Applied Catalysis A: General, 208:307–316, 2001. → pages 43, 52, 57, 82, 92 [101] OR. Stull, Jr. EF. Westrum, and GC. Sinke. pages 15–20, 1969. → pages 44 [102] E. Newson, TB. Truong, P. Hottinger, FV. Roth, and THH. Schucan. In Proc. of the 12th World Hydrog. Eng. Conf., Buenos Aires, Argentina, pages 935–42, 1998. → pages 44 [103] G. Cacciola, N. Giordano, and G. Restuccia. Cyclohexane as a Liquid Phase Carrier in Hydrogen Storage and Transport. International Journal of Hydrogen Energy, 9:411–419, 1984. → pages 51 [104] D. Klvana, A. Touzani, J. Chaouki, and G. Belanger. 1991. International Journal of Hydrogen Energy, 16:55–60, 1991. → pages 51 [105] A. Moores, M. Poyatos, Y. Luo, and R. Crabtree. Catalysed Low Temperature H2 Release from Nitrogen Heterocycles. New Journal of Chemistry, 30:1675–1678, 2006. → pages 51 [106] F. Sotoodeh, L. Zhao, and K.J. Smith. Kinetics of H2 Recovery from Dodecahydro-N-Ethylcarbazole over a Supported Pd Catalyst. Applied Catalysis A: General, 362:155–162, 2009. → pages 51, 53, 60, 62, 63, 66, 67, 68, 69, 70, 71, 72, 74, 75, 79, 82, 84, 92, 93, 103, 104, 121, 133, 182 [107] F. Sotoodeh and K.J. Smith. Kinetics of Hydrogen Uptake and Release from Heteroaromatic Compounds for Hydrogen Storage. Industrial and Engineering Chemistry Research, 49(1018-1026), 2010. → pages 51, 60, 62, 69, 72, 74, 82, 84, 86, 92, 93, 101, 103, 104, 184 [108] D. Suh, T. Park, and S. Ihm. Characteristics of Carbon-Supported Palladium Catalysts for Liquid-Phase Hydrogenation of Nitroaromatics. Industrial and Engineering Chemistry Research, 31:1849, 1992. → pages 52  146  [109] J.N. Kuhn and W. Huang and Ch.K. Tsung and Y. Zhang and G.A. Somorjai. Structure Sensitivity of Carbon-Nitrogen Ring Opening: Impact of Platinum Particle Size from below 1 to 5 nm upon Pyrrole Hydrogenation Product Selectivity over Monodisperse Platinum Nanoparticles Loaded onto Mesoporous Silica. Journal of American Chemical Society, 130:14026–14027, 2008. → pages 52 [110] Y. Lu, H. Zhang, Y. Xu, B. Song, H. Li, S.N Bao, and P. He. Adsorption Structures of Tetracene on the Ru(1 0 1 0) Surface. Applied Surface Science, 253:2025–2030, 2006. → pages 52, 97 [111] J.R. Regalbuto. Catalyst Preparation: Science and Engineering. Taylor and Francis group, LLC, 2007. → pages 53, 79 [112] N. Babu, N. Lingaiah, R. Gopinath, P. Reddy, and P. Prasad. Characterization and Reactivity of Alumina-Supported Pd Catalysts for the Room-Temperature Hydrodechlorination of Chlorobenzene. Journal of Physicaal Chemistry C, 111:6447–6453, 2007. → pages 55, 57, 73, 82 [113] C. Amairia, S. Fessi, A. Ghorbel, and A. Rives. Methane Oxidation Behaviour over SolGel Derived Pd/Al2O3-ZrO2 Materials: Influence of the Zirconium Precursor, journal = Journal of Molecular Catalysis A Chemical, year = 2010, volume = 332, pages = 25-31,. → pages 55, 82 [114] M. Gurrath, T. Kuretzky, H. Boehm, L. Okhlopkova, A. Lisitsyn, and V. Likholobov. Palladium Catalysts on Activated Carbon Supports: Influence of Reduction Temperature, Origin of the Support and Pretreatments of the Carbon Surface. Carbon, 38:1241–1255, 2000. → pages 92 [115] J. Bakker, A. Geert van der Neut, M. Kreutzer, J. Moulijn, and F. Kapteijn. Catalyst Performance Changes Induced by Palladium Phase Transformation in the Hydrogenation of Benzonitrile. Journal of Catalysis, 274:176–191, 2010. → pages 57, 92 [116] M. Myyrylainen and T.T. Rantala. Kinetic Monte Carlo Modeling of CO Desorption and Adsorption on Pd(1 1 0) Surface. Catal.sis Today, 100:413417, 2005. → pages 60 [117] D.Y. Murzin. Kinetic Analysis of Cluster Size Dependent Activity and Selectivity. Journal of Catalysis, 276:85–91, 2010. → pages 69, 76, 77, 133 147  [118] R. Iler. The Chemistry of Silica. John Wiley and Sons, New York, 1979. → pages 72 [119] Z. Li, Q. Zhao, C. Liu, and Y. Wang. → pages 72 [120] C. Contescu, D. Macovei, C. Craiu, C. Teodorescu, and J.A. Schwarz. Thermal Induced Evolution of Chlorine-Containing Precursors in Impregnated Pd/Al203 Catalysts. Langmuir, 11:2031, 1995. → pages 72, 73, 74, 75 [121] W. Zou and R. Gonzalez. The preparation of silica supported Pd catalysts: the effect of pretreatment variables on particle size. Catalysis Letters, 12:73–86, 1992. → pages 72, 74, 92 [122] T. Mang, H. Knozinger, J. Hightower, W. Delgass, E. Iglesia, and A. Bell. → pages 72 [123] E. Vilarrasa-Garcia, A. Infantes-Molina, R. Moreno-Tost, E. Rodriguez-Castellon, A. Jimenez-Lopez, Jr. C.L Cavalcante, and D.C.S. Azevedo. Thiophene Adsorption on Microporous Activated Carbons Impregnated with PdCl2. Energy Fuels, 24:3436–3442, 2010. → pages 73 [124] F.L. Normand, J. Barrault, R. Breault, L. Hilaire, and A. Kiennemann. Catalysis with Palladium Deposited on Rare Earth Oxides: Influence of the Support on Reforming and Syngas Activity and Selectivity. The Journal of Physical Chemistry, 95:257, 1991. → pages 73 [125] J. Murthy, S.Ch. Shekar, A. Padmasri, A. Venugopal, V. Kumar, B.M. Nagaraja, V. Shashikala, B.D. Raju, P. Kanta Rao, and K.S. Rama Rao. Promotional Effect of Magnesia Addition to Active Carbon Supported Pd Catalyst on the Characteristics and Hydrodechlorination Activity of CCl2F2. Catalysis Communications, 5:161–167, 2004. → pages 74 [126] W. Jozwiak and T. Maniecki. Influence of Atmosphere Kind on Temperature Programmed Decomposition of Noble Metal Chlorides. Thermochimica Acta, 435:151–161, 2005. → pages 74 [127] A. Binder, M. Seipenbusch, M. Muhler, and G. Kasper. Kinetics and Particle Size Effects in Ethene Hydrogenation over Supported Palladium Catalysts at Atmospheric Pressure. Journal of Catalysis, 268:150–155, 2009. → pages 76  148  [128] G.L. Bezemer, J.H. Bitter, H.P.C.E. Kuipers, H. Oosterbeek, J.E. Holewijn, X.Xu, F. Kapteijn, A.J. van Dillen, and K.P. de Jong. Cobalt Particle Size Effects in the Fischer-Tropsch Reaction Studied with Carbon Nanofiber Supported Catalysts. Journal of American Chemical Society, 128:3956, 2006. → pages 76 [129] D.Y. Murzin. Size-Dependent Heterogeneous Catalytic Kinetics. Journal of Molecular Catalysis A Chemical, 315:226–230, 2010. → pages 76, 77 [130] A. Cooper, A. Scott, D. Fowler, J. Cunningham, M.Ford, F. Wilhelm, V. Monk, H. Cheng, and G. Pez. Hydrogen Storage by Reversible Hydrogenation of Liquid-Phase Hydrogen Carriers. FY 2008 Annual Progress Report, DOE Hydrogen Program. Air Products and Chemicals, Inc., pages 602–606, 2008. → pages 88, 90, 94 [131] F. Sotoodeh and K.J. Smith. Structure sensitivity of dodecahydro-n-ethylcarbazole dehydrogenation over pd catalysts. Journal of Catalysis, Accepted on Dec. 30th 2010. → pages 91, 92 [132] H. Ohtani, M.A. Van Hove, and G.A. Somorjai. Molecular Structure of Benzene Coadsorbed with CO on Pd(ll1): A Dynamical Low-Energy Electron Diffraction Analysis. Journal of Physical Chemistry, 92:3974, 1988. → pages 97 [133] A. Barbieri, M.A. Van Hove, and G.A. Somorjai. Benzene coadsorbed with CO on Pd(111) and Rh(111): detailed molecular distortions and induced substrate relaxations. Surface Science, 306:261–268, 1994. → pages [134] H. Hoffmann, F. Zaera, R.M. Ormerod, R.M Lambert, L.P. Wang, and W.T. Tysoe. Discovery of a Tilted Form of Benzene Chemisorbed on Pd(111): As NEXAFS and Photoemission Investigation. Surface Science, 232:259–265, 1990. → pages [135] A.F. Lee, K. Wilson, R.M. Lambert, A. Goldoni, A. Baraldi, and G. Paolucci. On the Coverage-Dependent