CATALYSIS FOR FORMATION OF BIO - DERIVED ESTERS AS PRODUCTS AND AS INTERMEDIATES TO EPOXIDES By Arati Santhanakrishnan A DISSERTATION Submitted to Michigan State University in partial fulfilment of the requirements for the degree of Chemical Eng ineering - Doctor of Philosophy 201 6 ABSTRACT CATALYSIS FOR FORMATION OF BIO - DERIVED ESTERS AS PRODUCTS AND AS INTERMEDIATES TO EPOXIDES By Arati Santhanakrishnan The production of liquid fuels from renewable biomass resources will require multiple routes and feedstock sources beyond simple ethanol or biodiesel production. Esters are an important class of organic compounds that are used in several applications in the chemical industry as both end products and as intermediates to other value added products. Two reaction types have been studied in this dissertation that are very different in their chemistry, and thus their requirements, showing the range of processes involved in the design of bio - refineries. They are the liquid phase acid catalyzed parallel es terification reactions in batch reactor configurations, and vapor phase base catalyzed conversion of propylene glycol acetates to propylene oxide, a chemical with a fast growing global market , in a fixed bed reactor . Four results should be highlight ed in particular. The first is the development of a kinetic model for esterification that uses non ideal concentrations instead of conventional activity terms. The model enables the simulation of esterification occurring with feed streams of multiple alcohols an d acids under a wide range of conditions. Second, a study of the structure - reactivity relationship of over eighty reactions from literature helps predict the rates for a number of simple esterification reactions with a number of catalysts. Third, the reac tion system involving acylation of propylene glycol with acetic acid has been successfully modeled to obtain rate constant parameters for esterification, transesterification and hydrolysis using the non - ideal concentration model. Fourth , a detailed reactio n and catalyst properties study that optimized selectivity to nearly 90%, and establishes the chemical nature of the cata lyst under reaction conditions. These findings improve our understanding of chemical systems involving organic esters in a wide range of physical and chemical conditions, and will be useful in the design of complex processes for the production of esters and their use as intermediates. iv ACKNOWLEDGEMENTS I would like to thank Dr. Dennis J. Miller for being a patient teacher, role model a nd constant source of inspiration during the course of my dissertation work. I would like to express my gratitude to Dr. Carl T. Lira, Dr. James E. Jackson and Dr. David Hodge for serving on my committee, and for giving me their time and the benefit of the ir expertise and experience. I would also like to thank Dr. Lars Peereboom for training me in the art of experimental research, his patient explanations and all his help in the laboratory. I would like to express my sincere thanks to the National Corn Gr owers Association and Michigan State University for funding my research. I would also like to specially thank Dr. Patrick B. Smith and Dr. Adina Dumitrascu from the Michigan Molecular Institute for their guidance on the propylene oxide study and for provid ing kinetic data on propylene glycol esterification with acetic acid. I would like to thank Dr. Anne Lown, Dr. Alvaro Orjuelo, Dr. Xianfeng Ma, Dr. Xi Hong, Dr. Abu Hassan from whom I have learnt many useful things during the course of my research. I wou ld also like to thank my family for being a constant source of support and encouragement. v TABLE OF CONTENTS LIST OF TABLES ................................ ................................ ................................ ....................... viii LIST OF FIGURES ................................ ................................ ................................ ....................... xi LIST OF SCHEMES ................................ ................................ ................................ .................... xix KEY TO SYMBOLS AND ABBREVIATIONS ................................ ................................ ......... xx 1 Introduction ................................ ................................ ................................ ............................. 1 1.1 Esterification modeling ................................ ................................ ................................ .... 1 1.2 Catalysis for propylene glycol epoxidation ................................ ................................ ...... 2 2 Kinetics of mixed ethanol/ n - butanol esterification of butyric acid with Amberlyst 70 and para - toluene sulfonic acid ................................ ................................ ................................ ............... 6 2.1 Abstract ................................ ................................ ................................ ............................ 6 2.2 Introduction ................................ ................................ ................................ ...................... 6 2.3 Materials and Methods ................................ ................................ ................................ ..... 8 2.3.1 Materials ................................ ................................ ................................ ................... 8 2.3.2 Heterogeneous catalyst conditioning ................................ ................................ ........ 9 2.3.3 Kinetic experiments ................................ ................................ ................................ 10 2.3.4 Analysis ................................ ................................ ................................ ................... 10 2.4 Results ................................ ................................ ................................ ............................ 11 2.4.1 Mass transfer considerations ................................ ................................ ................... 14 2.4.2 Reaction eq uilibrium constants ................................ ................................ ............... 15 2.4.3 Kinetic Model Description ................................ ................................ ...................... 18 2.4.4 Application of Kinetic Model to Single Alcohol Esterification ............................. 19 2.4.5 Application of kinetic model to mixed alcohol esterification ................................ . 26 2.5 Conclusions ................................ ................................ ................................ .................... 27 APPENDICES ................................ ................................ ................................ .............................. 29 Appendix A: Calibration plots for gas chromatography analysis ................................ ............. 30 Appendix B. Reaction profiles of Amberlyst 70 catalyzed reactions ................................ ...... 33 Appendix C: Reaction profiles of p - TSA catalyzed reactions ................................ .................. 39 REFERENCES ................................ ................................ ................................ ............................. 44 3 Unification of esterification rates using non - ideal concentration model .............................. 47 3.1 Abstract ................................ ................................ ................................ .......................... 47 3.2 Introduction ................................ ................................ ................................ .................... 47 3.3 Calculation procedure ................................ ................................ ................................ .... 50 3.4 Results ................................ ................................ ................................ ............................ 52 3.4.1 Rate constants for straight chain alcohols with common acid and catalyst ............ 55 3.4.2 Rate constants for secondary alcohols with common acid and catalyst ................. 57 3.4.3 Different primary carboxylic acids esterified by common alcohol and catalyst .... 57 vi 3.4.4 Different branched carboxylic acids, same alcohol, same catalys t ......................... 58 3.4.5 Ratio of rate constants of the same reaction with different catalysts ...................... 60 3.5 Discussion ................................ ................................ ................................ ...................... 63 3.5.1 Taft equation ................................ ................................ ................................ ........... 65 3.6 Conclusions ................................ ................................ ................................ .................... 71 APPENDICES ................................ ................................ ................................ .............................. 72 Appendix D: Sample calculation of non - ideal concentration rate constant, k NIC ..................... 73 Appendix E: Properties of alcohols, carboxylic acids and acid catalysts used in calculat ions 75 REFERENCES ................................ ................................ ................................ ............................. 78 4 Kinetics of p - Touenesulfonic Acid - Catalyzed 1,2 - Propylene Glycol Acetylation .............. 82 4.1 Abstract ................................ ................................ ................................ .......................... 82 4.2 Introduction ................................ ................................ ................................ .................... 82 4.3 Materials and Methods ................................ ................................ ................................ ... 84 4.4 Results ................................ ................................ ................................ ............................ 85 4.4.1 Equilibrium constants ................................ ................................ ............................. 85 4.4.2 Kinetic modeling ................................ ................................ ................................ ..... 87 4.5 Conclusions ................................ ................................ ................................ .................... 96 APPENDICES ................................ ................................ ................................ .............................. 97 Appendix F: Predicted and experimental mole fraction prof iles of all experiments in PG esterification with acetic acid (Chapter 2) ................................ ................................ ................ 98 Appendix G. Raw data from propylene glycol esterification with acetic acid (conducted at Michigan Molecular Institut e) ................................ ................................ ................................ 105 REFERENCES ................................ ................................ ................................ ........................... 116 5 Catalytic epoxidation of propylene glycol and its acetates ................................ ................. 118 5.1 Abstract ................................ ................................ ................................ ........................ 118 5.2 Introduction ................................ ................................ ................................ .................. 118 5.3 Materials and Methods ................................ ................................ ................................ . 121 5.3.1 Materials ................................ ................................ ................................ ............... 121 5.3.2 Feed preparation ................................ ................................ ................................ .... 122 5.3.3 Catalyst preparation ................................ ................................ .............................. 122 5.3.4 Fixed bed reactor and condensation system ................................ .......................... 123 5.3.5 Analysis ................................ ................................ ................................ ................. 124 5.3.6 Catalyst characteriza tion ................................ ................................ ....................... 125 5.4 Results ................................ ................................ ................................ .......................... 126 5.4.1 Thermodynamic Analysis of Reaction Network ................................ ................... 126 5.4.2 Control Reactions ................................ ................................ ................................ .. 127 5.4.3 Cesium nitrate on silica gel ................................ ................................ ................... 129 5.4.4 Potassium salt catalysts ................................ ................................ ......................... 130 5.4.4.1 Mass transport limitations ................................ ................................ ............. 131 5.4.4.2 Effect of weight hour space velocity ................................ ............................. 131 5.4.4.3 Effect of temperature ................................ ................................ ..................... 133 5.4.4.4 Effect of catalyst loading ................................ ................................ ............... 134 5.4.4.5 Surface area studies ................................ ................................ ....................... 135 5.4.4.6 Chemical nature of catalyst during reaction ................................ .................. 136 5.4.4.7 Experiments with KOAc, K 2 CO 3 and K 2 SiO 3 based catalysts ...................... 139 vii 5.4.4.8 FTIR and XPS analysis of catalysts ................................ .............................. 140 5.4.4.9 Monolayer coverage calculations and controlled pore glass catalysts .......... 146 5.5 Conclusions and recommendations for future work ................................ .................... 152 APPENDICES ................................ ................................ ................................ ........................... 155 Appendix H: List of all e xperiments with reaction conditions, conversion, selectivity and carbon recovery ................................ ................................ ................................ ....................... 156 Appendix I: Calibration plots for all components in reaction system ................................ .... 158 Appendix J: Mechanisms for major reactions in reaction system ................................ .......... 162 Appendix K: Gibbs energy of reaction and equilibrium constants at reaction temperature ... 164 Appendix L: Weisz Prater calculation ................................ ................................ .................... 165 Appendix M: TGA curve calculations, atomic concentrations from XPS and EDS analyses 167 REFERENCES ................................ ................................ ................................ .......................... 172 viii LIST OF TABLES Table 2.1: Summary of experiments and conditions ................................ ................................ .... 12 Table 2.2 : Estimated effectiveness factors for Amberlyst 70 catalyzed esterification reactions . 15 Table 2.3: UNIFAC groups and their counts in reaction components ................................ .......... 16 Table 2.4: Group volume and surface area contributions ................................ ............................. 16 Table 2.5: UNIFAC group binary in teraction parameters ................................ ............................ 17 Table 2.6 : Optimized kinetic parameters with 95% confidence limits and equilibrium constants from experimental data (T in Kelvin) ................................ ................................ ........................... 21 Table 2.7: Absolute errors in model fits for each experiment ................................ ...................... 22 Table 3.1: Comparison of activity coefficient values estimated by NRTL - HOC and UNIFAC .. 51 Table 3.2 : Summary of reactions and calculated non - ideal concentration - based rate constants at ................................ ................................ ................................ ................................ .............. 53 Table 3.3: Structures of branched carboxylic acids in Figure 3.6. ................................ ............... 60 - ideal concentration model ................... 61 Table 3.5: Es values for each alcohol in Figure 3.7 ................................ ................................ ...... 70 T able 3.6: H + concentrations of catalysts ................................ ................................ ..................... 75 ................................ ................................ ....................... 7 6 ................................ ................................ ......... 77 Table 4.1: Summary of reactions and experimental conditions ................................ .................... 86 Table 4.2: UNIFAC groups and their counts in reaction components ................................ .......... 90 Table 4.3: Group volume and surface area contributions ................................ ............................. 90 Table 4.4: UNIFAC group binary interaction parameters ................................ ............................ 90 Table 4.5 Optimized kinetic parameters with 95% confidence limits for PG acetylation with p - TSA catalyst ................................ ................................ ................................ ................................ .. 92 Table 4.6: Absolute residuals for each experiment ................................ ................................ ....... 95 ix Tab le 4.7: Experiment 1 ................................ ................................ ................................ .............. 105 Table 4.8: Experiment 2 ................................ ................................ ................................ .............. 105 Table 4.9: Experiment 3 ................................ ................................ ................................ .............. 106 Table 4.10: Experiment 4 ................................ ................................ ................................ ............ 107 Table 4.11: Experiment 5 ................................ ................................ ................................ ............ 108 Table 4.12: Experiment 6 ................................ ................................ ................................ ............ 108 Table 4.13: Experiment 7 ................................ ................................ ................................ ............ 109 Table 4.14: Experiment 8 ................................ ................................ ................................ ............ 109 Table 4.15: Experiment 9 ................................ ................................ ................................ ............ 110 T able 4.16: Experiment 10 ................................ ................................ ................................ .......... 111 Table 4.17: Experiment 11 ................................ ................................ ................................ .......... 112 Table 4.18: Experiment 12 ................................ ................................ ................................ .......... 113 Table 4.19: Experiment 13 ................................ ................................ ................................ .......... 113 Table 4.20: Experiment 14 ................................ ................................ ................................ .......... 114 Table 4.21: Experiment 15 ................................ ................................ ................................ .......... 114 Table 4.22: Experiment 16 ................................ ................................ ................................ .......... 115 Table 4.23: Experiment 17 ................................ ................................ ................................ .......... 115 Table 5.1: List of reactions run with 1.5 mmol/g CsNO3 on silica gel catalyst with space velocity =1.7 g feed/g catalyst/h ................................ ................................ ................................ ............... 129 Table 5.2: List of experiments with KOH on silica gel ................................ .............................. 131 Table 5.3: Surface areas of KOH on silica gel catalysts and supports determined by N 2 adsorption (BET method) ................................ ................................ ................................ ........... 135 Table 5.4: Experiments conducted with KOAc at temperature of 400C and WHSV = 3.4 g feed/ g cat/h ................................ ................................ ................................ ................................ .......... 139 Table 5.5: Experiments conducted with 1.25 mmol K 2 CO 3 ....................... 140 Table 5.6: Binding energy peaks from XPS spectra for calcined support silica gel, neat K 2 O:SiO 2 and neat K 2 CO 3 . ................................ ................................ ................................ .......................... 143 x Table 5.7: Binding energy peaks from XPS spectra for 2.5 mmol K + /g catalyst made from KOH, KOAC and K 2 CO 3 . ................................ ................................ ................................ ..................... 144 Table 5.8: Bi nding energy peaks from XPS spectra for lower loadings of KOH on silica gel (0.5 and 1.5 mmol/g) ................................ ................................ ................................ .......................... 145 Table 5.9 ................................ ................................ .. 146 Table 5.10: Surface areas of catalysts and supports determined by N 2 adsorption (BET method) ................................ ................................ ................................ ................................ ..................... 147 Table 5.11: Experiments done to determine the effect of surface area with 1.25 mmol/g K 2 CO 3 ................................ ................................ ................................ ................................ ..................... 148 Table 5.12: Binding energy peaks from XPS spectra for K 2 CO 3 on CPG before and after use . 150 Table 5.13 : Monolayer coverage calculations ................................ ................................ ............ 152 Table 5.14: List of all experiments with reaction conditions, conversion, selectivity and carbon recovery ................................ ................................ ................................ ................................ ....... 156 Table 5.15 : Gibbs energy polynomial function values and Gibbs energy of formation at standard conditions (298 K) and at reaction temperature (673 K) ................................ ............................ 164 Table 5.16: Equilibrium constants estimated from Gibbs energy of reaction calculations at reaction temperature of 673 K ................................ ................................ ................................ .... 164 Table 5.17 : Atomic concentrations from XPS spectra for 2.5 mmol K + /g catalyst made from KOH, KOAC and K 2 CO 3 ................................ ................................ ................................ ............ 168 Table 5.18: XPS spectra for lower loadings of KOH on silica gel (0.5 and 1.5 mmol/g) .......... 169 Table 5.19: Atomic concentrations from XPS spectra for K 2 CO 3 on CPG before and after use 169 Table 5.20: Atomic concentrations from EDS spectra for 2.5 mmol K + /g catalyst made from KOH, KOAC and K 2 CO 3 . ................................ ................................ ................................ .......... 170 xi LIST OF FIGURES - (gray) and activity - (black) based esterification equilibrium Ethanol , ( + , + ) n - butanol .................... 17 Figure 2.2: Experimental and predicted concentration profiles of ethanol and n - butanol individual esterification in the b) Run 5 ( n - Activity based fits. ethyl butyrate; (×) n - - n - butyl butyrate. ................................ ................................ ................................ ................................ ........ 23 Figure 2.3: Experimental and predicted concentration profiles of ethanol and n - butanol individual esterification in the presence of 0.1 wt.% p - ; b) Run 23 ( n - Mole - Activity based fits. ethyl butyrate; (×) n - - n - butyl butyrate. ................................ ................................ ................................ ................................ ........ 23 Figure 2.4: Experimental and predicted concentration profiles of mixed alcohol esterification in the presence of 1 wt.% Amberlyst 70. a) Run 13 (Ethanol: n - Butanol = 1.15:1, T=80 o C); b) Run 11 (Ethanol: n - Butanol = 0.21:1, T = 60 o C ); c ) Run 9 (Ethanol: n - Butanol = 1:1, T = 60 o C). ethanol; ethyl butyrate; (×) n - - n - butyl butyrate. ................................ ...... 25 Figure 2.5: Experim ental and predicted concentration profiles of mixed alcohol esterification in the presence of 0.1 wt.% p - TSA. a) Run 31 (Ethanol: n - Butanol = 4.1:1 , T = 80°C) ; b) Run 29 (Ethanol: n - Butanol = 0.18:1,T = 80 °C); c) Run 30 (Ethanol: n - Butanol = 1:1,T =80°C) ethyl butyrate; (×) n - - n - butyl butyrate. ................................ ................ 26 Figure 2.6: Plot of a rea ratio vs weight ratio of ethanol over internal standard ethyl caprylate ... 30 Figure 2.7: Plot of area ratio vs weight ratio of ethyl butyrate over internal standard ethyl caprylate ................................ ................................ ................................ ................................ ........ 30 Figure 2.8: Plot of area ratio vs weight ratio of water over internal standard ethyl caprylate ...... 31 Figure 2.9: Plot of area ratio vs weigh t ratio of butanol over internal standard ethyl caprylate ... 31 Figure 2.10: Plot of area ratio vs weight ratio of butyl butyrate over internal standard ethyl caprylate ................................ ................................ ................................ ................................ ........ 32 xii Figure 2.11: Plot of area ratio vs weight ratio of butyric acid over internal standard ethyl caprylate ................................ ................................ ................................ ................................ ........ 32 Figure 2.12: Experimental and predicted concentr ation profiles of ethanol individual Mole - fraction ethyl butyrate; (×) water ................................ ................................ ................................ ................................ ............ 33 Figure 2.13: Experimental and predicted concentration profiles of ethanol individual Mole - fraction Activit ethyl butyrate; (×) water ................................ ................................ ................................ ................................ ............ 33 Figure 2.14: Experimental and predicted concentration profiles of ethanol individual esterification in the pr Mole - fraction ethyl butyrate; (×) water ................................ ................................ ................................ ................................ ...... 34 Figure 2. 