i C ONDENSED PHASE CONVERSION OF BIOETHANOL TO 1 - BUTANOL AND HIGHER ALCOHOLS By Tyler L Jordison A DISSERTATION Submitted to Michigan State University i n partial fulfillment of the requirements for the degree of Chemical Engineering D octor of Philosophy 201 6 ii A BSTRACT C ONDENSED PHASE CONVERSION OF BIOETHANOL TO 1 - BUTANOL AND HIGHER ALCOHOLS By Tyler L Jordison Higher a lcohols are important chemical feedstocks as well as potential fuels. With the recent surge in bioethanol pro duction, it would be advantageous to convert bioethanol to butanol and higher alcohols. Results by various authors for a wide range of reaction conditions are presented. For butanol specifically, the highest yields have been obtained with hydroxyapatite, hydrotalcite and alumina - supported nickel catalysts. The literature shows it is a challenge to convert ethanol to butanol, since no one has achieved butanol yields higher than ~30%. In this research project, attention is focused on alumina - supported ni ckel, since it is robust, stable, and well suited for condensed phase ethanol Guerbet chemistry. Catalyst screening of different compositions was performed and higher alcohol selectivities were analyzed. The 8%Ni/8%La - Al 2 O 3 was proven to produce over 80% selectivity to higher alcohols at 50% ethanol conversion . The impacts of water removal on target alcohol yield with the 8Ni/8La catalyst were investigated in a batch reactor. Removing water decreased selectivity to CH 4 and CO 2 from 15% without water rem oval to 8% with water removal. Preliminary kinetics of 1 - butanol and 1 - hexanol formation were investigated by looking at initial rates of their formation a t 215 °C , 230 °C , and 239°C. Runs with ethanol/acetaldehyde/H 2 were performed to investigate the step s of the ethanol Guerbet reaction mechanism. Runs completed at 150°C and 200°C were modeled and rate constants were determined for acetaldehyde hydrogenation, acetalde hyde condensation, and butyraldehyde hydrogenation. It was found ethanol dehydrogenatio n is in equilibrium and is the rate limiting iii step of the ethanol Guerbet mechanism. The activation energy for ethanol dehydrogenation was calculated to be 150 KJ/mol. Therefore, the effect of H 2 on a neat ethanol run was examined and found to have little effect on ethanol conversion rate. Ethanol conversion rates were the same due to the side reaction of H 2 with ethanol to CH 4 and water, which offsets the negative effect of hydrogen on acetaldehyde formation rate. The presence of excess H 2 was found to d ecrease 1 - butanol and 1 - hexanol formation rates. iv Copyright by TYLER L JORDISON 2016 v Dedicated to my parents, loving wife, and beautiful daughter Avery vi A CKNOWLEDGEMENTS I would like to tha nk my advisor, Dr. Dennis Miller, for giving me the opportunity to work on this project. I have learned a great amount about research from him during my studies and I am forever grateful for his mentorship. I would like to thank Dr. Carl Lira for his ass istance with my VLE modeling. His help with modeling the thermodynamics of my reaction system was a major contribution to helping me accurately characterize it. I want to thank Dr. Lars Peereboom for his training on instruments and also for guidance on e xperiments. I also want to thank my committee, Dr. David Hodge, Dr. James Jackson, and Dr. Chris Saffron for their support. I would like to thank Evan Wegener f or contributing to my catalyst characterization studies. I would like to thank the U.S. Depa rtment of Energy (Award no. DE - FG36 - 04GO14216) and the National Corn Growers Association for financial support of this work. To close, I want to thank my family and friends for their enduring support . for my wife being by my side from beginni ng to end of this wonderful journey. vii T ABLE OF CONTENTS LIST OF TABLES ......................................................................................................................... .ix LIST OF FIGURES .......................................................................................................................xii 1 Literature Review and Background ................................ ................................ ......................... 1 1.1 Introduction ................................ ................................ ................................ ................. 1 1.2 Economics and Energy Yield ................................ ................................ ...................... 5 1.3 Early Patents ................................ ................................ ................................ ............... 6 1.4 MgO, Mixed Oxide, and Hydroxyapatite Catalysts ................................ .................... 7 1.5 Homogeneous Catalysis ................................ ................................ ............................ 11 1.6 Condensed phase reactions ................................ ................................ ....................... 12 1.7 Alumina - Supported - Nickel Catalysts ................................ ................................ ....... 12 1.8 Ethanol Reaction with Methanol ................................ ................................ .............. 13 1.9 Research Objectives ................................ ................................ ................................ .. 14 1.9.1 Develop Efficient, Condensed Phase Alcohol Condensation Process Using Neat Ethanol ................................ ................................ ................................ .............................. 14 1.9.2 Identify Robust, Stable Catalyst Compositions ................................ .................... 15 1.9.3 Expand Understanding of Ethanol Guerbet Reaction Mechanism ....................... 15 16 2 Catalyst Screening ................................ ................................ ................................ ................. 20 2.1 Introduction ................................ ................................ ................................ ............... 20 2.2 Experimental ................................ ................................ ................................ ............. 23 2.2.1 Materials and Catalyst Preparation ................................ ................................ ....... 23 2.2.2 Catalyst Characterization ................................ ................................ ...................... 23 2.2.3 Batch Reaction Studies ................................ ................................ ......................... 24 2.3 Results and Discussion ................................ ................................ ............................. 29 2.3.1 Catal yst Characterization ................................ ................................ ...................... 29 2.3.2 Catalytic Reactions ................................ ................................ ............................... 30 APPENDIX REFERENCES ..42 3 VLE Modeling ................................ ................................ ................................ ....................... 45 3.1 Introduction ................................ ................................ ................................ ............... 45 3.2 Thermodynamic Modeling of Reaction ................................ ................................ .... 46 3.2.1 SR Polar EOS ................................ ................................ ................................ ....... 46 3.2.2 Parameter Estimation ................................ ................................ ............................ 49 3.2.3 Applying SR - Polar EOS to Batch Experiments ................................ ................... 52 3.2.4 Model Validation and Parameter Adjustment ................................ ...................... 56 3.2.5 Liquid and Vapor Densities ................................ ................................ .................. 59 3.2.6 Comparison of SR - Polar EOS analysis with conventional liquid phase analysis 60 3.2.7 Experimental Repeatability ................................ ................................ ................... 62 3.3 Conclusions ................................ ................................ ................................ ............... 64 viii REFERENCES 4 Impact of Water and its Removal on Etha nol Guerbet Reaction ................................ ........... 80 4.1 Introduction ................................ ................................ ................................ ............... 80 4.2 Experimental ................................ ................................ ................................ ............. 81 4.2 .1 Materials ................................ ................................ ................................ ............... 81 4.2.2 Catalytic reactions with water addition ................................ ................................ 82 4.2.3 Catalytic reactions with water removal ................................ ................................ 82 4.2.4 Modeling reaction system with water removal ................................ ..................... 85 4.3 Results and Discussion ................................ ................................ ............................. 88 4.3.1 Water addition runs ................................ ................................ ............................... 88 4.3.2 Water removal runs ................................ ................................ ............................... 91 4.4 Conclusions ................................ ................................ ................................ ............... 93 REFERENCES 5 Investigation of the Ethanol Guerbet Reaction Mechanism ................................ .................. 96 5.1 Introduction ................................ ................................ ................................ ............... 96 5.2 Experimental ................................ ................................ ................................ ............. 97 5.2.1 Materials ................................ ................................ ................................ ............... 97 5.2.2 Ethanol/ acetaldehyde /H 2 reactions ................................ ................................ ....... 98 5.2.3 Ethanol/H 2 reactions ................................ ................................ ............................. 98 5.3 Results and Discussion ................................ ................................ ............................. 98 5.3.1 Prel iminary Initial Rate Kinetics for 1 - Butanol and 1 - Hexanol ........................... 98 5.3.2 Effect of Catalyst Loading ................................ ................................ .................. 100 5.3.3 Kinetic modeling o f ethanol/acetaldehdye/H 2 reactions ................................ ..... 104 5.3.4 Effect of H2 on higher alcohol formation rate ................................ .................... 114 5.3.5 Reaction of neat 1 - b utanol ................................ ................................ .................. 118 5.3.6 Reaction of ethanol/butyraldehyde ................................ ................................ ..... 120 5.3.7 Reaction of ethanol/isobutyraldehyde ................................ ................................ 123 5.4 Conclusions ................................ ................................ ................................ ............. 126 REFERENCES 6 Summary and Recommendations for Fu ture Work ................................ ............................. 135 6.1 Catalyst Screening ................................ ................................ ................................ .. 135 6.2 VLE Modeling ................................ ................................ ................................ ........ 135 6.3 I mpact of Water ................................ ................................ ................................ ...... 136 6.4 Investigation of the Ethanol Guerbet Reaction Mechanism ................................ ... 136 6.5 Recommendations for Future Work ................................ ................................ ........ 137 6.5.1 Condensed - phase continuous reactions ................................ .............................. 137 6.5.2 Minimizing ethanol decomposition to CH 4 and CO 2 ................................ ......... 139 ix LIST OF TABLES Table 1 - 1 Lower heating values are presented higher alcohols in MJ/L. The solubilities of the alcohols in water, and solubilities of water in the alcohols are also shown. ................................ ... 1 Table 1 - 2 Reaction conditions, conversion percentages, and selectivities are reported for the conversion of ethanol to 1 - butanol for Kourtakis [18, 27 - 29]. ................................ ..................... 10 Table 2 - 1 Acid and base site densities from CO 2 and NH 3 chemisorption. ................................ 30 Table 2 - 2. Product selectivities are provided along with ethanol conversion and total car bon recoveries. Unidentified selectivity is based on total unidentified peak area using 1 - hexanol response factor. ................................ ................................ ................................ ............................. 31 Table 2 - 3 Preliminary catalyst screening experiments at 230°C, autogeneou s pressure, 0.093 g cat/ g EtOH, and 10 hr run time. HA = higher (C 4 - C 8 ) alcohols. ................................ ................. 32 Table A2 - 1 Components are identified from GC - MS for run 79TLJ022812 on DB wax column with TCD detector. GC - MS analysis was performed at the mass spec facility at MSU. Components with (*) identified by GC - MS and (**) indicates component was hidden and/or could not be located on chromatogram. ................................ ................................ ........................ 35 Ta ble A2 - 2 Components are identified for run 02 - 36TLJ101713. The sample was analyzed with a sol gel wax column and FID detection. Component with (*) is internal standard. 36 Table 3 - 1 Compariso n of error in vapor pressure calculated by PRWS, PSRK, and SR - Polar. PRWS - 1 denotes standard PR, PRWS - 2 denotes Boston Mathias, and PRWS - 3 denotes Schwartszentruber [3]. ................................ ................................ ................................ .................. 47 Table 3 - 2 Groupi ng of observed species into modeled components. ................................ .......... 48 Table 3 - 3 Binary parameters for the ethanol Guerbet system. ................................ .................... 50 Table 3 - 4 Volu me translation constants are listed for alcohols, water, methane, and CO2. The (*) indicates translation constant was fit to binary with ethanol and not pure component data [3]. ................................ ................................ ................................ ................................ ....................... 52 Table 3 - 5 C omparison of predicted and experimental gas quantities in SR - Polar validation experiments at 230°C with adjusted ethanol/CH 4 and ethanol/CO 2 binary parameters . .............. 58 Table 3 - 6 Hydrogen and oxygen recoveries are compared for the liquid - phase only method and SR - Polar model. ................................ ................................ ................................ ............................ 62 Table 4 - 1 Ethanol conversion and carbon recoveries are shown for runs with and without the water removal loop. ................................ ................................ ................................ ...................... 92 x Table 5 - 1 Gas phase enthalpies and free energies of formation (298 K) and equilibrium constants at 503 K [1]. ................................ ................................ ................................ .................. 96 Table 5 - 2 E thanol conversion based on ethanol concentration and carbon in the products is shown for the different loading runs. Carbon recoveries are also shown for the different loadings. All values shown are percentages. ................................ ................................ ............. 103 Table 5 - 3 Modeled ethanol concentrations with percent errors are shown at 150°C, 175°C, and 200°C. ................................ ................................ ................................ ................................ ......... 107 Table 5 - 4 Modeled acetaldehyde concentrations with percent erro rs are shown at 150°C, 175°C, and 200°C. ................................ ................................ ................................ ................................ .. 108 Table 5 - 5 Modeled butyraldehyde concentrations with percent errors are shown at 150°C, 175°C, and 200°C. ................................ ................................ ................................ ...................... 109 Table 5 - 6 Modeled 1 - butanol concentrations with percent errors are shown at 150°C, 175°C, and 200°C. ................................ ................................ ................................ ................................ .. 111 Table 5 - 7 Modeled water concentrations with percent errors ar e shown at 150°C, 175°C, and 200°C. ................................ ................................ ................................ ................................ ......... 111 Table 5 - 8 Modeled partial pressures of hydrogen with percent errors are shown at 150°C, 175°C, and 200°C. ................................ ................................ ................................ ...................... 112 Table 5 - 9 Rate constants and activation energies are shown for ethanol/acetaldehyde runs at 150°C and 200°C. ................................ ................................ ................................ ....................... 113 Table 5 - 10 Acetaldehyde conversion and carbon recoveries are shown for 20% acetaldehyde/80 % ethanol runs done at 150°C and 200°C. Ethanol conversion is not reported because ethanol was formed and not consumed. Catalyst loading was 0.04 g cat/g mixture. .............................. 114 Table 5 - 11 Ethanol conversion (%) and carbon recoveries (%) are shown for runs with and without H 2 . Carbon recoveries are based on liquid products only. ................................ ............. 116 Table 5 - 12 Molar composit ion of gas phase at end of run is shown with CH 4 /CO 2 ratio for the runs with and without H 2 . ................................ ................................ ................................ ........... 117 Table 5 - 13 Carbon recoveries and 1 - butanol conversion are shown for the neat 1 - butanol run at 2 30°C. ................................ ................................ ................................ ................................ ......... 119 Table 5 - 14 End of run gas analysis is shown for neat 1 - butanol run at 230°C. ........................ 120 Table 5 - 15 Butyraldehyde conversion a nd ethanol conversions are shown. Carbon recoveries are also shown. ................................ ................................ ................................ ............................ 123 xi Table C5 - 1 Part 1 of the master reaction list is shown. Runs with (*) are preliminary conti nuous runs with ethanol flow rate in ml/min under starting reactor mass column. ............................... 128 Table C5 - 2 Part 2 of the master reaction list is shown. Runs with (*) are preliminary continuous runs with catalyst bed weight under g cat/g react. column. ................................ ........................ 129 Table C5 - 3 Part 3 of the master reaction list. ................................ ................................ ............ 131 Table 6 - 1 Reaction cond itions are shown for preliminary trickle bed reactions . .................. 137 Table 6 - 2 Results summary for preliminary trickle bed reactions. ................................ ............ 138 xii LIST OF FIGURES Figure 1 - 1 Mechanism for the oxo process [5]. ................................ ................................ ............. 2 Figure 1 - 2 Reaction tree for the ethano l Guerbet reaction system. ................................ ............... 3 Figure 1 - 3 Hypothesized proton abstraction mechanism [11]. ................................ ...................... 4 Figure 1 - 4 Economics are shown for con verting ethanol to 1 - butanol and 1 - hexanol. Ethanol was set as 70% and 60% cost of total production. Ethanol conversion is set at 100%. Prices from icispricing.com for ethanol and products were: ethanol ($0.38 /lb), 1 - butanol ($0.97 /lb), 1 - heanol ($0.97 /lb). ................................ ................................ ................................ ........................... 6 Figure 2 - 1 Sample chromatogram of liquid sample at end of run for run 79TLJ022812 using TCD detection. All alcohol peaks and water peak are resolved. A more detailed sample analysis is shown in Table A2 - 1 of Appendix A. ................................ ................................ ....................... 27 Figure 2 - 2 Sample chromatogram is shown for run 02 - 36TLJ101713 using FID detection. The sample was injected on a sol gel wax column with FID detector. A more detailed sample analysis is shown in ................................ ................................ ................................ ...................... 28 Figure 2 - 3 Selectivities for 1 - butanol, C 6 + alcohols , acetaldehyde, ethyl acetate, and diethyl ether for 8Ni/Al and 8Ni/9La - Al catalysts at 23 0°C and 0.04 g cat/g EtOH loading. .................. 33 Figure A2 - 1 TCD response factors for ethanol, 1 - butanol, ethyl acetate, and C 6 + alcohols. ..... 37 Figure A2 - 2 TCD response factors for acetaldehyde, butyraldehyde, C 4 + esters, 4 - heptanone, and 1 - decanol. ................................ ................................ ................................ ............................... 37 Figure A2 - 3 TCD response factors for 1 - butanol and C 6 + alcohols with butyl hex anoate internal standard. ................................ ................................ ................................ ................................ ........ 38 Figure A2 - 4 FID response factors for ethyl acetate, diethyl ether, 1 - butanol, and C 6 + alcohols with butyl hexanoate internal standard. ................................ ................................ ........................ 38 Figure A2 - 5 FID response factors for 4 - heptanone, acetaldehyde, butyraldehyde, and C 6 + aldehydes. ................................ ................................ ................................ ................................ ...... 39 Figure A2 - 6 CO 2 - - alumin a - supported catalysts. ........................... 40 Figure A2 - 7 NH 3 - - alumina - supported catalysts. ......................... 40 Figure A2 - 8 Sample BET surface area plot for 8Ni/9La 2 O 3 - - alumina catalyst 41 Figure 3 - 1 Experimental vapor pressure data is compared with predicted vapor pressure by Peng Robinson and SR Polar [3]. ................................ ................................ ................................ .......... 47 xiii Figure 3 - 2 Predicted hase equilibria for ethanol - CH4 is shown with the regressed, temperature dependent ka,EtOH - CH4 and with the ka,EtOH - CH4 adjusted to fit our validation experiments [14]. ................................ ................................ ................................ ................................ ............... 50 Figure 3 - 3 SR - Polar translated density predictions are compared with experimental data [3, 14]. ................................ ................................ ................................ ................................ ....................... 52 Figure 3 - 4 Phase equilibria is show pressure also apply to multi - component mixtures. ................................ ................................ ...... 55 Figure 3 - 5 Flowsheet for SR - Polar model application to reaction data. ................................ ...... 56 Figure 3 - 6 Liquid and vapor phase densities for pure ethanol at 25°C (left), pure ethanol at 230°C (center), and for the reaction mixture at 41% ethanol conversion (right). ........................ 59 Figure 3 - 7 Comparison of EtOH conversion and 1 - butanol selectivity for conventional liquid phase analysis and SR - Polar EOS analysis. ................................ ................................ ................. 61 Figure 3 - 8 Comparison of C6 alcohol selectivity and carbon recovery for conventional liquid phase analysis and SR - Polar EOS analysis. ................................ ................................ ................ 61 Figure 3 - 9 Ethanol conversion and carbon recovery are shown for two runs with identical catalyst at 230°C and 0.04 g cat/g EtOH loading. ................................ ................................ ........ 63 Figure 3 - 10 Selectivity to 1 - butanol and C 6+ alcohols is shown for two runs with identical catalyst at 230°C an d 0.04 g cat/g EtOH loading. ................................ ................................ ........ 63 Figure 3 - 11 Experimental reactor pressure and gas selectivities are shown for two runs with identical catalyst at 230°C and 0.04 g cat/g EtOH loading. ................................ ......................... 63 Figure B3 - 1 Isothermal VLE data for ethanol - 1 - butanol at 323K, including SR - Polar prediction from regressed parameters [8]. ................................ ................................ ................................ ..... 66 Figure B 3 - 2 Isothermal VLE data for ethanol - 1 - butanol at 403K, including SR - Polar prediction from regressed parameters [8]. ................................ ................................ ................................ ..... 66 Figure B3 - 3 Isothermal VLE data for ethanol - water at 298K, including SR - Polar prediction from regressed parameters [10]. ................................ ................................ ................................ ... 67 Figure B3 - 4 Isothermal VLE data for ethanol - water at 348K, including SR - Polar prediction from regressed parameters [10]. ................................ ................................ ................................ ... 67 Figure B3 - 5 Isothermal VLE data for ethanol - water at 473K, including SR - Polar prediction from regressed parameters [9]. ................................ ................................ ................................ ..... 68 Figure B3 - 6 Isotherma l VLE data for ethanol - methane at 448K, including SR - Polar prediction from regressed parameters [14]. ................................ ................................ ................................ ... 68 xiv Figure B3 - 7 Isothermal VLE data for ethanol - methane at 498K, including SR - Polar predict ion from regressed parameters [14]. ................................ ................................ ................................ ... 69 Figure B3 - 8 Isothermal VLE data for ethanol - carbon dioxide at 304K, including SR - Polar prediction from regressed parameters [6]. ................................ ................................ .................... 69 Figure B3 - 9 Isothermal VLE data for ethanol - carbon dioxide at 353K, including SR - Polar prediction from regressed parameters [5]. ................................ ................................ .................... 70 Figure B3 - 10 Isot hermal VLE data for ethanol - carbon dioxide at 453K, including SR - Polar prediction from regressed parameters [7]. ................................ ................................ .................... 70 Figure B3 - 11 Isothermal VLE data for 1 - butanol - carbon dioxide at 313K, inclu ding SR - Polar prediction from regressed parameters [13]. ................................ ................................ .................. 71 Figure B3 - 12 Isothermal VLE data for 1 - butanol - carbon dioxide at 393K, including SR - Polar prediction from regressed parameters [12]. ................................ ................................ .................. 71 Figure B3 - 13 Isothermal VLE data for 1 - butanol - carbon dioxide at 430K, including SR - Polar prediction from regressed parameters [11]. ................................ ................................ .................. 72 Figure B3 - 14 Isothermal VLE data for 1 - butanol - water at 323K, including SR - Polar prediction from regressed parameters [8]. ................................ ................................ ................................ ..... 72 Figure B3 - 15 Isothermal VLE data for 1 - butanol - water a t 383K, including SR - Polar prediction from regressed parameters [8]. ................................ ................................ ................................ ..... 73 Figure B3 - 16 Isothermal VLE data for 1 - butanol - water at 403K, including SR - Polar prediction from regressed parameters [8 ]. ................................ ................................ ................................ ..... 73 Figure B3 - 17 Isothermal VLE data for carbon dioxide - water at 473K, including SR - Polar prediction from regressed parameters [15]. ................................ ................................ .................. 74 Figure B3 - 18 Isothermal VLE data for carbon dioxide - water at 523K, including SR - Polar prediction from regressed parameters [15]. ................................ ................................ .................. 74 Figure B3 - 19 SR - Polar liquid density prediction of ethanol with fitted density translation parameter [3]. ................................ ................................ ................................ ................................ 75 Figure B3 - 20 SR - Polar liquid density prediction of 1 - butanol with fitted density translation parameter [3]. ................................ ................................ ................................ ................................ 75 Figure B3 - 21 SR - Polar liquid density prediction of 1 - hexanol with fitted density translation parameter [3]. ................................ ................................ ................................ ................................ 76 Figure B3 - 22 SR - Polar liquid density pr ediction of water with fitted density translation parameter [3]. ................................ ................................ ................................ ................................ 76 xv Figure B3 - 23 SR - Polar liquid density prediction of ethanol - carbon dioxide with fitted density translation parameter [7] 77 Figure 4 - 1 Recirculation loop with drying bed is shown with attachments to a 300 ml Parr reactor. The loop includes a chiller that cools the mixture to ~40°C and a heater that heats it back up to ~220°C. ................................ ................................ ................................ ....................... 83 Figure 4 - 2 Schematic is shown for the recirculating pump, taken from Seifried [14].Circulating pump body parts: (a) pump cylinder, (b) solenoids (S1 and S2), (c) reducing union, (d) union cross, (e) plug, (f) check valve, (g) union - tee, (h) needle valve, (i) tubing, (j) piston, and (k) compression spring. ................................ ................................ ................................ ...................... 85 Figure 4 - 3 Gas selectivity (CH 4 + CO 2 ) is plotted vs time on the left and vs ethanol conversion on the right. ................................ ................................ ................................ ................................ ... 89 Figure 4 - 4 Gas selectivity is plotted vs water concentration on the left and 1 - butanol, C 4 - C 8 alcohol selectiv ity is plotted vs conversion on the right. ................................ .............................. 89 Figure 4 - 5 Accounted water ratio is plotted vs ethanol conversion on the left and the accounted water ratio including a correction for water forme d from formed CH 4 is plotted vs ethanol conversion on the right. ................................ ................................ ................................ ................. 90 Figure 4 - 6 Gas selectivites are plotted vs time (top left) and vs ethanol conversion (top right). Higher alcohol selectivit es are plotted vs time (bottom left) and vs ethanol conversion (bottom right). ................................ ................................ ................................ ................................ ............. 91 Figure 4 - 7 Gas (left) and higher alcohol (right) selectivities are plotted vs water concentration (wt%). ................................ ................................ ................................ ................................ ............ 92 Figure 5 - 1 Initial rate data for 1 - butanol and 1 - hexanol at 0.06 g cat/g EtOH loading. .............. 99 Figure 5 - 2 Arrhenius plots are shown for 1 - butanol and 1 - hexanol at 0.03 and 0.06 g cat/g EtOH loading. ................................ ................................ ................................ ................................ .......... 99 Figure 5 - 3 Initial reaction rates for 1 - butanol and 1 - hexanol formation at different catalyst loadings. ................................ ................................ ................................ ................................ ...... 