Adsorption Geometry of Benzene Adsorbed on Pd111: A Study by Fast XPS Photoemission and NEXAFS. Journal of Physical Chemistry B, 104:11729, 2000. → pages [136] W.T. Tysoe, R.M. Ormerod, R.M. Lambert, G.Zgrablich, and A. Ramirez Cuesta. Overlayer Structure and Kinetic Behavior of Benzene on Paladium. Journal of Physical Chemistry B, 97:3365, 1993. → pages  149  [137] H. Orita and N. Itoh. Simulation of Phenol Formation from Benzene with a Pd Membrane Reactor: Ab Initio Periodic Density Functional Study. Applied Catalysis A: General, 258(1):17–23, 2004. → pages 97 [138] M. Ohara, Y. Kim, S. Yanagisawa, Y. Morikawa, and M. Kawai. Role of Molecular Orbitals Near the Fermi Level in the Excitation of Vibrational Modes of a Single Molecule at a Scanning Tunneling Microscope Junction. Physical Review Letters, 100:136104, 2008. → pages 98 [139] W. Dou, D. Guan, F. Song, N. Li, H. Zhang, H. Li, P. He, H. Chen, S. Bao, and P. Hofmann. Valence Electronic Properties of n-Channel Organic Materials Based on Fluorinated Derivatives of Perylene Diimides. Journal of Chemical Physics, 244711:128, 2008. → pages 98 [140] B. Hammer, L.B. Hansen, and J.K. Norskov. Improved Adsorption Energetics within Density-Functional Theory Using Revised Perdew-Burke-Ernzerhof Functionals. Physical Review B, 59:74137421, 1999. → pages 98 [141] J.P. Perdew, K. Burke, and M. Ernzerhof. Generalized Gradient Approximation Made Simple. Physical Review Letters, 77:38653868, 1996. → pages 98 [142] J.L. Garnett and W.A. Sollich-Baumgartner. Pi Complex Adsorption in Hydrogen Exchange on Group VIII Transition Metal Catalysts. Advances in Catalysis, 16:95–121, 1996. → pages 104 [143] M. Saeys, M. Reyniers, M. Neurock, and G. Marin. Adsorption of Cyclohexadiene, Cyclohexene and Cyclohexane on Pt(111). Surface Science, 600:3121–3134, 2006. → pages 106, 108, 117, 118, 121 [144] M. Saeys, S.G. Marie-Francoise, M. Reynier, M. Neurock, and G.B. Marin. Density Functional Study of Benzene Adsorption on Pt(111), Density Functional Study of Benzene Adsorption on Pt(111). Journal of Physical Chemistry B, 106:7489–7498, 2002. → pages 107, 110, 118 [145] M. Saeys, M. Neurock, and G.B. Marin. Density Functional Theory Analysis of Benzene (De)hydrogenation on Pt(111): Addition and Removal of the First Two H-atoms. Journal Physical Chemistry, 107:3844–3855, 2003. → pages 109 [146] Alibaba.com, 1999. Founded in Hangzhou, China. → pages 136  150  [147] S. Brunauer, P.H. Emmett, and E. Teller. Am. Chem. Soc., 1938. → pages 155 [148] Micromeritics Instrument Corporation. AutoChem II 2920 Automated Catalyst Characterization System Operators Manual V3.05, 2006. → pages 156 [149] H.S. Fogler. Elements of Chemical Reaction Engineering. Printice Hall PTR, 4 edition, 2006. → pages 173, 174 [150] F. Weiyang and H.J. Bart. Prediction of Diffusivities in Liquids. Chem. Eng. Technol, 8(21):659, 1998. → pages 173 [151] P. Englezos and N. Kalogerakis. Applied Parameter Estimation for Chemical Engineers, chapter 8. Marcel-Dekker, New York, 2001. → pages 175 [152] S.T. Oyama. Preparation and Catalytic Properties of Transition Metal Carbides and Nitrides. Catalysis Today, 15:179–200, 1992. → pages 181, 182 [153] M. Espinnoza, J. Cruz-Reyes, M. Del Valle-Granados, E. Flores-Aquino, M. Avalos-Borja, and S. Fuentes-Moyado. Catalysis Letters, 120:137–142, 2008. → pages 181 [154] S.B. Derouane-Abd Hamid, J.R. Anderson, I. Schmidt, C. Bouchy, C.J.H. Jacobsen, and E. G. Derouane. Catalysis Today, 63:461–469, 2000. → pages 181 [155] C. Bouhy, S.B. Derouane-Abd Hamid, and E.G. Derouane. Chemical Communications, 2:125–126, 2000. → pages [156] S.Z. Li, J.S. Lee, T. Hyeon, and K.S. Suslick. Applied Catalysis A: general, 184:1–9, 1999. → pages [157] M. Nagai, T. Miyao, and T. Tuboi. Catalysis Letters, 18:9–14, 1993. → pages 181 [158] G. Djega-Mariadassou, M. Boudart, G. Bugli, and C. Sayag. Catalysis Letters, 31:411–420, 1995. → pages 181 [159] M. Xiang, D. Li, H. Qi, W. Li, B. Zhong, and Y. Sun. Fuel, 86(9):1128–1303, 2007. → pages 181, 182, 191  151  [160] V. Keller, P. Weher, F. Garin, R. Ducros, and G. Maire. Journal of Catalysis, 153(1):9–16, 1995. → pages 181 [161] M. Nagai, A.M. Zahidul, and K. Matsuda. Applied Catalysis A: general, 3113(2):137–145, 2006. → pages 181, 191 [162] A. Szechenyi and F. Solymosi. Journal of Physical Chemistry C., 111(26):9509–9515, 2007. → pages 181 [163] T.C. Xiao, A.P.E. York, V.C. Williams, H. Al-Megren, A. Hanif, Z-Y Zhou, and M.L.H. Green. Chemical Materials, 12(12):3896–3905, 2000. → pages 181, 182 [164] X. Li, D. Ma, L. Chen, and X. Bao. Catalysis Letters, 116(1-2):63–69, 2007. → pages 181 [165] J.M. Giraudon, L. Leclercq, G. Leclercq, A. Lofberg, and A. Frennet. Journal of Material Sciences, 28:2449–2454, 1993. → pages 181 [166] D. Zeng and M. Hampden-Smith. Journal of Chemical Materials, 5:681–689, 1993. → pages 181 [167] L.G. Rosa, J.C. Fernandes, and P.M. Amaral. International Journal of Refractory Metals and Hard Materials, 17:351–356, 1999. → pages 182 [168] P.M. Patterson, T.K. Das, and B.H. Davis. Applied Catalysis A: general, 251:449–455, 2003. → pages 182 [169] M.K. Neylon, S. Choi, H. Kwon, K.E. Curry, and L.T. Thompson. Applied Catalysis A: general, 183:253–263, 1999. → pages [170] J.B. Claridge, A.P.E. York, A.J. Brungs, and M.L.H. Green. Chemical Materials, 12:132–142, 2000. → pages 182 [171] M. Ogura, S. Shinomiya, J. Tateno, Y. Nara, M. Nomura, E. Kikuchi, and M. Matsukata. Applied Catalysis A: general, 219:33–43, 2001. → pages 182, 185 [172] T. Suzuki and T. Okuhara. Microporous Mesoporous Materials, 43:83–89, 2001. → pages 182, 185 [173] S.F. Zaman and K.J. Smith. Synthesis gas conversion over MoP catalysts. Catalysis Communications, 10:468–471, 2009. → pages 185  152  Appendices  153  Appendix A  Catalyst preparation: calculations of required chemicals Preparation of Pd/silica: Sample calculations for the required chemicals for preparing 5 wt% Pd on silica are as follows: 1. Required amount of Pd/silica catalyst =10 g 2. Amount of Pd in Pd/silica catalyst =  5 g Pd (10 g Pd/silica) = 0.5 g 100 g Pd/silica  (A.1)  3. Required amount of PdCl2 =  177.33 g/mol PdCl2 (0.5 g Pd) = 0.83 g 106.4 g Pd  (A.2)  4. Required amount of silica = (10 g Pd/silica) − (0.5 g Pd) = 9.5 g  (A.3)  5. Required amount of water solution = (1.15  ml )(9.5 g silica) = 10.925 ml g silica 154  (A.4)  Appendix B  Catalyst characterization techniques B.1 Total BET surface area Brunauer, Emmett and Teller in 1938 proposed a model that described the multilayer physisorption and calculated the monolayer coverage of the adsorbate [45, 147]. The BET equation is as follow: P/P0 1 C−1 P P = V C (P )+ V C V (1 − P0 ) m 0 m  (B.1)  where C is a constant, V is the total volume adsorbed (STP) at pressure P, Vm is the volume adsorbed (STP) at monolayer coverage, and Po is the saturation vapor pressure of the adsorbate gas (or vapor). A plot of the left-hand side in Equation B.1 vs. P/Po should give a linear plot with a slope of (C-1)/Vm C and an intercept of 1/Vm C, from which Vm and C can be determined and surface area is then calculated. The preferred range of P/Po for best results is 0.05 to 0.4.  