15: Experimental and predicted concentration profiles of n - butanol individual Mole - fraction n - - n - butyl butyrate; (×) water ................................ ................................ ................................ ................................ ...... 34 Figure 2.16: Experimental and predicted concentration profiles of n - butanol individual ) Mole - fraction n - - n - butyl butyrate; (×) water ................................ ................................ ................................ ................................ ...... 35 Figure 2.17: Experimental and predicted concentration prof iles of n - butanol individual Mole - fraction n - - n - butyl butyrate; (×) water ................................ ................................ ................................ ................................ ...... 35 Figure 2.18: Experimental and predicted concentration profiles of mixed alcohol esterification in the presence of 1 wt.% Amberlyst 70. Run 10 (Ethanol: n - Butanol =1:1, T=80 o mole ) ethyl butyrate; (×) n - - n - butyl butyrate ................................ .......................... 36 Figure 2.19: Experimental and predicted concentration profil es of mixed alcohol esterification in the presence of 0.5 wt.% Amberlyst 70. Run 12 (Ethanol: n - Butanol = 0.88:1, T=80 o ethyl butyrate; (×) n - b - n - butyl butyrate ................................ .......................... 36 xiii Figure 2.20: Experimental and predicted concentration profiles of mixed alcohol esterification in the presence of 0.75 wt.% Amberlyst 70. Run 14 (Ethanol: n - Buta nol = 1.05:1, T=80 o ethyl butyrate; (×) n - - n - butyl butyrate ................................ .......................... 37 Fig ure 2.21: Experimental and predicted concentration profiles of mixed alcohol esterification in the presence of 1 wt.% Amberlyst 70. Run 15 (Ethanol: n - Butanol = 0.21:1, T=80 o butyric acid; ethyl butyrate; (×) n - - n - butyl butyrate ................................ .......................... 37 Figure 2.22: Experimental and predicted concentration profiles of mixed alcohol esterification in the pre sence of 2 wt.% Amberlyst 70. Run 16 (Ethanol: n - Butanol = 0.86:1, T=80 o ethyl butyrate; (×) n - - n - butyl butyrate ................................ .......................... 38 Figure 2.23: Experimental and predicted concentration profiles of mixed alcohol esterification in the presence of 1 wt.% Amberlyst 70. Run 17 (Ethanol: n - Butanol = 0.2:1, T=80 o mole fraction based fit ethyl butyrate; (×) n - - n - butyl butyrate ................................ .......................... 38 Figure 2.24: Experimental and predicted concentration profiles of ethanol individual esterification in the presence of 0.1 wt.% p - Mole - fraction ethyl butyrate; (×) water ................................ ................................ ................................ ................................ ...... 39 Figure 2.25: Experimental and predicted concentration profiles of ethanol individual esterification in the presence of 0.1 wt.% p - Mole - fraction Activity based fits. ethyl butyrate; (×) water ................................ ................................ ................................ ................................ ...... 39 Figure 2.26: Experimental and predicted concentration profiles of ethanol individual esterification in the presence of 0 .1 wt.% p - Mole - fraction ethyl butyrate; (×) water ................................ ................................ ................................ ................................ ...... 40 Figure 2.27: Experimenta l and predicted concentration profiles of n - butanol individual esterification in the presence of 0.1 wt.% p - Mole - fraction based n - - n - butyl butyrate; (×) water ................................ ................................ ................................ ................................ .............. 40 Figure 2.28: Experimental and predicted concentration profiles of n - butanol individual esterification in the presence of 0.1 wt.% p - Mole - fraction xiv based f n - - n - butyl butyrate; (×) water ................................ ................................ ................................ ................................ ...... 41 Figure 2.29: Experimental and predicted concentration profiles of n - butanol individua l esterification in the presence of 0.1 wt.% p - Mole - fraction n - - n - butyl butyrate; (×) water ................................ ................................ ................................ ................................ ...... 41 Figure 2.30: Experimental and predicted concentration profiles of mixed alcohol esterification in the presence of 0.1 wt.% p - TSA. Run 26 (Ethanol: n - Butanol = 0.22:1, T=60 o mole but ethyl butyrate; (×) n - - n - butyl butyrate ................................ .......................... 42 Figure 2.31: Experimental and predicted concentration profiles of mixed alcohol esterification in the presence of 0.1 wt.% p - TSA. Run 27 (Ethanol: n - Butanol = 1:1, T=60 o mole ethyl butyrate; (×) n - - n - butyl butyrate ................................ .......................... 42 Figure 2.32: Experimental and predicted concentration profiles of mixed alcohol esterification in the presence of 0.1 wt.% p - TSA. Run 28 (Ethanol: n - Butanol = 5:1, T=60 o mole fraction based fits; ( ethyl butyrate; (×) n - - n - butyl butyrate ................................ .......................... 43 families with differing number of - Butyric acid, - Butyric acid, Amberlyst 15 (×) - - Propionic acid, Dowex W50×4 ................................ ................................ ................................ ..................... 56 - Acetic acid, Smopex - - Propio nic acid, Smopex - - Pentanoic acid, Smopex - 101 ..................... 56 common carboxylic acid and catalyst. ( ) - - Butyric acid, - Propionic acid, Smopex - 101 ................................ ................................ .......... 57 carbon - Methanol, SAC - - 2 - Propanol, Smopex - 101 ................................ ................................ ......................... 58 differing number of - Methanol, - Methanol, Hydrochloric acid ................................ ................................ ........... 58 xv Figure 3.6: Non ideal rate const carbons in branched carboxylic acid chain with methanol and hydrochloric acid catalyst .......... 59 Figure 3.7: Taft equation applied to a can didate family of reactions (Reactions based on study by Erdem and Cebe) 34 - ideal concentration plotted. ................................ ................................ ................................ ................................ .......... 69 - based equilibrium constants for the K a1 K a2 . ................................ ................................ ...... 87 Figure 4.2: Experimental and predicted concentration profiles of PG acetylation (Experiment 3: UNIFAC activity based propylene glyc propylene glycol diacetate; (×) water. ................................ ................................ .............. 93 Figure 4.3: Experimental and predicted concentration profiles of PGDA hydrolysis (Experiment UNIFAC activity propylene glycol diacetate; (×) water. ................................ ................................ ... 93 Figure 4.4:Experimental and predicted concentration profiles of PG - PGDA transacetylation UNIFAC activity based model prediction propylene glycol monoacetate; propylene glycol diacetate; (×) water. ................................ ........ 94 Figure 4.5: Parity plot of kinetic model fit for PGMA mole frac tions in experiments ................. 94 Figure 4.6: Experimental and predicted concentration profiles of PG acetylation (Experiment 1: UNIFAC activity based propylene glycol diacetate; (×) water ................................ ................................ ............... 98 Figure 4.7: Experimental and pred icted concentration profiles of PG acetylation (Experiment 2: UNIFAC activity based propylene glycol diacetate ; (×) water ................................ ................................ ............... 98 Figure 4.8: Experimental and predicted concentration profiles of PG acetylation (Experiment 4: UNIFAC activity based propylene glycol diacetate; (×) water ................................ ................................ ............... 99 xvi Figure 4.9: Exper imental and predicted concentration profiles of PG acetylation (Experiment 5: UNIFAC activity based propylene glycol diacetate; (×) water ................................ ................................ ............... 99 Figure 4.10: Experimental and predicted concentration profiles of PG acetylation (Experiment 6: UNIFAC activity based propylene glycol diacetate; (×) water ................................ ................................ ............ 100 Figure 4.11: Experimental and predicted concentration profiles of PG acetylation (Experiment 7: UNIFAC activity based propylene glycol monoaceta propylene glycol diacetate; (×) water ................................ ................................ ............ 100 Figure 4.12: Experimental and predicted concentration profiles of PG acetylation (Experiment 8: UNIFAC activity based propylene glycol diacetate; (×) water ................................ ................................ ............ 101 Figure 4.13: Experimental and predicted concentration profiles of PG acetylation (Experiment 9: UNIFAC activity based pro propylene glycol diacetate; (×) water ................................ ................................ ............ 101 Figure 4.14: Experimental and predicted concentration profiles of PG acetylation (Experiment UNIFAC activity based propylene glycol diacetate; (×) water ................................ ................................ ............ 102 Figure 4.15: Experimental and predicted concentration profiles of PG acetylation (Experiment UNIFAC activity based propylene glycol diacetate; (×) water ................................ ................................ ............ 102 Figure 4.16: Experimental and predicted concentration profiles of P GDA hydrolysis (Experiment UNIFAC activity propylene glycol diacetate; (×) propylene glycol water ................................ ................................ . 103 Figure 4.17: Experimental and predicted concentration profiles of PGDA hydrolysis (Experiment UNI FAC activity xvii propylene glycol diacetate; (×) water ................................ ................................ . 103 Figure 4.18: Experimental a nd predicted concentration profiles of PG transacetylation propylene glycol monoacetate; propylene glycol diacetate; (×) water ................................ ....... 104 Figure 4.19: Experimental and predicted concentration profiles of PG transacetylation propylene glycol monoacetate; propylene glycol diacetate; (×) water ................................ ....... 104 mmol/g of KOH in catalyst ................................ ................................ ................................ ......... 132 Figure 5.2: Rate constant vs contact time for a first order reaction A straight line through the origin fits the data with an R 2 value of 0.94 ................................ ................................ ............... 1 32 loading = 2.5 m mol/g of KOH in catalyst) (R14 R18) ................................ ............................ 133 Figure 5.4: Arrhenius plot of composite reaction of PGA to products (R14 - R18) ..................... 134 Figure feed/g cat/h) ................................ ................................ ................................ ................................ 135 ) neat KOAc ................................ ................................ ................................ ................................ .. 138 Figure 5.7: Absorbance FTIR analyses of fresh and used catalyst samples. (1: 1.25 mmol/g K 2 CO 3 on silica, 2: post reaction 2.5 mmol KOAc/g on silica, 3: post reaction 2.5 mmol /g KOH on silica, 4: unused 2.5 mmol/g KOH on silica) ................................ ................................ ......... 141 Figure 5.8: Absorbance FTIR analyses of fresh and used catalyst samples. (1: silica gel support, 2: calcined fresh 1.5 mmol KOH/g on si lica, 3: post reaction 1.5 mmol/g KOH on silica, 4: unused 1.25 mmol/g K 2 SiO 3 on silica, 5: neat K 2 SiO 3 ) ................................ .............................. 142 Figure 5.9: Absorbance FTIR analyses of fresh and used catalyst samples. (1: K 2 SiO 3 on silica gel, 2:1.25 mmol/g K 2 CO 3 on CPG post reaction, 3: 1.25 mmol/g K 2 CO 3 on CPG calcined, unused, 4: K 2 SiO 3 , neat) ................................ ................................ ................................ ............. 149 xviii coverage >100% if surface is fully saturate d, and more than one layer of potassium salt is present ................................ ................................ ................................ ................................ ..................... 151 Figure 5.11: Plot of area ratio vs weight ratio of PO over internal standard decanol ................. 158 Figure 5.12: Plot of area ratio vs weight ratio of propanal over internal standard decanol ........ 158 Figure 5.13: Plot of area ratio vs weight ratio of water over internal standard decanol ............. 159 Figure 5.14: Plot of area ratio vs weight ratio of allyl alcohol over internal standard decanol .. 159 Figure 5.15: Plot of area ratio vs weight ratio of propylene glycol over internal standard decanol ................................ ................................ ................................ ................................ ..................... 159 Figure 5.16: Plot of area ratio vs weight ratio of acetone over internal standard decanol .......... 160 Figure 5.17: Plot of area ratio vs weight ratio of acetic acid over internal standard decanol ..... 160 Figure 5.18: Plot of area ratio vs weight rat io of propylene glycol diacetate over internal standard decanol ................................ ................................ ................................ ................................ ........ 160 Figure 5.19: Plot of area ratio vs weight ratio of allyl acetate over internal standard decanol ... 161 Figure 5.20: Plot of area ratio vs weight ratio of 2 - ethyl,4 - methyl,1,3 - dioxolane over internal standard decanol ................................ ................................ ................................ .......................... 161 Figure 5.21: XPS C 1s spectra of 1.5 m mol/g KOH on silica gel post reaction ......................... 167 Figure 5.22: XPS O 1s spectra of 1.25 mmol/g K 2 CO 3 on CPG post reaction ........................... 168 Figure 5.23: SEM image of post reaction 2.5 mmol/g KOH on silica gel ................................ .. 170 Figure 5.24: SEM image of post reaction 2.5 mmol/g KOAc on silica gel ................................ 171 Figure 5.25: SEM image of post reaction 1.25 mmol/g K 2 CO 3 on silica gel .............................. 171 xix LIST OF SCHEMES Scheme 2.1: Esterification of butyric acid with ethanol and n - but anol ................................ .......... 8 Scheme 3.1: Rate limiting step of esterification reaction mechanism ................................ .......... 67 Scheme 4.1: Glycol - carboxylic acid system reactions ................................ ................................ . 83 Scheme 4.2: Acetylation of 1,2 - propylene glycol with acetic acid. ................................ ........... 84 Scheme 4.3: Simplified reaction scheme for kinetic model ................................ ......................... 86 Scheme 5.1: Chlorohydrin process ................................ ................................ ............................. 119 Scheme 5.2: tert - Butyl peroxide process ................................ ................................ .................... 119 Scheme 5.3: Routes from propylene glycol to propylene oxide ................................ ................. 120 Scheme 5.4: Propylene glycol acetates deacetoxylation system ................................ ................ 127 Scheme 5.5: Formation of potassium carbonate on surface of catalyst during reaction ............. 137 Scheme 5.6: Mechanisms for formation of PO and its isomers ................................ .................. 162 xx KEY TO SYMBOLS AND ABBREVIATIONS Nomenclature and Units C BA = concentration of butyric acid in liquid phase, kmol m - 3 C T = total molar density of the reacting fluid, kmol m - 3 D ef f = effective diffusivity of butyric acid, m 2 s - 1 D BA = bulk di ffusivity of butyric acid in alcohol, m 2 s - 1 d p = swelled diameter of catalyst in reaction conditions, m E a,m = the activation energy of the rate constant for the reaction m, kJ kmol - 1 F min = square root of mean of squared absolute residues K a,m = activit y based equilibrium constant for reaction m K x,m = mole fraction based equilibrium constant for reaction m k 0,m = the pre - 3 (kg catalyst) - 1 s - 1 kmol - 1 N C = number of components in reaction N T = total number of moles in the reactor, kmol r m = rate of reaction m per unit volume, kmol s - 1 m - 3 r* obs = observed rate of reaction per weight of catalyst, mol s - 1 kg - 1 V = reaction volume, m 3 V p,swollen = volume of swollen catalyst, m 3 V p,dry = dry volume of catalyst, m 3 w CAT = catalyst loading in the reaction mixture, kg catalyst/ kg solution x i = mole fraction of component i in t he reaction mixture at equilibrium x i = mole fraction of component i in the liquid mixture a i = activity of species i in solution xxi C T = total molar density of the reacting fluid, kmol /m 3 E a,m =activation energy for reaction m, kJ/kmol F min =objective funct ion for kinetic parameter fitting k m = rate constant for reaction m, m 3 k - m = reverse rate constant for reaction m, m 3 K a,m = activity based equilibrium constant for reaction m k 0,m = pre - exponential factor for reaction m, m 3 n = number of samples taken in an experiment N c = number of reacting species in experiment N T = total number of moles in the reactor, kmol r m = rate of reaction m per unit volume, kmol/s/m 3 V = volume of reacting phase, m 3 w CAT = catalyst loading in the reaction mixture, kg cat/kg soln) k 0,AC = activity based rate constant, (m 3 soln/(kmole soln) · (kg cat) · s) k 0,NIC = non ideal concentration model based rate constant, (m 6 soln/(kmole soln) · (kmole H + )· s) k 1 = concentration based rate constants of the reaction of interest k Ref = concentration based rate constants of reference reaction E S = steric effect of substituents C TS = concentration of the transition state, kmole/m 3 K TS = Equilibrium constant for formation of transition state from reactants k B Boltzmann constant T = temperature G TS = Gibbs free energy of reaction of reactants to form transition state, J/mole xxii C PGA = concentration of propylene glycol acetates in liquid phase, mole m - 3 Greek i = activity coefficient of component i in the reaction mixture at equilibrium = particle porosity w = Thiele modulus CAT = density of catalyst, kg m - 3 i,m = ratio of stoichiometric coeffic ient of component i with respect to the reference component in reaction m v i = stoichiometric coefficient of component i in the reaction mixture at equilibrium ing reaction that involves the reference reactant b = frequency of the transition state crossing the activation barrier (s - 1 ) i,m =ratio of stoichiometric coefficients of component i with respect to the reference component in reaction m me , s Abbreviations FTIR = Fourier transform infrared spectroscopy XPS = X - ray photoelectron spectroscopy EDS = energy - dispersive X - ray spectroscopy A - 70 = Amberlyst 70 p - TSA = para - t oluene sulfonic acid EQ = equilibrium UNIFAC = U NI versal Functional Act ivity Coefficient xxiii EB = ethyl butyrate BA = butyric acid Eth = ethanol W = water NRTL = Non - random two liquid model AA = acetic acid PG = propylene glycol PGDA = propylene glycol diacetate PGMA = propylene glycol monoacetate (1: primary, 2: secondary) PO = propylene oxide Conv. = Conversion Temp = Temperature CPG = Controlled Pore Glass 1 1 I ntroduction The production of fuels and specialty chemicals from renewable biomass sources will require multiple economic and environmentally friendly routes and feedsto ck sources. Side products of several b i o - refinery processes such as et hanol and biodiesel production present opportunities for further processing to produce value added products. Esters are one such important class of organic compounds used in a variety of applica tions in the chemical industry. This dissertation focuses on esters in particular as both end products and as intermediates to other products. 1.1 Esterification modeling In bio - refinery processes, streams with multiple carboxylic acids and multiple a lcohols are frequently encountered, for instance the Guerbet reaction of ethanol to higher alcohols, and fermentation to succinic acid along with acetic acid as a side product. The possibility of converting such mixed streams to ester mixtures for value ad dition and for easy separation by process integrated techniques such as reactive distillation is attractive, but hinges on answering certain questions. Does the presence of one alcohol inhibit the rate of another? Can we develop kinetic models that can pre dict the rate of reactions in a mixture using information from individual reactions? In Chapter 2 , as a case study, a detailed investigation was conducted of the kinetics of acylation of ethanol - butanol mixtures with butyric acid in both homogeneous and h eterogeneous catalysis conditions with the conventionally used mole fraction and activity based models, followed by attempts to fit the mixed reactions to them. There were large deviations from 2 experiment using both models , especially in comparing kinetic rates of different alcohol systems . It was also observed that purely activity based models do not predict reactions with varying initial molar ratios of reactants well irrespective of the activity coefficient model used. Therefore, both non ideality (e.g. activity) and differences in molar densities (e.g. absolute concentrations) were incorporated into a single kinetic model that fit the mixed simultaneous acylation of multiple alcohols. In Chapter 3 , it was observed that when normalizing rates to the mola r densities, the rates of many esterification reactions simplified to a single turn over number value for a specific catalyst. The implications of this are that essentially most alcohols and acids react at the same rate for a specific catalyst, and that th is single value may be used to predict most esterification reactions, including parallel reactions with any initial composition. The condensation of kinetic information for so many reactions also allows the study of structure - reactivity relationships of si milar esterification families. The comparatively higher rates for methanol and acetic acid shows that after an initial decrease in rate with addition of carbons, the length of the carbon chain is not an important factor for rate of reaction until it starts to inhibit the formation of a single phase. Trends have been studied of linear and branched alcohols and carboxylic acids, and with different homogeneous and heterogeneous catalysts and the results documented provide a useful database for estimating the r ate of future reactions of interest. 1.2 Catalysis for p ropylene glycol epoxidation Propylene oxide is a useful intermediate chemical for many everyday products such as defoamers, anti - freeze, and lubricants in the form of polyglycols, polyurethanes and polyg lycol ethers. Current commercial technologies rely on the use of propylene from petrochemical sources. 3 In the interest of reducing our reliance on fossil sources and shifting to more environmentally friendly processes, alternate routes to make propylene ox ide are being investigated. Processes to convert sorbitol and glycerol to propylene glycol have been commercialized . Instead of the traditional pathway to propylene glycol from propylene derived propylene oxide, the pathway to propylene oxide from bio - glyc erol derived propylene glycol becomes an attractive option. A number of catalyst screening studies are available in literature, but in order to design an economically viable process, the reaction pathways need to be understood, and the catalyst needs to be characterized. In Chapter 4 , the acylation of propylene glycol with acetic acid to give propylene glycol mono acetates and propylene glycol diacetate is modeled using the non - ideal concentration based model. The results align with the esterification mod el developed in Chapter 2 . The rates of esterification are the expected value predicted by findings in Chapter 1 and 2. In Chapter 5 , several candidate catalysts were chosen, and experiments were conducted to understand their activity in depth. Initial ru ns with pure propylene glycol show a large amount of di - propylene glycol formation, which was much reduced when a feed mixture of propylene glycol and its acetates were used instead. The effect of various reaction parameters such as temperature, catalyst l oading, contact time, and feed concentration has been evaluated in a gas phase fixed bed reactor configuration. The optimum conditions for the best propylene oxide yield have been found to be at 400 Extremely short contact times and low concentrations did not in effect increase selectivity above 90% by mole. 4 It is also important to understand the chemical nature of t he basic salt on the surface of the catalyst to see how it participates in the reaction. Alkali metal salts on silica can react to form the respective silicates or remain unchanged depending on a number of factors including temperature, basicity and loadin g of alkali metal, and type of support used. For this reason, the unused and post reaction catalysts were characterized by a number of techniques such as FTIR, XPS and EDS. The work in this dissertation expands the possibilities of esters as intermediates as well as products of bio - refineries. A more thorough understanding of the basics of esterification reaction kinetics is achieved, which in turn is useful for flexible process design. 5 Included in the following chapter is a copy of the paper: Kin etics of Mixed Ethanol/ n - Butanol Esterification of Butyric Acid with Amberlyst 70 and p - Toluene Sulfonic Acid, by Arati Santhanakrishnan , Abigail Shannon , Lars Peereboom , Carl T. Lira , and Dennis J. Miller Department of Chemical Engineering and Materials Science, Michigan State University , 2527 Engineering Building, East Lansing, Michigan 48824 - 1226, Uni ted States Reprinted with permission from Ind. Eng. Chem. Res. , 2013 , 52 (5), pp 1845 1853 Copyright © 2013 American Chemical Society 6 2 Kinetics of mixed ethanol/ n - butanol esterification of butyric acid with Amberlyst 70 and para - t oluene sulfonic acid 2.1 Ab stract Esterification of butyric acid with ethanol, n - butanol and ethanol/ n - butanol mixtures was studied using Amberlyst 70 cation exchange resin and homogeneous para - toluene sulfonic acid as catalysts. The kinetics of individual alcohol acylation were fir st examined in batch reactions at different temperatures and catalyst loadings, and then esterification in ethanol - n - butanol mixtures of varying co ncentration ratio s w as characterized . Both non - ideal solution and ideal solution kinetic models were develope d. The se models accurately predict the esterification of butyric acid by the individual alcohols, and a simple additive combination of the ind ividual kinetic models provides a good description of mixed alcohol esterification. Using non - ideal concentration as the measure of species activity, ethanol and n - butanol esterification kinetics are described by a common rate constant that is postulated to be universal for any normal alcohol esterification of butyric acid over a given catalyst. The kinetic models th us have broad application such as in simulatin g reactive distillation processes for mixed alcohol esterification. 