100 Figure 5 - 4 Selectivites for 1 - butanol and 1 - hexanol are plotted at different catalyst loadings. . 101 Figure 5 - 5 Plotted on left is ethanol convers ion vs time, based on carbon in ethanol. Plotted on right is ethanol conversion vs time based on ethanol carbon equivalence in the liquid product. 102 Figure 5 - 6 Selectivities for 1 - butanol a nd 1 - hexanol are plotted vs conversion at different catalyst loadings. Ethanol conversion was determined from reacted ethanol carbon based on GC analysis of the liquid product. ................................ ................................ ................................ ..... 102 Figure 5 - 7 Vapor liquid equilibrium is shown for ethanol and H 2 at 498 K [3]. ....................... 106 xvi Figure 5 - 8 Modeled ethanol concentration at 150°C, 175°C and 200°C. Initial charge of 80 mol% ethanol, 20 mol% acetaldeh yde with 0.04 g cat /g reactant loading. ............................... 106 Figure 5 - 9 Modeled acetaldehyde concentration at 150°C, 175°C, and 200°C. Initial charge of 80 mol% ethanol, 20 mol% acetaldehyde with 0.04 g cat / g reactant loading. .......................... 108 Figure 5 - 10 Modeled butyraldehyde concentration at 150°C, 175°C and 200°C. Initial charge of 80 mol% ethanol, 20 mol% acetaldehyde with 0.04 g cat /g reactant loading. .......................... 109 Figure 5 - 11 Modeled 1 - butanol and water concentration at 150°C, 175°C, and 200°C. Initial charge of 80 mol% ethanol, 20 mol% acetaldehyde with 0.04 g cat /g reactant loading. .......... 110 Figure 5 - 12 Modeled hydrogen partial pressure at 150°C, 175°C, and 200°C. Initial charge of 80 mol% ethanol, 20 mol% acetaldehyde with 0.04 g cat /g reactant loading. ............................... 112 Figure 5 - 13 Arrhenius plot is shown for the ethanol Guerbet reaction mechanism. ................. 113 Figure 5 - 14 Ethanol conversion and ethanol concentration are plotted for r uns with and without hydrogen. Reaction temperature was 230°C with 0.04 g cat/g ethanol loading. ....................... 115 Figure 5 - 15 Water concentration is plotted vs time for the reactions with and without H 2 a t 230°C and 0.04 g cat/ g ethanol loading. ................................ ................................ .................... 115 Figure 5 - 16 Concentration of 1 - butanol and C 6 alcohols are plotted for runs with and without hydrogen. Reaction temperature was 230°C with 0.04 g cat/g ethanol loading. ....................... 116 Figure 5 - 17 Selectivites of 1 - butanol and C 6 alcohols are plotted for runs with and without hydrogen. Reaction temperature was 230°C with 0.04 g cat/g ethanol loa ding. ....................... 117 Figure 5 - 18 Concentration of ethyl acetate and acetal are plotted for runs with and without hydrogen. Reaction temperature was 230°C with 0.04 g cat/g ethanol loading. ....................... 118 Figure 5 - 19 1 - Butanol concentration is plotted 1 - butanol concentration vs. time for the pure 1 - butanol run at 230°C on the left. Butyraldehyde and 2 - ethyl - hexanol concentrations are plotted vs time on right. ................................ ................................ ................................ .......................... 119 Figure 5 - 20 Ethanol and butyraldehyde concentration vs time is shown in the top left. 1 - Butanol and 2 - ethyl - hexanol concentration vs time is shown top right. Concentrations of hig her aldehydes hexenal (HEN), hexanal (HAN), 2 - ethyl - hexenal (2 - E - HEN), and 2 - ethyl - hexanal (2 - E - HAN) are shown in bottom left. Initial rates for higher aldehydes and 1 - butanol are shown in bottom right. ................................ ................................ ................................ ................................ 121 Figure 5 - 21 Gas chromatogram for ethanol/butyraldehyde reaction at end of run. Unmarked peaks are unidentified peaks. ................................ ................................ ................................ ...... 122 Figure 5 - 22 Ethanol and acetaldehyde concentration pro files are shown for a neat ethanol run and for the mixed 70 mol% EtOH/30 mol % isobutyraldehyde run at 230°C and 0.04 g cat/g mixture loading. ................................ ................................ ................................ .......................... 124 xvii Figure 5 - 23 Concentration profiles are shown for the neat ethanol run and the mixed 70 mol % EtOH/30 mol % isobutyraldehyde run at 230°C and 0.04 g cat/g mixture loading. .................. 124 Figure 5 - 24 Concentration profiles are shown for isobutyraldeh yde and isobutanol for the mixed 70 mol % EtOH/30 mol % isobutyraldehyde run at 230°C and 0.04 g cat/g mixture loading. .. 125 Figure 5 - 25 Chromatogram is shown for end of run of 70 mol% EtOH/30 mol% isobutyraldehdye at 230°C and 0.04 g cat/g mixture loading. ................................ .................... 126 1 1 Literature Review and Background 1.1 Introduction With the recent increase in demand for renewab le fuels, there has been a surge in ethanol production to meet that demand. There are markets that could capitalize on this increased ethanol production. One such market would require this ethanol to be converted to butanol and higher alcohols. In fact, higher alcohols have energy values close to gasoline, making them better than ethanol for fuels ( Table 1 - 1 ) [ 1 ] . Higher alcohols also have broad applications in the commercial chemical industry. Tsuchida et al. [ 2 ] - Butanol is an important chemical feedstock used as a solvent and in polymer materials such as butyl acrylate and but yl 1 - butanol and higher alcohols have a higher market value than ethanol [ 3 ] . My research project will focus on butanol production from ethanol, but alcohols higher than 1 - butanol can also be produc ed. Table 1 - 1 Lower heating values are presented higher alcohols in MJ/L. The solubilities of the alcohols in water, and solubilities of water in the alcohols are also shown. - - - - - - - - - The current method for producing 1 - butanol is the oxo process, which uses petroleum as the feedstock [ 2 ] . With the oxo process, 1 - butanol is produced from the hydroformylation of propylene [ 5 ] . A rhodium complex is used to combine hydrogen and carbon monoxide with 2 propylene to form butryaldehyde ( Figure 1 - 1 ). Figure 1 - 1 Mechanism for the oxo process [ 5 ] . Butyraldehyde is hydrogenated downstream to produce 1 - butanol. The oxo process is complicated, requires high energy input, and is costly [ 2 ] . Consequently, 1 - butanol prices fluctuate because of fluctuating propylene prices; these price fluctuations are a direct result of fluctuating oil prices. A greener process needs to be realized, which would directly convert ethanol to 1 - butanol. Not only would it be attractive to convert ethanol to 1 - butanol from an environmental standpoint, but it may prove to be more attractive economically. Ethanol technologies, whether ethano l is being produced from starches or cellulose, are constantly evolving. Ethanol prices always have the opportunity to go down, while oil prices will most likely increase with time. 3 Carbon - carbon coupling of alcohols is commonly regarded as the Guerbet r eaction [ 2 , 6 - 9 ] . This was first discovered by Guerbet in the late 19 th century; Guerbet performed his Figure 1 - 2 Reaction tree for the ethanol Guerbet reaction system. reaction of ethanol to 1 - butanol, but with little yield [ 10 ] . In t he Guerbet reaction, ethanol is oxidized to acetaldehyde, the acetaldehyde undergoes an aldol condensation to crotonaldehyde, and then the crotonaldehyde is hydrogenated to 1 - butanol [ 11 ] . One molecule of water is formed for every molecule of 1 - butanol formed. The formed 1 - butanol can then react with itself and ethanol to form 1 - hexanol, 2 - ethyl - 1 - butanol, 2 - ethyl - 1 - hexanol and other higher alcohols. A Guerbet reaction tree is shown in Figure 1 - 2 , which shows the aldol condensation mechanisms 4 for these higher alcohol reactions. Some of the literature supports the Guerbet mechanism [ 2 , 12 , 13 ] , and some of the literature supports a simpler proton abstraction mechanism [ 11 , 14 , 15 ] . In proton abstraction, a proton is extracted from the beta - carbon of one ethanol molecule which produces a nucleophile that attacks another ethanol molecule [ 11 ] . The coupled product Figure 1 - 3 Hypothesized proton abstraction mechanism [ 11 ] . dehydrates to form 1 - butanol and water ( Figure 1 - 3 ) . Experiments by Ndou, Yang C., and Yang K.W. with acetaldehyde support this mechanism [ 11 , 14 , 15 ] . It will be shown though that the literature more strongly supports the Guerbet mechanism [ 2 , 13 , 16 ] . Various researchers have made attempts to convert ethanol to 1 - butanol over the past 80 years and have obtained moderate ethanol conversion percentages and moderate 1 - butanol sel ectivities [ 2 , 9 , 11 , 15 , 17 - 20 ] . Others have used ethanol as a limiting reagent with methanol and/or propanol to attain moderate to high ethanol conversion percentages [ 3 , 8 , 21 - 23 ] . Of all reaction condition s, catalyst choice is the most important parameter. A heterogeneous catalyst is primarily looked at to keep process costs down. Many have used pure MgO or mixed with other basic cations as a solid - base catalyst for this reaction [ 9 , 11 , 12 , 18 - 20 , 22 , 23 ] . Alkali cation zeolites have also been used for this reaction [ 14 ] . The most recent l iterature describes using partially decomposed hydrotalcites, hydroxyapatites, and alumina - supported nickel, which have produced the highest 1 - butanol yields thus far [ 2 , 15 , 16 , 18 ] . 5 1.2 Economics and Energy Yield Historically the price of 1 - butanol has been roughly three times higher than the price of ethanol [ 1 ] . This makes it attractive to convert ethanol to 1 - butanol if 1 - butanol yields are high and the cost of production is minimal. Profit return (%) was calculated based on assuming two ethanol cost percentages (60% and 70%): Where P. Frac is the ethanol cost fraction. The ethanol cost fraction represents the fraction of total production cost that is from cost of the raw ethanol feed. The other fracti on represents processing cost. Higher alcohol yields were determined at target returns of 15% and 30% . At 70% ethanol cost percentage, it would take 82% higher alcohols yield to obtain 15% return. To achieve 30% return it would take 93% higher alcohol yi eld. As processing cost increases, higher alcohol yield also has to increase to achieve a return. In fact, at 60% ethanol cost, 30% return is not possible at 100% higher alcohol yield. A 15% return can be achieved at 95% higher alcohol yield. 6 Figure 1 - 4 Economics are shown for converting ethanol to 1 - butanol and 1 - hexanol. Ethanol was set as 70% and 60% cost of total production. Ethanol conversion is set at 100%. Price s from icispricing.com for ethanol and products were: ethanol ($0.38 /lb), 1 - butanol ($0.97 /lb), 1 - heanol ($0.97 /lb). Energy yield of converting ethanol to 1 - butanol was calculated based on assuming 100% conversion of ethanol to 1 - butanol. Two molecules of ethanol produce one molecule o f 1 - buta nol and one molecule of water. The lower heating value (LHV) for ethanol is 1278 kJ/mol and the LHV for 1 - butanol is 2,509 kJ/mol [ 24 ] . The theoreti cal maximum energy yield is: Despite losing one mole of hydrogen to water, 1 - butanol retains most of the energy from ethanol with 98% energy yield. 