B.2 Temperature-programmed reduction (TPR) Temperature-programmed reduction (TPR) of the calcined Pd/SiO2 catalyst precursors was performed using a Micromeritics Autochem II 2920 analyser fitted 155  with a thermal conductivity detector (TCD). About 0.2 g of each sample was loaded into a u-tube reactor and flushed for 30 min in a 50 cm3 (STP)/min flow of 10 % H2 in Ar at room temperature. The sample was then heated to 673 K at a rate of 10 K/min with the final temperature held for 1 h.  B.3 Temperature-programmed desorption (TPD) Temperature-programmed desorption (TPD) of CO from the Pd catalysts was also performed using the Micromeritics Autochem II 2920 analyser. About 0.2 g of the dried catalyst was loaded into the reactor and calcined at 773 K for 6 h in a flow of Ar at 30 cm3 (STP)/min, followed by TPR at 10 K/min to a final temperature of 673 K in a 30 cm3 (STP)/min flow of 10 % H2 in Ar, holding the final temperature for 1 h. The catalyst was then flushed in a 30 cm3 (STP)/min flow of Ar for 1 h and cooled to room temperature, before being exposed to a 50 cm3 (STP)/min flow of 10 % CO in He at 313 K for 1 h. The catalyst was subsequently flushed in 50 cm3 (STP)/min He for 2 h. The temperature-programmed desorption was then started in the same He flow by heating the sample to 773 K at a rate of 10 K/min and holding the final temperature for 1 h.  B.3.1 Injection loop calibration Injection loop must be calibrated before its first use to determine its precise volume under local conditions [148]. Calibration consists of: • Determining the average area of a the peaks generated by injections of a known volume of gas through the analyzer septum using a glass syringe. • Determining the average area of the peaks generated by injections of the same gas using the analyzers internal loop. • Calculating the volume of the loop by comparing the average peak area generated by the loop injections with that generated by the syringe injections  156  Volume at STP Volume at standard reduction of gas data to Standard Temperature Pressure (STP) conditions (273.13 K and 760 mmHg) is calculated as follows, VST P = Va (  Pa 273.15 )×( ) 273.15 + Ta 760 mmHg  (B.2)  where VST P =Volume at STP, cm3 Va =Volume at ambient temperature and pressure, cm3 Ta =Ambient temperature, ◦ C Pa =Ambient pressure, mmHg Injection loop calibration These equations are used during calibration of a dose loop, VSST P = VS (  273.15 Pa )×( ) 273.15 + Ta 760 mmHg  and Vl = (  Al ×VSST P ) AS  where VSST P =Volume of syringe at STP, cm3 VS =Physical volume of syringe, cm3 Ta =Ambient temperature, ◦ C Pa =Ambient pressure, mmHg Vl =Effective loop volume (at loop temperature), cm3 Al =Average peak area of loop injections AS =Average peak area of syringe injections  157  (B.3)  (B.4)  B.4 High resolution transmission electron microscopy (HRTEM) A FEI Tecnai G2 200 keV with a LaB6 filament capable of 1.4 A˚ point-to-point resolution was used for high resolution transmission electron microscopy (HRTEM). Pd/SiO2 catalyst samples with variable Pd loadings were ground for 20 min to a fine powder using an agate mortar and pestle. 1 mg of the sample was then dispersed in 10 ml ethanol ultrasonically for 5 min. A droplet of the suspension was then placed on a 200 mesh copper grid coated with formvar carbon and left to dry for 10 min before analysis.  158  Appendix C  Mass spectrometery C.1 Mass flow controllers (MFCs) calibration Mass spectrometer (MS) was used to continuously monitor the dehydrogenation reactor effluent. The MS was calibrated using a constant flow of He as a carrier gas, and stepwise increasing of known amount of H2 gas introduced into the He flow. He and H2 flow were controlled using a He 200 SCCM and a H2 20 SCCM mass flow controller (MFC), respectively, which were calibrated using a bubble flow meter prior to the MS calibration. Table C.1 and C.2 summarize the He and H2 MFC calibration results. Table C.1: He 200 SCCM MFC calibration using a bubble flow meter and stopwatch for 79.9 % of the total flow. Volume (ml) 25 25 25 25 25 25 25 25 25  Time (min) 0.1485 0.1463 0.1463 0.1475 0.1452 0.1473 0.1442 0.1433 0.1458 Average flow (ml/min)  159  flow (ml/min) 168.3502 170.8428 170.8428 169.4915 172.2158 169.6833 173.4104 174.4186 171.4286 171.1871  Table C.2: H2 20 SCCM MFC calibration using a bubble flow meter and stopwatch. Controller knob (%) 10 15 20 25 30 35 40 50  volume (ml) 5 10 10 15 15 25 25 25  Time (min) 1.87 2.51 1.90 2.30 1.91 2.72 2.40 1.93  flow (ml/min) 2.67 3.98 5.25 6.52 7.84 9.18 10.41 12.95  Figure C.1 shows the calibration curve obtained by linear fit of the obtained calibration data summarized in Table C.2.  Figure C.1: H2 20 SCCM MFC calibration curve obtained from the data shown in Table C.2.  C.2 Mass spectrometer (MS) calibration for H2 The He flow was kept constant at 171.19 ml/min as a carrier gas during the MS calibration for H2 . H2 flow was increased stepwise and the pressure of He and H2 160  was continuously monitored. Table C.3 summarizes the measured partial pressures as well as the H2 to He partial pressure ratio used to find the H2 gas calibration factor. Table C.3: MS calibration data for H2 using a constant He flow of 171.19 ml/min as the carrier gas. time (min) 0.00 5.00 41.00 93.00 140.00 168.00 198.00 232.00 267.00 303.00 332.00 365.02 407.02 436.02 467.02 a ratio  He (torr) 6.109×10−5 6.111×10−5 6.061×10−5 6.112×10−5 6.178×10−5 6.195×10−5 6.255×10−5 6.283×10−5 6.310×10−5 6.375×10−5 6.411×10−5 6.401×10−5 6.460×10−5 6.483×10−5 6.544×10−5  H2 (torr) 1.112×10−6 1.872×10−6 2.882×10−6 3.965×10−6 6.212×10−6 8.285×10−6 1.143×10−6 1.344×10−6 1.659×10−6 1.891×10−6 2.199×10−6 2.668×10−6 3.183×10−6 3.692×10−6 5.080×10−6  P(H2 )/P(He) 0.01820 0.03063 0.04756 0.06487 0.10055 0.13373 0.18269 0.21390 0.26291 0.29667 0.34299 0.41684 0.49273 0.56943 0.77637  Y(H2 )/Y(He)a 0.00000 0.00152 0.00304 0.00457 0.00761 0.01066 0.01522 0.01827 0.02283 0.02588 0.03045 0.03806 0.04567 0.05328 0.07612  of the moles of H2 to the moles of He.  The calibration factor, M, is obtained from the slope of the linear fit plot of P(H2 )/P(He) vs. Y(H2 )/Y(He), M.  Y (H2 ) F(H2 ) P(H2 ) = = Y (He) F(He) P(He)  (C.1)  where F(H2 ) and F(He) are the molar flow of H2 and He, respectively. The obtained result is shown in Figure C.2.  161  Figure C.2: MS calibration curve for H2 .  162  Appendix D  Gas chromatography A Shimadzu QP-2010S gas chromatograph with a mass spectrometer detector (GC/MS) was used to analyze the obtained liquid samples from hydrogenation and dehydrogenation reactions using a Restek RTX5 30M x 0.25 mm capillary column.  D.1  Temperature program  The temperature program used for the sample analysis is summarized in Table D.1. Table D.1: Tenperature program used in the Shimadzu QP-2010S GC/MS. Parameter Initial column temperature Initial column temperature hold time Temperature ramp Final column temperature Final column temperature hold time Total analysis time Injection volume  D.2  Unit K min K/min K min min µl  Value 393 5 10 543 2 22 1  Calibration  If the hydrogenation reactant concentration in feed is CA0 , and the obtained area from GC/MS analysis is A0 , then a response factor (R.F.) is defined as:  163  R.F. =  A0 CA0  (D.1)  Using the obtained response factor, the unknown concentration of the hydrogenation/dehydrogenation products can be calculated by dividing the obtained area for each compound by the response factor, as follows: Ccompound =  Acompound R.F.  (D.2)  where Ccompound is the concentration of compound of interest and Acompound the obtained area from GC/MS.  164  Appendix E  Sample calculations E.1 N-ethylcarbazole hydrogenation reaction at 413 K The following section shows sample calculations for hydrogenation of a 6 wt% solution of N-thylcarbazole in decalin at 413 K and 6 MPa over a 5 wt% Ru/alumina catalyst. The commercial Ru/alumina catalyst was received in a pre-reduced state and was pre-treated before the hydrogenation reaction.  E.1.1 Experiment procedure A solution of 6 g N-ethylcarbazole in 100 ml decalin was prepared and loaded to the reactor. 1 g of commercially reduced catalyst was added to the solution. The reactor was sealed properly and purged with N2 for 30 min followed by hydrogen for another 30 min, then heated up to 413 K while purging with hydrogen with very small flow rate. After the temperature reached 413 K and remained constant, the reactor outlet was closed and reactor was pressurized with H2 . The reaction started immediately. The hydrogen pressure was monitored and liquid samples were taken for rate analysis. Table E.1 shows the reactant and material information used for the hydrogenation reaction, as well as the reaction conditions. Tables E.2 and E.3 show the obtained results for H2 consumption and the liquid sample analysis. The final pressure took into account the pressure loos occurred in each sampling step. The total pressure loss was added to the obtained final pressure 165  Table E.1: Experiment information Reactant loaded, g Solvent loaded, ml Catalyst loaded, g Reactant loaded, mol Molarity in solvent, mol/ml Reaction pressure, MPa Reaction Temperature, K Stirrer speed, rpm Reaction time, h  6 100 1 0.0298 0.0003 6 413 600 24  of the reactor. Table E.2: Hydrogen consumption Initial pressure (at 413 K), psi Final pressure (at 413 K), psi Initial conc.a , mol/ml Initial conc., mol Final conc.a , mol/ml Final conc., mol Stoichiometric consumption of H2 for complete hydrogenation, mole Initial H2 , mol Final H2 , mol Total H2 consumed, mol Expected reactor pressure after complete hydrogenation, psi a Initial  856 373+33 (pressure loss due to sampling) = 406 0.000298 0.0298 0.000005254 0.0005254 6×0.0298 = 0.1788 846/145.0377×200/8.314/413 = 0.34 406/145.0377×212/8.314413 = 0.1728 0.34-0.1728 = 0.1672 (0.34-0.1788)×8.314×413/212 ×145.0377 = 378.67  and final concentrations are based on liquid samples analysis.  The obtained rate constant is based on a linear fit to the experimental concentration data of the reactant measured during the experiment and analyzed by the GC, according to the following rate equation:  − rA = dCA /dt = k CA ⇒ Ln(CA /CA0 ) = −k t  (E.1)  where CA0 and CA are the initial reactant concentration and concentration at time 166  Table E.3: N-ethylcarbazole conversion obtained by GC analysis of the samples Time (min) 0.00 17.17 43.17 80.17 123.17 175.17 442.17 534.17 630.17 718.17 1223.17 1452.17  N-ethylcarbazole (mol/ml) 0.000298049 0.000270185 0.000119580 0.000032065 0.000023546 0.000022684 0.000022355 0.000018888 0.000009432 0.000011548 0.000013130 0.000005254  X (%) 0.00 9.35 59.88 89.24 92.10 92.39 92.50 93.66 96.84 96.13 95.59 98.24  Ln(CA /CA0 ) 0.000163931 -0.097988121 -0.913106707 -2.229341315 -2.538119387 -2.575447246 -2.590031454 -2.758704721 -3.453176698 -3.250731977 -3.122338636 -4.038280266  Figure E.1: First-order rate constant calculations for the hydrogenation reaction at 413 K. t, respectively. K is the first-order rate constant obtained from the slope of the plot Ln(CA /CA0 ) vs. reaction time, t. Figure E.1 shows the rate constant calculations obtained from the liquid sample analysis. The reaction rate constant was found to be 0.0181 min−1 . The reactant initial concentration was 0.0002980 mol.ml−1 . 167  Therefore we have:  k CA0 = (0.0181 min−1 )(0.0002980 mol min−1 ) = 5.39 × 10−6 mol. ml −1 . min−1  (E.2)  E.2 Dodecahydro-N-ethylcarbazole dehydrogenation reaction The following section shows the details of calculations for the dehydrogenation reaction of dodecahydro-N-ethylcarbazole over a 4 wt% Pd/SiO2 catalyst at 423 K and 101 kPa.  E.2.1 Experiment procedure 1. Loading a 50 ml three-neck glass flask reactor with 10 ml of the hydrogenated product and diluting with 20 ml of decalin. 2. Purging the system with Helium to remove air. 3. Heating the reactor using a heating mantle and controller, and recording the temperature continuously. 4. Monitoring the outlet gas of the reactor continuously with mass spectrometer. 5. Adding 0.07 g of 4 wt% Pd/silica (prepared in the lab, pre-reduced in the quartz U-tube using the muffle furnace) to reactor by the catalyst injection glass tube after temperature reached the set-point. 6. Taking liquid samples. Table E.4 summarizes the reactant and products concentrations obtained from liquid sample analysis using the Shimadzu GCMS. Table E.5 summarizes the conversion of the reactant, selectivity to N-ethylcarbazole, as well as the gas phase Hydrogen production. 168  Figure E.2: H2 profile obtained from the gas phase analysis using MS. Table E.4: Concentrations of reactants and products in mol/ml. Sample 1 2 3 4 5 6 7 8 9 10  Time, min 0.00 4.67 9.33 14.50 22.33 29.67 37.17 50.50 81.00 119.50  N-ethylcarbazole 0.0000 3.3589 9.5660×10−6 2.9796×10−5 4.1174×10−5 4.7330×10−5 5.8841×10−5 8.1411×10−5 9.5924×10−5 9.9350×10−5  Tetrahydro0.0000 2.2840×10−5 4.6634×10−5 5.0183×10−5 5.2938×10−5 5.0594×10−5 4.0119×10−5 1.7938×10−5 3.4258×10−6 0.0000  Octahydro0.0000 2.6131×10−5 2.5082×10−5 1.4564×10−5 4.5891×10−6 1.4250×10−5 3.8946×10−5 0.0000 0.0000 0.0000  Dodecahydro9.9350×10−5 4.7020×10−5 1.8067×10−5 4.8058×10−6 6.4825×10−7 0.0000 0.0000 0.0000 0.0000 0.0000  Figure E.3 shows the obtained rate constant for the dehydrogenation reaction of dodecahydro-N-ethylcarbazole assuming a first-order reaction. Initial reaction rate for the dehydrogenation reaction as well as the reaction TOF were calculated using the initial reactant concentration and the obtained first-  169  Table E.5: Dehydrogenation reaction data analysis. Time, min 0 4.67 9.33 14.5 22.33 29.67 37.17 50.5 81 119.5  Conversion, % 0 52.67 81.81 95.16 99.35 100 100 100 100 100  N-ethylcarbazole selectivity, % 0 6.42 11.77 31.52 41.72 47.64 59.23 81.94 96.55 100  H2 production, mmol 0 4.9133 8.8229 12.2592 14.0393 14.6763 15.4290 16.8066 17.6774 17.8829  Ln(CA /CA0 ) 0 -0.7481 -1.7045 -3.0288 -5.0321  Figure E.3: First-order reaction rate constant calculated using the reactant concentration. order rate constant as follows: Initial rate = (k, min−1 )(CA0 , mol/ml)/(Catalyst, g)(1000, ml/L) = (0.2148, min−1 )(9.9350 × 10−5 , mol/ml)/(0.07g)(1000, ml/L) −1 −1 = 0.3049 mol.L−1 .min−1 .g−1 cat (M.min .gcat )  170  (E.3)  −1 T OF = (0.0921, L.min−1 .g−1 cat )/(0.04, wt% Pd loading)(106.4, g.mol )  /(0.11, Pd dispersion)(9.9350 × 10−5 , mol/ml)(1000, ml/L) = 221.16 min−1  (E.4)  E.3 Rate constant sample calculation reported in Table 4.3 Sample calculation for obtaining the rate constant reported in L.min−1 .gcat −1 from the rate constant reported in min−1 is as follows: For a heterogeneous catalytic system, 1st -order reaction rate is defined as, −1 − r = k C Ccat  (E.5)  where r in this equation is the reaction rate (mol.min−1 .gcat −1 ), k is the rate constant (min−1 ), C is the reactant concentration (mol/L), and Ccat is the catalyst concentration (gcat /L). Reaction rate in equation E.5 can also be written as, − r = k′ C  (E.6)  where k′ is the rate constant (L.min−1 .gcat −1 ) and C is the reactant concentration (mol/L). For the dehydrogenation reaction over the 0.5 wt % Pd/SiO2 , we have the following data: 1. catalyst amount = 0.38 g 2. reaction volume = 0.03 L 3. k = 0.0457 min−1 Therefore we have, Ccat = (0.38 g)/(0.03 L) = 12.6667 g/L  171  k′ = (0.0457 min−1 )/(12.6667 g/L) = 0.0036 L.min−1 .gcat −1  172  Appendix F  Internal diffusion F.1 Effective diffusivity Because of the tortuosity and interconnecting paths of pore bodies in the catalyst with varying cross-sectional areas, diffusion can not be described within each and every one of the pathways individually. Consequently, an effective diffusion coefficient, De , is defined to describe the average diffusion as follow [149]: De =  DAB ϕP σC τ˜  (F.1)  Where DAB is bulk or Knudsen diffusivity, τ˜ is tortuosity, ϕP is pellet porosity, and  σC is constriction factor. DAB for liquid phase can be obtained from the StokesEinstein equation [150]: kB T DAB = (F.2) 6π µ R0 where kB is the Boltzmann constant, µ is the solvent viscosity, R0 is the solute radius and T is the temperature in K.  F.2 Thiele modulus, effectiveness factor and Weisz-Prater criterion The dimensionless Thiele modulus value is a measure of the ratio of a surface reaction rate to a rate of diffusion through the catalyst. For a zero order reaction, 173  Thiele modulus is obtained as follow [149]: √  ϕ =R  k0 DeCA  (F.3)  where k0 is the zero order reaction rate constant, De is effective diffusivity and CA is concentration of the reactant or product. The magnitude of internal effectiveness factor (ranging from 0 to 1) indicates the relative importance of diffusion and reaction limitations and is defined as:  η=  3 (ϕ cothϕ − 1) ϕ2  (F.4)  The term ηϕ 2 is the Weisz-Prater parameter (CW P ) which shows the ratio of the actual reaction rate to a diffusion rate. Calculated results for internal diffusion of THCZ and H2 in mesitylene for THCZ dehydrogenation on Pd/silica at 140 ◦ C and 1 atm are summarized in Table F.1. Table F.1: THCZ, H2 internal diffusion in mesitylene on Pd/silica at T=140 ◦ C, P=1 atm Name Boltzmann Constant Temperature Solvent viscosity Solute radius Tortuosity Constriction factor Porosity Diffusivity Effective diffusivity Catalyst radius Reaction rate Concentration Thiele modulus Effectiveness factor  symbol kB T µ R0 τ˜ σC ϕP DAB De R k0 CA ϕ η  unit J.K−1 K kg/(m.sec) m dimensionless dimensionless dimensionless m2 /sec m2 /sec m mM/min.gcat mol/lit dimensionless dimensionless  value for THCZ 1.38E-23 413 7.77E-4 5E-10 3.0 0.8 0.4 7.78E-10 8.30E-11 1.14E-04 17 0.2 0.47 0.99  value for H2 1.38E-23 413 7.77E-4 5E-11 3.0 0.8 0.4 7.78E-09 8.30E-10 1.14E-04 17 0.26 0.13 1  The obtained effectiveness factor for both THCZ and H2 shows that the reaction is not diffusion-limited within the catalyst. 174  Appendix G  Reaction modeling G.1  Standard error of the parameter estimation  The parameter estimation and the standard error calculations have been done using the method described by Englezos et al. [151]. The covariance matrix of distribution, denoted as Cov(k*), was calculated to measure the reliability of the estimated k parameters as follows: COV (k∗ ) = σε2 [A∗ ]−1  (G.1)  The matrix A* in this equation is calculated at k*. σε2 is the variance and is calculated using the following equation:  σε2 =  S(k∗ ) S(k∗ ) = degree o f f reedom Nm − P  (G.2)  in which degree of freedom is defined as the number of measurements minus the number of unknown parameters. The standard error associated with the estimated parameter is denoted as σKi and is calculated using the diagonal elements of the A* matrix, defined as:  σKi = σε  √ {[A∗ ]−1 }ii 175  (G.3)  This method have been used for calculating the standard errors of the estimated rate constants as parameters, shown in the following section.  G.2  Codes  The following codes were used to Model the hydrogenation reactions of N-ethylcarbazole at 423 K using MATLAB software. ===============================================================  G.2.1  Main body  FARNAZ SOTOODEH, JULY 2009 clear all; clc % Order of reactions: n1=1; n2=1; n3=1; n4=1; n5=1; n6=1; n7=1; y60= 2.98027E-1; h=1e-10; global n1 n2 n3 n4 n5 n6 n7 DATA=[ 0  0  0  0  0  0  2.98E-1  2.58  8.18E-4 1.72E-3 2.22E-4 1.50E-5 4.60E-4 2.77E-1  7.58  2.57E-3 2.05E-2 2.12E-3 1.06E-4 4.86E-3 1.68E-1  26.58  2.13E-3  5.25E-2 3.20E-3 5.00E-4 5.59E-2 9.52E-2  34.58  12.00E-5 7.11E-2 3.79E-3 9.73E-4 7.51E-2 8.00E-2  53.58  6.40E-5  6.20E-2 4.15E-3 1.44E-3 1.06E-1  114.58  9.50E-5  4.00E-2 17.21E-4 1.57E-3 2.47E-1 7.61E-4 176  5.00E-2  275.58  4.00E-6  0.90E-3 4.44E-4 1.00E-3 2.69E-1 3.43E-4  416.58  3.00E-6  0.70E-3 3.10E-4  5.00E-4 2.79E-1 2.71E-4];  T =DATA(:,1)’; CAX=DATA(:,2)’; CBX=DATA(:,3)’; CCX=DATA(:,4)’; CDX=DATA(:,5)’; CEX=DATA(:,6)’; CFX=DATA(:,7)’; t=T; options=optimset(’display’, ’iter’,’LargeScale’, ’off’,’MaxIter’, 10000, ’MaxFunEvals’, 10000); lb=[0 0 0 0 0 0 0 ]; ub=[10 10 10 10 10 10 10 ]; % inital guess for parameters k0=[0.1 0.1 0.1 0.1 0.1 0.1 0.1]; [k,S]=fminsearchbnd(@OBJECTIVE150Cmech8, k0,lb,ub, options,T,CAX,CBX,CCX,CDX,CEX,CFX); x0 = [ 0 0 0 0 0]; tspan=0:1:500; [t,yk1r]=ode45(’ODE150Cmech8’,tspan,x0,[],[k(1)+h k(2) k(3) k(4) k(5) k(6) k(7)]); [t,yk2r]=ode45(’ODE150Cmech8’,tspan,x0,[],[k(1) k(2)+h k(3) k(4) k(5) k(6) k(7)]); [t,yk3r]=ode45(’ODE150Cmech8’,tspan,x0,[],[k(1) k(2) k(3)+h k(4) k(5) k(6) k(7)]); [t,yk4r]=ode45(’ODE150Cmech8’,tspan,x0,[],[k(1) k(2) k(3) k(4)+h k(5) k(6) k(7)]); [t,yk5r]=ode45(’ODE150Cmech8’,tspan,x0,[],[k(1) k(2) k(3) k(4) k(5)+h k(6) k(7)]); [t,yk6r]=ode45(’ODE150Cmech8’,tspan,x0,[],[k(1) k(2) k(3) k(4) k(5) k(6)+h k(7)]); [t,yk7r]=ode45(’ODE150Cmech8’,tspan,x0,[],[k(1) k(2) k(3) k(4) k(5) k(6) k(7)+h]); [t,yk1l]=ode45(’ODE150Cmech8’,tspan,x0,[],[k(1)-h k(2) k(3) k(4) k(5) k(6) k(7)]); [t,yk2l]=ode45(’ODE150Cmech8’,tspan,x0,[],[k(1) k(2)-h k(3) k(4) k(5) k(6) k(7)]); [t,yk3l]=ode45(’ODE150Cmech8’,tspan,x0,[],[k(1) k(2) k(3)-h k(4) k(5) k(6) k(7)]); [t,yk4l]=ode45(’ODE150Cmech8’,tspan,x0,[],[k(1) k(2) k(3) k(4)-h k(5) k(6) k(7)]); [t,yk5l]=ode45(’ODE150Cmech8’,tspan,x0,[],[k(1) k(2) k(3) k(4) k(5)-h k(6) k(7)]); [t,yk6l]=ode45(’ODE150Cmech8’,tspan,x0,[],[k(1) k(2) k(3) k(4) k(5) k(6)-h k(7)]); [t,yk7l]=ode45(’ODE150Cmech8’,tspan,x0,[],[k(1) k(2) k(3) k(4) k(5) k(6) k(7)177  h]); df1dk1=(yk1r(:,1)-yk1l(:,1))/2/h; df1dk2=(yk2r(:,1)-yk2l(:,1))/2/h; df1dk3=(yk3r(:,1)-yk3l(:,1))/2/h; df1dk4=(yk4r(:,1)-yk4l(:,1))/2/h; df1dk5=(yk5r(:,1)-yk5l(:,1))/2/h; df1dk6=(yk6r(:,1)-yk6l(:,1))/2/h; df1dk7=(yk7r(:,1)-yk7l(:,1))/2/h; df2dk1=(yk1r(:,2)-yk1l(:,2))/2/h; df2dk2=(yk2r(:,2)-yk2l(:,2))/2/h; df2dk3=(yk3r(:,2)-yk3l(:,2))/2/h; df2dk4=(yk4r(:,2)-yk4l(:,2))/2/h; df2dk5=(yk5r(:,2)-yk5l(:,2))/2/h; df2dk6=(yk6r(:,2)-yk6l(:,2))/2/h; df2dk7=(yk7r(:,2)-yk7l(:,2))/2/h; df3dk1=(yk1r(:,3)-yk1l(:,3))/2/h; df3dk2=(yk2r(:,3)-yk2l(:,3))/2/h; df3dk3=(yk3r(:,3)-yk3l(:,3))/2/h; df3dk4=(yk4r(:,3)-yk4l(:,3))/2/h; df3dk5=(yk5r(:,3)-yk5l(:,3))/2/h; df3dk6=(yk6r(:,3)-yk6l(:,3))/2/h; df3dk7=(yk7r(:,3)-yk7l(:,3))/2/h; df4dk1=(yk1r(:,4)-yk1l(:,4))/2/h; df4dk2=(yk2r(:,4)-yk2l(:,4))/2/h; df4dk3=(yk3r(:,4)-yk3l(:,4))/2/h; df4dk4=(yk4r(:,4)-yk4l(:,4))/2/h; df4dk5=(yk5r(:,4)-yk5l(:,4))/2/h; df4dk6=(yk6r(:,4)-yk6l(:,4))/2/h; df4dk7=(yk7r(:,4)-yk7l(:,4))/2/h; df5dk1=(yk1r(:,5)-yk1l(:,5))/2/h; df5dk2=(yk2r(:,5)-yk2l(:,5))/2/h; df5dk3=(yk3r(:,5)-yk3l(:,5))/2/h; df5dk4=(yk4r(:,5)-yk4l(:,5))/2/h; df5dk5=(yk5r(:,5)-yk5l(:,5))/2/h; df5dk6=(yk6r(:,5)-yk6l(:,5))/2/h; df5dk7=(yk7r(:,5)-yk7l(:,5))/2/h; Jac1= [df1dk1 