2.2 Introduction The growing need to reduce dependence on fossil sources for fuels and chemicals has led to the exploration of pathways for their manufacture from renewable sources. Esters of higher alcohols (alcohols with more than two carbons) are one such industrially important class of compounds , and economically viable processes for making them need to be designed . Blends of esters are being considered as attractive solvents or as additives to biofuels because of their high energy densi ty and favorable fuel propertie s . The most common route for producing esters is the 7 direct esterification of carboxylic acids with alcohols 1 using either homogeneous acid catalysts such as sulfuric acid or para - toluene sulfonic acid ( p - TSA) or solid h eterogeneous acid catalysts such as cationic exchange resins. Mixed alcohol streams from biomass can be obtained in several ways: from condensation of lower alcohols to higher alcohols via the Guerbet reaction, 2 4 via Fischer - Trop sch synthesis of alcohols from syn thesis gas 5 8 , or from fus el alcohols 9 prod uced in ethanol fermentation . E sterification for biofuel or solvent applications is an attractive use of these mixed alcohol streams , as it would lead to value - added products without the need for separation into individual components. Simu ltaneous esterification with multiple alcohols is described in the patent literature. 10 13 Since esterification reactions are thermodynamically limited, reactive distillation is a viable option for mixed alcohol processing . 14,15 In simulations of reactive distillation, esterification of a mixture of amyl alcohol and n - butanol with acetic acid has been examined to compare separation - first and reaction - first scheme s. 16 Reaction first schemes were determined to be more economical. The design of such reactive distillation schemes for mixed alcohols requires a good understanding of the kinetics of the reaction system, as it is generally not known whether the presence of one alcohol accelerates or inhibits the rate of reaction of another, or if the formation of m ixed esters leads to transesterification that could overcomplicate the recovery process. Recent advances in producing butyric acid via fermentation of biomass carbohydrates has sparked interest in using b utyric acid as a building block via esterification and other reactions . 17 Apart from their potential as biofuel components, ethyl butyrate and n - butyl butyrate serve as food flavo ring agents and green solvents 18 . The kinetics of n - butyl butyrate formation using Dowex 19 8 as an esterification catalyst has b een previously studied and modeled using quasi - homogeneous , Eley - Rideal , and Langmuir - Hinshelwood rate models . The kinetics of ethyl butyrate formation ha ve not been previously reported . In this study, the kinetic behavior of butyric acid esterification w ith mixed e thanol and n - butanol ( Scheme 2 . 1 ) is investigated using homogeneous ( p - TSA) and heterogeneous (Amberlyst 70 ion exchange resin ) catalys t s. K inetics of individual ethanol and n - butanol esterification reac tions are first presented and then reaction kinetics for mi xtures of varying ethanol and n - butanol compositions are reported . The kinetic model fitted to experimental data is useful in designing and characterizing reactive distillation columns for mixed al cohol esterification reactions. 2.3 Materials and Methods 2.3.1 Materials Reagent grade ethanol (200 Proof, Decon Labs, Inc., King of Prussia, P ennsylvania ), n - butanol (99.9%, Sigma Aldrich Corp., St. Louis, M issouri ), n - butyl butyrate (>98%, Sigma A ldrich Corp., St. Louis, M issouri ) ethyl butyrate (>99%, Sigma Aldrich Corp., St. Louis, M issouri ), water (HPLC solvent, JT Baker Reagent Chemicals. Phillipsburg, N ew Jersey ), p - Scheme 2 . 1 : Esterification of butyric acid with ethanol and n - butanol 9 toluene sulfonic acid monohydrate (Spectrum Quality Products, Inc., Gardena, C alifornia ), butyric acid (>99%, natural, Sigma Aldrich Corp., St. Louis, M issouri ), acetonitrile (HPLC grade, Emanuel Merck Damstadt Chemicals, Phila delphia, P ennsylvania ), methanol (Sigma Aldrich Corp., St. Louis, M issouri ), and ethyl octanoate (Sigma Aldrich Corp., St. Louis, M issouri ) were used without further purification. Gas chromatograph ic (GC) analysis of the aforementioned chemicals showed no significant presence of impurities except for trace amounts of water. Hydranal - coulomat E solution (Riedel - de Haën, Seelze, Germany) was used in Karl - Fisher titrations. Helium (99.995%, AirGas , USA) was used as carrier gas for GC. The properties of the het erogeneous cat ion exchange resin catalyst Amberlyst 70 ® (Dow Chemical Company, Midland, M ichigan ) are reported in the literatur e. 20 2.3.2 Heterogeneous catalyst conditioning As - received Amberlyst 70 (A - 70) was sieved in a series of US - standard sieves (Dual Manufacturing Company, Chicago, I llinois ) and t he - 45 +60 mesh (0.25 0.35 mm diameter) fraction was used in kinetic experiments. The resin was washed with methanol multiple times until the supernatant liquid was colorless, and then filtered to remove excess methanol. The resin was then dried in an oven at 373 K for 2 days. The dried resin was stored in a sealed container in a desiccator and removed in required amounts for reactions. Fresh catalyst was used for each experiment. To find the ion exchange capacity of the catalyst, a known quantity of dry A - 70 was submerged in ethan ol for 4 - 5 hours and then titrated with NaOH . The average ion exchange capacity was found to be 2.35 ± 0.1 eq uivalents H + /kg, in reasonable agreement with the value of reported by the manufacturer . 20 10 2.3.3 Kinetic experiments Isothermal kinetic experiments were carried out in 75 ml batch reactors in a Parr 5000 M ultireactor system (Parr Instrument Co., Moline, I llinois ). The reactor system is equipped with temperature and stirring speed control , and with a dip tube on each reactor to collect liquid samples durin g reaction . The end of the dip tube is fitted with a 2 µm stainless steel filter to avoid withdrawing solid catalyst along with liquid sample. To begin an experiment, e thanol and n - butanol were weighed out alone or in predetermined molar ratios and added to the reactor with a known amount of catalyst (Amberlyst 70 for heterogeneous catalysis and p - toluene sulfonic acid for homogeneous catalysis). The reactor was sealed and heated until it stabilized at the desired reaction temperature. Stirring was set to 800 rpm unless otherwise specified. Once the desired temperature was reached, a specified amount of butyric acid was added to the reactor through the sample port in a single shot ; the moment of addition was taken as time zero of the reaction. Total reactan t weight came up to approximately 0.04 0 kg per reaction. Samples of 0.5 1 mL were withdrawn at specified time intervals during the kinetic regime (0 - 6 hr) using a 3 mL Luer - lok tip syringes (Becton Dickson and Co., Franklin Lakes, NJ) and stored in her metically sealed vials in a standard refrigerator at 277 K. Samples to characterize reaction e quilibrium were taken 24 - 36 hours after the start of the reaction. 2.3.4 Analysis The initial water concentration in the reactants was determined by Karl Fischer titra tion in an Aquacounter coulometric titrator AQ - 2100 ® (JM Science Inc., Grand Island, NY) and taken into account for calculations. 11 Analysis of reaction samples was carried out in a Varian 450 gas chromatograph outfitted with a thermal conductivity detector (Varian Medical Systems Inc., Palo Alto, CA). Reaction samples were diluted 10 - fold in acetonitrile containing 11.11 wt. % ethyl octanoate as an internal standard. Separation was done o n a 0.53mm ID Aquawax - DA 30 m capillary column with 1.0µm film thicknes s. Helium carrier gas flow rate was set to 10 mL min - 1 . The following temperature program was used: initial column temperature 313 K for 2 min, ramp at 10 K min - 1 to 423 K, ramp at 30 K min - 1 to 503 K , hold 2 min. The detector temperature was held at 513 K . Standards of known composition in the range of interest were prepared and run in the chromatograph before and after reaction samples to calibrate the response factor of each component of the reaction . 2.4 Results A list of all experiments conducted, along wi th their conditions (temperature, initial reactant molar ratio, weight fract ion of catalyst) is given in Table 2 . 1 . Control experiments to investigate autocatalysis of the reaction showed negligible rates over the temperature range studied . Although etherification side reactions have been observed in other studies involving the acylation of n - butanol 7,13 and ethanol 8 at temperatures above 386 K, no ethers ( di - n - butyl ether, diethyl ether or ethyl n - butyl ether) wer e observed in the reaction samples in this study. This is because low loadings of catalyst (~0.1 w t. % p - TSA and 1 wt. % A - 70) , low ratios of alcohol to acid (~3:1 - 5:1), and relatively low temperatures were used in the experiments . 12 Table 2 . 1 : Summary of experiments and conditions No. Temperature Catalyst M olar feed ratio s Solution Density (C T ) (kmol m - 3 ) Catalyst loading (kg cat / kg soln) Figure No. ethanol: acid n - butanol : acid 1 60 A - 70 4.3 14.9 0.01 2.12 2 80 A - 70 2.8 15.0 0.01 2 .2 3 100 A - 70 8.7 15.0 0.0075 2.13 4 120 A - 70 6.6 14.9 0.0099 2.14 5 60 A - 70 3.7 10.9 0.0096 2 .2 6 80 A - 70 4.4 10.9 0.0101 2.15 7 100 A - 70 4.5 10.9 0.0101 2.16 8 120 A - 70 3.3 10.9 0.0115 2.17 9 60 A - 70 0.5 4.7 14.9 0.0109 2 .4 10 60 A - 70 2.6 2.6 12.8 0.0085 2.18 11 60 A - 70 0. 9 4.2 11.5 0.0099 2 .4 12 80 A - 70 1. 5 1.7 12.6 0.005 2.19 13 80 A - 70 4. 5 3. 9 12.9 0.0105 2 .4 14 80 A - 70 2.0 1. 9 12.6 0.0077 2.20 15 80 A - 70 0. 8 3.8 11.5 0.0093 2.21 16 80 A - 70 1. 9 2.2 12.6 0.0196 2.22 17 80 A - 70 0. 6 3.0 11.5 0.0097 2.23 18 60 p - TSA 2.9 14.8 0.0012 2.24 19 80 p - TSA 2. 8 14.9 0.0013 2.25 20 100 p - TSA 3.7 15.0 0.0012 2 .3 21 120 p - TSA 3.7 15.0 0.0013 2.26 22 60 p - TSA 2. 0 10.9 0.0013 2.27 23 80 p - TSA 3.0 10.9 0.0013 2 .3 24 100 p - TSA 2. 8 10.9 0.0014 2.28 25 120 p - TSA 3.7 10.9 0.0013 2.29 26 60 p - TSA 0. 4 1. 8 11.4 0.0012 2.30 13 27 60 p - TSA 3. 6 3.3 13.0 0.0013 2.31 28 60 p - TSA 5.0 1.0 14.4 0.0013 2.32 29 80 p - TSA 0.5 2.7 11.4 0.0013 2 .5 30 80 p - TSA 1. 9 1.9 12.6 0.0014 2 .5 31 80 p - TSA 3.7 0.9 14.0 0.0014 2 .5 14 2.4.1 Mass transfer considerations Accurate characterization of reaction kinetics requires that the experiments be conducted in the kinetic regime; i.e . at conditions where external and internal mass transfer resistances do not affect reaction rate. Preliminary experiments a t varying stirring speeds showed that conversion rates were unaffected above 600 rpm, implying that the external mass transfer resistances are negligible at a stirring speed of 800 rpm. To estimate the influence of intra - particular mass transfer resistance in this heterogeneous catalyst reaction , the Weisz - Prater criterion was used . 21 The observable modulus wa s first calculated (Eq. 2 . 1 ) . ( 2 . 1 ) where r* obs is the observed rate of reaction per weight of catalyst, CAT is density of catalyst (assumed to be 1000 kg m - 3 ), C BA is the liquid p hase concentration of butyric acid, d p is swelled diameter of catalyst at reaction conditions ( E q . 2 . 2 ) ( 2 . 2 ) where d p is 0.30 mm and V p,swollen /V p,dry is determined from a simple measurement to be 2.0 for both alcohols . The effective diffusivity D ef f of butyric acid in alcohol is estimated ( Eq . 2 . 3 ) , where pore tortuosity is assumed to be equ al to the invers e of particle porosity , and D BA is bulk diffusivity of butyric acid in alcohol estimated from the Wilke - Chang equation . 22 ( 2 . 3 ) 15 The Thiele modulus ( ) and effectiveness factor ( ) for butyric acid esterification are calculate d from the observable modulus assuming the reaction is pseudo first order in butyric acid ( i.e . excess alcohol) and thus . Values of were evaluated for butyric acid in each of the alcohols ( Table 2 . 2 ) and were found to be ~0.9 3 and ~0.96 for ethanol and n - butanol esterification, respectively. These values of indicat e that intra - particular resistances can be neglected . Table 2 . 2 : Estimated effectiveness factors for Amberlyst 70 catalyzed esterification reactions Run Temperature M olar feed ratio ethanol: acid n - butanol : acid 1 60 4.3 0.94 2 80 2.8 0.94 3 100 8.7 0.96 4 12 0 6.6 0.89 5 60 3.7 0.97 6 80 4.4 0.9 2 7 100 4.5 0.9 4 2.4.2 Reaction equilibrium constants Equilibrium constants were determined for each reaction by sampling the reactor contents after 24 to 48 hours of reaction. The activity - based equilibrium constan t K a,m for reaction m is given in Eq. 2 . 4 . ( 2 . 4 ) 16 Here x i , i , and v i represent the mole fraction, activity coefficient, and stoichiometric coefficient of component i in the reacti on mixtu re at equilibrium . The ratio of activity coefficients ( K , m ) accounts for deviations from ideal behavior ; values of activity coefficients were estimated using UNIFAC ( U NI versal Functional Activity Coefficient) . 23 The group type counts for each component in the reaction system, the group surface area and volume contributions, and the group binary interaction parameters are recorded in Table 2 . 3 , Table 2 . 4 , and Table 2 . 5 , respectively . Table 2 . 3 : UNIFAC groups and their counts in reaction components Component name Group Count butyric acid CH3 1 CH2 2 COOH 1 ethanol CH3 1 CH2 1 OH 1 ethyl butyra te CH3 2 CH2 2 CH2COO 1 water H2O 1 n - butanol CH3 1 CH2 2 OH 1 butyl butyrate CH3 2 CH2 4 CH2COO 1 Table 2 . 4 : Group volume and surface area contributions Group R k Q k CH3 0.9011 0.848 CH2 0.67 54 0.54 COOH 1.3013 1.224 CH2COO 1.6764 1.42 H2O 0.92 1.4 OH 1 1.2 17 Table 2 . 5 : UNIFAC group binary interaction parameters Name CH2 OH H2O CH2COO COOH CH2 0 986.5 1318 232.1 663.5 OH 156.4 0 353.5 101.1 199 H2O 300 - 229.1 0 72.87 - 14.09 CH2COO 114.8 245.4 200.8 0 660.2 COOH 315.3 - 151 - 66.17 - 256.3 0 The m ole - fraction based equilibrium constants ( K x , m ) were calculated from the composition of the reaction mixture. The experimental mole fraction - based a nd activity - based equilibrium constants for butyric acid with ethanol and n - butanol are presented in Figure 2 . 1 . The enthalpy of reaction r ), obtained from the slope of the trend line in Figure 2 . 1 , is 17 k J /mole for ethanol and 19 kJ /mole for n - butanol esterification of butyric acid. Although previous studies have reported temperature independent values of equil ibr ium constants in kinetic models 19 here t he data in Figure 2 . 1 were used to calculate temperature - dependent equilibrium constants in the kinetic model described below. Figure 2 . 1 - (gray) and activity - (black) based esterification equilibrium constants from experimental data. ( Ethanol , ( + , + ) n - butanol 18 2.4.3 Kinetic Model Description In a batch reactor, the change in number of moles of component i participating in M reactions can be expressed as ( 2 . 5 ) where is the total number of moles in the reactor, M is the number of reactions in the system, is the reaction volume, is the rate of reaction m per unit volume, and is the mole fraction of component i in the liquid mixture. The parameter is the ratio of stoichiometric coefficients of component i with respect to the reference component in reaction m . For this esterification system, Eq . 2 . 5 can be expressed in terms of total molar density of the liquid phase ( = N T /V ) because the total number of moles is conserved and rea ction volume is thus assumed constant during reaction. ( 2 . 6 ) With a neat mixture (e.g., no solvent) of reactants that constitute a non - ideal liquid phase, the r ate of formation of ester in reaction m, r m , - ideal concentration equal to C T x i i (kmol/m 3 ) representing the activity of each species in the rate expression. T his non - ideal concentrat ion has been previously defined and used for liquid phase reactions . 24,25 Its use is required here in order to compare butyric acid esterification rates in neat mixtures of different alcohols, because the usual form of thermodynamic 19 activity ( a i = x i i ) do es not account for differences in the overall solution density ( C T , Table 2 . 1 ) and thus differences in the absolute concentration of each species in the mixture for different alcohol species . T he rate of formation of ethyl butyrate ( EB ) in the batch reaction can thus be expressed in terms of the mole fraction s and activity coefficient s of butyric acid ( BA ), ethanol ( Eth ), ethyl butyrate , and water ( W ); w CAT , the catalyst loading in the reaction mixture ; C T , the tot al molar density of the reacting fluid; k 0,1 , the pre - exponential factor, and E a,1 , the activation energy of the rate constant for the reaction. ( 2 . 7 ) A similar expression is derived for the formation of n - butyl butyrate ( BB ) by esterification with n - butanol ( But ) . ( 2 . 8 ) U sing the rate expressions (Eq. 2 . 7 and 2 . 8 ), Eq. 2 . 6 ca n be written for every species in the reaction mixture to give six ordinary differential equations describing the esterification system. 2.4.4 Application of Kinetic Model to Single Alcohol Esterification Data from individual esterification reactions involvin g only one alcoholic species were first fit to obtain reliable kinetic models to describe their behavior , and then parameters obtained from individual esterification reactions were used to predict mixed alcohol esterification behavior. The set of ordinary differential equations (Eq. 2 .6) for each species in the reaction mixture were integrated numerically using the functions nlinfit and ode23 from the optimization toolbox in Matlab 7.12.0. Both non - activity - based and ideal con centration 20 (herein mole fraction - based all activity coefficients defined as unity) models were regressed. Optimization was done by minimizing F min (Eq. 2 . 9 ) defined as the sum of squared differences between calculated and experimental species mole fractions for all species in all esterification reaction s conducted . ( 2 . 9 ) In Eq. 2 . 9 , n is the number of experimental samples withdrawn in the experiments regressed, and N c is the number of reacting components in those experiments. Optimized kinetic parameters with 95% confidence limi ts are reported in Table 2 . 6 for both A - 70 and p - TSA catalysts . A ctivation energies are in the expected range of values (45 ± 10 k J mol - 1 ) for esterification of small aliphatic carboxylic acids with aliphatic alcoh ols. Absolute errors for each experiment were calculated (Eq. 2 . 10 ) and are presented i n Table 2 . 7 . ( 2 . 10 ) 21 Table 2 . 6 : Optimized kinetic parameters with 95% confidence limits and equilibrium constants from experimental data (T in Kelvin) Alcohol (Catalyst) Mole fraction model Acti vity model Ethanol ( A - 70 ) k 0 ( 3 (kg catalyst) - 1 s - 1 kmol - 1 ) 4.1±0.5×10 3 6.2±1.6×10 3 E a (kJ k mol - 1 ) K 1 45900±3400 exp( - 2119/T + 6.80) 47300±6100 exp( - 2219/T + 8.83) Rate constant value at 60 2.6 ±3.5 ×10 - 4 2.4 ±7.8 ×10 - 4 n - Butanol ( A - 70 ) k 0 ( 3 (kg catalyst) - 1 s - 1 km ol - 1 ) 6.8±1.4×10 3 11.9±0.5×10 3 E a (kJ k mol - 1 ) K 2 46800±290 exp( - 2339/T + 7.5) 48300±155 exp( - 1921/T + 8.8) Rate constant value at 60 3.1 ±0.3 ×10 - 4 3.2 ±0.04 ×10 - 4 Ethanol ( p - TSA) k 0 ( 3 (kg catalyst) - 1 s - 1 kmol - 1 ) 22.9 ±1.4×10 3 26.2±1.7×10 3 E a (kJ k mol - 1 ) 44900±182 45200±192 Rate constant value at 60 2.1 ±0.1 ×10 - 3 2.1 ±0.1 ×10 - 3 n - Butanol ( p - TSA) k 0 ( 3 (kg catalyst) - 1 s - 1 kmol - 1 ) 21.3 ±2.6 × 10 3 71.8 ± 6.8× 10 3 E a (kJ k mol - 1 ) 44900±375 48300±300 Rate constant value at 60 1.9 ± 0.2 ×10 - 3 1.9 ±0.3 ×10 - 3 22 Table 2 . 7 : Absolute errors in model fits for each experiment Run Temperature Catalyst Absolute Error Mole fraction F abs Activity F abs 1 60 A - 70 0.043 0.05 1 2 80 A - 70 0.045 0.045 3 100 A - 70 0.04 9 0.04 6 4 120 A - 70 0.05 1 0.05 3 5 60 A - 70 0.01 9 0.02 4 6 80 A - 70 0.03 1 0.0 60 7 100 A - 70 0.0 40 0.063 8 120 A - 70 0.077 0.11 9 60 A - 70 0.058 0.051 10 60 A - 70 0.044 0.045 11 60 A - 70 0.051 0.062 12 80 A - 70 0.071 0.082 13 80 A - 70 0. 068 0.069 14 80 A - 70 0.072 0.087 15 80 A - 70 0.098 0.099 16 80 A - 70 0.028 0.036 17 80 A - 70 0.042 0.048 18 60 p - TSA 0.048 0.048 19 80 p - TSA 0.065 0.065 20 100 p - TSA 0.083 0.083 21 120 p - TSA 0.046 0.046 22 60 p - TSA 0.055 0.095 23 80 p - TSA 0.107 0.06 9 24 100 p - TSA 0.117 0.084 25 120 p - TSA 0.134 0.080 26 60 p - TSA 0.052 0.188 27 60 p - TSA 0.029 0.035 28 60 p - TSA 0 .074 0.069 29 80 p - TSA 0.033 0.100 30 80 p - TSA 0.018 0.054 31 80 p - TSA 0.013 0.029 23 Select mole fraction profiles of activity - based and mole fraction based models fits with experimental data for in dividual ethanol and n - butanol esterification experiments are shown in Figure 2 . 2 : Experimental and predicted concentration profiles of ethanol and n - butanol individual esterification in the presence of 1 wt. % Amberlyst 70: a) Run 2 (Ethanol, T = 80 n - Butanol, T = 60 Activity ethyl butyrate; (×) n - butanol; - n - butyl butyrate. Figure 2 . 3 : Experimental and predicted concentration profiles of ethanol and n - butanol individual esterification in the presence of 0.1 wt. % p - TSA: a) Run 20 (Ethanol, T = 100 b) Run 23 ( n - Butanol, T = 80 Mole - fraction based fit Activity based ethyl butyrate; (×) water; n - - n - butyl butyrate. a) b) a) b) 24 Figure 2 . 2 for A - 70 catalyst and in Figure 2 . 3 for p - TSA. All remaining mole fraction profiles for both catalysts are shown in Figure 2 . 12 through Figure 2 . 32 in Appendix B and C . These plots indicate that the models fit experimental data reasonably well. The rate of esterification with homogeneous p - TSA catalyst is substantially high er than with heterogeneous A - 70 resin. Using catalyst loadings, the A - 70 acid site density of 2.35 eq H + /kg, and initial concentrations, the initial turn over number (TON, kmol BA/kmol H + /hr) for ethanol on p - TSA is 70.8 and on A - 70 is 16.8. For butanol, the initial TON is 40.7 on p - TSA and 11.3 on A - 70. The ratio of TON for the two catalysts is approximately four for both alcohols. We attribute the lower TON in A - 70 not to mass transport effects (see Table 2 . 2 ) but to steric e ffects associated with limited access of acid and alcohol to an anchored acid site in the porous solid catalyst versus the unhindered access acid and alcohol have to free acid in solution. The similarity in activation energies over the two catalysts suppor ts this postulate and that the reaction mechanism is the same on homogeneous p - TSA and heterogeneous A - 70. The absolute rate of butyric acid esterification (kmol/m 3 /sec) in ethanol is higher than in n - butanol for both catalysts . This is expecte d because t he absolute concentration (kmol/m 3 ) of ethanol in solution is higher than n - butanol, and thus the hydroxyl group concentration is higher. 26 However, in the rate expression using non - ideal concentration (C T x i i ) for each of the catalysts, the rate constants for ethanol and butanol esterification ( Table 2 . 6 ) have the same value within experimental uncertainty. The values given in Table 2 . 6 are thus for butyric acid esterification with alcohols over the respective catalysts, provided non - ideal concentration is used as the measure of species activity. This result is consistent with earlier results obtained in a broader study in our laboratory, 26 where a common rate constant for C2 to C8 25 alcohols was observed over A - 70 catalyst. Interestingly, methanol showed a substantially higher rate constant on the same basis. Figure 2 . 4 : Experimental and p redicted concentration profiles of mixed alcohol esterification in the presence of 1 wt. % Amberlyst 70. a) Run 13 ( Ethanol : n - Butanol = 1.15:1, T=80 o C); b) Run 11 (Ethanol: n - Butanol = 0.21:1, T = 60 o C ); c) Run 9 (Ethanol: n - Butanol = 1:1, T = 60 o ethyl butyrate; (×) n - - n - butyl butyrate. a) b) c) 26 Figure 2 . 5 : Experimental and predicted concentration profiles of mixed alcohol esterification in the presence of 0.1 wt. % p - TSA. a) Run 31 (Ethanol: n - Butanol = 4.1:1 , T = 80°C) ; b) Run 29 (Ethanol: n - Butanol = 0.18:1,T = 80 °C); c) Run 30 (Ethanol: n - Butanol = 1:1,T =80°C). ) ethanol; ethyl butyrate; (×) n - - n - butyl butyrate. 2.4.5 Application of kinetic model to mixed alcohol esterification The parameters from the activi ty and mole fraction based models from the two individual esterification studies were used to predict mixed alcohol behavior in initially ethanol rich, n - butanol rich and equimolar reactant mixtures. Figure 2 . 4 compares the predic tion of the two models with experimental data from mixed alcohol esterification runs for A - 70. Figure 2 . 5 does the same for p - TSA catalyzed mixed alcohol reactions. The additive combination of individual alcohol esterification rat e models predicts the mixed ethanol/ n - butanol esterification system well a) b) c) 27 for both catalysts . This result, observed over a wide range of alcohol molar ratios and supported by the common value of rate constant, is strong evidence that the alcohols are not co mpeting to adsorb on the acid catalyst sites in either solid Amberlyst 70 or in solution with p - toluenesulfonic acid. Rather, the alcohols behave similarly, and the rate determining step for esterification must be related to formation of an intermediate f rom butyric acid, or must take place apart from the active catalyst site. To the best of our knowledge, t here have been no previous experimental kinetic studies on mixed alcohol esterification using homogeneous or heterogeneous catalysis. However, a simi lar study examining mixed acid esterification found that the additive combination of both esterification reactions also predicted the kinetics of the mixed acid system well. 27 The ability to predict mixed alcohol or acid esterification rates based on indi vidual species rates, along with the commonality of rate constants for analogous species, greatly simplifies the characterization of complex esterification systems likely to be encountered in the bio - refinery, where either or both mixed alcohols and mixed acids are present. 2.5 Conclusions The kinetics of ethanol and n - butanol esterification of butyric acid in the presence of homogeneous p - toluene sulfonic acid and heterogeneous ion exchange resin catalyst Amberlyst 70 have been accurately described by ide al (mole fraction - based) and non - ideal (activity - based) models. Mixed alcohol esterification kinetics are accurately predicted from an additive combination of esterification rates of the individual alcohols, Further, by using non - ideal concentration as the measure of species activity, ethanol and n - butanol esterification kinetics are esterification over a given catalyst. The additivity of rates and common rat e constant greatly 28 simplify the simulation of actual biorefinery processes where multiple alcohol species are forming esters. 29 APPENDICES 30 Appendix A : Calibrati on plots for gas chromatography analysis The following plots ( Figure 2 . 