1.3 Early Patents In U.S. patent 2,971,033 Fa rrar [ 17 ] discusses the condensation of ethanol over a catalyst consisting of potassium carbonate, magnesium oxide, and copper chromite. Reactions were carried out in an autoclave with pressures between 900 - 1000psi. Farrar had an ethanol 7 conversion of 32.5% at a temperature of 227 ° C and a pressure of 950 PSI. The selectivity to 1 - butanol was 47%. In U.S. patent 1,910,582 [ 9 ] a process is described by Wibaut for converting ethanol to 1 - butanol over a magnesium oxide ca talyst in a batch reactor. Wibaut states temperatures around 325°C and pressures around 450 - 1500 psi are favorable for ethanol conversion to 1 - butanol. In all of the experiments described in the invention, ethanol was only in the vapor phase. At 275 ° C a nd a pressure of 1470 PSI, 26.8% ethanol conversion and 34% selectivity to 1 - butanol was obtained. In U.S. patent 1,992,480, [ 20 ] Otto describes using magnesium oxide, copper oxide, chromium oxide, and silver oxide to convert ethanol to 1 - butanol. Hydrogen was used as a carrier gas in a flow reactor. Ethanol conversion was 60% and 1 - butanol selectivity was 16% at 260 ° C and 1300PSI. 1.4 MgO, Mixed Oxide, and Hydroxyapatite Catalysts Ndou et al. [ 11 ] investigated various solid base cata lysts for the conversion of ethanol to 1 - butanol and found magnesium oxide to produce the best results in a vertical, fixed bed reactor. Nitrogen was used as the carrier gas and reactions were performed at atmospheric pressure. Ethanol was fed into the r eactor at a relatively high temperature of 475 ° C. Though ethanol conversion was at 56%, 1 - butanol selectivity was minimal at 18%. When pure acetaldehyde was passed over MgO, less 1 - butanol was produced than when pure ethanol was passed over MgO. Researc hers have recently hypothesized MgO itself is not active enough to accomplish high 1 - butanol formation rates [ 7 , 13 , 25 ] . Cosimo et al. [ 25 ] de scribes how formation rates of 1 - butanol and isobutanol are slower than dehydrogenation reactions because acid - base surface properties are important and specific surface atom arrangements are needed to adsorb adjacent species. A strongly basic catalys t has to be tailored to improve high alcohol yields by 8 incorporating a specific type and amount of acid sites; the atoms of the catalyst also have to be arranged in a way that enables the maximum amount of aldol condensations to take place. Gines et al. [ 7 ] illustrated this concept by incorporating copper and potassium into a MgCeOx catalyst. Gines postulates copper increases aldol condensation by recombining hydrog en atoms to form hydrogen gas, which hydrogenates aldol species after they have coupled. Copper also prevents aldol species from hydrogenating back to ethanol molecules, which decreases ethanol conversion. Marcu et al. [ 19 ] looked at combinations of Cu - Mg - Al mixed oxide catalysts and achieved low ethanol conversion percentages with moderate 1 - butanol selectivities. Contrary to Ndou and Yang, C., Marcu had significant acetaldehyde selectivities, thereby contributing to the Guerbet mechanism. It was also shown that water had a negative impact on catalyst activity. When the reaction was carried out at 100 hours, the selectivity to 1 - butanol was 80% and the ethanol conversion was 9%. Marcu recently published a paper in 2013 [ 26 ] where they describe substituting Pd, Ag, Mn, Fe, Cu, Sm, and Yb as a component in their layered double hydroxide (LDH) precursor to their Mg - Al mixed oxide catalysts. They found a Pd - Mg - Al catalyst to have the highest 1 - butanol selecti vity at 73%, but ethanol conversion was low at 4%. The Pd - Mg - Al catalyst was also found to have relatively strong basicity, but low acidity when compared with other mixed oxide catalysts in the study. The addition of water significantly decreased 1 - butan ol selectivity to 48%, with an increase of selectivity towards C 6 , C 8 acetals and acetaldehyde. Tsuchida et al. [ 2 ] have shown that a nonstoichiometric hydroxyapatite (HAP) catalyst, which is a calcium phosphate compound (Ca 5 (PO 4 ) 3 OH), had the best 1 - butanol yields when compared with various alkali oxides. Like Marcu, Tsuc 9 mechanism on both HAP and MgO. At 300 ° C, the 1 - butanol selectivity increased and acetaldehyde selectivity decreased as the contact time increased over HAP. Tsuchida hypothesized HAP is better at trapping molecular hydrog en than MgO, which allows more aldol Tsuchida achieved significantly higher yields in US patent 6,323,383 using a CaCe/P catalyst [ 12 ] . Results were also obtained with other metals substituted for Ce. Tsuchida has achieved 1 - butanol yield of 26.9%. Kourtakis et al. have also achieved significant 1 - butanol yields [ 18 , 27 - 29 ] . In U.S. patent 7,700,810, Kourtakis et al. [ 18 ] discussed the conversion of ethanol to 1 - butanol over a hydrotalcite catalyst. Hydrotalcites are classified under a broad range [ 18 ] . Th e general formula of a LDH is (M 2+ 1 - x M 3+ x (OH) 2 )(A n - x/n ) yH 2 O where M 2+ is any divalent cation (Mg 2+ for a hydrotalcite), M 3+ is any trivalent cation, and A n - is any anion [ 18 ] mixed oxides [ 18 ] . In this specific patent the hydrotalcite had the formula (Mg .753 Al .247 (OH) 2) (OH - .247 ).(.5H 2 O). Kourtakis reports partially decomposing their hydrotalcite catalyst before employing it in the conversion of ethanol to 1 - butanol. Kourtakis also used similar hydrotalcites with different interlayer anions in U.S. patents 7,700,810; 7,700,811; 7,700,812; and 7,807,857 [ 18 , 27 - 29 ] . Reaction data for these patents are summarized in Table 1 - 2 under the last three digits of each patent. Not only has Kourtaki s done extensive work with partially decomposed hydrotalcites, they also have a patent application using lanthanum stabilized aluminum supported basic oxides [ 30 ] . Selectivity towards alcohols reached a maximum at 50% at 22% conversion with a 33.5 wt% K 2 O/La - Al 2 O 3 catalyst. 10 Table 1 - 2 Reaction conditions, conversion percentages, and selectivities are reported for the conversion of ethanol to 1 - butanol for Kourtakis [ 18 , 27 - 29 ] . Patent 810 811 812 857 Reactor Type Packed bed S.S. Packed Bed, S.S Packed Bed S.S Packed Bed, S.S. Catalyst (Mg .753 Al .247 (OH) 2) (OH - .247 ).(.5H 2 O) [(Mg .75 Al .25 (OH) 2 )(CO 3 .125 )].yH 2 O[CuC O 3 ] .013 [Mg .75 Al .25 (OH) 2 ][(CoEDTA 4 - ) 2 - } .44 OH - (.56)2 ].yH 2 O Ca .13 [Al .987 La .01 3 ][CO 3 2 - or O 2 - ] 1.63 Temp. ( ° C) 300 400 350 400 Pressure (psi) 14.7 - 3000 14.7 - 3000 14.7 - 3000 14.7 - 3000 Carrier gas N 2 N 2 N 2 N 2 Etoh Conv. (%) 44.1 30.1 36.3 30 1 - Butanol Sel. (%) 44.6 52.7 38.5 35 1 - Butanol Yield (%) 19.7 15.9 14 10.5 Building on the work of Tsuchida, Ogo et al. [ 31 , 32 ] investigated substituting strontium for calcium in the hydroxyapatite structure. They also investigated substituting vanadium as vanadate, which replaces phosphate in hydroxyapatite. Their highest 1 - butanol selectivity of 81% was obtained using a Sr - P hydroxyapatite in the vapor phase at 8% conversion at 300°C. This was higher than the previous selectivities by Tsuchida with a Ca - P hydroxyapatite. Like Tsuchida, the conversion is still too low to make the process practical. If higher conversions are achieved, 1 - butanol selectivity will likely drop significantly. Tsuchida demonstrated this when they made biogasoline from ethanol [ 16 ] . They had 64% conversion at a high temperature (>400°C) but many components were made su ch as C 6 , C 8 alcohols, hydrocarbons, and some aromatics. Carvalho et al. [ 33 ] suggested that the re was disagreement among the authors over how the physiochemical properties of Mg - Al mixed oxide catalysts related to 1 - butanol formation. Therefore, Carvalho investigated these properties of a typical Mg - Al mixed for 1 - butanol production from ethanol. Carvalho found adjacent acid and medium strength base sites are 11 needed to produce the intermediates needed for 1 - butanol production, as well as accomplish the condensation. These pairs of acid/base sites are important, but strong base sites are not needed and there does not have to be a special atom arrangement in the catalyst, as others have suggested. 1.5 Homogeneous Catalysis Homogenous catalysts were developed for the oxo process but also have been used for ethanol Guerbet chemistry. Koda et al. [ 34 ] used an iridium complex, dppp ligand, 1,7 octadiene, and sodium ethoxide at 120°C for 15 hours. They achieved 58% selectivity to 1 - butanol at 38% ethanol conversion. Other products formed were 2 - ethyl - 1 - butanol, n - hexanol, 2 - ethyl - 1 - hexanol, and n - octanol, which is characteristic of the Guerbet reaction mechanism. The temperature used was much lower (<100°C) than others in the literature. However, the ir idium complex and ligand must be separated, which might not be economical at a large scale. Also, if sodium ethoxide and octadiene are required for the reaction, additional separation steps will be required. Considering the costs and complications assoc iated with homogeneous ethanol Guerbet reactions, it would make more sense to use an active heterogeneous catalyst that can obtain the same 1 - butanol yield. However, at the 2013 Spring ACS meeting, Wass et al. [ 35 ] presented a high butanol s electivity of 95% with a ruthenium diphosphine catalyst. This is a significant selectivity that no one has presented in the literature to this date. In their paper published in Angewandte Chemie, they reported 82% 1 - butanol selectivity at 34% ethanol co nversion [ 36 ] . The concentration of the ruthenium was 0.1 mol % with 0.1 mol % ligand. There was also 5 mol% sodium ethoxide used as a base catalyst. Reactions were pe rformed at 150°C. 12 1.6 Condensed phase reactions Most of the literature has been concerned with Guerbet reactions in the vapor phase. - Al 2 O 3 - supported metal catalysts in a batch reactor at 250°C and pressures up to 100 bar. [ 37 ] Ethanol conversion and 1 - butanol selectivity reached maximum values of 25% and 80%, respectively. The same authors then carried out continuous liquid - phase ethanol c onversion to 1 - butanol at 240°C and 70 bar. [ 38 ] With continuous operation, 1 - butanol - Al 2 O 3 catalyst and ethanol conversion was between 10 and 30%. Cobalt - Al 2 O 3 produced the highest ethanol conversion. Ghaziaskar et al achieved 35% ethanol conversion and 83% C 4 + alcohol selectivity at 250°C and 176 bar w ith 8% Ni/ - Al 2 O 3 . [ 39 ] 1.7 Alumina - Supported - Nickel Catalysts Yang, K.W. et al. [ 15 ] studied et hanol condensation over alumina - supported metal catalysts and obtained results that agreed with Ndou and Yang, C. An 2 O 3 catalyst was found t o be the most effective at producing 1 - butanol with 64.3% selectivity at 19.1% conversion. Selectivity to acetaldehyde, butyraldehdye, and ethyl acetate was 6%, 4%, and 