df1dk2 df1dk3 df1dk4 df1dk5 df1dk6 df1dk7 df2dk1 df2dk2 df2dk3 df2dk4 df2dk5 df2dk6 df2dk7 df3dk1 df3dk2 df3dk3 df3dk4 df3dk5 df3dk6 df3dk7 df4dk1 df4dk2 df4dk3 df4dk4 df4dk5 df4dk6 df4dk7 df5dk1 df5dk2 df5dk3 df5dk4 df5dk5 df5dk6 df5dk7]; [t,YM]=ode45(’ODE150Cmech8’,tspan,x0,[],[k(1) k(2) k(3) k(4) k(5) k(6) k(7)]); CA =real(abs(YM(:,1)))’; CB =real(abs(YM(:,2)))’; CC=real(abs(YM(:,3)))’; CD =real(abs(YM(:,4)))’; CE=real(abs(YM(:,5)))’; CF=y60-CA-CB-CC-CD-CE; DF=5*9-7; Astar=Jac1’*Jac1; Astarinv=inv(Astar); std-error=SE*sqrt(diag(Astarinv)) alpha=0.05; 178  tval=tinv(1-alpha/2,DF); k1CI-left=k(1)-tval*std-error(1) k1CI-right=k(1)+tval*std-error(1) k2CI-left=k(2)-tval*std-error(2) k2CI-right=k(2)+tval*std-error(2) k3CI-left=k(3)-tval*std-error(3) k3CI-right=k(3)+tval*std-error(3) k4CI-left=k(4)-tval*std-error(4) k4CI-right=k(4)+tval*std-error(4) k5CI-left=k(5)-tval*std-error(5) k5CI-right=k(5)+tval*std-error(5) k6CI-left=k(6)-tval*std-error(6) k6CI-right=k(6)+tval*std-error(6) k7CI-left=k(7)-tval*std-error(7) k7CI-right=k(7)+tval*std-error(7) fprintf(’k1 Confidence interval=%.2f =< k1 <= %.2f \ n’, k1CI-left, k1CI-right); fprintf(’k2 Confidence interval= %.2f =< k2 <= %.2f\ n’, k2CI-left, k2CI-right); fprintf(’k3 Confidence interval= %.2f =< k3 <= %.2f\ n’, k3CI-left, k3CI-right); fprintf(’k4 Confidence interval= %.2f =< k4 <= %.2f\ n’, k4CI-left, k4CI-right); fprintf(’k5 Confidence interval= %.2f =< k5 <= %.2f\ n’, k5CI-left, k5CI-right); fprintf(’k6 Confidence interval= %.2f =< k6 <= %.2f\ n’, k6CI-left, k6CI-right); fprintf(’k7 Confidence interval= %.2f =< k7 <= %.2f\ n’, k7CI-left, k7CI-right); subplot(2,3,2),plot(T,CAX,’o’,tspan,CA) title(’tetra’) ylabel(’Concentration, M’) subplot(2,3,3), plot(T,CBX,’o’,tspan,CB) title(’octa’) subplot(2,3,4), plot(T,CCX,’o’,tspan,CC) title(’hexa’) subplot(2,3,5), plot(T,CDX,’o’,tspan,CD) title(’deca’) xlabel(’Time, min’) ylabel(’Concentration, M’) subplot(2,3,6), plot(T,CEX,’o’,tspan,CE) title(’dodeca’) xlabel(’Time, min’) subplot(2,3,1), plot(T,CFX,’o’,tspan,CF) title(’ECZ’) xlabel(’Time, min’) ===============================================================  G.2.2  Objective function  function OBJ=OBJECTIVE150Cmech8(x, T,CAX,CBX,CCX,CDX,CEX,CFX) x0 = [ 2.98027E-1 0 0 0 0 0 ]; [T,Y]=ode45(′ ODE150Cmech8′ ,T,x0,[],abs(x)); CA =real(abs(Y(:,1))); CB =real(abs(Y(:,2)));  179  CC =real(abs(Y(:,3))); CD =real(abs(Y(:,4))); CE =real(abs(Y(:,5))); CF =real(abs(Y(:,6))); CA=CA(2:length(CA)); CB=CB(2:length(CB)); CC=CC(2:length(CC)); CD=CD(2:length(CD)); CE=CE(2:length(CE)); CF=CF(2:length(CF)); CAX=CAX(2:length(CAX)); CBX=CBX(2:length(CBX)); CCX=CCX(2:length(CCX)); CDX=CDX(2:length(CDX)); CEX=CEX(2:length(CEX)); CFX=CFX(2:length(CFX)); OBJ=sum( 25 ∗(CA-CAX′).2 + 800∗(CB-CBX′ ).2 + 100∗(CC-CCX′ ).2 + 5000∗(CDCDX′ ).2 + 20000∗(CE-CEX′ ).2 + 5∗(CF-CFX′ ).2 ); ===============================================================  G.2.3  Ordinary differential equations  function dy = ODE150Cmech8(t,y,flag,p) global n1 n2 n3 n4 n5 n6 n7 dy=zeros(6,1); k1=p(1); k2=p(2); k3=p(3); k4=p(4); k5=p(5); k6=p(6); k7=p(7); dy(1)=-k1*y(1)n1 -k2*y(1)n2 ; dy(2)=k1*y(1)n1 -k3*y(2)n3 ; dy(3)=k2*y(1)n2 -k5*y(3)n5 -k6*y(3)n6 +k4*y(4)n4 ; dy(4)=k3*y(2)n3 -k4*y(4)n4 ; dy(5)=k5*y(3)n5 -k7*y(5)n7 ; dy(6)=k6*y(3)n6 +k7*y(5)n7 ; ===============================================================  180  Appendix H  Additional publications This appendix explains the following manuscript: L. Zhao, F. Sotoodeh and K.J. Smith, Increased surface area of unsupported Mo2 C catalyst by alkali-treatment, Catalysis Communications 11 (2010) 391 - 395. Preparation of the alkali-treated Mo2 C for 12 h, catalyst characterizations including XRD, BET surface area and pore size distributions measurements of the 6 h and 12 h alkali-treated Mo2 C, as well as dehydrogenation reactions of THCZ over a 5 wt % Pd/SiO2 were done by Farnaz Sotoodeh.  H.1  Introduction  Molybdenum carbide is known to have catalytic properties similar to those of noble metals. Consequently, there is a significant interest in Mo2 C as a catalyst for a wide range of reactions, including hydrogenation [152, 153], dehydrogenation [154–157], synthesis gas conversion [158, 159], hydrocarbon isomerization [160], watergas-shift [161] and hydrogen production from methanol and ethanol [162]. Several methods have been developed for the synthesis of Mo2 C, including the most often used temperature-programmed reduction (TPR) of oxide precursors using CH4 or CH4 /H2 mixtures [75]. Other approaches have also been reported such as TPR using butane [163], carbothermal hydrogen reduction [164], the pyrolysis of metal precursors [165], solution reactions [166], and photochemically promoted  181  formation of carbide from metallic molybdenum and graphite [167]. The goal of almost all these synthesis methods is to enhance the catalyst surface area and hence increase the activity of Mo2 C catalysts. However, some of the known preparation methods are subject to limitations associated with cost and convenience. Furthermore, the surface area of the Mo2 C produced by temperature- programmed reduction of oxide precursors is greatly influenced by preparation conditions. The temperature ramp rate, the flow rate and composition of the reducing gas (CH4 , C2 H6 , H2 ), and the precursor, all impact the product surface area and consequently, reported surface areas of unsupported Mo2 C vary widely [75, 152–163, 168–170]. An alternative, simple method of producing high surface area Mo2 C would be of significant benefit. Alkali-treatment is a well known post-treatment technique used in zeolite synthesis, which introduces mesopore and macropore structure to the zeolite pore system [171, 172]. The alkali-treated zeolites show higher catalytic activity and increased surface area compared to un-treated zeolites. Although this technique has been used to prepare high surface area zeolites, it has not been applied to Mo2 C catalysts. In the present study, we report a simple post-treatment of Mo2 C using a hot alkali solution that enhances the Mo2 C surface area while retaining the structural integrity of the Mo2 C. The obtained high surface area Mo2 C has been used as a catalyst for the dehydrogenation of 1,2,3,4-tetrahydrocarbazole (THCZ) and the hydrogenation of CO. THCZ is a candidate heteroaromatic compound for hydrogen storage. Hydrogen is released through catalytic dehydrogenation at low temperature (< 423 K), using platinum group metal catalysts (Pd, Pt, Rh and Ru) [53, 54, 106]. Because of the cost of the platinum group metals, metal carbides are possible alternatives for this reaction. Metal carbides have also been shown to be active in several CO hydrogenation reactions aimed at producing higher hydrocarbons [168] and alcohols [159] and consequently we have tested the prepared Mo2 C for syngas conversion as well.  182  H.2  Experimental  H.2.1  Catalyst preparation  Unsupported Mo2 C was synthesized by temperature-programmed reduction of an oxide precursor. Approximately 3 g of MoO3 (MoO3 , +99.5 %, SigmaAldrich) was