6 through Figure 2 . 11 ) show calibrations of standards prepared to determine response factors of each component in the reaction system, determined by the slope of area ratio over weight ratio of component over int ernal standard ethyl caprylate. Figure 2 . 6 : Plot of area ratio vs weight ratio of ethanol over internal standard ethyl caprylate Figure 2 . 7 : Plot of a rea ratio vs weight ratio of ethyl butyrate over internal standard ethyl caprylate y = 0.99x R² = 0.9996 0 0.05 0.1 0.15 0.2 0.25 0.3 0 0.05 0.1 0.15 0.2 0.25 0.3 area ethyl butyrate/ area ethyl ccaprylate weight ethyl butyrate/ weight ethyl caprylate ethyl butyrate 31 Figure 2 . 8 : Plot of area ratio vs weight ratio of water over internal standard ethyl caprylate Figure 2 . 9 : Plot of area ratio vs weight ratio of butanol over internal standard ethyl caprylate y = 1.8046x R² = 0.997 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0 0.2 0.4 0.6 0.8 1 area water/ area ethyl caprylate weight water/ weight ethyl caprylate water y = 1.1271x R² = 0.9987 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 area butanol/ area ethyl caprylate weight butanol/ weight ethyl caprylate butanol 32 Figure 2 . 10 : Plot of area ratio vs weight ratio of butyl butyrate over interna l standard ethyl caprylate y = 0.9841x R² = 0.9983 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0 0.05 0.1 0.15 0.2 area butyl butyrate/ area ethyl caprylate weight butyl butyrate/ weight ethyl caprylate butyl butyate Figure 2 . 11 : Plot of area ratio vs weight ratio of butyric acid over internal standard ethyl caprylate y = 1.3175x R² = 0.9967 0 0.05 0.1 0.15 0.2 0.25 0.3 0 0.05 0.1 0.15 0.2 area butyric acid/ area ethyl caprylate weight butyric acid/ weight ethyl caprylate butyric acid 33 Appendix B. Reaction profiles of Amberlyst 70 catalyzed reactions Figure 2 . 12 : Experimental and predicted concentration profiles of ethanol individual est erification in the presence of 1 wt. % Amberlyst 70 : Run 1, T = 60 ( ) Mole - Activity ethanol; ( ethyl butyrate; (×) water Figure 2 . 13 : Experimental and predicted concentration profiles of eth anol individual este rification in the presence of 1 wt . % Amberlyst 70 : Run 3, T = 100 ( ) Mole - ethanol; ( ethyl butyrate; (×) water 34 Figure 2 . 14 : Experimental and predicte d concentration profiles of ethanol individual este rification in the presence of 1 wt. % Amberlyst 70 : Run 4, T = 120 ( ) Mole - Activity based fits. ethanol; ( ethyl butyrate; (×) water Figure 2 . 15 : Experimental and predicted concentration profiles of n - butanol individual este rification in the presence of 1 wt. % Amberlyst 70 : Run 6 , T = 8 0 ( ) Mole - Activi butyric acid; n - - n - butyl butyrate ; (×) water 35 Figure 2 . 16 : Experimental and predicted concentration profiles of n - butanol individual este rification in the presence of 1 wt. % Amberlyst 70 : Run 7, T = 100 ( ) Mole - n - - n - butyl butyrate ; (×) water Figure 2 . 17 : Experimental and predicted concentration profiles of n - butanol individual este rification in the presence of 1 wt. % Amberlyst 70 : Run 8, T = 120 ( ) Mole - n - - n - butyl butyrate ; (×) water 36 Figure 2 . 18 : Experimental and predicted concentration profiles of mixed alcohol esterification in the presence of 1 wt. % Amberlyst 70. Run 10 (Ethanol : n - Butanol = 1:1, T=8 0 o C ) mole fract ethyl butyrate; (×) n - - n - butyl butyrate Figure 2 . 19 : Experimental and predicted concentration profiles of mixed alcohol esterification in the presence of 0.5 wt. % Amberlyst 70. Run 1 2 (Ethanol : n - Butanol = 0.88:1, T=8 0 o C ) ethanol; ethyl butyrate; (×) ) n - - n - butyl butyrate 37 Figure 2 . 20 : Experimental and predicted concentration profiles of mixed alcohol esterification in the presence of 0.75 wt. % Amberlyst 70. Run 14 (Ethanol : n - Butanol = 1 .05:1, T=8 0 o C ) ethanol; ethyl butyrate; (×) n - - n - butyl butyrate Figure 2 . 21 : Experiment al and predicted concentration profiles of mixed alcohol esterification in the presence of 1 wt. % Amberlyst 70. Run 15 (Ethanol : n - Butanol = 0.21:1, T=8 0 o C ) e thyl butyrate; (×) n - - n - butyl butyrate 38 Figure 2 . 22 : Experimental and predicted concentration profiles of mixed alcohol esterification in the presence of 2 wt. % Amberlyst 70. Run 16 (Ethanol : n - Butanol = 0.86:1, T=8 0 o C ) ethyl butyrate; (×) n - - n - butyl butyrate Figure 2 . 23 : Experimental and predicted concentration profiles of mixed alcohol esterification in the presence of 1 wt. % Amberlyst 70. Run 17 (Ethanol : n - Butanol = 0.2:1, T=8 0 o C ) butyric a ethyl butyrate; (×) n - - n - butyl butyrate 39 Appendix C: Reaction profiles of p - TSA catalyzed reactions Figure 2 . 24 : Experimental and predicted co ncentration profiles of ethanol individual esterification in the presence of 0.1 wt. % p - TSA : Run 18, T = 60 ( ) Mole - fraction ethanol; ( ethyl butyrate; (×) water Figure 2 . 25 : Experimental and predicted concentration profiles of ethanol individual esterification in the presence of 0.1 wt. % p - TSA : Run 19, T = 80 ( ) Mole - fraction Activity based fits. ethanol; ( ethyl butyrate; (×) water 40 Figure 2 . 26 : Experimental and predicted concentration profiles of ethanol individual esterification in the presence of 0.1 wt. % p - TSA : Run 21, T = 120 ( ) Mole - fraction ethanol; ( ethyl butyrate; (×) water Figure 2 . 27 : Experimental and predicted concentrati on profiles of n - butanol individual esterification in the presence of 0.1 wt. % p - ( ) Mole - fraction butyric acid; n - - n - butyl butyrate; (×) water 41 Figure 2 . 28 : Experimental and predicted concentration profiles of n - butanol individual esterification in the presence of 0.1 wt. % p - ( ) Mole - fraction ) n - - n - butyl butyrate; (×) water Figure 2 . 29 : Experimental and predicted concentration profiles of n - butanol individual esterification in the presence of 0.1 wt. % p - ( ) Mole - fraction n - - n - butyl butyrate; (×) water 42 Figure 2 . 30 : Experimental and predicted concent ration profiles of mixed alcohol esterification in the presence of 0.1 wt. % p - TSA . Run 26 (Ethanol : n - Butanol = 0. 22 :1 , T=6 0 o C ) ethyl butyrate; (×) ) n - - n - butyl butyrate Figure 2 . 31 : Experimental and predicted concentration profiles of mixed alcohol esterification in the presence of 0.1 wt. % p - TSA . Run 27 (Ethanol : n - Butanol = 1 :1, T=6 0 o C ) mole ethyl butyrate; (×) n - - n - butyl butyrate 43 Figure 2 . 32 : Experimental and pred icted concentration profiles of mixed alcohol esterification in the presence of 0.1 wt. % p - TSA . Run 28 (Ethanol : n - Butanol = 5 :1, T= 6 0 o C ) mole ethyl butyrate; (×) n - - n - butyl butyrate 44 REFERENCES 45 REFERENCES (1) Iwasaki, T.; Maegawa, Y.; Ohshima, T.; Mashima, K. In Kirk - Othmer Encyclopedia of Chemical Technology ; John Wiley & Sons, Inc., 2000. (2) Ueda, W.; Ohshida, T.; Kuwabara, T.; Morikawa, Y. Catal. Lett. 1992 , 12 , 97 104. (3) Carlini, C.; Girolamo, M. D.; Macinai, A.; Marchionna, M.; Noviello, M.; Galletti, A. M. R.; Sbrana, G. J. Mol. Catal. Chem. 2003 , 204 205 , 721 728. (4) Carlini, C.; Macinai, A.; Raspolli Galletti, A. M.; Sbrana, G. J. Mol. Catal. Chem. 2004 , 212 , 65 70. (5) Mazanec, T. J. J. Catal. 1986 , 98 , 115 125. (6) Mills, G. A. Fuel 1994 , 73 , 1243 1279. (7) Xiang, M.; Li, D.; Xiao, H.; Zhang, J.; Qi, H.; Li, W.; Zhong, B.; Sun, Y. Fuel 2008 , 87 , 599 603. (8) Dry, M. E. Cata l. Today 2002 , 71 , 227 241. (9) Peralta - Yahya, P. P.; Keasling, J. D. Biotechnol. J. 2010 , 5 , 147 162. (10) Schulz, E.; Bauer, H.; Merscher, K. D. Coupled production of two esters. US7115773 B2, October 3, 2006. (11) Thurman, L. R.; Harris, J. B.; Mcate e, M. R. Coproduction of low molecular weight esters of alkanols. EP0260572 A1, March 23, 1988. (12) Warner, R. J.; Murphy, C. D.; Grajales, G. A. R.; Santiago, F. J. S.; Solorzano, F. A.; Torres, J. A. Process for the simultaneous coproduction and purifi cation of ethyl acetate and isopropyl acetate. US6765110 B2, July 20, 2004. (13) Acker, P. E. van; Mathieu, O.; Milner, R. J.; Pacynko, W. F. Ester co - production. US6093845 A, July 25, 2000. (14) Harmsen, G. J. Chem. Eng. Process. Process Intensif. 2007 , 46 , 774 780. (15) Stankiewicz, A. Chem. Eng. Process. Process Intensif. 2003 , 42 , 137 144. (16) Lee, H. - Y.; Yen, L. - T.; Chien, I. - L.; Huang, H. - P. Ind. Eng. Chem. Res. 2009 , 48 , 7186 7204. (17) Zhang, C.; Yang, H.; Yang, F.; Ma, Y. Curr. Microbiol. 200 9 , 59 , 656 663. (18) Armstrong, D. W.; Yamazaki, H. Trends Biotechnol. 1986 , 4 , 264 268. (19) Ju, I. B.; Lim, H. - W.; Jeon, W.; Suh, D. J.; Park, M. - J.; Suh, Y. - W. Chem. Eng. J. 2011 , 168 , 293 302. 46 (20) Amber lyst 70 product data sheet, 2005. (21) Weisz, P. B.; Hicks, J. S. Chem. Eng. Sci. 1995 , 50 , 3951 3958. ( 22) Wilke, C. R.; Chang, P. AIChE J. 1955 , 1 , 264 270. (23) Hansen, H. K.; Rasmussen, P.; Fredenslund, A.; Schiller, M.; Gmehling, J. Ind. Eng. Chem. Res. 1991 , 30 , 2352 2355. (24) Eckert, C. A. ; Hsieh, C. K.; McCabe, J. R. AIChE J. 1974 , 20 , 20 36. (25) Eckert, C. A. Ind. Eng. Chem. 1967 , 59 , 20 32. (26) Pappu, V. K. S.; Kanyi, V.; Santhanakrishnan, A.; Lira, C. T.; Miller, D. J. Bioresour. Technol. 2013 , 130 , 793 797. (27) Orjuela, A.; Yanez , A. J.; Santhanakrishnan, A.; Lira, C. T.; Miller, D. J. Chem. Eng. J. 2012 , 188 , 98 107. 47 3 Unification of esterification rates using non - ideal concentration model 3.1 Abstract A number of liquid phase esterification systems from the literature with differen t carboxylic acids, alcohols and acid catalysts (homogeneous and heterogeneous) have been considered in the framework of a non - ideal concentration model to ascertain if an estimate of most esterification reaction rate s is possible from the data already ava ilable. Trends observed from comparing the non - ideal concentration model - based rate constants of five different types of reaction families are reported. Different primary alcohols of varying carbon chain lengths were found to have similar rate constants. T he same was observed for different primary carboxylic acids. For branched alcohols and carboxylic acids, the rates were lower than the rates of their primary counterparts, but within a level of branching, a group of reactants have comparable rate constants . Rate constants of reactions involving the same reactants but different acid catalysts, homogeneous and heterogeneous, were also examined. It was found that the ratios of the rate constants of reactions of two different catalysts is the same for multiple reaction sets involving the same reactants. This could be an effective test for similarity of mechanisms between two catalysts and offers the possibility of estimating the rate of one catalyst provided the rate of another and properties of both. The Taft e quation is related to non - ideal concentration based rates. 3.2 Introduction There is a preponderance of literature on acid catalyzed esterification reactions. As an increasing number of esters are becoming commercially important chemicals 1 , the possibility of manufacturing multiple esters with product distributions depending on market needs is being 48 explored. 2 7 Carboxylic acids and alcohols form a significant portion of biomass derived esters for use as biofuel additives, green solvents, and food flavoring ag ents. There is therefore a need to gain a deeper understanding of the factors that affect the rate of reaction to facilitate flexible process design. Multiple attempts have been made to determine commonalities in rates of similar reaction systems, whether from investigating structure - reactivity relationships 8 18 or by examining the surface reaction mechanism involved 19 . Conventionally, kinetics of heterogeneously catalyzed esterification reactions are reported as either concentration or activity based mod els, sometimes with modifications for local sorption equilibria at the surface of the catalyst due to swelling, 20,21 and for adsorption and desorption at the surface, such as Eley - Rideal and Langmuir Hinshelwood mechanisms. 22 Activity based models are usua lly found to predict kinetic behavior as well as or better than purely concentration based models, as they account for deviations from ideality of the reaction mixture. Activity based rate expressions for pseudo - homogeneous models typically take the form: (3. 1 ) Activities here are calculated based on mole fractions and activity coefficients estimated by a suitable model ( ). In designing processes where similar initial compositions will be used as those used in the kinetic study, such as when reactant ratios are limited to those involving a large excess of alcohol due to solubility limitations, this expression has worked well. 23,24 This is because the total solution density does not vary appreciably and a composite rate constant may be used without error. 49 Ideally, kinetic parameters reported should work for any composition, tempera ture and catalyst concentration. To overcome the limitation of kinetic rate expressions such as that - . 3.2 below) that accounts for both non - ideality and variations in initial molar c oncentrations for a wide range of initial compositions of its reactants. The non - ideal concentration model reflects the reduced number of moles per volume of higher molecular weight reactants for the same molar reaction mixture composition, and enables us to make some observations on structure - reactivity relationships as well as on reaction mechanisms. The inclusion of solution density makes the rate dependent on the absolute concentration of the alcoholic and carboxylic species ( - OH and COOH) in the solut ion. The need for this became more noticeable in a previous work conducted to study the behavior of two alcohols, ethanol and butanol , reacting simultaneously with butyric acid, where it became important to also distinguish between cases where the ratio o f alcohol to acid was the same but the ratio of the individual alcohols was different (2 moles of ethanol and 1 mole of butanol vs 1 mole of ethanol and 2 moles of butanol). 25 With the non - ideal concentration model, both ethanol and n - butanol were found to have the same rate constant within experimental uncertainty. This has prompted a more expansive study to look at other esterification systems in the literature. Rather than deal with a rigorous derivation of activities from molarities instead of mole fra ctions, which would involve calculation of excess volume dependency on composition, one can still empirically represent varying initial effective functional group concentrations by an expression of the form in Eq. 3. 2 : (3. 2 ) 50 It should be noted that, fortuit ously, this empirical modification works for reactions in which the total number of moles on the reactant and product sides is the same (in this case, two). If this were not the case, it would not be possible to factorize out C T , leading to less elegant ex pressions, with the value of C T changing through the course of the experiment. Equilibrium constants would have to be defined in terms of effective molarities. Nevertheless, this expression accounts for both non ideality and variations in concentrations, a nd has been used before. 26 28 Catalyst weight fraction w CAT is converted to molar hydronium concentration [H + ] to enable comparisons between different reactions. 3.3 Calculation procedure A total of 86 esterification systems from our laboratories and from the literature consisting of various alcohols, acids and catalysts have been c onsidered. The vast majority of the kinetic studies surveyed did not consider changes in (actual species or functional group concentrations) , leading to reported values being a comp osite of the reactant ratios used in each individual study. Calculations of non - ideal rate constants were therefore done from initial stages of reaction . A reference experiment at 60°C was chosen for each reaction system . Initial rate s w ere calculated base d on a first or second order fit of the first several data points extrapolated to time=0 from available information (graphs of conversion or reaction profile data ) . Confidence intervals were calculated based on the fits using MATLAB R2013a and are reporte d (Table 3.1 in Section 3 .4). NRTL - HOC was used to find the activity coefficients of the acid and alcohol where binary interaction parameters were available from literature (accessed via ASPEN v8.7); in the absence of binary interaction parameters, UNIFAC was used to estimate activity coefficients . Activity coefficients for both models were calculated using a MATLAB program developed by Dr. Carl T. 51 Lira of the department of Chemical Engineering and Materials Science at Michigan State University. The metho d used for calculating activity coefficients is reported (Table 3.1 in Section 3.4). NRTL - HOC is a modification of the conventional NRTL model that is recommended for systems that include carboxylic acids, and accounts for dimerization of carboxylic acids in the vapor phase. A comparison of activity coefficient values calculated by NRTL - HOC and UNIFAC for several reaction systems is reported (Table 3. x ) . It should be noted that only reaction systems which ha ve data available for both NRTL - HOC and UNIFAC are included. The difference in activity coefficients of alcohols between the two models ran ges from 1% to 15%. The average difference for propanol is much higher than for ethanol and butanol. In the case of carboxylic acids, the average difference between the two models varies by 8 - 10% . More data is required to ascertain the suitability of UNIFA C for carboxylic acids with non - linear structures. Table 3 . 1 : Comparison of activity coefficient values estimated by NRTL - HOC and UNIFAC Reaction system: NRTL - HOC activity coefficients UNIFAC Alcohol Carboxy lic acid Catalyst Alcohol Carboxylic acid Alcohol Carboxylic acid alcohol acid alcohol acid Methanol Butyric Acid Amberlyst 70 0.9902 1.0193 0.9890 1.0980 Methanol Butyric Acid Amberlyst 15 0.9902 1.0193 0.9890 1.0980 Methanol Butyric Acid Amber lyst 36 0.9902 1.0193 0.9890 1.0980 Methanol Acetic Acid Amberlyst 15 0.9524 0.9894 0.9561 0.9506 Methanol Acetic Acid HI 0.9621 0.9780 0.9680 0.9374 Butanol Acetic Acid Smopex - 101 0.9805 1.0233 0.9888 1.0302 Ethanol Acetic Acid Smopex - 101 0.9944 1.005 4 0.9348 0.9509 Methanol Acetic Acid Smopex - 101 0.9530 0.9888 0.9568 0.9499 Ethanol Acetic Acid Amberlyst 15 0.9944 1.0054 0.9348 0.9509 52 Methanol Propionic Acid Smopex - 101 0.9652 1.0087 0.9074 1.0340 Ethanol Propionic Acid Smopex - 10 1 0.9970 0.9977 0. 9956 1.0060 Propanol Acetic Acid Smopex - 101 0.9867 1.0127 0.9373 0.9524 Propanol Propionic Acid Smopex - 101 0.9833 0.9933 0.8317 0.8895 Methanol Propionic Acid Dowex 50Wx8 - 400 0.9652 1.0087 0.9074 1.0340 Ethanol Propionic Acid Dowex 50 Wx8 - 400 0.9970 0.9977 0. 9947 0 .9962 Propanol Propionic Acid Dowex 50Wx8 - 400 0.9833 0.9933 0.8317 0.8895 Methanol Acetic Acid SAC - 13 0.9757 0.9562 0.9816 0.9166 Methanol Propionic Acid SAC - 13 0.9783 0.9890 0.9538 0.9632 Methanol Butyric Acid SAC - 13 0.98 73 1.0270 0.9833 1.1137 Methanol Acetic Acid s ulfuric Acid 0.9757 0.9562 0.9816 0.9166 Methanol Propionic Acid s ulfuric Acid 0.9783 0.9890 0.9538 0.9632 Methanol Butyric Acid s ulfuric Acid 0.9873 1.0270 0.9833 1.1137 Methanol Acetic Acid HC l 0.9999 0.8 480 0.9999 0.8431 Methanol Propionic Acid H Cl 0.9998 0.8792 0.9998 0.7734 Methanol Butyric Acid H Cl 0.9998 0.9448 0.9999 1.0197 Butanol Acetic Acid Amberlyst 15 0.9805 1.0233 0 .9888 1.0302 Methanol Propionic Acid Amberlyst 15 0.9652 1.0087 0.9074 1.034 0 Methanol Propionic Acid Smopex - 101 0.9652 1.0087 0.9074 1.0340 Densities and van der Waal volumes of compounds were obtained from DIPPR. A sample cal culation is shown for the system 2 - propanol - acetic acid in presence of Amberlyst 36 in Appendix D . 19 3.4 Results The results of non - ideal concentration rate analysis for several reactions computed from literature data are reported ( Table 3 . 2 ). Families of reactions where one component was varied 53 while keeping the other two constant ( alcohol, carboxylic acid, acid catalyst) were used to study the trends of rate constants to shed light on structure - reactivity relationships after accounting for differences in solution density and non - ideal behavior. Table 3 . 2 : Summary of reactions and calculated non - ideal concentration - based rate constants at 60 No. Alcohol Carboxylic acid Catalyst Rate constant k NIC (m6 soln/(kmole soln) · (kmole H+)·s) Ref. 1 Methanol Acetic Acid Amberlyst 15 2.74±1.9E - 04 20 2 2 - Butanol Butyric Acid Amberlyst 70 6.09±0.20E - 05 29 3 2 - Butanol Propionic Acid Dowex 50W - 4 7.72 ±1.10E - 05 34 4 2 - Butanol Propionic Acid Smopex - 101 2.69±0.32E - 05 34 5 2 - Ethyl Hexanol Butyric Acid Amberlyst 70 1.32±0.16E - 04 29 6 2 - Propanol Acetic Acid Amberlite IRA 120 6.09±0.60E - 05 19 7 2 - Propanol Acetic Acid Amberlyst 15 6.22±0.20E - 05 19 8 2 - Pro panol Acetic Acid Dowex 50Wx8 - 400 1.36±0.80E - 04 19 9 2 - Propanol Propionic Acid Dowex 50W - 4 1.00±0.09E - 04 34 10 2 - Propanol Acetic Acid Smopex - 101 5.62±0.20E - 05 33 11 2 - Propanol Pentanoic Acid Smopex - 101 1.74±0.33E - 05 33 12 2 - Propanol Propionic Acid Smop ex - 101 2.45±0.31E - 05 33 13 4 - Heptanol Butyric Acid Amberlyst 70 1.09±0.51E - 04 29 14 4 - Heptanol Butyric Acid p - TSA 5.88±2.79E - 04 29 15 Amyl Alcohol Acetic Acid Amberlite IRA 120 1.41±0.07E - 05 32 16 Amyl Alcohol Acetic Acid Amberlyst 15 1.86±0.10E - 05 32 17 Amyl Alcohol Acetic Acid Amberlyst 36 1.71±0.04E - 05 32 18 Benzyl Alcohol Acetic Acid Amberlyst 15 7.57±0.81E - 05 38 19 Butanol Butyric Acid Amberlyst 15 5.59±0.43E - 05 29 20 Butanol Butyric Acid Amberlyst 36 3.74±0.62E - 05 29 21 Butanol Butyric Acid A mberlyst 70 1.80±0.16E - 04 29 22 Butanol Butyric Acid Amberlyst 70 3.1±0.30E - 04 25 23 Butanol Propionic Acid Dowex 50W - 4 3.26±1.40E - 04 34 24 Butanol Butyric Acid p - TSA 1.9±0.20E - 03 25 25 Butanol Acetic Acid Smopex - 101 1.08±0.28E - 04 33 26 Butanol Pentan oic Acid Smopex - 101 5.74±0.81E - 05 33 54 27 Butanol Propionic Acid Smopex - 101 7.70±6.40E - 05 33 28 Ethanol Acetic Acid Amberlyst 15 9.44±0.26E - 05 21 29 Ethanol Butyric Acid Amberlyst 15 6.02±0.32E - 05 29 30 Ethanol Butyric Acid Amberlyst 36 5.87±0.40E - 05 29 31 Ethanol Butyric Acid Amberlyst 70 1.73±0.44E - 04 29 32 Ethanol Butyric Acid Amberlyst 70 2.60±3.50E - 04 25 33 Ethanol Propionic Acid Dowex 50W - 4 2.96±1.40E - 04 34 34 Ethanol Butyric Acid p - TSA 2.14±0.24E - 04 25 35 Ethanol Acetic Aci d Smopex - 101 3.88±0.45E - 04 33 36 Ethanol Pentanoic Acid Smopex - 101 8.21±2.80E - 05 33 37 Ethanol Propionic Acid Smopex - 101 1.65±0.77E - 04 33 38 Ethylene Glycol Acetic Acid Amberlyst 36 3.02±0.07E - 05 39 39 Ethylene Glycol Mono Acetate Acetic Acid Amberlyst 36 6.58±0.41E - 04 39 40 Heptanol Butyric Acid Amberlyst 70 9.76±14.8E - 05 29 41 Heptanol Butyric Acid p - TSA 9.61±6.95E - 04 29 42 Isobutanol Butyric Acid Amberlyst 15 5.81±0.22E - 05 29 43 Isobutanol Butyric Acid Amberlyst 70 9.66±0.39E - 05 29 44 Isobutanol Propionic Acid Dowex 50W - 4 1.12±0.31E - 04 22 45 Methanol Butyric Acid Amberlyst 15 2.27±0.68E - 04 29 46 Methanol Butyric Acid Amberlyst 36 1.49±0.45E - 04 29 47 Methanol Butyric Acid Amberlyst 70 4.29±0.75E - 04 29 48 Methanol Propionic Acid Dowex 50W - 4 4.9 0±0.38E - 04 34 49 Methanol Formic Acid HCl 6.70±0.11E - 02 30 50 Methanol Acetic Acid HCl 1.70±0.03E - 02 30 51 Methanol Propionic Acid HCl 1.56±0.04E - 02 30 52 Methanol Butyric Acid HCl 6.82±0.32E - 03 30 53 Methanol Pentanoic Acid HCl 6.95±0.16E - 03 30 54 M ethanol Octanoic Acid HCl 7.12±0.19E - 03 30 55 Methanol Nonanoic Acid HCl 7.19±0.19E - 03 30 56 Methanol Lauric Acid HCl 7.53±0.19E - 03 30 57 Methanol Trimethyl Acetic Acid HCl 6.74± 1.12 E - 04 31 58 Methanol Diethyl Acetic Acid HCl 2.04±0. 95 E - 04 31 59 Metha nol Dipropyl Acetic Acid HCl 1.91±0.10 E - 04 31 60 Methanol Dibutyl Acetic Acid HCl 1.82±0.39 E - 04 31 61 Methanol Diisobutyl Acetic Acid HCl 9.01± 2.38E - 05 31 55 Table 3. 2 62 Methanol Betamethyl Valeric Acid HCl 1.78 ±1.07 E - 03 31 63 Methanol Acetic Acid SAC - 13 2.83±0.17 E - 04 35 64 Methanol Propionic Acid SAC - 13 1.58±0.09E - 04 35 65 Methanol Butyric Acid SAC - 13 8.87±0.53E - 05 35 66 Methanol Hexanoic Acid SAC - 13 8.69±0.52E - 05 35 67 Methanol Caprylic Acid SAC - 13 6.51±0.41E - 05 35 68 Methanol Acetic Acid Smopex - 101 8.58±0.1 3E - 04 33 69 Methanol Pentanoic Acid Smopex - 101 1.49±0.17E - 04 33 70 Methanol Propionic Acid Smopex - 101 4.50±2.60E - 04 33 71 Methanol Acetic Acid sulfuric acid 1.92±0.09E - 02 35 72 Methanol Propionic Acid sulfuric acid 8.95±0.05E - 03 35 73 Methanol Butyric Acid sulfuric acid 5.10±0.30E - 03 35 74 Methanol Caprylic Acid sulfuric acid 8.33±0.41E - 03 35 75 Methanol Hexanoic Acid sulfuric acid 6.95±0.49E - 03 35 76 Octanol Butyric Acid Amberlyst 70 9.00±10.8E - 05 29 77 Amyl Alcohol Propionic A cid Dowex 50Wx8 - 400 2.88±0.17E - 04 34 78 Propanol Butyric Acid Amberlyst 15 5.69±0.22E - 05 29 79 Propanol Butyric Acid Amberlyst 36 4.05±0.88E - 05 29 80 Propanol Butyric Acid Amberlyst 70 1.39±0.08E - 04 29 81 Propanol Propionic Acid Dowex 50W - 4 3.32±1.60E - 04 40 82 Propanol Acetic Acid Smopex - 101 1.32±0.65E - 04 33 83 Propanol Pentanoic Acid Smopex - 101 7.53±0.52E - 05 33 84 Propanol Propionic Acid Smopex - 101 9.64±2.40E - 05 33 85 Propylene Glycol Acetic Acid p - TSA 2.70±0.19E - 3 41 86 Propylene Glycol monoaceta te Acetic Acid p - TSA 2.40±0.21E - 3 41 3.4.1 Rate constants for straight chain alcohols with common acid and catalyst With the exception of methanol, varying the length of carbon chain for ethanol and higher alcohols does not change the rate constant significant l y ( Figure 3 . 1 , Figure 3 . 2 ). As higher alcohols such as heptanol and octanol are used at low alcohol:acid ratios (~3:1), the rate constant slightly drops due to phase separa tion of the heavier esters formed from water. 29 The rate constant of methanol for both homogeneous and heterogeneous catalysis is considerably higher (2 - 2.5 times) 56 than the rate constants of C2 - C8 linear primary alcohols, as might be expected because of t he lack of steric hindrance from alkyl groups. Figure 3 . 1 : Non ideal rate constants at 60 - - Butyric acid, Amberlyst 15 (×) - Butyric acid, Amberlyst - Propionic acid, Dowex W50×4 Figure 3 . 2 : Non ideal rate constants at 60 number of carbons in primary alcohol chain with common carboxylic acid and - Acetic acid, Smopex - - Prop ionic acid, Smopex - - Pentanoic acid, Smopex - 101 57 3.4.2 Rate constants for secondary alco hols with common acid and catalyst Branched alcohols react slower than their primary isomers as expected due to increased steric hindrance ( F igure 3 . 3 ) . Within a level of branching the rates are similar, as seen with rates of acylation of 2 - propanol and 2 - butanol with propionic acid in presence of Dowex Wx4 - 500, and propionic acid with Smopex - 101. Smopex - 101 is a fibrous polymer supported sulfo nic acid catalyst with a site concentration of ~3.2 equivalents per gram conventionally used as a metal scavenger. Isobutanol is not as sterically hindered as 2 - butanol and has a higher rate constant, as is the case for 2 - ethylhexanol and 4 - heptanol. F igure 3 . 3 : Non ideal rate constants at 60 alcohols with common carboxylic acid and catalyst. ( ) - Propionic acid, Dowex - - Propionic acid, Smopex - 101 3.4.3 Different primary carboxylic acid s esterified by common alcohol and catalyst Non ideal rate constants for different primary acids reach a plateau for acids higher than propionic acid for both homogeneous and heterogeneous acids ( Figure 3 . 4 , Figure 3 . 