3% respectively. No further characterization was carried out over these side products with time. These experiments were vapor phase reactions at 200°C. When acetaldehyde was passed over the alumina catalyst, less 1 - butanol was produced than when pure ethanol was reacted over the alumina catalysts. The addition of crotonaldehdye also decreased 1 - butanol formation. Yang, K.W. was the first to report ethanol Guerbet reactions with an aluminum - supported nickel catalyst. Supported Ni catalysts have been used extensively for ethanol steam reforming reactions [ 40 - 43 ] . Nickel is becoming desirable as an effective ethanol Guerbet catalyst. Riittonen et al. [ 37 ] surveyed Ni and other metals supported on alumina in a batch reactor system 13 at 250°C and autogeneous pressure. Nickel was found to be most active over platinum, silver, and gold. Highest 1 - butanol selectivity reached was 80% at 20% conversion after 24 hours run time. While selectivity was high, ethanol conversion was too low for the long run time. Riitto nen also investigated hydrogen addition and found a decreasing 1 - butanol formation rate with increasing initial hydrogen pressure. Water removal was briefly examined by placing molecular sieves directly in the reactor. Ethanol conversion increased from 2 0% to 30%, but no further characterization or studies on water removal were performed. Ghaziaskar et al. [ 39 ] performed ethano l Guerbet reactions to 1 - butanol and 1 - hexanol at 250°C in a continuous reactor. Sub and super critical ethanol reactions were investigated. The - Al 2 O 3 catalyst first de scribed by Yang, K.W. was used. Temperature was varied from 135 - 300°C with we ight hourly space velocities varied from 6.4 - 15.6 h - 1 . Conversion reached a maximum at 35% with 62% and 21% selectivity towards 1 - butanol and 1 - hexanol respectively. Other products in the reaction were acetaldehyde, butyraldehdye, ethyl acetate and 2 - pent anone. Increasing the reaction pressure increased conversion by increasing the concentration and residence time. 1.8 Ethanol Reaction with Methanol In U.S. patent 5,095,156 Radlowski [ 23 ] compared the conversion of methanol/ethanol mixtures over neat MgO, alumina - MgO, and charcoal - MgO. Copper, stainless steel, and quartz reactors were also investigated. A stainless steel reactor utilizing a charcoal MgO catalyst produced the best conversion percentages, but had high selectivities towards CO/CO 2 . Olson et al. [ 3 ] has shown that high conversions of ethanol and high selectivities of 2 - methyl - 1 - propanol could be obtained by using MgO - loaded carbon catalysts. It is apparent from Radlowski and O lson that carbon may be a significant factor in the Guerbet reaction of ethanol 14 and methanol. Reactions were done at 360°C in a continuous reactor at atmospheric pressure. Nitrogen was the carrier gas with a methanol/ethanol ratio of 7. Conversion for et hanol was 100% with 30% methanol conversion. There was 90% selectivity to 2 - methyl - 1 - propanol. In U.S. patent 5,559, 275 Barger [ 21 ] describes the conversion of methanol and ethanol to higher branched oxygenates by using a catalyst consisting of zinc, magnesium, zirconia, titanium, manganese, chromium, and lanthanide oxi des. Much like Radlowski, the catalyst gave high conversion percentages, but also high selectivities towards CO/CO 2 . In patent 4,935,538, Budge et al. [ 44 ] achieved high C 6 aldol species yields from propanol. The procedure involved impregnating a United catalyst with cesium nitrate and bismuth nitrate, which helped improve yields. Ueda et a l. [ 8 ] explains a process for converting methanol and ethanol to higher alcohols using MgO. Ueda reporte d moderate conversions (60%) of ethanol, but 29% and 46% selectivity to n - propanol and 2 - methyl - 1 - propanol respectively. The 2 - methyl - 1propanol was produced from formed n - propanol reacting with excess methanol, which is characteristic of the Guerbet mecha nism. A continuous fixed bed quartz reactor was used for catalyst evaluation. Reactions were at 390°C and atmospheric pressure. 1.9 Research Objectives 1.9.1 Develop Efficient, Condensed Phase Alcohol Condensation Process Using Neat Ethanol The literature predomin antly reports ethanol Guerbet chemistry in the vapor phase at relatively high temperatures (>300°C). In this research project a condensed phase reaction process will be developed in a batch reactor to convert ethanol to higher alcohols. Lowering the reac tion temperature will minimize side reactions and lower energy costs. Longer contact times are another advantage to liquid phase reactions, thereby increasing conversion. Running at the 15 needed higher pressures will also increase local gas phase concentrat ions and improve mass transfer. 1.9.2 Identify Robust, Stable Catalyst Compositions This research will primarily focus on using heterogeneous catalysts to convert ethanol to 1 - butanol. Many researchers have utilized hydrotalcite/LDH derived Mg - Al mixed oxides as well as hydroxyapatites to catalyze butanol production from ethanol. Mixed oxides are advantageous in that they contain the acid/base pairs necessary for aldol condensations to take place as well as the basicity required to perform ethanol dehydrogenat ion and butyraldehyde hydrogenation. This functionality can be replicated by a more stable and active alumina - supported nickel catalyst. Nickel metal provides dehydrogenation/hydrogenation functionality, while the specific acid sites on the gamma alumina work in conjunction with the nickel to perform the aldol condensation step in the Guerbet mechanism. Therefore, various nickel catalysts will be screened using lanthanum oxide to modify the acidity/basicity ratio to maximize target alcohol yield. Nickel content will be moderated to minimize ethanol decomposition. 1.9.3 Expand Understanding of Ethanol Guerbet Reaction Mechanism There is disagreement in the literature on the mechanism for the ethanol to 1 - butanol reaction. Some believe ethanol dehydrates and dire ctly dimerizes to 1 - butanol while others believe ethanol proceeds via an acetaldehyde intermediate as described by the Guerbet mechanism. Side products such as acetaldehyde, butyraldehyde and higher branched alcohols would support the Guerbet mechanism. One would not expect to see the presence of branched higher alcohols if ethanol directly dehydrates to 1 - butanol. Reaction samples will be carefully analyzed for these intermediates and quantified in this research project to improve understanding of the e thanol Guerbet reacti on 16 REFERENCES 17 R EFERENCES [1] R. Cascone, Chemical Engineering Progress, 104 (2008) S4 - S9. [2] T. Tsuchida, S. Sakuma, T. Takeguchi, W. Ueda, Industrial & Engineering Chemistry Researc h, 45 (2006) 8634 - 8642. [3] E.S. Olson, R.K. Sharma, T.R. Aulich, Applied Biochemistry and Biotechnology, 113 (2004) 913 - 932. [4] E.W. Flick, Industrial Solvents Handbook, 5 ed., William Andrew Publishing/Noyes, 1998. [5] T. Kim, F.E. Celik, D.G. Hanna, S. Shylesh, S. Werner, A.T. Bell, Topics in Catalysis, 54 (2011) 299 - 307. [6] R.T. Clark, US Patent 3,972,952 1976. [7] M.J.L. Gines, E. Iglesia, Journal of Catalysis, 176 (1998) 155 - 172. [8] W. Ueda, T. Kuwabara, T. Ohshida, Y. Morikawa, Journal of the Chem ical Society - Chemical Communications, (1990) 1558 - 1559. [9] J.P. Wibaut, US Patent 1,910,582 1933. [10] M. Guerbet, Comptes Rendus Hebdomadaires Des Seances De L Academie Des Sciences, 133 (1901) 300 - 302. [11] A.S. Ndou, N. Plint, N.J. Coville, Applied Cat alysis a - General, 251 (2003) 337 - 345. [12] T. Tsuchida, K. Atsumi, S. Sakuma, T. Inui, US Patent 6,323,383 2001. [13] T. Tsuchida, J. Kubo, T. Yoshioka, S. Sakuma, T. Takeguchi, W. Ueda, Journal of Catalysis, 259 (2008) 183 - 189. [14] C. Yang, Z.Y. Meng, Jo urnal of Catalysis, 142 (1993) 37 - 44. [15] K.W. Yang, X.Z. Jiang, W.C. Zhang, Chinese Chemical Letters, 15 (2004) 1497 - 1500. [16] T. Tsuchida, T. Yoshioka, S. Sakuma, T. Takeguchi, W. Ueda, Industrial & Engineering Chemistry Research, 47 (2008) 1443 - 1452. [17] M.W. Farrar, US Patent 2,971,033 1961. [18] K. Kourtakis, M.B.D. Amore, L.E. Manzer, US Patent 7,700,810 2010. [19] I.C. Marcu, D. Tichit, F. Fajula, N. Tanchoux, Catalysis Today, 147 (2009) 231 - 238. 18 [20] W.F. Otto, US Patent 1,992,480 1935. [21] P.T. Barger, US Patent 5,559,275 1996. [22] A.S. Ndou, N.J. Coville, Applied Catalysis A - General, 275 (2004) 103 - 110. [23] C.A. Radlowski, G.P. Hagen, US Patent 5,095,156 1992. [24] C.L. Yaws, Knovel, 2012. [25] J.I.D. Cosimo, C.R. Apesteguia, M.J.L. Gines, E. Iglesia, Journal of Catalysis, 190 (2000) 261 - 275. [26] I. - C. Marcu, N. Tanchoux, F. Fajula, D. Tichit, Catalysis Letters, 143 (2013) 23 - 30. [27] K. Kourtakis, M.B. D'Amore, L.E. Manzer, US Patent 7,700,811 2010. [28] K. Kourtakis, M.B. D'Amore, L.E. Manz er, US Patent 7,700,812 2010. [29] K. Kourtakis, R. Ozer, US Patent 7,807,857 2010. [30] K. Kourtakis, R. Orzer, US 2010/0160692 A1 2010. [31] S. Ogo, A. Onda, Y. Iwasa, K. Hara, A. Fukuoka, K. Yanagisawa, Journal of Catalysis, 296 (2012) 24 - 30. [32] S. Og o, A. Onda, K. Yanagisawa, Applied Catalysis a - General, 402 (2011) 188 - 195. [33] D.L. Carvalho, R.R. de Avillez, M.T. Rodrigues, L.E.P. Borges, L.G. Appel, Applied Catalysis a - General, 415 (2012) 96 - 100. [34] K. Koda, T. Matsu - Ura, Y. Obora, Y. Ishii, Chem istry Letters, 38 (2009) 838 - 839. [35] R.L. Wingard, D.F. Wass, G. Dowson, J. Lee, American Chemical Society Spring Meeting, 2013. [36] D. Wass, R. Wingad, G. Dowson, J. Lee, Abstracts of Papers of the American Chemical Society, 245 (2013). [37] T. Riitt onen, E. Toukoniitty, D.K. Madnani, A. - R. Leino, K. Kordas, M. Szabo, A. Sapi, K. Arve, J. Warna, J. - P. Mikkola, Catalysts, 2 (2012) 68 - 84. [38] T. Riittonen, K. Eranen, P. Maki - Arvela, A. Shchukarev, A. - R. Rautio, K. Kordas, N. Kumar, T. Salmi, J. - P. Mikk ola, Renewable Energy, 74 (2015) 369 - 378. [39] H.S. Ghaziaskar, C. Xu, Rsc Advances, 3 (2013) 4271 - 4280. [40] C.K.S. Choong, L. Huang, Z. Zhong, J. Lin, L. Hong, L. Chen, Applied Catalysis a - General, 407 (2011) 155 - 162. 19 [41] C.K.S. Choong, Z. Zhong, L. Hua ng, Z. Wang, T.P. Ang, A. Borgna, J. Lin, L. Hong, L. Chen, Applied Catalysis a - General, 407 (2011) 145 - 154. [42] M.C. Sanchez - Sanchez, R.M. Navarro, J.L.G. Fierro, Catalysis Today, 129 (2007) 336 - 345. [43] J. Sun, X.P. Qiu, F. Wu, W.T. Zhu, International Journal of Hydrogen Energy, 30 (2005) 437 - 445. [44] J.R. Budge, S.V. Compton, US Patent 4,935,538 1990. 