placed in a quartz tubular reactor and heated at a rate of 5 K/min from room temperature to 573 K, followed by heating at a rate of 1 K/min from 573 K to 948 K in 100 ml(STP)/min of CH4 /H2 (1:4 vol). The final temperature was held for 2 h in the same CH4 /H2 flow. The sample was then rapidly cooled to room temperature in a 80 ml(STP)/min flow of H2 . The sample was passivated by exposure to a mixture of 1 % O2 in He at a flow rate of 133 ml(STP)/min for 4 h before being removed from the reactor. Approximately 3 g of the passivated Mo2 C was placed in 90 ml of a 5 M NaOH solution and stirred for 2, 6 and 12 h at 363 K. The mixture was subsequently cooled in an ice bath. The liquid and solid were separated by centrifuging at 12,000 rpm for 20 min at 288 K. The solid was washed with deionized water until the washings showed a neutral pH. The catalyst was then dried at room temperature and activated in H2 (510 ml(STP)/min) at 573 K for 2 h before being used for reaction. The catalysts are designated AT-2, AT-6 and AT-12 to identify the duration of the alkali-treatment in each case.  H.2.2  Characterization  The passivated catalysts were characterized by powder X-ray diffraction (XRD), scanning electron microscopy (SEM), energy dispersive X-ray analysis (EDX), N2 adsorption/desorption at 77 K and thermogravimetric analysis. Prior to N2 adsorption, the samples were degassed at 423 K in 30 mol % N2 /He (15 ml(STP)/min) for 2 h to remove moisture. All other analyses were carried out on the passivated samples without further treatment. The XRD analysis was performed using a Siemens ˚ The scan rate was 2◦ per D500 Cu Kα X-ray source of wavelength (λ ) 1.54 A. min for a range of 580◦ to confirm the structure of Mo2 C, and 0.5◦ per min for the range of 3242◦ to estimate the crystallite size (dc ) of the Mo2 C using the Scherrer equation, dc = Kλ /β cos θ , where the constant K was considered 0.9, λ is the wavelength of radiation, β is the peak width in radians and θ is the angle of diffraction. 183  N2 - adsorption/desorption isotherms were recorded at 77 K using a Micromeritics ASAP 2020 analyzer. The total surface area was calculated according to the BET isotherm, and the micropore volume, the mesopore volume, and external surface area were evaluated by the t-plot method. The pore size was reported as the average pore diameter. SEM/EDX analysis was performed using a Hitachi S-3000 N electron microscope operated with a 150 kV electron beam acceleration voltage. Thermogravimetric analysis of the catalysts was done using a TGA-50 thermogravimetric analyzer (Shimadzu, Japan). About 5 to 10 mg of the catalyst was loaded into an alumina crucible and heated to 1000 K at a rate of 5 K/min in an air atmosphere (flow rate = 15 ml(STP)/min). The weight change were quantified and differential TGA (DTG) was obtained.  H.2.3  Catalyst activity  Dehydrogenation of 1,2,3,4-tetrahydrocarbazole (THCZ) was performed at 423 K using a solution of 2.0 g THCZ dissolved in 30 ml mesitylene. The solution was loaded into a 50 ml glass flask reactor. The 1.0 g alkali-treated Mo2 C was added to the solution after activation in hydrogen. The reactor was sealed, stirred, purged in a flow of He (100 ml(STP)/min) and heated to 423 K. The He was used as a carrier gas for continuous removal of the produced hydrogen and the exit gas composition was continuously monitored using a quadrupole mass spectrometer. Catalyst activity was determined by the production of H2 and confirmed by periodic liquid analysis using a gas chromatograph (Shimadzu GC-14A) with a flame ionization detector and an ATTM-5 25 M × 0.53 mm capillary column. Product identities were also confirmed by a Shimadzu QP-2010S GC/MS using a Restek RTX5 30 M × 0.25 mm capillary column. The absence of internal and external mass transfer effects on the reaction was confirmed experimentally and by theoretical analysis [107]. The activity of the Mo2 C catalysts for synthesis gas conversion was also measured using a fixed-bed micro reactor (o.d. = 9.53 mm and i.d. = 6.35 mm, copper lined stainless steel tube). The catalyst was loaded into the reactor and fixed on a packed bed of quartz wool and quartz beads. The catalyst bed depth was 5 to 7 cm to ensure plug-flow in the reactor. Synthesis gas (H2 :CO = 1) was reacted at  184  573 K, a pressure of 8.3 MPa and a GHSV = 3960 h−1 . Details of experimental procedure have been discussed elsewhere [173]. Conversion and selectivity data were confirmed to be within ±10 % of the reported results by repeating the experiments. Internal and external heat and mass transfer rates at the conditions of the experiments were found to be negligible according to the calculations based on the Mears and Weisz−Prater criteria.  H.3  Results and discussion  H.3.1  Mo2 C properties  The properties of the Mo2 C and the alkali-treated Mo2 C, reported in Table H.1, show that several changes occurred to the Mo2 C after alkali-treatment. Compared to the un-treated Mo2 C, the alkali-treated Mo2 C showed a significant increase in surface area, an increase in the micropore area and a decrease in the average pore size of the Mo2 C. The BET surface area of the alkali-treated Mo2 C increased from 29 m2 /g to 40 to 47 m2 /g, an increase of about 50 %. The N2 adsorption and desorption isotherms of the Mo2 C and the alkali-treated Mo2 C (Figure H.1) showed increased hysteresis effects with increased time of the alkali-treatment. Figure H.2 presents the pore size distribution of the Mo2 C and the alkali-treated Mo2 C. Clearly, the alkali-treatment increased the fraction of pores < 3 nm in size and decreased the average pore size (d p ) of the Mo2 C treated with NaOH for 2 and 6 h. After 12 h, Figure H.2 shows that a significant increase in pores of size 20 - 30 nm had also occurred. The data of Table H.1 shows that as the alkalitreatment time increased, the micropore area also increased. These results suggest that the alkali-treatment initially removed material from the Mo2 C to generate new pores of < 3 nm in diameter, resulting in a decrease in the average pore size. With increased treatment time (12 h) the removal of material from larger pores also occurred. These observations are quite different to changes observed in zeolites treated with alkali. Typically with zeolites, the microporosity decreases and the meso- and macroporosity increases, leading to an increase in the average pore size of the alkali-treated zeolite compared to the un-treated zeolite [171, 172]. Figure H.3 shows the XRD patterns of the Mo2 C and the alkali-treated Mo2 C. 185  Table H.1: Properties of the unsupported Mo2 C catalysts. Property N2 adsorption− desorption SBET (m2 /g) St−micro (m2 /g) St−ext (m2 /g) V p (cm3 /g) d p (nm) XRD analysis dc (1 0 1) (nm) dc (1 0 0) (nm) ˚ a (A) ˚ c (A)  Alkali-treated Mo2 Ca 2 h AT-2  6 h AT-6  12 h AT-12  29.4 ± 1.1b 0.06 ± 0.05 29.9 ± 1.9 0.031 ± 0.005 4.19 ± 0.5  44.5 0.05 44.4 0.043 3.82  47.3 ± 3.1b 4.8 ± 2.5 42.5 ± 0.6 0.036 ± 0.001 3.01 ± 0.12  40.0 10.4 29.6 0.030 2.98  13.6 15.2 3.005 4.742  13.0 14.5 3.005 4.782  13.1 15.2 3.009 4.764  13.9 14.6 3.012 4.746  Mo2 C  a Mo C treated in 5 M NaOH for the time period shown. 2 b Standard deviation from three repeat measurements.  