5 ). Formic acid reacts over 10 times faster than acetic or propionic acid with methanol (rate constant value of 0.19 (units), not shown in Figure 3 . 5 ) 30 . In the p resence of a strong mineral acid, the dissociation 58 of formic acid is greatly suppressed, so the increase in rate may be attributed to reactivity differences and not to significant autocatalysis. Figure 3 . 4 : Non ideal rate constants at 60 number of carbons in primary carboxylic acid chain with common alcohol and - Methanol, SAC - - 2 - Propanol, Smopex - 101 Figure 3 . 5 : Non ideal rate con stants at 60 number of carbons in primary carboxylic acid chain with common alcohol and catalyst. ( ) - Methanol, Sulfuric acid ( ) - Methanol, Hydrochloric acid 3.4.4 Different branched carboxylic acids, same alcohol, sa me catalyst Non ideal rate constants for carboxylic acids increase with the increase in branching. However, increasing the length of those branched chains does not have a significant steric effect as seen in the rate constants of diethyl acetic acid, dipro pyl acetic acid and dibutyl acetic acid ( Figure 3 . 6 ) . 31 59 Figure 3 . 6 : Non ideal rate constants at 60 differing number of carbons in branched carboxylic acid chain with methanol and hydrochloric acid catalyst 60 Table 3 . 3 : Structures of branched carboxylic acids in Figure 3 . 6 . beta - Methylvaleric acid Trimethylacetic acid Diethyl acetic acid Dipropyl acetic acid Dibutyl acetic acid Diisobutyl acetic acid 3.4.5 Ratio of rate constants of the same reaction with different catalysts Several example s of similar reactions carried out with different catalysts (homogeneous and heterogeneous) show that the ratio of turn over numbers (TON) for two catalysts is the same irrespective of the reaction. For example, methanol esterification of various acids sho ws the same turn - over ratio for sulfuric acid and SAC - 13( Table 3 . 4 ). 61 Table 3 . 4 : Ratio of rate constants at 60 - ideal concentration model Alcohol Carbo xylic acid Ratio of rate constants of two catalysts for different alcohols and carboxylic acids at 60 non - ideal concentration model Ref. Amberlyst 36/Amberlyst 15 Methanol Butyric acid 0.66 29 , 29 Ethanol Butyric acid 0.98 29 , 29 Propanol But yric acid 0.70 29 , 29 Butanol Butyric acid 0.68 29 , 25 Amyl Alcohol Acetic acid 0.92 32 , 32 Smopex - 101/Dowex W×4 2 - Butanol Propionic acid 2.87 33 , 34 2 - Propanol Propionic acid 4.09 33 , 34 Butanol Propionic acid 4.24 33 , 34 Ethanol Propionic acid 1.79 33 , 3 4 Propanol Propionic acid 3.45 33 , 34 Methanol Propionic acid 1.31 33 , 34 SAC - 13/sulfuric acid Methanol Acetic acid 0.34 35 , 35 Methanol Propionic acid 0.28 35 , 35 Methanol Butyric acid 0.40 35 , 35 Methanol Hexanoic acid 0.22 35 , 35 Methanol Octanoic aci d 0.17 35 , 35 Amberlyst 70/ p - TSA 4 - Heptanol Butyric acid 0.18 29 , 29 Butanol Butyric acid 0.20 29 , 29 Butanol Butyric acid 0.17 29 , 25 Ethanol Butyric acid 0.24 29 , 29 Ethanol Butyric acid 0.19 29 , 25 Amberlyst 70/Amberlyst 15 Ethanol Butyric acid 3.01 2 9 , 29 Ethanol Butyric acid 2.40 29 , 25 Propanol Butyric acid 2.44 29 , 29 Butanol Butyric acid 2.96 29 , 29 Butanol Butyric acid 3.46 29 , 25 The same is observed in the case of ethanol and butanol esterifying butyric acid in presence of Amberlyst 70 and para - Toluene sulfonic acid ( p - TSA). Both cases indicate a similar reaction 62 mechanism for both homogeneous and the particular heterogeneous catalysts . The higher rate constant of p - TSA over Amberlyst 70 may be explained by the increased steric hindrance to the protonation of carboxylic acids by the bulky polymer backbone of Amberlyst 70 compared to acid sites in free solution. This does not however explain the higher rate constants of Amberlyst 70 over Amberlyst 15. Amberlyst 15 has a higher cation exchange ca pacity, higher surface area and bigger pores. 42 However , the activity of Amberlyst 70 is higher per acid site (Table 3.3) . This may be due to a combination of the effect of resin swelling and the fact that the acid sites are not saturated, i.e there are at any instant of time, enough available protonated carboxylic acid sites for alcohol molecules that diffuse into the catalyst. Resin catalysts are known to swell in the presence of certain polar solvents such as alcohols. Aliphatic and aromatic alcohols are able to diffuse through the bulk of the catalyst structure, thus making porosity not as important a factor as acid site strength. 43 The polymer backbone of Amberlyst 70 is halogenated. The presence of electron withdrawing groups is known to enhance the ac id strength of Amberlyst 70 acid sites , making the turn over frequency of Amberlyst 70 higher . 4 4 In reactions with non - swelling solvents, one might expect the activity of Amberlyst 15 to be higher than Amberlyst 70. 42 The activity of Smopex - 101, a fibrou s rather than resin catalyst with a lower acid site concentration with the same active group as Amberlyst 15 shows a higher activity. In this case, higher accessibility of acid sites in the fibrous catalyst rather than polymer resin may be said to cause th e higher activity of Smopex - 101. 63 3.5 Discussion The non - ideal concentration model has been previously evaluated 25 for esterification reactions in Chapter 2 and is a useful method of reporting rate constants for maximum flexibility in reaction conditions of t emperature, initial reactant composition and catalyst concentration. It has now been applied to several reactions with a view of comparing different reactants after normalizing for differences in solution density and accounting for non - ideal behavior. For a majority of reactions, with the exception of methanol and carboxylic acids smaller than butyric acid, increasing the number of alkyl groups in a carbon chain of either reactant does not have a significant polar or steric effect on reactivity. R easonable estimates of rate constants for new reactions of interest involving several primary and secondary alcohols and carboxylic acids may be obtained from the data already available. The variation in rate constant values within the plateau region of reactivity f or each reaction family ranges from 1 to 3% of the mean value, except for acetic acid esterification in the presence of Smopex - 101, which had a n error of 12% of the mean value. This is unsurprising as the confidence limits for reactions with Smopex - 101 hav e relatively larger values (Table 3.1). The activation energies for a wide variety of carboxylic acids and alcohols is 50±10 kJ/mol, meaning the maximum error in the estimate of the rate constant is within 50% of its true value in the temperature range of 25 This predictive capability enables the easy design of parallel reaction systems with multiple alcohols streams and multiple carboxylic acid streams commonly found in biomass conversion processes. Calculating the ratio of rate constants for tw o different catalysts for a number of reactions offers a simple way of evaluating similarity of mechanism between the two catalysts, and a number of such pairs of catalysts have been presented. 64 Esterification and hydrolysis reactions are considered to be predominantly sterically controlled. 13 15,36 The rate limiting step involves the attack of the alcohol on the carbonyl carbon of the protonated carboxylic acid. The reactivity of the carboxylic acid depends on the electrophilicity of the partially positiv e carbonyl carbon. Alkyl groups added to the carbon chain have an electron donating effect, effectively reducing the electrophilicity of the carbonyl carbon. The carbonyl carbon is sp 2 hybridized and the angle of attack of the nucleophile is affected by tw o factors. First, the repulsion of the high electron density region on the carbonyl oxygen atom, and second, the steric hindrance and repulsion of the OH and R group attached to the trigonal carbon atom. If portions of the carboxylic acid offer any steric hindrance to the angle of attack, the rate is reduced. Formic acid, having only a hydrogen atom, not only does not donate electron density to the carbonyl carbon, it has virtually no repulsive influence on the nucleophile, giving the alcohol a large geomet ric region of possible attack, thereby increasing the fraction of fruitful collisions between itself and formic acid. The methyl group on acetic acid donates electron density to the carbonyl carbon, and also reduces the region of successful attack of the n ucleophile. The inductive effect drops in strength with increase in chain length, and steric effects dominate the reactivity of the bigger propionic acid. Butyric acid and higher acids show little difference in reactivity. The reactivity of the alcohol dep ends on its ability to attack the protonated carboxylic acid. Methanol, having only one alkyl carbon, offers the least steric hindrance and therefore has more success than ethanol and higher alcohols. Little information was found on the structure - reactivi ty relationship of polyols such as ethylene glycol, propylene glycol, sorbitol, and glycerol which form an important portion of useful intermediates from biomass breakdown processes. However, the r esults of the study in Chapter 4 of propylene glycol esteri fication in the presence of p - TSA when normalized to the increased 65 concentration of hydroxyls in a diol, are close to the expected rate constant value for simple esterifications in p - TSA. Dicarboxylic acid esterification reactions have also not been consi dered as they are strongly autocatalytic reactions even at room temperature and expressing catalyst concentration in the kinetic equation presents a challenge for most systems involving them. Additionally, obtaining accurate kinetic data for alcohol - dicarb oxylic acid mixtures is made difficult by the almost instantaneous conversion of significant amounts of both to ester product. Although the non - ideal concentration model works for a wide range of reaction conditions, in some cases where the reaction is co nducted in a regime where the carboxylic acid is in excess of the alcohol and plays an important role in the catalysis, it fails to predict rates accurately irrespective of the activity coefficient model used when the reaction is started in excess ester or excess water. 37 3.5.1 Taft equation Linear free energy relationships have been used extensively to quantify the link between the structure of a reactant in a reaction and the rate of the reaction. The Taft equation in particular has been used frequently to emp irically relate rates of esterification reactions to steric effects in series of increasing alkyl chains in reactants, especially alcohols. 25 , 33,34 In this section, the non - ideal concentration based rate constants are related to the Taft steric equation, w hich relies on concentration based rates. 3. 3 . 66 (3. 3 ) Where k 1 and k Ref are concentration based rate constants of the reaction of interest and a reference reaction. E S is the steric effect of substituents in reactants (relative to a refe rence reactant that has a short alkyl chain, usually the methyl group), is substituent specific, and the product of regression substituent relative to the corr esponding reaction that involves the reference reactant, and is a reaction specific parameter. The rate constants used in these studies are concentration based rate constants that do not explicitly account for non - idealities. An attempt to derive a relatio nship between non - ideal concentration rates and the steric effect of substituents follows: Rate of a reaction is written as Eq. 3. 4 : (3. 4 ) In Eq. 3. 4 b is the frequency of the transition state crossing the activation barrier, C TS and TS are the concentration and activity coefficient of the tra nsition state , respectively . According to transition state theory, the transition state is considered to be in constant thermodynamic equilibrium with the reactants. The transition state formed in esterification is shown in Scheme 3 . 1 . 67 Scheme 3 . 1 : Rate limiting step of esterification reaction mechanism The equilibrium constant for transition state and reactants is expressed as in Eq . 3. 5 . 26 (3. 5 ) C TS , C and C RCOOH are the concentrations of the transition state, alcohol and carboxylic acid respectively, while TS , , and RCOOH are their activity coefficients. Substituting for C TS form Eq. 3. 5 into the equation for rate (Eq. 3. 4 ), we get Eq. 3. 6 . (3. 6 ) The frequency of the transition state crossing the barrier is expanded as Eq. 3. 7 . (3. 7 ) In Eq. 3. 7 , k B and h T is temperature in Kelvin. The thermodynamic definition of the reaction equilibrium constant is : (3. 8 ) Substituting for K TS in the expression for rate (Eq. 3. 6 and Eq. 3. 7 ), we get Eq. 3. 9 , 68 (3. 9 ) In cases where the reaction solution is close to ideal, this expression becomes the law of mass action (Eq. 3. 10 ). (3. 10 ) In Eq. 3. 10 above, the activity coefficients are all assumed to be unity and are lumped in with the rate constant k. Several kinetic studies of esterification reactions have concluded that activity based mod el better predicts experimental behavior, the rate constant is therefore expressed as Eq. 3. 11 . (3. 11 ) It should be noted that Eq. 3. 11 above is for the forward reaction alone. The rate constants used in the Taft equatio n are concentration based (mol e /volume), which means in the application of the Taft equation to esterification, the activity coefficients are lumped in with the rate constants. Separating them from the rate constants, we get Eq . 3. 12 , which is rearranged to give Eq. 3. 13 . (3. 12 ) Rate constant k 69 (3. 13 ) The non - ideal concentration based rates and concentration based rates are then related as in Eq. 3. 14 . (3. 14 ) Or, in terms of the steric parameters, as in Eq . 3. 15 : (3. 15 ) Figure 3 . 7 shows the Taft equation plotted both as concentration and non - ideal concentration based rate constant ratios , to the reference reactant ethanol. E S values used are reported in Table 3 . 5 . Figure 3 . 7 : Taft equation applied to a candidate family of re actions (Reactions bas ed on study by Erdem and Cebe) 34 - ideal 70 concentration based rate constant ratios of methanol ( ), primary alcohols ( ), and branched alcohols ( ) are plotted. The dotted line shows the linear fit of the Taft equation for concentratio n based rate constant ratios to the ethanol rate co nstant from Erdem and Cebe 34 . The data in grey are the non - ideal concentration based rate constant ratios to ethanol rate constant. Table 3 . 5 : Es values for each alcohol in Figure 3 . 7 Alcohol Es Methanol 1.24 Ethanol 0 n - Propyl alcohol - 0.07 n - B utyl alcohol - 0.36 n - P entyl alcohol - 0.39 iso b utyl acohol - 0.93 i so p ropyl alcohol - 0.47 2 - B utyl alcohol - 1.13 The h orizontal dashed lines show regions where there is little deviation in values of the two reaction families: esterification of propanoic acid with primary alcohols, and esterification of propanoic acid with branched alcohols. Both sets have almost no change in rate constant ratios. For a reference reactant within a family, is close to zero. This simple derivation - ideality. (3. 16 ) 71 3.6 Conclusions Using a non - ideal concentration model accounts for both, non - ideality and differences in initial molar compositions. Within a reactivity plateau, it is now possible now to have a quick theoretical estimate of reaction rate for a particular system from the data of a reference reaction. This makes process design easier for the flexible formation of esters in biorefineries. Reactants such as formic acid, acetic acid, and methanol, which lack attached alkyl groups , have much higher reaction r ates in comparison to their longer hydrocarbon chain counterparts. 72 APPENDI CES 73 Appendix D : Sample calculation of non - ideal concentration rate constant, k NIC Non ideal rate constant calculated for acylation of propanol with butyric acid in presence of Am berlyst 36 from initial stages as: , where dx/dt is the absolute value of initial mole fraction change of any one reactant with respect to time, are mole fraction activities of carbox ylic acid and alcohol, and w CAT is the weight fraction of catalyst in the solution. Experiment was conducted at 60°C, with an initial catalyst loading w CAT = 1 wt. % (0.01 kg cat/kg soln), and an initial molar ratio of propanol: butyric acid = 3:1 . Initia l molar fractions of alcohol and carboxylic acid are 0.75 and 0.25 respectively. Ion exchange capacity of Amberlyst 36 IE = 5.4×10 - 3 kmoles H + /kg cat . 61 The initial rate was determined by a second order fit from da ta extrapolated to t=0 . Solution molar density , where MW propanol and MW butyric acid propanol butyric acid are mass densities at 60 °C. Volume change on mixing, and due to reaction is considered to be negligible at 60 °C. Mole frac tions are calculated from the ratio of reactants to be 0.25 for butyric acid and 0.75 for propanol . Activity coefficients for butyric acid and propanol are 0.9784, 0.9966 using 74 UNIFAC . A ctivities are calculated to be at 0.244 for butyric acid and 0.747 for propanol using UNIFAC. Hence k = 1.78×10 - 4 kg soln · m 3 soln · (kg catalyst) - 1 · s - 1 · kmol - 1 . In terms of active site concentration of catalyst, k NIC = k×IE× SOL = 4.1×10 - 5 m 6 soln· (kmoles H + ) - 1 · (kmol soln) - 1 · s - 1 . 75 Appendix E : Properties of alc o hols, carboxylic acids and acid catalysts used in calculations The following tables are properties used for calculations Table 3 . 6 : H + concentrations of catalysts Catalyst kmol H+/kg cat alyst Dowex 50Wx8 - 400 2.12E - 03 Amberlite IRA 120 4.40E - 03 Amberlyst 15 4.70E - 03 Amberlyst 70 2.55E - 03 Amberlyst 36 5.40E - 03 sulfuric acid 1.02E - 02 Amberlyst 39 5.00E - 03 Amberlyst 3 5 5.20E - 03 HCl 2.74E - 02 Purolite 4.90E - 03 HI 7.82E - 03 SAC - 13 1.31E - 04 p - TSA 5.26E - 03 Dowex 50W - 4 4.55E - 03 Smopex - 101 3.20E - 03 76 Table 3 . 7 : Densities of alcohols at 60 Alcohols Density at 60 (kg/m 3 ) Molar Mass (kg/kmol) Methanol 754.7 32.0 Ethanol 753.4 46.1 Propanol 768.3 60.1 2 - Propanol 745.9 60.1 Isopropanol 746.1 60.1 Butanol 773.8 74.1 2 - Butanol 767.2 74.1 Isobutanol 767.8 74.1 Pentanol 748.8 88.1 2 - Ethyl Hexanol 802.8 130.2 4 - Heptanol 787.2 116.2 Amyl Alcohol 784.0 88.2 Octanol 798.9 130.2 Ethylene Glycol 1100.0 62.1 Diethylene Glycol 1200.0 106.1 Triethylene Glycol 1150.0 150.2 Isoamyl Alcohol 774.5 88.2 Heptanol 787.2 116.2 Ethylene Glycol 1110.0 62.1 Benzyl Alcohol 1014.4 108.1 Propylene Glycol 1005.3 76.1 77 Table 3 . 8 : Densities of carboxylic acids at 60 Carboxylic acids Density at 60 (kg/m 3 ) Molar Mass (kg/kmol) Acetic Acid 1005.588 60.05 Propionic Acid 952.62679 74.07548 Butyric Acid 921.03584 88.11 Pentanoic Acid 904.14871 102.13 Hexanoic Acid 892.30483 116.16 Caprylic Acid 910 144.21 Acryl ic Acid 1005.69 72.06 Lactic Acid 1187.892 90.08 Octanoic Acid 910 144.21 Formic Acid 1220 46 Nonanoic Acid 900 158.23 Levulinic Acid 1107.891 116.11 Lauric Acid 800 200.3 Betamethyl Valeric Acid 896.15 132.158 Trimethyl Acetic Acid 884.6 102.132 Diethyl Acetic Acid 887.3 116.16 Dipropyl Acetic Acid 864 144.211 Methacrylic Acid 1020 86.06 Dibutyl Acetic Acid 840.7 172 Di - isobutyl Acetic Acid 840.7 172 Citric Acid 1104.142 192.124 Maleic Acid 1358.189 116.07 Decanoic Acid 893 172.26 Tartaric Acid 1745.109 150.087 Succinic Acid 1340.086 118.09 78 REFERENCES 79 REFERENCES (1) Larock, R. C. Comprehensive organic transformations ; VCH New York, 1989; Vol. 315. (2) Acker, P. E. van; Mathieu, O.; Milner, R. J.; Pacynko, W. F. Ester co - product ion. US6093845 A, July 25, 2000. (3) Bam, N.; Drown, D. C.; Korus, R.; Hoffman, D. S.; Johnson, T. G.; Washam, J. M. Method for purifying alcohol esters. US5424467 A, June 13, 1995. (4) Masanori, O.; Seishiro, Y. Mixed sugar alcohol esters of higher and lower saturated fatty acids. US3649647 A, March 14, 1972. (5) M, G. C. Preparation of mixed esters of polyhydric alcohols. US2309949 A, February 2, 1943. (6) Drent, E. Process for the co - production of carboxylic acids and carboxylic acid esters. US466485 1 A, May 12, 1987. (7) Warner, R. J.; Murphy, C. D.; Grajales, G. A. R.; Santiago, F. J. S.; Solorzano, F. A.; Torres, J. A. Process for the simultaneous coproduction and purification of ethyl acetate and isopropyl acetate. US6765110 B2, July 20, 2004. (8 ) Hinshelwood, C. N.; Legard, A. R. J. Chem. Soc. Resumed 1935 , 1588 1591. (9) DeTar, D. F. J. Org. Chem. 1980 , 45 , 5166 5174. (10) Eckey, E. W. J. Am. Oil Chem. Soc. 1956 , 33 , 575 579. (11) Pfeffer, P. E.; Silbert, L. S. J. Org. Chem. 1976 , 41 , 1373 1 379. (12) Srilatha, K.; Lingaiah, N.; Sai Prasad, P. S.; Prabhavathi Devi, B. L. A.; Prasad, R. B. N.; Venkateswar, S. Ind. Eng. Chem. Res. 2009 , 48 , 10816 10819. (13) Taft, R. W. J. Am. Chem. Soc. 1953 , 75 , 4538 4539. (14) Charton, M. J. Org. Chem. 197 8 , 43 , 3995 4001. (15) Charton, M. J. Am. Chem. Soc. 1975 , 97 , 1552 1556. (16) Loening, K. L.; Garrett, A. B.; Newman, M. S. J. Am. Chem. Soc. 1952 , 74 , 3929 3931. (17) Charton, M. J. Am. Chem. Soc. 1975 , 97 , 3694 3697. (18) Morgan, K. J.; Unwin, N. J. Chem. Soc. B Phys. Org. 1968 , 880 884. (19) Ali, S. H.; Merchant, S. Q. Int. J. Chem. Kinet. 2006 , 38 , 593 612. (20) Pöpken, T.; Götze, L.; Gmehling, J. Ind. Eng. Chem. Res. 2000 , 39 , 2601 2611. 80 (21) Mazzotti, M.; Neri, B.; Gelosa, D.; Kruglov, A.; Mor bidelli, M. Ind. Eng. Chem. Res. 1997 , 36 , 3 10. (22) Ju, I. B.; Lim, H. - W.; Jeon, W.; Suh, D. J.; Park, M. - J.; Suh, Y. - W. Chem. Eng. J. 2011 , 168 , 293 302. (23) Orjuela, A.; Yanez, A. J.; Santhanakrishnan, A.; Lira, C. T.; Miller, D. J. Chem. Eng. J. 20 12 , 188 , 98 107. (24) Kolah, A. K.; Asthana, N. S.; Vu, D. T.; Lira, C. T.; Miller, D. J. Ind. Eng. Chem. Res. 2008 , 47 , 5313 5317. (25) Santhanakrishnan, A.; Shannon, A.; Peereboom, L.; Lira, C. T.; Miller, D. J. Ind. Eng. Chem. Res. 2012 , 52 , 1845 1853 . (26) Eckert, C. A.; Hsieh, C. K.; McCabe, J. R. AIChE J. 1974 , 20 , 20 36. (27) Eckert, C. A. Ind. Eng. Chem. 1967 , 59 , 20 32. (28) Madon, R. J.; Iglesia, E. J. Mol. Catal. Chem. 2000 , 163 , 189 204. (29) Pappu, V. K. S.; Kanyi, V.; Santhanakrishnan, A .; Lira, C. T.; Miller, D. J. Bioresour. Technol. 2013 , 130 , 793 797. (30) Smith, H. A. J. Am. Chem. Soc. 1939 , 61 , 254 260. (31) Smith, H. A. J. Am. Chem. Soc. 1940 , 62 , 1136 1140. (32) Appl. Catal. Gen. 2011 , 396 , 14 19. (33) Lilja, J.; Murzin, D. Y.; Salmi, T.; Aumo, J.; Mäki - Arvela, P.; Sundell, M. J. Mol. Catal. Chem. 2002 , 182 183 , 555 563. (34) Erdem B.; Cebe M, Zeitschrift für Physikalische , 20 11 , 225 , 125 - 136. (35) Liu, Y.; Lotero, E.; Goodwin Jr., J. G. J. Catal. 2006 , 243 , 221 228. (36) MacPhee, J. A.; Panaye, A.; Dubois, J. - E. Tetrahedron 1978 , 34 , 3553 3562. (37) Grob, S.; Hasse, H. Ind. Eng. Chem. Res. 2006 , 45 , 1869 1874. (38) Roy, R .; Bhatia, S. J. Chem. Technol. Biotechnol. 1987 , 37 , 1 10. (39) Schmid, B.; Döker, M.; Gmehling, J. Ind. Eng. Chem. Res. 2008 , 47 , 698 703. (40) Ali, S. H.; Tarakmah, A.; Merchant, S. Q.; Al - Sahhaf, T. Chem. Eng. Sci. 2007 , 62 , 3197 3217. (41) Santhana krishnan, A.; Peereboom, L.; Miller, D. J.; Dumitrascu, A.; Smith, P. B. Ind. Eng. Chem. Res. 2013 , 52 , 9337 9342. 81 Included in this Section is a copy of the paper: Measurement of p - Toluenesulfonic Acid - Catalyzed Reaction Kinetics of 1,2 - Propylene Glycol Acetylation Using In Situ 1 H NMR Spectroscopy , by Arati Santhanakrishnan , Lars Peereboo m , Dennis J. Miller * , Adina Dumitrascu , and Patrick B. Smith Department of Chemical Engineering and Materials Science, Michigan State University , East Lansing, Michigan 48823, United State s Michigan Molecular Institute , 1910 West Saint Andrews Road, Midland, Michigan 48640, United States Reprinted with permission from Ind. Eng. Chem. Res. , 2013 , 52 (27), pp 9337 9342 Copyright © 2013 American Chemical Society 82 4 Kinetics of p - Touenesulfoni c Acid - Catalyzed 1,2 - Propylene Glycol Acetylation 4.1 Abstract The reaction kinetics of the acetylation of 1,2 - propylene glycol (PG) catalyzed by p - t oluenesulfonic acid ( p - TSA) determined using an in - situ 1 H NMR method by collaborating researchers were mod eled . Both primary and secondary mono - acetate esters of PG were observed as well as the di - ester. The reaction kinetics were characterized as a function of PG to acetic acid (AA) stoichiometry, p - TSA concentration , and temperature. The acetylation reacti ons and the trans acetylation of diester with PG were modeled as reversible second order reactions. Equilibrium constants were determined experimentally for each reaction. Activation energies for the consecutive acetylation of PG and PGMA are 56 and 47 kJ/ mol, respectively. The activation energy for intermolecular trans acetylation of PG with its diacetate ester to form PG monoacetate i s 63 kJ/mol. The rate constants obtained from the model, when normalized to the two hydroxyls present per molecule of diol, are close to the expected value of the universal rate constant for p - TSA esterification calculated in Chapter 2 . 4.2 Introduction Esters of glycols and other polyols have many industrial applications including as coalescing agents, 1 lubricants, 2 intermediates to polyurethane foams, 3 plasticizers, 4 and emulsifiers in the food processing industry. 5 The conventional route for their synthesis is the acylation of polyols with a carboxylic acid in the presence of acid catalysts. Typically, the reaction system consis ts of the esterification reactions of alcohols and subsequent hydroxyl esters with carboxylic acids to form esters and water, and disproportionation reactions involving the transesterification of the 83 monoesters to form diesters and glycol ( Scheme 4 . 1 ). The usage of glycol esters in a wide variety of applications requires the optimization of esterification processes for high yields of either monoesters or diesters with high purity, warranting a thorough understanding of the kinetics of glycol esterification systems. Scheme 4 . 1 : Glycol - carboxylic acid system reactions Glycol + Carboxylic acid Monoester + Water Monoester + Carboxylic acid Diester + Water 2Monoesters Diester + Glycol In this study, a candidate reaction system consisting of propylene glycol reacting with acetic acid in the presence of para - toluene sulfonic acid has be en studied to model the effect of initial reactan t molar ratios, temperature, catalyst loading and non - ideality of solutions on the rate of reaction and equilibrium composition with a kinetic model that uses non - ideal concentrations. Using propylene glycol instead of ethylene glycol enables differentiation between primary and secondary monoesters and to understand the ir interaction s and individual rates of further esterification . Propylene glycol (PG) is a large commodity chemical with annual global sales o f about $ 1 50 million 6 . Acetate esters of PG are excellent model compounds for further chemistry , and their formation has not yet been studied in detail . The acetylation of PG with acetic acid (AA) to form either primary (PGMA1) or secondary (PGMA2) mono - acetate esters and the subsequent acetylation with another acetic acid molecule to form propylene glycol diacetate (PGDA) ( Scheme 4 . 2 ) is a commercially valuable process. Commercial processes now exist to make pr opylene glycol from glycerol, which in turn is a side product of biodiesel synthesis. Bio - based acetic acid may be produced readily from ethanol. Kinetic models of the esterification of acetic acid with propylene glycol that account for both non 84 ideality o f the solutions and a wide range of reaction conditions enable simulations for the design of future biorefinery processes. Scheme 4 . 