20 2 Catalyst Screening 2.1 Introduction The continuing trend in U.S. policy toward production of renewable fuels has led to the maturation of corn ethanol p roduction and the recent emergence of commercial cellulosic ethanol production. In addition to being a biofuel, ethanol is a valuable chemical feed stock that could penetrate existing markets as its production increases. One such market is the productio n of 1 - butanol and higher alcohols. These alcohols have broad applications in the commercial chemical industry as feed stocks and components for solvents, consumer goods, and materials [ 1 ] . As fuels, higher alcohols have energy values closer to gasoline than ethanol and have lower affinity for water, making them superio r as fuel components to ethanol , as was shown in chapter 1 ( Table 1 - 1 ) [ 2 ] . It is no surprise, therefore, that 1 - butanol and higher alcohols have h igher market value than ethanol [ 3 ] . The current method for producing 1 - butanol is the oxo process, in which petroleu m - derived propylene [ 1 ] is hydroformylated (CO + H 2 ) over a homogeneous rhodium catalyst t o form butryaldehyde, which is then hydrogenated to produce 1 - butanol. The oxo process is complicated, requires high energy input, and is costly [ 1 ] , and butanol prices fluctuate with propylene prices. The direct conversion of ethanol to 1 - butanol would be a renewable, environmentally cally than the oxo process. Ethanol production technologies, whether utilizing sugar cane, starches, or cellulose, are constantly becoming more efficient and less expensive. Hence, ethanol prices will continue to decline over time, while oil (and propylen e) prices will almost inevitably increase. 21 Reactions involving carbon - carbon coupling of alcohols are commonl y regarded as Guerbet reactions [ 4 - 7 ] . Guerbet first reported the reaction of ethanol to butanol in the late 19 th century, but achieved only a low yield [ 8 ] In the Guerbet reaction, ethanol is oxidized to acetaldehyde; acetaldehyd e undergoes an aldol condensation to crotonaldehyde, and then the crotonaldehyde is hydrogenated to 1 - butanol [ 9 ] . One molecule of water is formed for every molecule of bu tanol formed. The formed butanol can then react with itself or with ethanol to form 1 - hexanol, 2 - ethyl - 1 - butanol, 2 - ethyl - 1 - hexanol, and other higher alcohols. It is noted that some literature supports this aldol condensation (Guerbet) mechanism, [ 1 , 10 - 12 ] while other authors support a simpl er proton abstraction mechanism [ 9 , 13 , 14 ] . In proton abstraction, a proton is extracted from the beta - carbon o f one ethanol molecule, which produces a nucleophile that a ttacks another ethanol molecule [ 9 ] . The coupled product dehydrates to form 1 - butanol and water. Many researche rs have converted ethanol to 1 - butanol in the vapor phase at relatively high temperatures (>300°C), achieving ethanol conversions of 7 - 80% and 1 - butanol selectivities of 10 - 70% [ 1 , 7 , 9 , 11 , 14 - 27 ] . Others have used ethanol as a limiting reagent with methanol and/or propanol to attain mode rate to high ethanol conversion [ 3 , 6 , 21 , 28 , 29 ] . Catalyst composit ion is the most important factor in higher alcohol yields - many studies used MgO in pure form or mixed with other basic oxides as a solid base catalyst for this reaction [ 6 , 7 , 9 , 15 , 20 , 21 , 24 , 29 ] . Alkali cation - exchang ed zeolit es have also been used [ 13 ] . The most recent - Al 2 O 3 - supported nickel, which have produced the highest 1 - butan ol yields thus far [ 10 , 14 , 17 ] . Few researchers have performed ethanol Guerbet reactions in the condensed phase. - Al 2 O 3 - supported metal catalysts in a batch reactor at 250°C and 22 pressures up to 100 bar [ 30 ] . Ethanol conversion and 1 - butanol selectivity reached maximum values of 25% and 80%, respectively. The same authors then carried out continuous liquid - phase ethanol conversion t o 1 - butanol at 240°C and 70 bar [ 25 ] . With continuous operation, 1 - butanol - Al 2 O 3 catalyst and ethanol conversion was between - Al 2 O 3 produced the highest eth anol conversion. Ghaziaskar et al achieved 35% ethanol conversion and 83% C 4 + alcohol selectivity at 250°C and 176 bar - Al 2 O 3 [ 16 ] . Although heterogeneous catalysts have been most widely studied, Wass et al. claimed 95% selectivity to 1 - butanol with a homogeneous ruthenium diphosphine catalyst [ 31 ] . Mixed oxides, such as Mg x AlO y, contain weak Lewis acid - strong Bronstead base site pairs necessary for ethanol dehydrogenation, aldol condensation, and butyraldehyde hydrogenation [ 5 , 32 ] . - Al 2 O 3 - supported nickel cataly - Al 2 O 3 to achieve a vapor phase ethanol conversion of 19% wi th 64% selectivity to 1 - butanol [ 14 ] . The nickel metal provides dehydrogenation/hydrogenation functi onality, while the - Al 2 O 3 work in conjunction with the nickel to perform the aldol condensatio n step in the Guerbet mechanism [ 5 , 32 ] . - Al 2 O 3 catalyst makes it a good starting point for this work, which investigates modification of the nickel catalyst with lanthanum oxide for condensed - phase ethanol Guerbet reactions. It is hypothesized that lanthanum will inhibit - Al 2 O 3 . Liquid phase processing is intrinsically preferable to the vapor phase route, as i t involves smaller reaction vessels for the same throughput and saves energy by not requiring feed vaporization. 23 2.2 Experimental 2.2.1 Materials and Catalyst Preparation Ni(NO 3 ) 2 2 O (Reagent Grade, Jade Scientific), and La(NO 3 ) 2 2 O (99%, Fluka) were used as - Al 2 O 3 cylindrical extrudates (Johnson Matthey) and - Al 2 O 3 spheres (Strem Chemical). Anhydrous ethanol (Koptec, 200 proof) was used as the initial reactor charge. The catalysts p repared (identifier) - Al 2 O 3 - Al 2 O 3 (10Ni/Al), 8 wt% Ni/7 wt% La 2 O 3 - - Al 2 O 3 (8Ni/7La - Al), 14 wt% La 2 O 3 - - Al 2 O 3 (14La - Al), 8 wt% Ni/9 wt% La 2 O 3 - - Al 2 O 3 (8Ni/9La - Al), and 8 wt% Ni/ 10 wt% CeO 2 - - Al 2 O 3 (8Ni/10Ce - Al). Ni - Al 2 O 3 were prepared by incipient wetness impregnation - Al 2 O 3 using a pre - - Al 2 O 3 pore volume. The catalysts (30 g per batch) were dried at 130°C for 18 hours and then reduced at 525°C and 1 atm in a tubular flow reactor for 20 hours in 35 ml (STP) H 2 /min. Nickel catalysts modified with La 2 O 3 were prepared in the same fashion as above, except that La(NO 3 ) 3 was deposited first by - Al 2 O 3 supp ort followed by drying at 130°C for 18 hours, and calcining at 600°C for 20 hours in 35 ml/min N 2 flow. This assured there was La 2 O 3 on the - Al 2 O 3 surface before the impregnation of the other metals. Most of the catalysts for screening studies were prep crushed into smaller particles (0.3 - 0.8 mm) before use. Several later catalysts were prepared on - Al 2 O 3 spheres and used without crushing in batch reactions. 2.2.2 Catalyst Characterization Acid and base site densities on prepared catalysts were measured with a Micromeritics Autochem II chemisorption analyzer. Ammonia and CO 2 were used for acid and base site 24 adsorption, respectively. After loading and establishing a stable baseline, catalysts were degassed by ramping th e temperature to 600°C at 10°C/min under helium and holding at 600°C for 60 minutes. Outgassed samples were cooled to 25°C at 10°C/min and held for 10 minutes. Ammonia or CO 2 was then passed across the catalyst at a flow of 50 ml/min for 30 minutes. Gas f low was then changed to helium (50 ml/min) for 90 minutes to remove weakly bound gases. Desorption of CO 2 or ammonia was carried out by ramping temperature to 600°C at 10 °C/min and holding at 600°C for 30 minutes. BET surface area measurements were done by nitrogen adsorption at 78K with Micormeritics ASAP 2010. Before analysis, samples were degassed in the degas port of the instrument at 120°C for 24 hr. 2.2.3 Batch Reaction Studies Ethanol Guerbet reactions were performed in a 300ml Parr reactor (Model 484 2, Parr Instruments, Chicago, Illinois) with reaction times between two and ten hours. Typically, 110 g of pure ethanol (or the desired feed mixture) was placed into the reactor along with the desired amount of catalyst. The reactor was purged with nitrog en and sealed with 0.1MPa of nitrogen overpressure. The reactions were carried out at autogeneous pressure. calibrated with the boiling points at 745 mm Hg absolute pre ssure of water (99.45 o C), 1,2 propylene glycol 187.0 o C), and ethyl nonanoate (226 o C) [ 33 ] . The controller measured a temperature of 98°C for water, 185°C for 1,2 propylene glycol, and 225 o C for ethyl nonanoate. Based on these results, a correction of +1 o C was applied to the temperature measured during experiment. This correction was included when comparing experimental results with results obtained from phase equilibrium modeling using the SR - Polar equation of state. 25 Pressure measurements for all reactions us ed a pressure gauge on the reactor head. The pressure gauge had increments of 50 psi for pressures up to 3000 psig. The accuracy of this gauge was checked with a large 10 psi increment dial test gauge. To measure pressures in the range of reaction condi tions, the reactor was evacuated to 14 torr and then filled with ethanol. Ethanol vapor pressure was measured from 190°C to 230°C, accounting for the reactor temperature correction mentioned above. The dial test gauge consistently read 1.5 ± 0.5% (5 - 10 ± 2 - [ 33 ] and the reactor head gauge consistently read 2.8 ± 1.5% (10 - 20 ± 10 psi) below the calculated ethanol vapor pressure. A check of the effect of this error on product yields and sel ectivities from experiment showed less than a one percentage point difference in selectivity in the worst case, so no correction to measured pressure was invoked in analysis of the experiments. Initial catalyst screening experiments were performed at a catalyst loading of 0.093 g catalyst/g ethanol, a 280 rpm stir rate, and 230 °C reaction temperature with only the final reaction mixture analyzed. In all later experiments, however, c oncentration profiles of key species over time were established by withd rawing liquid samples periodically from the reaction end to isolate the liquid sample from the reaction vessel. The sample tube was vented after isolating and the liquid sample was subjected to analysis by gas chromatography. Reactor pressure was monitored during reaction and after cooling at the end of reaction to aid in determining product compositions and quantities of gas formed. The quantity of gas produc ed in reaction was determined by weighing the entire cooled reactor with contents both before and after depressurization; the gas exhausted during depressurization was collected in a gas bag and analyzed by gas chromatography. 26 Liquid phase reaction samples were diluted 10 - fold in acetonitrile and then analyzed on a Varian 450 gas chromatograph configured with a DB wax column (30 m x 0.53 mm ID, 1.0 m film) and TCD detection. The temperature program for GC separation of liquid samples was to hold at 40 °C f or 2 min, ramp to 150°C at 10 °C/min, ramp to 250°C at 30°C/min, and then hold at 250°C for 2.00 min. Chromatographic response factors were determined by injecting calibration samples; in later experiments butyl hexanoate was used as an internal standard t o improve analytical accuracy. A sample chromatogram is shown in Figure 2 - 1 . The full component analysis is provided in Table A2 - 1, in Appendix A. Liquid sample analysis was then changed to a 30 meter sol gel wax column with FI D detection and split injection starting with run 02 - 29TLJ092013 on September 30 th , 2013 . This enabled better separation of components off the column. A split injector was used with a split ratio of 100:1. The temperature program was also changed to holdi ng for 4 minutes at 37 °C, ramp to 90°C at 10°C/min, hold at 90°C for 3 minutes, ramp to 150°C at 10 °C/min, ramp to 230°C at 30 °C/min, and then hold at 230°C for 2 minutes. A sample ch romatogram is shown in Figure 2 - 2 . A detailed component table for this run can be found in Appendix A, Table A2 - 2. Gas phase samples collected during depressurization at the end of experiment were analyzed on a Varian 3300 gas chromatograph with 60/80 Carboxen - 1000 column (15 ft x 1 SS, 2.1mm ID) and argon carrier gas. The temperature program for GC separation of gas samples was to hold at 40 °C for 2.0 min, ramp to 250°C at 20 °C/min, and hold at 250°C for 5.0 min. A calibration gas mixture containing 2.0 vol% each of CO, CH 4 , an d CO 2 , along with 100% CO 2 and 100% CH 4 , were used to develop response factors for the gas analysis. 