The eight peaks at 2θ equal to 34.48, 37.96, 39.52, 52.28, 61.76, 69.56, 74.80, and 75.72 can be indexed to hexagonal Mo2 C (1 0 0), (0 0 2), (1 0 1), (1 0 2), (1 1 0), (1 0 3), (1 1 2), and (2 0 1). There were no differences between the peaks before and after alkali-treatment, indicating that the crystal structure of the Mo2 C was not changed by the alkali-treatment. The Mo2 C crystallite size was calculated from the Scherrer equation and the results, summarized in Table H.1, suggest that no significant change in crystallite size occurred after alkali-treatment of the Mo2 C either. The lattice parameters of the catalysts were also calculated using the Bragg equation and the results are listed in Table H.1. According to the JCPDS card No. 11-0680, the lattice parameters of Mo2 C are a = 3.006 A˚ and c = 4.733 A˚ and the results obtained in the present work (Table H.1) are in good agreement with these values. Hence we conclude that the crystal structure of the Mo2 C was not affected by alkali-treatment. Also, the SEM micrographs of Figure H.4 show that the Mo2 C and the alkali-treated Mo2 C had very similar macro-morphology. The materials consisted of platelet-like grains, < 5 µ m in size, and the alkali-treatment had no discernable effect on this Mo2 C morphology. Figure H.5 shows the differential TGA curves of the Mo2 C and the 2 h alkalitreated Mo2 C. Both samples had very similar characteristics, with two main peaks 186  Figure H.1: N2 adsorptiondesorption isotherms measured at 77 K for Mo2 C and alkali-treated Mo2 C catalysts. generated as the Mo2 C was oxidized to MoO2 and then MoO3 . The total weight gain in both cases was 95 % of the expected gain for pure Mo2 C conversion to MoO3 , confirming that the bulk of the material was phase pure Mo2 C. These results indicate that the new pores of < 3 nm diameter generated by alkali-treatment were a consequence of Mo2 C dissolution, rather than simply due to the selective removal of excess carbon or other impurity from the pores of the catalyst.  H.3.2  THCZ dehydrogenation  Since THCZ is a solid powder at ambient conditions with high boiling point (598 K) and melting point (391 K), the dehydrogenation reaction was carried out in the liquid phase by dissolving the THCZ in mesitylene. Over supported Pd catalysts, the reaction was zero-order in substrate (THCZ) concentration, and the catalysts 187  Figure H.2: Pore size distribution of Mo2 C and alkali-treated Mo2 C catalysts. showed 100 % selectivity to the desired carbazole product. However, it was shown from experiment that product inhibition slowed the rate of reaction and 81 % conversion was obtained after about 27 h at 101 kPa and 413 K. The initial rate of THCZ dehydrogenation at these conditions was 10 mM.gcat −1 .min−1 [53]. The dehydrogenation activity of Mo2 C and the 2 h alkali-treated Mo2 C was determined at 423 K. The data for the two catalysts presented in Table H.2, show an initial THCZ dehydrogenation rate of 0.014 and 0.048 mM.gcat −1 .min−1 , respectively. The alkali-treated Mo2 C had about 3× the initial activity of the un-treated Mo2 C for THCZ conversion. The H2 produced in the first 30 min of reaction was similarly higher on the alkali-treated catalyst. The activity (on a mass basis) of both Mo2 C catalysts was approximately two orders of magnitude lower than that reported for supported Pd catalysts used for the dehydrogenation reaction. 188  Figure H.3: X-ray diffractograms of Mo2 C and alkali-treated Mo2 C catalysts.  189  Figure H.4: SEM micrographs of (A) Mo2 C and (B) 2 h alkali-treated Mo2 C. Table H.2: THCZ dehydrogenation activity measured in a batch reactor at 101 kPa and 423 K over unsupported Mo2 C catalysts. Catalyst Mo2 C AT-2 Mo2 C a After  H.3.3  Reaction temperature (K) 423 423  Initial rate (mM.gcat −1 .min−1 )a 0.014 0.048  H2 produced/gcat (mmol/gcat )a 0.056 0.184  30 min reaction.  CO hydrogenation reaction  Several Mo compounds have also been investigated for synthesis gas conversion. MoS2 , Mo2 O3 and Mo2 C show high selectivity towards liquid oxygenates, especially ethanol, when doped with alkali earth metals. In addition, the Mo2 C has high CO2 yield rather than water, a consequence of high watergas-shift activity  190  Figure H.5: Temperature-programmed oxidation of Mo2 C and 2 h alkalitreated Mo2 C catalysts. [161]. For example, the rate of CO consumption over Mo2 C with low surface area (approximately 5 m2 /g) at 573 K and 8.0 MPa with a H2 :CO ratio 1:1, was approximately 0.4 mmol.gcat −1 .min−1 based on a reported CO conversion of 58 % at a GHSV of 2000 h−1 [159]. In the present work, Mo2 C and the 2 h alkalitreated Mo2 C were tested for synthesis gas conversion at 573 K and 8.0 MPa with a H2 :CO ratio 1:1 and a GHSV of 3960 h−1 . The initial catalyst activities (after a 2 h stabilization period) are reported in Table H.3. The data show that alkalitreated Mo2 C increased the CO consumption rate by about 10 % compared to the un-treated Mo2 C. The data show higher CO consumption rates than that reported by Xiang et al. [159] on both the Mo2 C and the alkali-treated Mo2 C. High CH4 and CO2 selectivities were also obtained on the two catalysts of the present study. Typically, Mo2 C catalysts are promoted with alkali metal to reduce selectivity 191  Table H.3: Syngas conversion on Mo2 C and alkali-treated Mo2 C. Reaction conditions: pressure = 8.2 MPa, temperature = 573 K, CO:H2 = 1 and GHSV = 3960 h−1 . Catalyst  Mo2 C AT-2 Mo2 C a Measured  Initial CO consumption rate (mmol.gcat −1 .min−1 )a 2.06 2.24  Initial product selectivitya C2 + alcohols and HCs C atom% 25.4 27.5  CH4  CO2  26.0 20.9  48.6 51.7  2 h after the start of the experiment.  to CH4 so the high selectivity obtained in the present work was not unexpected. The alkali-treated catalysts had been thoroughly washed prior to use and EDX confirmed that the Mo2 C was Na-free. The product distribution also showed enhanced watergas-shift activity on the alkali-treated sample compared to the untreated Mo2 C. Figure H.6 shows the change in CO consumption rate and the product selectivity obtained over a period of approximately 10 h. The product selectivities remained relatively constant during the experiments, whereas the alkalitreated catalyst showed some deactivation, possibly because of carbon deposition. A loss in activity in the first 2 h of reaction may also explain why the differences in the CO consumption rates between the un-treated and alkali-treated Mo2 C were not as great as observed for the THCZ dehydrogenation reaction.  H.4  Conclusion  The surface area of an unsupported Mo2 C catalyst was increased by a simple alkalitreatment of the Mo2 C that had been prepared by temperature-programmed reaction of a MoO3 precursor. The alkali- treatment increased the catalyst surface area by 50 %, primarily through the creation of new pores of size < 3 nm. The alkalitreatment did not change the Mo2 C morphology nor the catalyst composition. The alkali-treated Mo2 C showed enhanced activity for the dehydrogenation of THCZ and for CO hydrogenation, compared to the un-treated Mo2 C.  192  Figure H.6: CO consumption rate and product selectivities obtained over the un-treated Mo2 C (open symbols) and the 2 h alkali-treated Mo2 C (solid symbols) measured at 8.2 MPa, 573 K, CO:H2 = 1 and GHSV = 3960 h−1 .  193  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0059178/manifest

Comment

Related Items