2 : Acetylation of 1,2 - propylene glycol with acetic acid. 4.3 Materials an d Methods Acetic acid (AA), 1,2 - propylene glycol (PG), propylene glycol diacetate (PGDA), and p - toluene sulfonic acid monohydrate ( p - TSA) were obtained from Sigma Aldrich and used without further purification. As part of a collaborative research project, k inetic experiments for acetylation of PG were conducted at Michigan Molecular Institute by Dr. Patrick B. Smith and Dr. Adina Dumitrascu using an in situ H 1 NMR method that enabled real time sample measurements and the accurate quantification of the prim ary and secondary monoesters. 8 5 Transacetylation reactions were carried out at Michigan State University in 75 - ml stainless steel batch autoclaves in a Parr 5000 Multireactor system. PG and PGDA were loaded into the autoclave and the mixture was heated to reaction temperature, at which time p - TSA catalyst was injected into the reaction fluid. Samples (~1.0 ml) were taken periodically, diluted ten - fold in acetonitrile prior to injection, and then analyzed in a Varian 450 gas chromatograph fitted with a ther mal conductivity detector (Varian Medical Systems Inc., Palo Alto, CA). Separation was done on a 0.53 mm ID Aquawax - DA 30 m capillary column with 1.0µm film thickness, a helium carrier gas flow rate of 10 ml/min, an injector temperature 250 o C, and detecto r temperature 240 o C. The initial column temperature was held at 40 o C for 1 min, ramped to 130 o C at 30 o C/min, ramped to 150 o C at 5 o C/min, and finally ramped to 200 o C at 30 o C/min and held for 2 min. Standards of known composition in the range of in terest were prepared and run in the chromatograph before and after reaction samples to calibrate the response factor of each component of the reaction. PGMA was determined stoichiometrically from the amount of PGDA consumed in the reaction mixture. 4.4 Results The full list of experiments conducted either at the Michigan Molecular Institute or at Michigan State Un iversity are recorded in Table 4 . 1 . 4.4.1 Equilibrium constants The activity - based equilibrium constants for Reactions 1 an d 2 in Scheme 4 . 3 (Eq. 4 . 1 ) were obt ained from experimental data at long reaction times where equilibrium was reached ( Figure 4 . 1 ). 86 Scheme 4 . 3 : Simplified reaction scheme for kinetic model Reaction 1 Reaction 2 Reaction 3 Table 4 . 1 : Summary of reactions and experimental conditions Experiment Temperature catalyst loading (kg cat/ kg soln) initial molar ratio Figure No. Esterification PG:AA 1 63 0.012 1: 1 4.6 2 63 0.016 2: 1 4.7 3 63 0.008 1: 2 4.2 4 75 0.008 1: 2 4.8 5 75 0.012 1: 1 4.9 6 75 0.006 1: 1 4.10 7 75 0.025 1: 1 4.11 8 75 0.06 3 1: 1 4.12 9 86 0.012 1: 1 4.13 10 86 0.016 2: 1 4.14 11 86 0.008 1:2 4.15 Hydrolysis PGDA: Water 12 64 0.01 3: 1 4.16 13 75 0.01 3: 1 4.3 14 86 0.01 3: 1 4.17 Transacetylation PGDA:PG 15 85 0.007 1: 1 4 .4 16 75 0.004 3: 1 4.18 17 65 0.00 7 1: 3 4.19 ( 4 . 1 ) Here x i , i , and v i represent the mole fraction, activity coefficient, and stoichiometric coefficient of component i in the reaction mixtur e at equilibrium. The equilibrium constant for the 87 transacetylation reaction , K a,3 , is equal to the ratio of the other two equilibrium constants (i.e K a,3 = K a,1 /K a,2 ). UNIFAC (UNIversal Functional Activity Coefficient) was used to estimate activity coeffi cients of components 7 . ( exothermic) for K a,1 , but a constant value for K a,2 ( Figure 4 . 1 ). Although only two temperatures were used for this analysis, a number of values at each temperature show a consistent difference for K a,1 , for a relatively small difference in temperature (~10°C). This warranted the inclusion of a temperature dependence for K a,1 , and consequently K a,3 , in the regression analysis discussed in the next section. The heat of reaction of PG acetylation (Reaction 1 of Scheme 4 . 3 ) and thus PG - PGDA transacetylation (Reaction 3 of Scheme 4. 3 ) is - 10 kJ/mol. Figure 4 . 1 - based equilibrium constants for the K a1 K a2 . 4.4.2 Kinetic modeling The NMR kinetic data for PG acetylation were modeled as a set of elem entary, reversible, consecutive reactions ( Scheme 4 . 3 ). The reaction of PG with PGDA to give PGMA ( propylene 88 glycol monoacetate ) ( k 3 ) in an intermolecular transacetylation is also included in the model to account for interconversion of the esters. Reverse reaction rates (hydrolysis of PGMA and PGDA, and conversion of PGMAs to PGDA and PG) are accounted for by the equilibrium constants and forward rate constants ( k - m = k m /K m for reaction m in Scheme 4 . 3 ). Initially, a more complex reaction network that treated the primary and secondary PGMA esters as separate species was considered ( Scheme 4 . 2 ); however, experimental concentration pro files showed a constant molar ratio of primary to secondary monoesters of approximately 1.7 at all stages of reaction, evidence that interconversion of the two PGMA esters is significantly faster than PG acetylation. This observation , along with the challe nge of preparing pure primary or secondary PGMA standards, led to the decision to treat the primary and secondary monoacetates as one entity. The sealed NMR tube used in reaction studies is modeled as a stirred batch reactor. The change in number of moles of component i ( N i ) participating in m reactions can be expressed as ( 4 . 2 ) where is the total number of moles in the reactor, is the volume of the reacting phase, is the rate of reaction m per unit volume, and is the mole fraction of component i in the liquid mixture. The parameter is the ratio of stoichiometric coefficients of component i w ith respect to the reference component in reaction m . Eq. 4 . 2 can be simplified by assuming total molar concentration of the liquid phase C T to be constant ( V / N T = 1/ C T ), as total number of moles is conserved in the reaction system a nd reaction volume is assumed constant during the run under isothermal conditions. 89 ( 4 . 3 ) Reaction rate r m is expressed in terms of an elementary rate law with t he product of solution density C T and species activity a i representing each species in the rate expression. This product, - C T x i i or C i i ), has been found in earlier work to be the basis for a universal approach to esterif ication kinetics. 8 10 Rates of formation of propylene glycol mono acetate ( PGMA ) and propylene glycol diacetate (PGDA), and the transacetylation of PG and PGDA to PGMA, can thus be expressed in terms of the activities of propylene glycol (PG), acetic acid ( AA) , PGMA, and water ( W ); K a,1 , K a,2 , and K a,3 , the thermodynamic (activity - based) equilibrium constants; w CAT , the concentration of catalyst in the reaction mixture; the solution density C T , and k 1, k 2 , and k 3 , the forward rate constants. The number o f moles and thus activity of water in the reaction mixture is calculated from the reaction stoichiometry. (N W = N W0 + N PGMA + 2N PGDA ). The UNIFAC activity coefficient model is used to calculate the activity coefficients of each species in the system. 90 Tabl e 4 . 2 : UNIFAC groups and their counts in reaction components Component name Group Count PG CH3 1 CH2 1 CH 1 OH 2 Acetic acid CH3 1 C OOH 1 PGMA CH3 1 CH2 1 CH 1 CH3 COO 1 OH 1 PGDA CH3 1 C H2 1 CH 1 CH3COO 2 Water H2O 1 Table 4 . 3 : Group volume and surface area contributions Group R k Q k CH3 0.9011 0.848 CH2 0.6754 0.54 COOH 1.3013 1.224 CH 3 COO 1.9031 1.7280 H2O 0.92 00 1.4 000 OH 1 .000 0 1.2 000 CH 0.4469 0.2280 Table 4 . 4 : UNIFAC group binary interaction parameters Name CH2 OH H2O CH3COO COOH CH2 0 986.5 1318 232.1 663.5 OH 156.4 0 353.5 101.1 199 H2O 300 - 229.1 0 72.87 - 14.09 CH3COO 1 14.8 245.4 200.8 0 660.2 COOH 315.3 - 151 - 66.17 - 256.3 0 91 ( 4 . 4 ) ( 4 . 5 ) ( 4 . 6 ) Five ordinary differential equations (one for each ch emical species in the system) were assembled from Eq . 4 . 3 through Eq. 4 . 6 ; these equations were simultaneously integrated and regressed to experimental data to estimate kinetic parameters using the functions nlinfi t and ode23 from the optimization toolbox of Matlab 7.12.0. To simplify parameter fitting, the rate constant k 3 for the transacetylation reaction of PG with PGDA to give PGMA was first determined using only experimental data for that reaction (Experiments 15 - 17 of Table 4 . 1 ). Then, using the value for k 3 , the hydrolysis and acetylation reactions were fit together to give the values of k 1 and k 2 (Experiments 1 - 14). Optimization was done by minimizing the sum of squared differences between calculated and experimental species mole fractions (Eq. 4 . 7 ), ( 4 . 7 ) where n is the number of experimental samples withdrawn in each experiment and N c is the number of reacting components in each experiment. Optimized kinetic parameters are reported in Error! Reference source not found. . 92 Table 4 . 5 Optimized kinetic parameters with 95% confidence limits for PG acetylation with p - TSA catalyst R eaction 1: Acetylation of PG to PGMA k 0,1 3.13 ± 0.05 × 10 6 m 3 E a,1 55700 ± 50 kJ / kmol K a,1 exp(1255/T - 1.91) Reaction 2: Acetylation of PGMA to PGDA k 0,2 5.22 ± 0.25 × 10 4 m 3 E a,2 47000 ± 140 kJ / kmol K a,2 1.52 Reaction 3: Transacetylation of PG + PGDA to PGMA k 0,3 9.05 ± 0.34 × 10 6 m 3 E a,3 63900 ± 630 kJ / mol K a,3 K a,1 / K a,2 The rate constant values at 60 parameters in Error! Reference source not found. are found to be 2.4 ×10 - 3 and 2.2×10 - 3 m 3 . T he se value s are very close to the rate constant value at 60 acylation of ethanol and n - butanol with butyric acid in the presence of p - TSA (2.1×10 - 3 ) . The value s of E a,1 and E a,2 are in the range of 50±10 kJ/mole, which is typical of esterification reaction activation energies. Figure 4 . 2 through Figure 4 . 4 compa re the predicted a nd experimental species mole fractions vs. time for acetylation, hydrolysis, and transacetylation reactions. 93 Figure 4 . 2 : Experimental and predicted concentration profiles of PG acetylation (Experiment 3: 6 3 UNIFAC activity based propylene glycol diacetate; (×) water. Figure 4 . 3 : Experimental and predicted concentration profiles of PGDA hydrolysis (Experiment 13: 75 UNIFAC activity acetic acid; propylene glycol diacetate; (×) water. 94 Figure 4 . 4 :Experimental and predicted concentration profiles of PG - PGDA transacetylation (Experime nt 15: 85 propylene glycol monoacetate; propylene glycol diacetate; (×) water. The model was found to predict experimental behavior accurately with the value of absolute residuals (predicted minus experimental mole fraction) averaged over all species at all data points for all experiments equal to 0.0098. A parity plot of PGMA experimental and p redicted mole fractions for every data point in every experiment, which demonstrates the overall quality of model fit to data, is shown in Figure 4 . 5 . Figure 4 . 5 : Parity plot of kinetic model fit for PGMA mole fractions in experiments 95 The absolute residual for each experiment are shown in Table 4 . 6 . Predicted and experimental mole fraction profiles for all ex periments are g iven in Fig ure 4 . 6 through Figure 4 . 19 of Appendix F. Raw NMR data from e xperiments conducted at Michigan Molecular Institute and Michigan State University are given in Table 4 . 7 through Tabl e 4 . 23 of Appendix G . Table 4 . 6 : Absolute residuals for each experiment Experiment Temperature Cata lyst loading (kg cat/ kg soln) Initial molar ratio F abs a Acetylation PG:AA 1 63 0.012 1: 1 0.016 2 63 0.016 2: 1 0.011 3 63 0.008 1: 2 0.009 4 75 0.008 1: 2 0.009 5 75 0.012 1: 1 0.009 6 75 0.006 1: 1 0.008 7 75 0.025 1: 1 0.009 8 75 0.063 1: 1 0.008 9 86 0.012 1: 1 0.009 10 86 0.016 2: 1 0.010 11 86 0.008 1:02 0.009 Hydrolysis PGDA: Water 12 64 0.01 3: 1 0.008 13 75 0.01 3: 1 0.013 14 86 0.01 3: 1 0.016 Transacetylation PGDA:PG 15 85 0.007 1: 1 0.006 16 75 0.004 3: 1 0.009 17 65 0.007 1: 3 0.008 a F abs is the absolute value of predicted minus experimental mole fraction averaged over all species and all data points for the experiment. 96 4.5 Conclusions Reaction kinetics and equilibrium constants for the reversible liquid phase acet ylation of PG with AA using p - toluenesulfonic acid as catalyst have been determined in the range of 63 °C 85 °C. The experiments were conducted and analyzed using an in - situ 1 H NMR technique that quantified all reactants and products except water in the reaction mixture over time. The experimental data were fit to a homogeneous activity - based model that describes the experimental behavior. Heats of reaction were calculated from experimentally determined equilibrium constants. The reaction of PG with PGD A to form PGMA was found to proceed at rates comparable to the acetylation reactions; interconversion of PGMA primary and secondary isomers was rapid, giving an equilibrium molar ratio of primary:secondary of 1.7:1 under all reaction conditions. The non - i deal concentration model rate constants were calculated at 60 formation from PGMA and acetic acid gives rate constant values similar to those obtained for butyric acid esterification with ethanol and n - butanol in the presence of p - TSA. This sim ilarity provides further impetus to use the non - ideal concentration model for esterification kinetics. 97 APPENDICES 98 Appendix F : Predicted and experimental mole fraction profiles of all experiments in PG esterification with acetic acid (Chapter 2) Fig ure 4 . 6 : Experimental and predicted concentration profiles of PG acetylation (Experiment 1: 63 UNIFAC activity based propylene glycol diacetate; (×) water Figure 4 . 7 : Experimental and predicted concentration profiles of PG acetylation (Experiment 2: 63 UNIFAC activity based propyl propylene glycol diacetate; (×) water 99 Figure 4 . 8 : Experimental and predicted concentration profiles of PG acetylation (Experiment 4: 75 UNIFAC activity based propylene glycol diacetate; (×) water Figure 4 . 9 : Experimental and predicted concentration profiles of PG acetylation (Experiment 5: 75 UNIFAC activity based propylene glycol propylene glycol diacetate; (×) water 100 Figure 4 . 10 : Experimental and predicted concentration profiles of PG acetylation (Experiment 6: 75 UNIFAC activity based propylene glycol diacetate; (×) water Figure 4 . 11 : Experimental and predicted concentration profiles of PG acetylation (Experiment 7: 75 UNIFAC activity based propylene glycol monoacetate; propylene glycol diacetate; (×) water 101 Figure 4 . 12 : Experimental and predicted concentration profiles of PG acetylation (Experiment 8: 75 UNIFAC activity based propylene glycol diacetate; (×) water Figure 4 . 13 : Experimental and predicted concentration profiles of PG acetylation (Experiment 9: 86 UNIFAC activity based pr opylene glycol diacetate; (×) water 102 Figure 4 . 14 : Experimental and predicted concentration profiles of PG acetylation (Experiment 10: 86 UNIFAC activity based propylene glycol diacetate; (×) water Figure 4 . 15 : Experime ntal and predicted concentration profiles of PG acetylation (Experiment 11: 86 UNIFAC activity based propylene g lycol diacetate; (×) water 103 Figure 4 . 16 : Experimental and predicted concentration profiles of PGDA hydrolysis (Experiment 12: 63 UNIFAC activity propylene glycol diacetate; (×) water Figure 4 . 17 : Experimen tal and predicted concentration profiles of PGDA hydrolysis (Experiment 14: 86 UNIFAC activity propy lene glycol diacetate; (×) water 104 Figure 4 . 18 : Experimental and predicted concentration profiles of PG transacetylation (Experiment 16: 75 propylene glycol monoacetate; propylene glycol diacetate; (×) water Figure 4 . 19 : Exp erimental and predicted concentration profiles of PG transacetylation (Experiment 17: 65 propylene glycol monoacetate; propylene glycol diacetate; (×) water 105 Appendix G . Raw data from propylene glycol esterification with acetic acid (conducted at Michigan Molecular Institute) Data in Table 4 . 7 through Tabl e 4 . 23 p resent relative molar quantities of each species as determined from in - situ NMR spectra of the reaction mixture. Table 4 . 7 : Experiment 1 Time (min) AA PGMA1 PGMA2 PGDA PG 0 100 0.001 0.001 0.001 100 7 84 10 5 0.001 85 17 69 20 10 0.001 70 27 57.002 26 12 2 60 37 52.002 27 16 2 55 47 48.002 30 17 2 51 150 37.002 31 19 6 44 Table 4 . 8 : Experiment 2 Time (mi n) AA PGMA1 PGMA2 PGDA PG 5 83 12 5 0.001 183 15 55.002 27 14 2 157 25 42.002 34 20 2 144 35 37.002 38 21 2 139 45 32.002 40 24 2 134 150 24.002 46 24 3 127 106 Table 4 . 9 : Experiment 3 Time (min) AA PGMA1 PGMA2 PGDA PG 0 100 0 0 0 50 5 91.998 6 2 0.001 42 15 82 12 6 0.001 32 25 71.002 17 8 2 23 35 69 19 10 2 19 45 66 20 10 3 17 150 59 19 11 7 13 107 Table 4 . 10 : Experiment 4 Time (min) AA PGMA1 PGMA2 PGDA P G 4 187 11 6 1 82 8 163 26 11 2 61 12 150 33 13 4 50 16 141 35 17 6 43 22 133 37 19 7 37 33 122 39 20 11 30 45 120 39 19 11 31 53 118 41 19 14 26 60 116 41 19 15 25 72 116 42 19 15 24 82 112 38 19 16 27 90 109 40 19 19 22 112 111 39 19 19 22 108 Table 4 . 11 : Experiment 5 Time (min) AA PGMA1 PGMA2 PGDA PG 4 84 11 6 0 83 8 67 21 9 2 68 12 59 25 12 3 61 17 56 28 14 3 55 24 49 30 16 4 50 30 47 31 16 5 48 45 43 33 17 6 44 64 45 33 16 6 46 69 44 33 16 7 44 100 42 33 16 8 43 Table 4 . 12 : Experiment 6 Time (min) AA PGMA1 PGMA2 PGDA PG 6 88 9 4 0 87 19 64 23 11 2 64 24 59 26 12 2 60 30 55 27 14 3 56 35 52 29 14 4 53 42 49 30 15 4 52 54 45 31 16 5 48 59 46 30 17 5 48 109 Table 4 . 13 : Experiment 7 Time (min) AA PGMA1 PGMA2 PGDA PG 4 81 12 6 1 81 8 59 25 11 2 62 14 48 31 14 4 51 19 42 32 15 6 47 28 39 32 17 6 45 38 38 32 17 7 44 48 38 32 17 8 43 58 36 33 16 8 43 73 35 32 16 8 44 Table 4 . 14 : Experiment 8 Time (min) AA PGMA1 PGMA2 PGDA PG 4 66 22 10 2 66 7 49 30 15 4 51 11 42 32 17 6 45 16 37 33 17 7 43 24 36 32 17 8 43 33 37 33 15 9 43 53 37 31 17 9 43 57 35 32 17 9 42 110 Table 4 . 15 : Experiment 9 Time (min) AA PGMA1 PGMA2 PGDA PG 6 79 15 8 1 76 9 61 23 15 2 60 11 57 27 14 3 56 18 47 31 17 5 47 23 44 31 18 6 45 36 44 33 16 5 46 41 41 32 18 7 44 52 42 30 20 7 44 67 41 31 18 8 43 76 44 30 16 7 47 90 41 29 19 8 44 110 42 31 16 6 47 114 41 29 19 8 44 111 Table 4 . 16 : Experiment 10 Time (min) AA PGMA1 PGMA2 PGDA PG 4 44 4 2 0 94 8 27 14 8 1 77 11 21 14 9 2 70 15 16 19 11 2 68 20 15 21 11 2 66 25 12 22 11 2 65 35 13 23 11 3 63 41 12 21 14 2 63 55 12 23 12 2 63 90 11 22 12 3 63 111 10 22 12 3 63 112 Table 4 . 17 : Experiment 11 Time (min) AA PGMA PGM A2 PGDA PG 5 192 10 5 0.1 85 11 139 31 17 5 48 15 125 33 25 7 36 19 114 36 23 11 30 27 105 36 25 14 25 32 104 36 23 16 25 40 107 37 19 16 28 48 104 36 23 18 22 58 107 34 21 19 25 66 100 36 18 20 25 75 105 35 22 20 22 87 108 37 17 19 27 98 102 34 20 22 24 110 102 33 22 22 24 125 102 36 18 22 24 113 Table 4 . 18 : Experiment 12 Time (min) AA PGMA1 PGMA2 PGDA PG 5.0 5.2 1.5 2.4 95.5 0.7 9.0 8.4 3.7 3.0 92.5 0.9 14.0 14.3 6.1 4.8 87.4 1.7 22.0 17.4 8. 4 5.5 84.3 1.7 40.0 25.7 12.7 7.7 76.9 2.6 50.0 28.5 14.2 8.6 74.3 2.8 62.0 28.9 15.0 8.5 73.8 2.7 Table 4 . 19 : Experiment 13 Time(min) AA PGMA1 PGMA2 PGDA PG 4.0 6.9 2.4 2.6 94.0 0.9 10.0 14.5 6.7 4.9 8 6.9 1.4 20.0 22.5 11.1 6.7 79.8 2.4 30.0 26.5 13.4 7.6 76.3 2.8 50.0 30.0 15.4 9.0 72.8 2.8 62.0 30.0 15.8 9.4 72.4 2.4 114 Table 4 . 20 : Experiment 14 Time (min) AA PGMA1 PGMA2 PGDA PG 5 10.6 4.9 3.6 90.5 1. 0 10 21.8 10.7 6.3 80.6 2.3 18 29.0 14.5 8.2 74.2 3.2 26 31.6 16.1 9.3 71.6 3.0 36 33.6 17.3 9.6 69.8 3.3 51 33.7 18.0 10.2 69.1 2.7 64 34.3 18.3 10.3 68.6 2.8 Table 4 . 21 : Experiment 15 Time (min) mole fractions PG PGMA1 PGDA PGMA2 0 0.71 0.049 0.213 0.024 15 0.69 0.079 0.189 0.039 30 0.66 0.112 0.173 0.055 45 0.64 0.137 0.158 0.067 60 0.62 0.156 0.147 0.077 75 0.61 0.172 0.136 0.085 85 0.59 0.186 0.126 0.095 115 Table 4 . 22 : Experiment 16 Time (min) mole fractions PG PGMA1 PGDA PGMA2 0 0.228 0.032 0.72 0.015 22 0.206 0.068 0.69 0.033 37 0.190 0.094 0.67 0.046 52 0.177 0.116 0.65 0.058 67 0.166 0.138 0.63 0.068 82 0.157 0.148 0.62 0.075 Tabl e 4 . 23 : Experiment 17 Time (min) mole fractions PG PGMA1 PGDA PGMA2 0 0.373 0.171 0.369 0.087 15 0.336 0.223 0.328 0.113 30 0.306 0.263 0.298 0.132 45 0.283 0.296 0.272 0.148 60 0.266 0.319 0.255 0.160 75 0.252 0.339 0.240 0.169 116 REFERENCES 117 REFERENCES (1) Vahteristo, K.; Laari, A.; Haario, H.; Solonen, A. Chem. Eng. Sci. 2008 , 63 , 587 598. (2) Gryglewicz, S.; Piechocki, W.; Gryglewicz, G. Bioresour. Technol. 2003 , 87 , 35 39. (3) Ehrlich, A.; S mith, M. K.; Patton, T. C. J. Am. Oil Chem. Soc. 1959 , 36 , 149 154. (4) Morgan, P. W. Ind. Eng. Chem. Anal. Ed. 1946 , 18 , 500 504. (5) Seiden, P. Purifying propylene glycol monoesters using vacuum distillation. US3669848 A, June 13, 1972. (6) Shelley, S . Chem. Eng. Prog. 2007 , 103 , 6 9. (7) Fredenslund, A.; Jones, R. L.; Prausnitz, J. M. AIChE J. 1975 , 21 , 1086 1099. (8) Eckert, C. A. Ind. Eng. Chem. 1967 , 59 , 20 32. (9) Eckert, C. A.; Hsieh, C. K.; McCabe, J. R. AIChE J. 1974 , 20 , 20 36. (10) Santha nakrishnan, A.; Peereboom, L.; Miller, D. J.; Dumitrascu, A.; Smith, P. B. Ind. Eng. Chem. Res. 2013 , 52 , 9337 9342. 118 5 Catalytic epoxidation of propylene glycol and its acetates 5.1 Abstract The base - catalyzed, gas - phase epoxidation of propylene glycol and i ts acetates to propylene oxide (PO) is investigated in a laboratory - scale fixed - bed reactor. Potassium salts (0.5 - 2.5 mmol/g) on silica gel support are identified as selective catalysts for the reaction. A temperature of 400 been found to be optimal, giving a maximum PO selectivity of 88% and 50% conversion of propylene glycol acetates. Higher temperatures and higher potassium loadings on silica gel lead to collapse of the support, resulting in reactor plugging. Pre - and post - reaction catalysts have been characterized by N 2 adsorption, XPS, EDS and FTIR. Bulk potassium carbonate has been found to be the stable active species on the support surface during reaction for loadings greater than 1.5 mmol/g K + on silica. At lower load ings, potassium silicate is the active component. The use of a stable, low surface area silica support prevents the collapse of the support structure, leading to higher conversions without sacrificing selectivity. 5.2 Introduction Propylene oxide (PO) is a va luable intermediate chemical with a global market of more than 9.2 million tonnes in 2014 that is expected to steadily increase in the future. 1 PO is used to make many everyday products such as defoamers, anti - freeze, and lubricants in the form of polyglyc ols, polyurethanes and polyglycol ethers. Conventionally, PO is manufactured by the chlorohydrin process ( Scheme 5 . 1 ) or by organic per oxidation ( Scheme 5 . 2 ) ; both processes rely heavily o n the availability of petroleum derived propylene. The chlorohydrin process involves 119 addition of stoichiometric amounts of chlorine and requires the disposal of aqueous solutions of s odium or calcium chloride of forty times the volume of PO produced, makin g the process a waste water burden . 2 The organic peroxidation route produces either tert - butanol or 1 - phenyl ethanol as co - products, depending on whether the organic peroxide used is synthesized from tert - butane or ethyl benzene, respectively. These proces ses also produce side products such as methyl formate, which has a boiling point close to PO, making it difficult to separate. In the interest of reducing our reliance on fossil sources and shifting to more environmentally friendly processes, alternate rou tes to make propylene oxide are being investigated. Scheme 5 . 1 : Chlorohydrin process Scheme 5 . 2 : t ert - B utyl peroxide process Although routes to make biom ass based propylene are being investigated , 3 6 clean and economical alternatives to the above mentioned processes for PO production are still under development, the most promising being the HPPO process (Dow Chemical) . Biodiesel production by transesterifi cation of triglycerides produces glycerol as a side product, providing the 120 opportunity to build reaction networks with glycerol as a building block. One such reaction of glycerol is hydrogenolysis to propylene glycol, which is being extensively studied 7 10 and is commercially practiced by Archer Daniels Midland . 1 Thus, i nstead of the traditional pathway to propylene glycol from propylene derived propylene oxide, the pathway to propylene oxide from bio - glycerol derived propylene glycol becomes an attractive option. Studies focusing on producing propylene oxide from propylene glycol in literature are broadly of two categories ( Scheme 5 . 3 ) . One route is the direct dehydration of propylene glycol to propylene oxide. 11 13 The other is i ndirect; propylene glycol is converted to its esters with an abundantly available carboxylic acid , and the ester - glycol mixture formed is t hen converted to propylene oxide . 14 17 The majority of catalysts tested for both routes were alkali or alkaline earth metal salts on neutral or weakly acidic supports such as silica and alumina. Other supports such as magnesium oxide , carbon and germanium dioxide loaded with similar base salts gave much lower yields (<7% by mole). 13 Scheme 5 . 3 : Routes from propylene glycol to propylene oxide 1) Direct base catalyzed dehydration to propylene oxide 2) Propylene glycol esterification to intermediate esters The optimized yield reported in multiple studies from both these rou tes using a variety of c atalysts and processes is ~30 %. The highest reported PG/PGMA conversions were for c esium on silica at 52%. 7 Sodium and lithium gave maximum yields of 20 % and 3%, respectively. 121 The reaction of PG and its acetates to PO occurs at hi gh temperature (>300 and proceeds with the formation of various side products ( Scheme 5 . 4 ) . Most of these studies screen a number of catalysts, but good process design requires a thorough understanding of the effec t of process parameters as well as fundamental catalyst properties. In this study, several candidate catalysts were chosen and a detailed investigation of the effect of reaction temperature from 340 o C to 420 o C, feed material (PG vs. PG acetates), contact time, and the loading of catalytic salt on the support was conducted. Characterization of catalysts via N 2 BET surface area measurement and FTIR, XPS, and EDS surface analysis has also been done. 5.3 Materials and Methods 5.3.1 Materials Reagent grade propylen e oxide ( 99%, Sigma Aldrich Corp., St. Louis, Missouri ), n - p ropanal ( 97%, Sigma Aldrich Corp., St. Louis, Missouri ), PGDA ( 99.7%, Sigma Aldrich Corp., St. Louis, Missouri ) , acetic acid (glacial, Emanuel Merck Damstadt Chemicals, Philadelphia, P ennsylvania ) , PG ( A.C.S Reagent, Jade Scientific Inc., Westland, Michigan ), allyl alcohol (99%, ACROS Organics, New Jersey) , 2 - ethyl - 4 - methyl - 1,3 - dioxolane (Alfa Aesar, Ward Hill, Massachusetts), allyl acetate (99%, Sigma Aldrich Corp., St. Louis, Missouri ) , water (HPL C solvent, JT Baker Reagent Chemicals , Phillipsburg, N ew Jersey ), acetone (JT Baker Reagent Chemicals, Phillipsburg, N ew Jersey) , cesium nitrate , potassium hydroxide (pellets, A.C.S Reagent, >85%, Sigma Aldrich Corp., St. Louis, Missouri ), potassium acetat e (>98%, Sigma Aldrich Corp., St. Louis, Missouri ), potassium carbonate (99.99%, Sigma Aldrich Corp., St. Louis, Missouri ), po tassium silicate (99%, anhydrou s, Alfa Aesar, Ward Hill, Massachusetts), silica gel 122 (high purity grade, 60 Å, 35 - 70 mesh, 0.212 - 0. 5 mm diameter, Sigma Aldrich Corp., St. Louis, Missouri ), and controlled pore glass (507 Å, 200 - 400 mesh, 40 - Lincoln Park, New Jersey) were used without further purification. The properties of Amberlyst 15 used to prepar e feed are reported in the literature . 20 5.3.2 Feed preparation Water content of propylene glycol and propylene glycol diacetate was measured by Karl Fischer titration and gas chromatography. Traces of water (<0.1 weight % ) were found in both chemicals. Small amounts ( <1% on basis of weight ) of acetic acid were found in propylene glycol diacetate. Weighed amounts of propylene glycol and propylene glycol diacetate were dried with activated D rierite in their respective containers. After allowing the containers to sit overnight, the resulting solutions were again analyzed for water content and impurities using Karl Fischer and gas chromatography. PGDA and PG were then mixed in a 1:1 molar ratio with 2% by weight Amberlyst 15 catalyst (dried in a vacuum oven at 100 ° C for well over 2 days), and freshly activated drierite granules in a batch reactor set up at 70 ° C. The final mixture was composed of 55 mol% primary and secondary propylene glycol monoacetates (in a ratio of primary:secondary = 1.7:1) and 22.5 mol % each of propylene glycol and propylene glycol diacetate. 5.3.3 Catalyst preparation Catalysts were prepared by incipient wetness. 12 Pore volume of fresh silica support was measured to be 0.69 cc/g. Basic salt catalyst was impregnated by filling the pore volume wi th a solution containing the desired loading of the catalyst material. The catalyst was dried at 12 0 ° C overnight, and then calcined for 6 hrs. at 550 ° C under a steady flow of nitrogen gas prior to use . 123 5.3.4 Fixed bed reactor and condensation system The reactor used was either a stainless steel or quartz tube of 0.5 in outer diameter and 43 cc volume. A thermoprobe (stainless steel or quartz) well wa s fitted axially along the length of the reactor to facilitate temperature readings at different positions during t he course of the reaction. The furnace was a brass tube 12 inches in length with high performance ( 313 W , 120V ) electrical heating tape wrapped snugly around it, and the rate of heating was regulated by a PID controller . Temperature readings inside the rea ctor were recorded in a data logger , initially at different positions along the length of the reactor and the pre - heat zone to get the temperature profile, and then during the remainder of the experiment, at the middle of the reactor bed zone. Borosilicate beads (2mm) or silicon carbide were used as packing material below and above the catalyst bed. Liquid feed was introduced at the top of the reactor through a tube just above the pre - heating zone. Liquid flow rate was maintained by a n HPLC pump (Varian), or in the case of experiments with PO as feed , a syringe pump was used. Inert gas nitrogen (unless otherwise mentioned) flow ed concentrically around the feed tube. Inert gas flow rate was mainta ined by a mass flow controller. The rea ctor outlet connect s to a two - chamber condenser system fitted with a three - way valve to direct product mixture to either chamber to enable continuous operation. The collection vessels (50 cc) were fitted with dip tubes and valves to collect samples during reaction. Depending on the cooling requirements of the product mixture, one or two stage condensation systems were used with either ice - water baths or at ~ - 79°C in dry ice - acetone baths. Non - condensable gases flow ed out of a vent at the top of the collecti on vessels. Flow rate out of the reactor was measured by a gas bubble meter and recorded at regular intervals to ensure continuity of flow. The reactor 124 was also fitted with a pressure ga u ge (0 - 100psig) to m onitor fluctuations in pressure in the reactor . T he reaction was run at 1 atm, but a sudden rise indicated reactor plugging . Before and after a reaction, weights of the reactor system and collected samples were recorded to calculate total mass recovery. The r eactor was heated to the required temperatur e with a steady nitrogen flow rate. After the reactor stabilize d at the required temperature, liquid feed flow was started (~0.1 mL/min). The first sample was collected after 3 h ou rs, and 3 - 4 subsequent samples were collected at 1.5 - 2 h our intervals. Sampl e collection was alternated between the two collection vessels. 5.3.5 Analysis Samples collected were diluted 10 - fold in a solution of acetonitrile containing 5wt% decanol as the internal standard. All products were quantified by a Varian 450 GC ( Varian Medica l Systems Inc., Palo Alto, CA) fitted with two detectors, a thermal conductivity detector and a flame ionization detector. Separation was done on a 0.53mm ID Aquawax - DA 30 m capillary column with 1.0µm film thickness. Helium carrier gas flow rate was set t o 10 mL min - 1 . Injector temperature was 250 of 37 30 of the TCD and FID were 240 Mass balances, propy lene glycol backbone balance (Eq. 5.1), ace tate species balances (Eq. 5.2) and overall carbon balances were done to ensure that the sample composition is representative of the reaction ; these balances were consistently above 93% closure. Carbon b alance values are reported in the Appendix H for each reaction in Error! Reference source not found. . 125 (5.1) (5.2) Conversion, PO selectivity, and PO yield were calculated from Eq. 5.3 , 5.4 , and 5.5 respectively. (5.3) (5.4) (5.5) 5.3.6 Catalyst characterization Qualitative FTIR studies of prepared and post - reaction catalysts were done. The catalyst samples were ground together with KBr in a 2% by weight mixture and pressed into a transparent tablet at 10000 psi in a dye press for approximately 5 minutes. The tablet was inserted into the FTIR spectrometer that was c ontinually purged with nitrogen . Mult iple spectra from 4000 cm - 1 to 5 00 cm - 1 were taken with each sample . Thermogravimetric analyses were done in a TGA Q500 using a Hi - Res TGA furnace control method to determine decomposition temperatures. 126 Surface areas and pore size distributions were obtained from nitrogen adsorption and desorpti on isotherms obtained from an ASAP 2010 (Micromeritics Instrument Corporation, Norcross, G eorgia). Base and acid site concentrations were determined from carbon dioxide and ammonia tempera ture programmed desorptions in an Autochem II Chemisorption Analyzer (Micromeritics Instrument Corporation, Norcross, Georgia) . SEM ima ges were obtained in a n EVO - LS - 25 (Carl Zeiss Microscopy, Thornwood, New York) variable pressure scanning electron microscope with a beam energy of 15 - 20 kV and an EDAX Pegasus camera. EDS analysis was done using TEAM software. XPS analysis was done using a Magnesium non chroma tic source with a pass energy of 189 eV for survey scans and 29.5 eV for spectra. Spectra were collected at a base pressure of 5×10 - 8 Torr . Spectra wer e fitted and analyzed using PHI Multipak (Physical Electronics Inc.) software. A take - off energy of 45 used. 5.4 Results 5.4.1 Thermodynamic Analysis of Reaction Network An estimate of the feasibility of the reaction was obtained from Gibbs energies of reactions. The details of the parameters used are reported in Appendix K . Speculative mechanisms of the majo r reactions in the s ystem are reported in Appendix J in Scheme 5 . 6 . PO is formed from PG and its acetates by intramolecular substitution. Propanal, acetone and allyl alcohol are formed from PG by elimination mechanisms. Di - propyle ne glycol and other ethers are formed by intermolecular substitution. Propanal further reacts with PG to form a secondary side product, a 1,3 - dioxolane. Allyl alcohol esterifies acetic acid to form allyl acetate. The isomerization of PO, a strained molecu le, to its side products propanal and acetone by 1,2 H shifts , and to allyl alcohol by 1,4 H 127 shifts , is also highly favorable according to Gibbs energy estimates from this study and from literature. 18,19 5.4.2 Control Reactions Preliminary reactions were carried out to understand the e ffect of thermal conditions and catalytic activity of the support. No reaction of PGMA was observed in reactions at 42 0 when the reaction was run with silica beads but no catalyst or support , indicating that the reaction is catalytic, and that the nature of the heating material used is inert. Experiments with PO as feed in the absence of any catalyst were conducted to ga u ge if there is a thermal limit to selectivity. They showed little to no conversion of PO at reaction Scheme 5 . 4 : Propylene glycol acetates deacetoxylation system 128 conditions, a result in agreement with literature. 18 However, the mass balances for these reactions were poor, owing to the high volatility of both the reactants and the products. Experiments were also conducted with PO feed with catalyst, which showed up to 35% conversion to side products at 450 but quantification of conversion rate is difficult without better product collection techniques . The boiling points of PO, propanal, acetone and allyl alcohol are 34 respectively, whereas the boiling points of PG, PGMA and PGDA are all within the range of 189 - 191 was much lower than the boiling poi nt of reaction mixtures with PG - PGMA feed. (Mass balances for experiments starting from PGMA feed were >93 %). Liquid nitrogen and dry - ice acetone baths proved unsuccessful in achieving total carbon recovery for PO isomerization experiments. Future attempt s at doing PO isomerization kinetic experiments might include trapping PO chemically. Mass left behind in reactor was calculated from weight measurements before and after reaction, and did not account for a significant fraction of the feed. The presence o f acid sites is known to strongly favor the formation of propanal from PG, as well as the isomerization of PO. Less than 1% by mole of products , consisting only of propanal , were formed with reaction or PO over untreated silica gel support . This is expecte d because of the presence of a few acidic silanol groups on silica gel , leading to acid - catalyzed elimination . This was confirmed from ammonia TPD of the silica gel support , showing 1.8 m m ol/g of acid sites for untreated silica gel, while calcined silica g el had an acid site density of 0.02 mmol/g rendering the support inert for all practical purposes . Further evidence of the near absence of hydroxyl groups in calcined silica gel support is provided f rom XPS spectra in Section 5.4.4. 8 . 129 5.4.3 Cesium nitrate on si lica gel The highest conversions reported in the literature were from cesium nitrate on silica, with a reported conversion of 52%. 12 Early runs were conducted to attempt to replicate the results from literature. Cesium nitrate on silica catalyst was prepar ed by a method as close a s possible to the literature description. 12 Reaction conditions w ere maintained at reported conditions of a weight hour space velocity of 1.7 hr - 1 and a reaction temperature of 400 . However, the reactor was repeatedly found to p lug at temperatures higher than 380 This is probably due to fusion of cesium nitrate at around 400 reacting with alkali salts . 21 Experiments conducted at or below 375 are reported in Table 5 . 1 . Table 5 . 1 : List of reactions run with 1.5 mmol/g CsNO3 on silica gel catalyst with space velocity =1.7 g feed/g catalyst/h No. Temp ( C) Conv. (%) Selectivity (%) Yield PO PO Propanal Acet one Allyl alcohol 1 351 36 10 13 18 5 4 2 351 18 29 29 16 11 5 3 352 20 26 31 17 11 5 4 a 374 46 9 34 18 9 4 5 384 29 13 43 28 12 4 6 b 375 44 13 6 4 3 5 a: partial pressure of feed reduced by half b: PG only feed use d The reactions show result s similar to other alkali metal catalysts at those conditions. Experiments at identical conditions to check the reproducibility (R 2, R3 ) demonstrate good reproducibility. As temperature rises, the selectivity of feed to propylene oxide rises and then drop s off, while selectivity to propana l increases. An increase in feed concentration in the gas phase leads to a decrease in selectivity of propylene oxide (R4), as expected . The overall yield of propylene oxide did not rise above ~ 6 % on a molar basis. 130 A co mparison experiment done at 375°C with pure propylene glycol as feed (R6) showed that the selectivity to propylene oxide was much higher than propanal selectivity. However, up to 72% of the reacted propylene glycol was consumed in the formation of ethers s uch as di propylene glycol and others, whereas with a mixture with propylene glycol acetates, selectivity to ethers was minimal. Therefore, a feed of a mixture of propylene glycol and its acetate mono and di - esters was used as feed in all successive reacti ons. Since the optimum results obtained for cesium catalysts were no higher than 5% yield, weaker base metal potassium was considered. Sodium and lithium showed poor results in literature. 13 5.4.4 Potassium salt catalysts Potassium catalysts on silica gave the next best conversion of ~45% in literature, in addition to being less expensive than cesium and rubidium. Detailed reaction studies were therefore conducted with potassium hydroxide catalyst to understand the effect of space velocity, temperature and conc entration of base on silica gel ( Table 5 . 2 ). A later experiment conducted with potassium catalyst on a stabilized, low surface area silica support, Controlled Pore Gl ass (Controlled Pore Glass Inc. ) was also compared with the sili ca gel based catalysts to determine the effect of surface area, discussed in Section 5.4.4 . 8 . 131 Table 5 . 2 : List of experiments with KOH on silica gel No. Loading (mmol K + / g catalyst) Temp ( C) WHSV (h - 1 ) Con v. (%) Selectivity (%) Yield PO PO Propanal Acetone Allyl alcohol 7 2.5 400 1.7 37 81 11 6 2 30 8 2.5 400 1.7 34 83 7 8 2 28 9 0.5 400 1.7 50 52 28 10 10 26 10 0.5 400 3.4 44 64 22 9 5 28 11 c 0.5 400 3.4 45 65 22 9 4 29 12 1.5 400 1.7 50 38 37 14 11 19 13 0.5 400 6.7 33 71 22 5 3 23 14 2.5 400 3.4 20 70 13 10 7 14 15 2.5 400 3.4 22 80 9 5 6 18 16 2.5 420 3.4 36 67 18 10 5 24 17 2.5 370 3.4 12 50 26 13 11 6 18 2.5 440 3.4 30 64 14 17 5 19 c: Methyl ethyl ketone promoter added, no improvem ent observed 5.4.4.1 Mass transport limitations The effect of intraparticle mass transport limitation was determined by calculating the observable modulus and invoking the Weisz Prater criterion 22 using data from a reaction at the lowest space velocity (highest contact time). The observable modulus 1.11×10 - 4 , rates. Details of calculations are reported in Appendix L. 5.4.4.2 Effect of weight hour space velocity Weight hour space velocity (WHSV, g feed/ g catalyst/hr), which is inversely related to contact time, was varied by increasing the amount of catalyst loaded into the reactor (R9 - 11, R 13). There is an incr ease of conversion with decrease in WHSV as expected, but selectivity increases with increasin g WHSV ( Figure 5 . 1 ), indicating that isomerization of propylene oxide to side products propanal, acetone and allyl alcohol plays an important role in selectivity. PO 132 isomerization experiments were not successful du e to due to difficulty in product collection. Additional experiments done at much higher space velocities are discussed later in Section 5.4.4 .7. Figure 5 . 1 ur space velocity. (T= 400 mmol/g of KOH in catalyst The order of reaction was estimated using reaction runs at different contact times using integrated rate equations. A first order reaction rate fits the data reasonably well; corresponding zero - order and second - order fits gave poorer results. ( Figure 5 . 2 ). Figure 5 . 2 : Rate constant vs contact time for a first order reaction A straight line through the origin fits the d ata with an R 2 value of 0 .94 133 5.4.4.3 Effect of temperature Figure 5 . 3 shows the dependence of conversion and selectivity on temperature of reaction. Conversion increases with temperature as expected until 420 to drop due to agglomeration of catalyst particles that results in poor accessibility to active sites in the catalyst. Figure 5 . 3 : emperature. (WHSV= 1.7 g feed/g cat/ h), loading = 2.5 mmol/g of KOH in catalyst) (R14 R18) Selectivity to propylene oxide is similarly shown to increase up to ~80% at lower temperatures and then decrease at higher temperatures. This is expected to resu lt from higher rates of isomerization of propylene oxide and a preference for elimination reactions to propanal, acetone , and allyl alcohol over substitution reactions to propylene oxide at higher temperatures. Optimum temperature for selectivity is 400 , whereas overall yield is highest at 420 composite activation energy is obtained from an Arrhenius plot ( Figure 5 . 4 ) of reactions over the temperature range of . 134 E a = 89.9 kJ/mole, k 0 = 3. 08×10 7 s - 1 Figure 5 . 4 : Arrhenius plot of composite reaction of PGA to prod ucts (R14 - R18) The rate constant in Figure 5.4 is derived from first order kinetics in a plug flow reactor. Ln k = ln(1 - X)/ where X is conversion on basis of PG and PGA, while is the contact time of the feed in the reaction bed. Activation energy is obtained from the slope to be 89.9 kJ/mole, while a composite pre - exponential factor is obtained from the exponent of the intercept as 3.08×10 7 s - 1 . 5.4.4.4 Effect of catalyst loading The effect of the loading of KOH on silica gel was studied (R7 R9, R12 of Table 5 . 2 ). At higher loadings, conversion was found to significantly drop , although there was a sharp increase in PO selectivity ( Figure 5 . 5 ). This is a result of the silica pore structure collapsing at higher base loading, as described below. 0 1 2 3 4 1.35 1.45 1.55 - ln(k) Temperature - 1 10 - 3 (K) 135 Figure 5 . 5 : 1.7 g feed/g cat/h) 5.4.4.5 Surface area studies Surface areas measured by BET/nitrogen method at various stages of catalyst preparation and post reaction are reported ( Table 5 . 3 ). As catalyst loading is increased, the silica support is collapsed to greater degrees, leading to a corresponding decrease in surface area of catalyst. Table 5 . 3 : Surface areas of KOH on s ilica gel catalysts and supports determined by N 2 adsorption (BET method) Catalyst Surface area (m 2 /g) S ilica gel uncalcined 5 14 Silica gel calcined 474 2.5 mmol KOH /g on SiOx ( impregnated, dried at 100 27 2.5 mmol/g KOH on SiOx calcined (fresh) 3 .0 2.5 mmol/g KOH on SiOx calcined (used) 0.2 1.5 mmol KOH/g catalyst used 12 0.5 mmol KOH/g catalyst used 94 0.5 mmol/g KOH on silica gel ( fresh, calcined ) 115 136 The compounded effect of surface area an d catalyst loading is considered in greater detail in Section 5.4. 4 .8. 5.4.4.6 Chemical nature of catalyst during reaction Alkali metal salts on silica can react or remain unchanged, depending on a number of factors including temperature, basicity and loading of alkali metal, and type of support used , which in turn affects the activity and selectivity of the catalyst . Acetic acid is liberated during the conversion of propylene glycol acetates. It may therefore be hypothesized that potassium hydroxide on the suppo rt is neutralized to potassium acetate, which in turn ketonizes to potassium carbonate with the liberation of acetone at ~400 Scheme 5 . 5 . Potassium carbonate may be the stable form of the cataly st under reaction conditions. 137 Scheme 5 . 5 : Formation of potassium carbonate on surface of catalyst during reaction Thermogravimetric analysis of neat potassium acetate shows a sharp weight loss at approx imately 400 Figure 5 . 6 ). 138 Figure 5 . 6 :Thermogravimetric analyses 1. ( ) KOAc on silica, 2. silica gel , 3. ( ) neat KOAc The weight loss recorded is equal to the expected stoich iometric loss of acetone in the conversion of potassium acetate to potassium carbonate via ketonization (Reaction 3 of Scheme 5 . 5 ). Silica gel loaded with potassium acetate undergoes sharp weight losses at 120 250 380 loss peak in silica gel alone at the same temperature. The second drop is the loss of water remaining from the preparation stage, and the third i s the ketonization of potassium acetate to potassium carbonate. Calculations show that the third peak is equal to the expected stoichiometric loss in weight from acetone (29.3%) (calculations in Appendix M ). Unfortunately, the amount of acetone liberated from the catalyst duri ng experiments (expected to be ~0.5 - 2.5 mmol) was miniscule comparison to the acetone produced as a side product of formation of propylene oxide to be accurately quantified. The TGA of silica gel shows the loss of hydroxyl groups with increase in temperature. K 2 CO 3 has a decomposition temperature higher than 800 21 Acetic acid liberated 139 during reaction apparently reacts with potassium hydroxide to form potassium acetate, which subsequently becomes potassium carbonate after ketoniza tion ( Scheme 5 . 5 ). 5.4.4.7 E xperiments with KOAc, K 2 CO 3 and K 2 S iO 3 based catalysts At loadings of catalyst higher than 2.5 mmol KOH /g, the catalyst was found to plug the reactor similar to cesium nitrate from dissolution of silica, forming glassy non porous clumps at room temperature at loadings higher than 5 mmol KOH/g. In order to make catalysts with higher concentrations of base salt on support, catalysts made with weaker basic salt KOAc were tested and are reported in Table 5 . 4 . A loading of 5 mmol/g KOAc on silica was found to have no significant improvement over 2.5 mmol/g KOAc on silica (R19 and R20). It is also observed that the activity of the KOAc catalyst is very similar to a KOH cataly st of the same loading (R14 and R15 of Table 5 . 2 ). Table 5 . 4 : Experiments conducted with KOAc at temperature of 400C and WHSV = 3.4 g feed/ g cat/h No. Loading (mmol K + / g catalyst) Temp ( C) WHSV (h - 1 ) Conv. (%) Selectivity (%) Yield PO PO Propanal Acetone Allyl alcohol 19 2.5 400 3.4 21 72 12 16 0 15 20 5.0 400 3.4 22 60 14 16 10 13 Reactions using K 2 CO 3 as base salts on support were then carried out to compare their act ivity and selectivity with the KOH and KOAc catalysts. These along with other reactions using K 2 CO 3 are reported in Table 5 . 5 . 140 Table 5 . 5 : Experiments conducted with 1.25 mmol K 2 CO 3 /g catalyst at 400 No. Catalyst base on silica gel WHSV (h - 1 ) Conv. (%) Selectivity (%) Yield PO PO Propanal Acetone Allyl alcohol 21 K 2 CO 3 3.4 25 67 25 7 3 17 22 d K 2 CO 3 3.4 30 63 26 9 3 19 23 K 2 CO 3 17.1 12 84 12 3 2 29 24 d K 2 CO 3 17.1 8 88 9 2 0.2 20 d: carbon dioxide used as feed carrier gas instead of nitrogen e: partial pressure of feed reduced by half R14, R19 and R21 with KOH, KOAc and K 2 CO 3 loaded on silica show similar conversion and selectivity at identical reaction conditions. An additional experime nt carried out (R 22) to determine the effect of carbon dioxide as feed carrier gas was done, however, no significant improvement was observed in the reaction yield or conversion. Additional experiments done with very high weight hour space velocities of 17.1 h - 1 , or short contact times of 0.56 s (R23, R24 in Table 5 . 5 ) yielded selectivity as high as 88%, but with correspondingly lower PGMA conversion of 8% by mole. 5.4.4.8 FTIR and XPS analysis of catalysts To ascertain the presence o f carbonate species, FTIR spectra of fresh and used KOH and KOAc catalysts were obtained ( Figure 5 . 7 ). 141 Figure 5 . 7 : Absorbance FTIR analyses of fresh and used catalyst samples. (1: 1.25 mmol/g K 2 CO 3 on silica, 2: post reaction 2.5 mmol KOAc/g on silica, 3: post reaction 2.5 mmol/g KOH on silica, 4: unused 2.5 mmol/g KOH on silica) The sharp peak at 138 c m - 1 has been previously characterized as the presence of bulk carbonate, 23 while 1541 cm - 1 is associated with supported carbonate. No peak is observed at 1541 cm - 1 , associated with bicarbonate or supported carbonate species on silica. 23 Si - O - Si bending in silica is observed at 810 cm - 1 in the silica gel spectrum. The large peak at 1091 cm - 1 is attributed to Si - O - Si bond stretching, 23 26 with the shoulder being attributed to skeletal stretching. 26 The presence of Si - O - K bonds in potassium silicate would shift the stretching peak to lower wave numbers due to the lengthening of the bond, 26,27 but the Si - O - Si stretching peak remains at 1091 cm - 1 indicating that no significant potassium silicate is present. IR spectra of neat potassium silicate and potassium silicate on silica gel show a characteristic peak at 958 cm - 1 , not seen in fre sh 142 or used spectra of any loading of KOH, KOAc or K 2 CO 3 catalysts. There is, however, a shift to lower wave numbers of the Si - O - Si bending peak to 788 cm - 1 , indicating an increase in proportion of non - bonding oxygens. 26 FTIR spectra of used samples of low er loadings of KOH (0.5 mmol and 1.5 mmol /g catalyst), and K2CO3 on CPG are shown in Figure 5 . 8 . Figure 5 . 8 : Absorbance FTIR analyses of fresh and used catalys t samples. (1: silica gel support, 2: calcined fresh 1.5 mmol KOH/g on silica, 3: post reaction 1.5 mmol/g KOH on silica, 4: unused 1.25 mmol/g K 2 SiO 3 on silica, 5: neat K 2 SiO 3 ) There is no bulk carbonate peak observed in lower loading KOH catalysts and in potassium carbonate on CPG. Calculations done to estimate monolayer coverage of catalyst show that there is less than monolayer coverage of potassium salts on the surface for the catalysts with lower loadings and C ontrolled Pore G lass due to their much higher surface areas. 143 No peak is observed at 1541 cm - 1 , associated with bicarbonate or supported carbonate species on silica. 23 It therefore becomes probable that there are some Si - O - K bonds formed in the micropores of the silica gel, but not enough to be visible on the IR spectra. There is no prominent silicate peak observed on these catalysts either, although there is a slight bump on the lower wavenumber side of the Si - O - Si peak at 1091 cm - 1 . There is no significant shift in Si - O - Si stretching or bendin g peaks. In order to ascertain the presence of Si - O - K bonds on lower loading KOH catalysts or on K 2 CO 3 on controlled pore glass, XPS spectra of post reaction samples of those catalysts were obtained. Table 5 . 6 : Binding energy peaks from XPS spectra for calcined support silica gel, neat K 2 O:SiO 2 and neat K 2 CO 3 . Sample K 2 CO 3 K 2 O:SiO 2 Silica gel calcined Binding Energy Area % Binding Energy Area % Binding Energy Area % C 1s 284.4 14.7 283.4 7.1 28 5.3 32.7 284.8 83.6 287.1 51.4 286.3 0.6 288.8 1.1 287.8 8.8 O 1s 531.7 16.8 530.6 43.7 533.1 533.1 64.3 532 56.4 534.8 18.9 Si 2p NA NA 103.9 K 2p 292.5 63.6 292.6 295.3 28.7 295.4 Silica gel calcined at 550 for 6 hours ( Table 5 . 6 ) shows only one peak in its Si 2p spectra at 103.9 eV, and one peak in its O 1s spectra at 533.1 eV. Both these are attributed to vitreous silica with no non - bridging oxygens. 28 The binding energy of Si atoms depends on the nature of the O atoms bonded to it. Since K atoms are much less electronegative than Si atoms, the valence 144 electron density around a non - bridging O atom is higher. The mutual screening leads to destabilization, leading to lower bin ding energies. The Si 2p spectra for K 2 O:SiO 2 shows lower values of binding energy for Si atoms in the presence of K 2 O. The O 1s values of neat potassium silicate show a peak for bridging O (533.1 eV) and another peak for non - bridging O attached to K atoms at lower binding energies of 528 - 530 eV. 29 Carbonate and bicarbonate peaks for C 1s spectra are at 287 eV and 289 eV respectively. 30 This is confirmed from the samples of neat K 2 O:SiO 2 and neat K 2 CO 3 . The C1s spectra of K 2 O:SiO 2 shows some adventitious ca rbon probably deposited during analysis at 284 - 285 eV. Table 5 . 7 : Binding energy peaks from XPS spectra for 2.5 mmol K + /g catalyst made from KOH, KOAC and K 2 CO 3 . Sample 2.5 mmol/g KOH on silica gel 2.5 mmo l/g KOAc on silica gel 1.25 mmol/g K 2 CO 3 on silica gel 1.25 mmol/g K 2 CO 3 on silica gel Post reaction Post reaction Calcined, unused Post reaction Binding Energy Area % Binding Energy Area % Binding Energy Area % Binding Energy Area % C 1s 285. 3 54.6 284.9 29.4 285.2 15.6 283.9 15.8 286.4 34.6 286.8 70.6 286.9 84.4 284.9 47.4 289.2 10.8 286.9 27.9 288.4 5.4 289.8 3.4 O 1s 530.7 4.2 530 1 530.8 7.7 531.1 8.6 532.2 56 532.3 18 532.1 31 532.5 27.2 5 33.7 34.5 533.7 40.9 533.4 44.6 533.7 19.7 534.5 5.4 534.5 40.2 534.5 16.8 534.5 44.6 Table 5 . 7 shows the binding energies from XPS spectra for 2.5 mmol K + /g catalyst made from KOH, KOAC and K 2 CO 3 . The peaks from C 1s sp ectra at 286.4 - 286.9 eV may be attributed to the bulk carbonate found on these catalysts from post reaction FTIR runs, along with lower peaks 284 - 285 eV attributed to lower oxidation states of C 1s, are probably due to carbonaceous 145 deposits during reaction or contamination during XPS analysis. Strong peaks at 532.2 - 534.5 eV are attributed to bridging O atoms in the silica support bulk and the bulk carbonate species. There are probably minor amounts of Si - O - K bonds present, as visible from lower peaks at ~5 30 eV 26 . Table 5 . 8 shows binding energy peaks for lower loadings of KOH on silica gel (0.5 and 1.5 mmol/g). The C 1s spectra for 0.5 mmol/g KOH do not show any significant amounts of peaks at 287 eV or higher both before and after reaction, whereas 1.5 mmol/g KOH on silica gel shows small amounts of carbonate at 286.5 eV. Definite peaks at 528 - 530 eV confirms the presence of Si - O - K bonds that prevail during reaction for both lower loadings on silica gel and controlled pore glass ( CPG) catalysts. Table 5 . 8 : Binding energy peaks from XPS spectra for lower loadings of KOH on silica gel (0.5 and 1.5 mmol/g) Sample 0.5 mmol/g KOH on silica gel 0.5 mmol/g KOH on silica gel 1.5 mmol/g KOH on silica gel Calcined, unused Post reaction Post reaction Binding Energy Area % Binding Energy Area % Binding Energy Area % C 1s 281.7 86.8 282.5 29.4 283.3 41.8 285.1 13.2 284.9 27.5 285 40.1 286.4 2.8 286.5 13.5 287.9 1. 9 288 4.7 O 1s 528.5 4.7 528.7 3.5 528.9 2 531.2 6.2 530.1 11.2 530.7 8.3 532.8 89.1 531.4 23.1 532.1 32.5 532.5 53 533.1 47.6 533.6 9.2 534.3 9.6 A potassium silicate catalyst of loading 1.5 mmol/g shows similar activities as the lower loadings of KOH on silica gel (R25 of Table 5 . 9 ). This is further proof of the chemical nature of those catalysts. 146 Table 5 . 9 : Potassium silicate on silica gel at 400 N o. Catalyst base on silica gel Loading (mmol K + / g catalyst) WHSV (h - 1 ) Conv. (%) Selectivity ( % ) Yield PO PO Propanal Acetone Allyl alcohol 25 K 2 SiO 3 1.5 3.4 53 42 33 10 15 22 The above findings indicate that in the case of lower loadings o f KOH on silica gel, the primary active component is potassium silicate. At higher loadings (>1.5 mmol K/g catalyst), there is a substantial change to potassium carbonate, which is present as bulk on the surface of the catalyst, and is the primary active c omponent. 5.4.4.9 Monolayer coverage calculations and controlled pore glass catalysts The effect of the loading of KOH on silica gel was studied earlier on in our experiments (R7 - 9, and R12 of Table 5 . 2 ). At higher loadings, conversion wa s found to significantly drop , although there was a sharp increase in PO selectivity ( Figure 5 . 5 ). The effect of catalyst loading is compounded with the effect of catalyst surface area. As catalyst loading is increased, the silica support is collapsed to greater degrees, leading to a corresponding decrease in surface area of catalyst ( Table 5 . 3 in section 5.4. 4 .5 and Table 5 . 10 below). This prompted a more detailed look into the n ature of the catalysts when impacted by one of the two or both factors. 147 Table 5 . 10 : Surface areas of catalysts and supports determined by N 2 adsorption (BET method) Catalyst Surface area (m 2 /g) 1.25 mmol/g K 2 CO 3 on SiOx calcined (fresh) 2.5 Unsupported K 2 CO 3 3.0 Controlled pore glass (CPG) 37 1.25 mmol/g K 2 CO 3 on CPG calcined (fresh) 34 1.25 mmol/g K 2 CO 3 on CPG calcined (used ) 32 Among the silica gel catalysts, simple calculations show that at the loa ding equal to full monolayer coverage of the two dimensional silica surface, the conversion is the highest. Assuming a hexagonal or cuboid arrangement of silica molecules, the number of moles of Si - O - bonds available on the surface approximated to a plain two dimensional areas was calculated. This gives us the maximum loading for a mo nolayer coverage. This optimal monolayer value is a loading of 1.5 mmol/g KOH, in which the surface area of the support is decreased to 12 m 2 /g, which is seen in Figure 5 . 10 . Surface area measurements of catalyst and support show an almost complete collapse of silica surface area (from 514 m 2 /g to 0.2 m 2 /g) at high (2.5 mmol/g) loadings of KOH. There is some collapse to 27 m 2 /g on impre gnation of KOH, but after calcination, the surface area is further reduced to 3.0 m 2 /g. This is attributed to temperature dependent structural modification of silica in presence of alkali salts. 31 The surface area of the neat potassium carbonate salt used in experiments was measured to be 2.5 m 2 /g, nearly identical to loaded potassium carbonate. Reactions R21in Table 5 . 5 and R25 in Table 5 . 11 compare the activity and selectivity of the same 148 molar loading o f neat K 2 CO 3 and 1.25 mmol/g K 2 CO 3 on silica with identical contact time and temperature. Table 5 . 11 : Experiments done to determine the effect of surface area with 1.25 mmol/g K 2 CO 3 No. Support Temp ( C) WHSV (h - 1 ) Conv. (%) Selectivity (%) Yield PO PO Propanal Acetone Allyl alcohol 26 f Neat 400 1.7 29 53 30 15 2 12 27 g CPG 400 3.4 44 75 14 7 4 33 f: g: neat potassium carbonate used without silica support potassium carbonate on controlled pore glass with higher surface area support Turn over numbers are a measure of moles reacted per mole of base site on catalyst in unit time, and are used to compare the activity of different catalysts. Neat K 2 CO 3 has a turn over number of ~3.40 h - 1 and loaded K 2 CO 3 has a turn over number of ~7.1 h - 1 . Although neat K 2 CO 3 has more moles of K + by weight, they are less dispersed. The surface density of neat K 2 CO 3 is 0.19 - 0.2 mmol/g of catalyst, whereas the saturated monolayer surface density of supported K 2 CO 3 is 0.25 - 0 .29 mmol/g. Reactions R24 and R26 compare the effect of support surface area: both activity and selectivity of the controlled pore glass is higher, indicating that higher surface areas and higher loadings are optimal for selectivity. Separate experiments with controlled pore glass were done (R 26) to separate the effect of surface area and catalyst loading. In order to disperse higher loadings, higher surface area supports that are chemically stable were required. Controlled pore glass was much more stabl e towards the attack of potassium base salts, with a surface area of 32 m 2 /g such that at loadings as high as 2.5 mmoles/g K+, the surface of the catalyst was not fully covered, and no drop in conversion was observed (R 26), in fact, the advantage of highe r loadings, the higher selectivity is also achieved. 149 The turn over number for K 2 CO 3 on CPG is 10.4 h - 1 . This is much higher than the aforementioned turn over numbers for neat K 2 CO 3 and loaded K 2 CO 3 on silica gel. FTIR spectra of CPG catalysts are shown in Figure 5 . 9 . There is a distinct peak at the same wavenumber as the peak attributed to bulk potassium silicate at 964 cm - 1 . There is no shift in the Si - O - Si stretching peak at 1091 cm - 1 . This indicates that the K 2 S iO 3 observed on the post reaction sample of K 2 CO 3 supported on CPG is bulk silicate. There is no visible peak for bulk carbonate at 1385 cm - 1 for spectra of fresh or post reaction catalyst. Figure 5 . 9 : A bsorbance FTIR analyses of fresh and used catalyst samples. (1: K 2 SiO 3 on silica gel, 2:1.25 mmol/g K 2 CO 3 on CPG post reaction, 3: 1.25 mmol/g K 2 CO 3 on CPG calcined, unused, 4: K 2 SiO 3 , neat) 150 Table 5 . 12 : Bind ing energy peaks from XPS spectra for K 2 CO 3 on CPG before and after use Sample 1.25 mmol/g K 2 CO 3 on CPG 1.25 mmol/g K 2 CO 3 on CPG Calcined, unused Post reaction Binding Energy Area % Binding Energy Area % C 1s 281.4 31.2 281.4 14.9 284.9 61.9 2 84.8 33.9 286.4 3.2 286.3 3.1 287.9 3.8 287.8 2.7 O 1s 530 5.4 528.5 1.7 531.8 37.8 530.2 7 532.6 56.8 531.6 28.5 532.5 54.7 533.7 8.2 Si 2p 101.6 7.4 103.2 92.7 Table 5 . 12 shows binding energy p eaks for K 2 CO 3 on CPG before and after use. As expected, K 2 CO 3 on CPG shows small amounts of carbonate on its surface. Peaks at 528 - 530 eV post reaction confirms the presence of Si - O - K bonds that prevail during reaction for controlled pore glass (CPG) cata lyst. 151 Figure 5 . 10 : of potassium base salt on silica surface. (T= 400 ). Percentage coverage >100% if surface is fully saturated, and more than one layer of potassium salt is present Figure 5 . 10 shows the dependence of conversion and PO selectivity on the coverage of potassium salt on the support. At coverage of more than a monolayer, conversion steadily decreases while PO selectivity increases. This effect has been recorded previously for alkali metal salts on silica. 32,33 As was seen in Section 5.4.4 . 8 , the catalysts loaded with m ore than a monolayer show the formation of bulk carbonate species, while those that have less than the full coverage do not show the presence of a bulk carbonate peak. This indicates that the active component of catalyst in the cases of more than monolayer coverage is potassium carbonate, which is a selective catalyst for PO, but has poor activity. PO selectivity is not merely a function of conversion. Potassium carbonate loaded on controlled pore glass shows similar high selectivity at high conversions. C onversion appears to be a strong function of percentage surface area of support covered. Irrespective of the loading, conversion is high ~50%, as long as surface coverage < 100%. In the case of silica gel supported KOH catalysts, it is possible that only t he base on the outside of the 152 catalyst is converted to K 2 CO 3 , while the basic species in the first layer is K - O - Si bonds, visible as 528 - 530 eV peak in O 1s XPS spectra. Table 5 . 13 shows the calculations for Figure 5 . 10 . Table 5 . 13 : Monolayer coverage calculations Catalyst Surface area (m 2 /g) Monolayer surface loading (mmol/g) No. of layers of base deposited monolayer surface density (mmol /m 2 ) Conversion PO selectivity 2.5 mmol/g KOH on silica gel 2 0.265 9 .4 0.13 22 80 Neat potassium carbonate 3 0.4 6.3 0.13 29 53 1.5 mmol/g KOH on silica gel 12 1.615 0.93 0.13 50 38 0.5 mmol/g KOH on silica gel 85 9.4 0.05 0.0059 50 52 1.25 mmoles K2 CO3 on CPG 32 4.3 0.58 0.078 44 75 EDS and XPS atomic concentrations provide atomic concentration data for the post reaction catalysts and are reported in Table 5 . 18 through Table 5 . 20 in Appendix M . 5.5 Conclusions and recommendations for future work The reaction system of propylene glycol and its acetates to propylene oxide and its side products in the presences of alkali metal salts on silica gel has been characterized. Early reactions with cesium nitra te on silica gel were unsuccessful , due to the dissolution of silica gel at high temperatures. A maximum yield of 6% was achieved at lower temperatures. Potassium base salts were found to be less corrosive to the support up to loadings of 3 mmol/g. The eff ect of weight 153 hour space velocity was studied, and the reaction was determined to be first order with respect to CPGA. Conversion decreases with decreasing contact time, however, selectivity increases, indicating that PO isomerization is an important react ion in the system. The effect of temperature was studied; conversion increases with temperature up to 420 and then drops due to accelerated agglomeration of catalyst particles. A composite activation energy of Ea = 89.9 kJ/mole was obtained based on first order kinetics. Selectivity decreases as temperature is increased, this could be due to the domination of elimination reactions to propanal, acetone and allyl alcohol over substitution reaction to propylene oxide, and increased PO isomerization. The trade - off between selectivity and conversion reaches an optimum value of 32% at 420 C. The activity of KOH, KOAc and K 2 CO 3 based catalysts on silica gel are similar. FTIR and XPS studies confirm that for loadings of higher than monolayer coverage the active component at reaction conditions is bulk carbonate, irrespective of the startin g material. For lower loadings that are sub monolayer, the active component consists of Si - O - K bonds. Surface area studies of the catalyst at different stages of preparation and post reaction have been studied. Monolayer coverage calculations of different catalysts show that sub monolayer coverage loadings are optimal for high conversions. Controlled pore glass is a stable high surface area support that allows high conversions without sacrificing selectivity. FTIR and XPS spectra show that the active compon ent of K 2 CO 3 loaded CPG catalyst is potassium silicate at reaction conditions. The high conversion and selectivity of catalysts with CPG as support is promising for further optimization of PO yields. High surface areas allow for monolayer coverage even at high alkali loadings, making high conversions possible along with high selectivity. Future studies at higher loadings of potassium for full coverage, and higher surface areas might be conducted. 154 Characterizing PO isomerization kinetics would be helpful to understand the limit of PO yield achievable. 155 APPENDICES 156 Appendix H : List of all experiments with reaction conditions, conversion, selectivity and carbon recovery Table 5 . 14 : List of all experiments with reaction conditions, conversion, selectivity and carbon recovery No Metal Loading Feed rate Wt. Carrier gas Temp WHSV Conv. Selectivity Yield PO C (mmol M/ g catalyst) (g/min) (g) (g/min) ( C) (g feed/g cat/h) (%) (%) rec N 2 unless otherwise stated PO Propanal Acetone Allyl alcohol (%) Cesium nitrate catalyst runs 1 1.5 0.1 3.5 0.1 351 1.7 36 10 13 18 5 4 86 2 1.5 0.1 3.5 0.1 351 1.7 18 29 29 16 11 5 95 3 1.5 0.1 3.5 0.1 352 1.7 20 26 31 17 11 5 90 4 a 1.5 0.1 3.5 0.2 374 1.7 46 9 34 18 9 4 97 5 1.5 0.1 3.5 0.1 384 1.7 29 13 43 28 12 4 96 6 b 1.5 0.1 3.5 0.1 375 1.7 44 13 6 4 3 5 78 Potassium hydroxide catalyst runs 7 2.5 0.1 3.5 0.1 400 1.7 37 81 11 6 2 30 101 8 2.5 0.1 3.5 0.1 400 1.7 34 83 7 8 2 28 99 9 0.5 0.1 3.5 0.1 400 1.7 50 52 28 10 10 26 93 10 0.5 0.1 1.8 0.1 400 3.4 44 64 22 9 5 28 98 11 c 0.5 0.1 1.8 0.1 400 3.4 45 65 22 9 4 29 97 12 1.5 0.1 1.8 0.1 400 3.4 50 38 37 14 11 19 100 13 0.5 0.1 0.9 0.1 400 6 .7 33 71 22 5 3 23 99 14 2.5 0.1 1.8 0.1 400 3.4 20 70 13 10 7 14 99 15 2.5 0.1 1.8 0.1 400 3.4 22 80 9 5 6 18 99 16 2.5 0.1 1.8 0.1 420 3.4 36 67 18 10 5 24 100 17 2.5 0.1 1.8 0.1 370 3.4 12 50 26 13 11 6 100 18 2.5 0.1 1.8 0.1 440 3.4 30 64 14 17 5 19 101 157 Potassium acetate catalyst runs 19 2.5 0.1 1.8 0.1 400 3.4 21 72 12 16 0 15 95 20 5 0.1 1.8 0.1 400 3.4 22 60 14 16 10 13 101 Potassium carbonate runs 21 2.5 0.1 1.8 0.1 400 3.4 25 67 25 7 3 17 98 22 d 2.5 0.1 1.8 0.1 400 3. 4 30 63 26 9 3 19 114 23 2.5 0.1 0.4 0.1 400 17.1 12 84 12 3 2 29 109 24 e 2.5 0.1 0.4 0.1 400 17.1 8 88 9 2 0.2 20 94 25 2.5 0.1 1.8 0.1 400 3.4 53 42 33 10 15 22 98 26 f 2.5 0.1 3.5 0.1 400 1.7 29 53 30 15 2 12 96 27 g 2.5 0.1 1.8 0.1 400 3.4 44 75 14 7 4 33 100 a: partial pressure of feed reduced by half b: PG only feed used c: feed mixed with 40% by mole methyl ethyl ketone found to promote reaction in previous study. 13 No significant improvement found in present study. d: carbon dioxide used as fe ed carrier gas instead of nitrogen e: partial pressure of feed reduced by half f: neat potassium carbonate used without silica support g: potassium carbonate on controlled pore glass with higher surface area support 158 Appendix I: Calibration plots for all c omponents in reaction system The following plots ( Figure 5 . 11 through Figure 5 . 20 ) show calibrations of standards prepared to determine response factors of each component in the reaction system, determine d by the slope of area ratio over weight ratio of component over internal standard decanol. Figure 5 . 11 : Plot of area ratio vs weight ratio of PO over internal standard decanol Figure 5 . 12 : Plot of area ratio vs weight ratio of propana l over internal standard decanol 159 Figure 5 . 13 : Plot of area ratio vs weight ratio of water over internal standard deca nol Figure 5 . 14 : Plot of area ratio vs weight ratio of allyl alcohol over internal standard decanol Figure 5 . 15 : Plot of area ratio vs weight ratio of propylene glycol over internal standard decanol 160 Figure 5 . 16 : Plot of area ratio vs weight ratio of acetone over internal standard decanol Figure 5 . 17 : Plot of area ratio vs weight ratio of acetic acid over internal standard decanol Figure 5 . 18 : Plot of area ratio vs weight ratio of propylene glycol diacetate over internal standard decanol 161 Figure 5 . 19 : Plot of area ratio vs weight ratio of allyl acetate over internal standard decanol Figure 5 . 20 : Plot of area ratio vs weight ratio of 2 - ethyl,4 - methy l,1,3 - dioxolane over internal standard decanol 162 Appendix J: Mechanisms for major reactions in reaction system Scheme 5 . 6 : M echanisms for formation of PO and its isomers F ormation of PO from PGA Formation of acetone from PGA Formation of propanal from PGA Formation of allyl alcohol from PGA PO isomerization to propanal PO isomerization to acetone 163 PO isomerization to allyl alcohol 164 Appendix K: Gibbs energy of reaction and equilibrium constants at reaction temperature Gibbs energies of formation in the gas phase were calculated for each component in the reaction system by a polynomial function (Eq. 5.6 ) found in literature ( Table 5 . 15 ). 34 The Gibbs energies of the major reactions and equilibrium constants for major reactions in the system were then calculated at reaction temperature of 673 K ( Table 5 . 16 ). (5.6) Table 5 . 15 : Gibbs energy polynomial function values and Gibbs energy of formation at standard conditions (298 K) and at reaction temperature (673 K) comp onent A B C D E G f (298 K ) ( kJ/mol ) G f (673 K ) (kJ/mol) PO - 8.2E+01 1.4E - 01 1.8E - 04 - 1.7E - 07 2.8E - 11 - 27.28 49.41 PG - 4.2E+02 3.3E - 01 2.8E - 04 - 1.9E - 07 3.6E - 11 - 300.78 - 120.39 Propanal - 1.8E+02 1.4E - 01 1.5E - 04 - 9.8E - 08 2.3E - 11 - 124.23 - 39.58 Acetone - 2.1E+02 1.4E - 01 1.7E - 04 - 1.1E - 07 2.5E - 11 - 152.63 - 63.64 Water - 2.4E+02 3.5E - 02 2.0E - 05 - 9.3E - 09 1.8E - 12 - 228.59 - 210.3 acetic acid - 4.2E+02 1.0E - 01 1.4E - 04 9.5E - 08 2.3E - 11 - 377.89 - 257.18 allyl alcohol - 1.2E+02 1.3E - 01 1.7E - 04 - 1.1E - 07 2.6E - 11 - 71.23 12 .45 PGDA - 7.9E+02 4.5E - 01 3.0E - 04 - 2.1E - 07 5.3E - 11 - 631.09 - 398.28 PGMA a - 6.0E+02 3.9E - 01 2.9E - 04 - 2.0E - 07 4.5E - 11 - 465.93 - 259.33 diPG 6.0E+02 5.7E - 01 3.9E - 04 - 2.7E - 07 6.6E - 11 802.40 1097.05 a:Approximated with average for PG and PGDA Table 5 . 16 : Equilibrium constants estimated from Gibbs energy of reaction calculations at reaction temperature of 673 K Reaction K eq at 673 K PG PO +water 1 . 3 ×10 3 PG propanal + water 1.1 ×10 10 PG allyl alcohol +wat er 1 .0×10 6 PG acetone +water 8. 3×10 11 PO propanal 8 .2×10 6 PO acetone 6.0×10 8 PO allyl alcohol 7 40 165 Appendix L: Weisz Prater calculation Mass transport limitations were calculated using the Weisz - Prater criterion. The observable modulus was fi rst calculated by Eq. 5.7 . (5.7) Here, r* obs is the observed rate of reaction per weight of catalyst calculated by moles reacted/time/weight of catalyst bed = 2.5×10 - 6 moles/g cat/s for a conversion of 40% at a WHSV of 1.7 h - 1 . The bulk density of the catalyst was measured in the laboratory to be 774 kg/m 3 . Assuming a void volume of 0.5, the particle density CAT , is then 1548 kg/m 3 . The particle diameter, d p , of the silica gel catalysts was between 0.2 to 0.5 mm, with an average value of 0.35 mm. C PGA is calculated as F PGA PGA is the feed rate of propylene glycol and its acetates the inert carrier gas nitrogen. The value of C PGA = 8.41 moles/m 3 . The effective diffusivity D eff of feed in nitrogen is estimated (Eq. 5.8 ), where pore tortuosity is assumed to be equ al to the inverse of particle porosity , and D BA is bulk diffusivity of feed in nitrogen. (5.8) D PGA is determined from the Chapman - Enskog relation expressed in Eq. 5.9 : 166 (5.9) where 1 and 2 are the two molecules present in the gas mixture, in this case PGMA and nitrogen respectively, T is temperature in Kelvin (673 K), M 1 and M 2 are the molar masses of the components in g/mole (118 and 28 g/mole re 2 12 is the average collision diameter (calculated from molecular volumes to be 3.11 Å). o of void volume in the catalyst particle over the total volume of the catalyst particle = 0.422. The calculated value of D eff is 9.6×10 - 6 m 2 /s. The observable modulus, 1.1×10 - 7 . The Thiele modulus ( ) and effectiveness factor ( ) for PGA epoxid ation are calculate d from the observable modulus assuming the reaction is first order in PGA and thus . Value of calculated is 0.99, which is within the range of negligible intraparticle mass transport limitations, and th e observed rate is close to the intrinsic value. 167 Appendix M: TGA curve calculations, atomic concentrations from XPS and EDS analyses Calculations from TGA curves shown in Figure 5 . 6 show that potassium acetate ketonizes to form potassium carbonate at reaction conditions. Neat potassium acetate weight before : 22.05 mg Neat potassium acetate weight after:15.07 mg Percentage weight loss = 29.51% Supported potassium acetate before: 44.23 mg Supported potassium acetate after: 41.3459 mg Percentage weight loss observed = 6.5% Percentage expected weight loss = ( total weight) ( loading of KOAc per gram of ca talyst) * .2959/total weight = 7.25% Sample XPS spectra are shown in Figure 5 . 21 and Figure 5 . 22 . Figure 5 . 21 : XPS C 1s spectra of 1.5 mmol/g KOH on silica gel post reaction 168 Figure 5 . 22 : XPS O 1s spectra of 1.25 mmol/g K 2 CO 3 on CPG post reaction Atomic concentrations from XPS and EDS analyses are reported below in Table 5 . 17 through Table 5 . 19 . Table 5 . 17 : Atomic concentrations from XPS spectra for 2.5 mmol K + /g catalyst made from KOH, KOAC and K 2 CO 3 Silica gel calcined 2.5 mmol/g KOH on silica gel Post reaction 2.5 mmol/g KOAc on silica gel Post reaction 1.25 mmol/g K 2 CO 3 on silica gel Calcined, unused 1.25 mmol/g K 2 CO 3 on silica gel Post reaction C 0 60.33 22.65 30.77 52.44 O 70.79 28.3 52.41 45.68 32.94 Si 29.21 4.67 17.05 14.78 10.51 K 0 6.34 7.89 5.75 3.92 O/Si 2.4 6.1 3.1 3.1 3.1 169 Table 5 . 18 : XPS spectra for lower loadings of KOH on silica gel (0.5 and 1.5 mmol/g) 0.5 mmol/g KOH on silica gel Calcined, unused 0.5 mmol/g KOH on silica gel Post reaction 1.5 mmol/g KOH on silica gel Post reaction C 7.28 32.56 16.61 O 66.52 45.86 58.03 Si 24.50 17.48 23.72 K 1.7 4.10 1.64 O/Si 2.7 2.6 2.4 Table 5 . 19 : Atomic concentrations from X PS spectra for K 2 CO 3 on CPG before and after use 1.25 mmol/g K 2 CO 3 on CPG 1.25 mmol/g K 2 CO 3 on CPG Calcined, unused Post reaction C 12.86 16.56 O 58.22 56.35 Si 25.04 24.22 K 3.88 2.88 O/Si 2.3 2.3 Atomic concentrations obtained from EDS analyses in 3 - 4 selected areas as highlighted in SEM images ( Table 5 . 20 ). Figure 5 . 23 through Figure 5 . 25 show SEM images for post reactio n catalyst samples. 170 Table 5 . 20 : Atomic concentrations from EDS spectra for 2.5 mmol K + /g catalyst made from KOH, KOAC and K 2 CO 3 . 2.5 mmol/g KOH on silica gel Calcined, unused 2.5 mmol/g KOH on silica gel Post reaction 2.5 mmol/g KOH on silica gel Calcined, unused 2.5 mmol/g KOH on silica gel Post reaction 1.25 mmol/g K 2 CO 3 on silica gel Calcined, unused 1.25 mmol/g K 2 CO 3 on silica gel Post reaction C 4.5±4.5 22.9±5.9 22.9±10.3 24.0±5.4 17.2±18.8 19.6±5.7 O 61.4±0.04 53.2±3.3 52.4±7.1 53.4±2.8 53.1±8.2 51.0±3.2 Si 31.7±1.4 20.2±2.7 16.5±4.7 15.6±3.7 2.7±1.9 25.1±2.4 K 4.5±0.5 3.7±0.8 5.6±1.1 7.0±1.1 0.5±0.7 4.9±1.2 Figure 5 . 23 : SEM image of post reacti on 2.5 mmol/g KOH on silica gel 171 Figure 5 . 24 : SEM image of post reaction 2.5 mmol/g KOAc on silica gel Figure 5 . 25 : SEM image of post reaction 1.25 mmo l/g K 2 CO 3 on silica gel 172 REFERENCES 173 REFERENCES (1) Lab Breakthrough: ADM Leads to Petroleum - Free Glycol Production Facility http:/ /energy.gov/articles/lab - breakthrough - adm - leads - petroleum - free - glycol - production - facility (accessed Mar 25, 2015). (2) Trent, D. L. In Kirk - Othmer Encyclopedia of Chemical Technology ; John Wiley & Sons, Inc., 2000. (3) Barbelli, M. L.; Santori, G. F.; N ichio, N. N. Bioresour. Technol. 2012 , 111 , 500 503. (4) Sheldon, R. A. Green Chem. 2014 , 16 (3), 950 963. (5) Chen, G. - Q.; Patel, M. K. Chem. Rev. 2011 , 112 (4), 2082 2099. (6) Rodriguez, B. A.; Stowers, C. C.; Pham, V.; Cox, B. M. Green Chem. 2014 , 16 (3), 1066 1076. (7) Zuwei, X.; Ning, Z.; Yu, S.; Kunlan, L. Science 2001 , 292 (5519), 1139 1141. (8) Binning, R. C.; Null, H. R. Production and recovery of propylene oxide by plural stage distillation and caustic addition. US3350417 A, October 31, 1967. (9) Meiers, R.; Dingerdissen, U.; Hölderich, W. F. J. Catal. 1998 , 176 (2), 376 386. (10) Clerici, M. G.; Bellussi, G.; Romano, U. J. Catal. 1991 , 129 (1), 159 167. (11) Costantini, M.; Jouffret, M. Production of epoxides from lower alkane - 1,2 - diols. U S4297287 A, October 27, 1981. (12) Yu, Z.; Xu, L.; Wei, Y.; Wang, Y.; He, Y.; Xia, Q.; Zhang, X.; Liu, Z. Chem. Commun. 2009 , No. 26, 3934 3936. (13) Ryu, J. - Y. Process for epoxidizing vicinal diols. EP0145447 A1, June 19, 1985. (14) Sherwin, M. B.; Fra nk, M. E. Process for preparing oxirane compounds. US4399295 A, August 16, 1983. (15) Sherwin, M. B.; Peress, J. Cracking of mixtures containing hydroxyesters. US4012424 A, March 15, 1977. (16) Weitz, H. - M.; Hartig, J.; Platz, R. Basic decarboxylation of propylene glycol esters in a high boiling solvent. US4158008 A, June 12, 1979. (17) Fouquet, G.; Merger, F.; Baer, K. From propylene glycol, catalyst, alkali metal carboxylate. US4226780 A, October 7, 1980. (18) Dubnikova, F.; Lifshitz, A. J. Phys. Chem . A 2000 , 104 (19), 4489 4496. 174 (19) Namuangruk, S.; Khongpracha, P.; Pantu, P.; Limtrakul, J. J. Phys. Chem. B 2006 , 110 (51), 25950 25957. (20) Amberlite and Amberlyst Resins http://www.sigmaaldrich.com/chemistry/chemical - synthesis/learning - center/techn ical - bulletins/al - 142/amberlite - amberlyst.html (accessed Mar 25, 2015). (21) Dent Glasser, L. S.; Kataoka, N. Cem. Concr. Res. 1981 , 11 (1), 1 9. (22) Weisz, P. B.; Hicks, J. S. Chem. Eng. Sci. 1995 , 50 (24), 3951 3958. (23) Bioresour. Technol. 2009 , 100 (20), 4690 4696. (24) Rees, C. A.; Provis, J. L.; Lukey, G. C.; van Deventer, J. S. J. Langmuir 2007 , 23 (15), 8170 8179. (25) Rees, C. A.; Provis, J. L.; Lukey, G. C.; van Deventer, J. S. J. Langmuir 200 7 , 23 (17), 9076 9082. (26) Simonsen, M. E.; Sønderby, C.; Li, Z.; Søgaard, E. G. J. Mater. Sci. 2009 , 44 (8), 2079 2088. (27) Bal, R.; Tope, B. B.; Das, T. K.; Hegde, S. G.; Sivasanker, S. J. Catal. 2001 , 204 (2), 358 363. (28) Nesbitt, H. W.; Bancroft , G. M.; Henderson, G. S.; Ho, R.; Dalby, K. N.; Huang, Y.; Yan, Z. J. Non - Cryst. Solids 2011 , 357 (1), 170 180. (29) Sawyer, R.; Nesbitt, H. W.; Secco, R. A. J. Non - Cryst. Solids 2012 , 358 (2), 290 302. (30) Shchukarev, A. V.; Korolkov, D. V. Cent. Eur. J. Chem. 2004 , 2 (2), 347 362. (31) Zemek, J.; Jiricek, P.; Gedeon, O.; Lesiak, B.; Jozwik, A. J. Non - Cryst. Solids 2005 , 351 (19 20), 1665 1674. (32) Bailey, O. H.; Montag, R. A.; Yoo, J. S. Appl. Catal. Gen. 1992 , 88 (2), 163 177. (33) Malinowski, S. ; Basinski, S. J. Catal. 1963 , 2 (3), 203 207. (34)