27 Figure 2 - 1 Sample chromatogram of liquid sample at end of run for run 79TLJ022812 using TCD detection . All alcohol peaks and water peak are resolved. A more detailed sample analysis is s hown in Error! Reference source ot found. of Appendix A. 28 Figure 2 - 2 Sample ch romatogram is shown for run 02 - 36TLJ101713 using FID detection. The sample was injected on a sol gel wax column with FID detector. A more detailed sample analysis is shown in E rror! Reference source ot found. of Appendix A. 29 C hromatographic data were entered into an Excel spreadsheet where calculations of species mass fractions in each liquid sample were carried out. This information was then incorporated into the thermodynamic modeling of the reaction system as d escribed below, ultimately leading to determination of ethanol conversion and product selectivities. 2.3 Results and Discussion 2.3.1 Catalyst Characterization Catalysts used in reaction studies were subject to measurement of BET surface area via N 2 adsorption, aci d site density via NH 3 chemisorption, and basic site density via CO 2 chemisorption. Results are presented in Table 6; profiles from temperature programmed desorption (TPD) experiments are given in Figure 5 for both NH 3 and CO 2 . NH 3 - Al 2 O 3 characterizes weak (35 - 150°C), medium (150 - 300°C), and strong (300 - 600°C) acid sites, and CO 2 - Al 2 O 3 gives information on weak (35 - 150°C), medium (150 - 300°C), and strong (300 - 600°C) basic sites. These temperature r anges for relative base and acid site strengths have been assigned in prior work. [ 12 ] Temperature p rogrammed desorption results in Table 2 - 1 show that total acid site density is relatively constant on the several catalysts examined, although the distribution of acid site strengths shifts toward weaker acidity with addition of La 2 O 3 - Al 2 O 3 . Basic site concentration measured by CO 2 adsorption (Figure 10), in contrast, shows the expected strong increase upon addition of La 2 O 3 , with a concurrent shift from weaker (35 - 150°C) to stronger (> 150°C ) basic sites at higher temperatures. 30 Table 2 - 1 Acid and base site densities from CO 2 and NH 3 chemisorption. 2.3.2 Catalytic Reactions Analysis of reaction equilibria shows that all reactions are favorable except for dehydrogenation of ethanol to acetaldehyde, for which the equilibrium constant at reaction conditions is ~0. 1 (see Supplementary Information). The SR - Polar equation of state has been applied to all reactions carried out and reported in this study. A typical experiment carried out at 230 o C with 8Ni/Al catalyst is analyzed here to illustrate application of the mo del. In this run, samples were taken at ten time points. Ethanol conversion, product selectivities, and carbon recoveries were calculated after application of the SR - Polar equation to each time point ( Table 2 - 2 ). The carbon recovery was greater than 95% for all time points. The 1 - butanol selectivity starts low at 6%, but increases to a maximum of 51%. The selectivity towards CH 4 and CO 2 is high early in the reaction (<60 min), levels out, and then increase again a t 600 minutes. Initial experiments with Ni/Al 2 O 3 and Ni/La 2 O 3 /Al 2 O 3 catalysts were carried out at 230°C, autogeneous pressure, and a catalyst loading of 0.093 g catalyst/g ethanol. In all reactions, the SR - Polar model was applied to most accurately char acterize ethanol conversion and product selectivities. Initial results, shown in Table 2 - 3 - Al 2 O 3 is - 1 ) - 1 ) Surface Area Catalyst Weak Medium Strong Total Weak Medium Strong Total BET (m 2 /g) - Al 2 O 3 188 238 167 593 61 32 2 95 153 8Ni/Al 173 232 284 688 84 51 7 142 152 9 La - Al 185 295 114 594 120 146 49 315 145 8Ni/9La - Al 190 260 125 575 98 126 61 285 124 31 an active catalyst for the ethanol Guerbet reaction. For a 10 hour reaction at 230 °C, 46% ethanol conversion was achieved, higher than the conversion achieved by Yang, [ 14 ] with Table 2 - 2 . Product selectivities are provided along with ethanol conversion and total carbon recoveries. Unidentified selectiv ity is based on total unidentified peak area using 1 - hexanol response factor. Time (min) 0 20 40 60 120 180 240 300 664 1344 Ethanol (conv. %) 2.7 4.2 6.5 7.9 8.9 11.8 14.8 16.4 25.3 41.0 1 - Butanol 5.8 11.5 15.5 21.2 34.7 42.4 46.0 51.6 52.4 47.5 1 - He xanol 0.5 0.0 0.6 1.1 2.2 3.6 4.5 5.5 7.0 7.9 CH 4 4.5 11.1 10.5 10.9 10.1 9.1 8.3 7.6 10.0 14.2 CO 2 0.9 2.2 2.1 2.2 2.0 1.8 1.7 1.5 2.0 2.9 2 - Ethyl - 1 - Butanol 0.0 0.0 0.0 0.0 0.5 0.9 1.4 1.7 2.6 3.6 1 - Octanol 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.7 1.0 2 - E thyl - 1 - Hexanol 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.3 0.7 Diethyl Ether 2.3 3.0 4.4 4.5 6.1 5.4 4.8 4.4 3.2 2.0 Ethyl Acetate 36.8 35.5 22.5 16.8 14.1 9.7 7.4 6.6 4.7 2.6 Acetaldehyde 16.4 14.4 10.4 8.4 7.7 4.4 4.7 4.2 2.5 1.2 Butyraldehyde 0.0 1.5 1.8 1. 8 1.9 1.8 1.7 1.6 1.2 0.6 Unidentified 0.0 0.0 25.4 28.7 15.2 24.1 17.4 11.2 11.2 11.6 Carbon Recov. (%) 99.1 99.1 97.9 97.4 98.2 97.5 97.3 98.2 97.8 95.1 46% selectivity to 1 - butanol and 13% selectivity to C 6 + alcohols. Total higher alcohols include 1 - butanol, 2 - ethyl - 1 - butanol, 1 - hexanol, 1 - octanol, 2 - ethyl - 1 - hexanol, and 1 - decanol. Methane was the major byproduct gas (20% selectivity), with minor amounts of carbon monoxide and carbon dioxide. Other products include acetaldehyde, ethyl acetate, an d diethyl ether. These results are consistent with the carbon - carbon bond cleaving capability of nickel - Al 2 O 3 . Increasing the Ni content to 10 wt% lowered overall alcohol selectivity because of increased ethanol decomposition to gases and possib le steam reforming reactions. 32 Table 2 - 3 Preliminary catalyst screening experiments at 230°C, autogeneous pressure, 0.093 g cat/ g EtOH, and 10 hr run time. HA = higher (C 4 - C 8 ) alcohols. Catalyst HA Sel. (%) Conv. (%) HA Yield (%) 8Ni - Al 57 46 26 10Ni - Al 47 52 24 8Ni/7La - Al 65 50 33 14La - Al 67 16 11 8Ni/9La - Al 71 55 39 8Ni/10Ce - Al 71 50 35 - Al 2 O 3 support, lanthanum oxide (La 2 O 3 ) was chosen as a catalyst modifier since it has been shown to have promoting effects on supported nickel catalysts. [ 34 , 35 ] Alkali earth metal oxides supported by La 2 O 3 - - Al 2 O 3 [ 36 , 37 ] have also been used for ethanol conversion to butanol. Addition of La 2 O 3 - Al 2 O 3 did not directly improve 1 - butanol selectivity, but it did improve the overall total higher alcohol yield. Interestingly, the highest 1 - butanol selectivity was achieved with 14 wt% La 2 O 3 - Al 2 O 3 material, but this catalyst had very low activit y. The catalyst screening showed 8 wt% Ni/9 wt% La 2 O 3 - Al 2 O 3 to be a preferred catalyst, since it gave the highest total higher alcohol (HA) yield of 38% at 57% ethanol conversion. A set of experiments was performed at 230 o - Al 2 O 3 suppo rt and 0.04 g catalyst/g ethanol catalyst loading to establish concentration profiles of product species over the course of reaction. Selectivities to the various products are plotted in Figure 2 - 3 for nickel catalyst with and wit hout added lanthanum oxide. 33 Figure 2 - 3 Selectivities for 1 - butanol, C 6 + alcohols , acetaldehyde, ethyl acetate, and diethyl ether for 8Ni/Al and 8Ni/9La - Al catalysts at 230°C and 0.04 g cat/g EtOH loadin g. - Al 2 O 3 , catalyst, initial selectivity is directed toward acetaldehyde, and ethyl acetate formation, with 1 - butanol selectivity increasing gradually to 50% at 20% ethanol conversion. The addition of lanthanum oxide to the catalyst r emarkably decreases acid - catalyzed ethyl acetate and diethyl ether formation, increases ethanol conversion, and essentially doubles 1 - hexanol selectivity. Because no acetic acid is observed, it is assumed ethyl acetate is a product of the Tis chenko reacti on of acetaldehyde [ 38 ] . Ethyl acetate formation rate is initially rapid and quickly declines to zero - this is likely indicative of a conditioning period of the catalys t, as calculations show the equilibrium constant of ethyl acetate formation from acetaldehyde is large (~800 ). 34 APPENDIX 35 APPENDIX A.1: Sample component tables for TCD and FID for GC analysis Table A 2 - 1 Components are identified from GC - MS for run 79TLJ022812 on DB wax column with TCD detector. GC - MS analysis was performed at the mass spec facility at MSU. Components with (*) identified by GC - MS and (**) indicates component wa s hidden and/or could not be located on chromatogram. Retention Time (min) Chemical ID Peak Area 0.92 Dietheylether 10101 1.10 Acetaldehyde 67549 2.08 Butyraldehyde 27243 2.20 Ethyl Acetate 197058 2.87 Ethanol 3741478 ** 2 - Pentanone* ** ** 2 - Ethyl - B utanal* ** ** Decane* ** 4.25 Ethyl Butyrate* 52692 ** 3 - Hexanone* ** 4.64 Water 2801538 4.71 Butyl Acetate 89989 4.84 Undecane 22841 5.49 2 - pentyl methoxyacetate* 117297 5.91 1 - Butanol 2489708 6.35 2 - Heptanone* 54781 6.86 Butyl Butyrate* 48488 7.08 Ethyl Hexanoate* 16763 7.64 Hexyl Acetate* 22306 7.83 4 - Heptanol* 13899 8.2 2 - ethyl - 1 - butanol 386047 8.41 4 - nonanone* 24286 8.83 1 - Hexanol 649619 9.15 2 - ethylhexyl acetate* 4584 9.26 2 - nonanone* 13771 9.55 Hexyl Butyrate* 16069 9.82 Ethyl oct anoate* 4722 10.56 2 - ethyl - 1 - hexanol 140734 11.01 2 - nonanol* 2716 11.1 Unidentified 24675 11.41 1 - octanol 161617 12.21 2 - undecanone* 16622 12.89 Unidentifed 33246 36 Table A2 - 1 (cont ) 13.60 Unidentified 39883 14.21 2 - butyl - 1 - octanol* 5292 14.40 2 - ethyl - 1 - dodecanol* 8046 Total Unidentified 97804 Table A 2 - 2 Components are identified for run 02 - 36TLJ101713. The sample was analyzed with a sol gel wax column and FID detection. Component with (*) is internal standard. Retention Time (min) Chemical ID Peak Area 2.10 Dietheylether 2485 2.30 Acetaldehyde 7539 2.77 Unidentifed 1747 2.95 Unidentified 3321 3.57 Butyraldehyde 2122 3.73 Ethyl Acetate 4338 3.86 Acetal 1881 3.52 Ethanol 169829 5.38 Un identified 6354 6.57 Crotonaldehyde 2498 7.30 Unidentifed 3582 8.23 Unidentified 2125 8.30 4 - Heptanone 3451 8.66 1 - Butanol 172274 9.34 2 - Ethylhexanal 2263 10.00 Hexenal 1744 11.16 Unidentified 1155 12.01 2 - ethyl - 1 - butanol 32484 12.55 Unidentifie d 1155 13.19 1 - Hexanol 55153 14.57 Butyl Hexanoate* 94096 16.00 2 - ethyl - 1 - hexanol 12968 16.72 Unidentified 2749 17.12 1 - octanol 14665 18.60 Unidentifed 1024 18.87 Unidentified 3100 19.56 1 - Decanol 3700 Total Unidentified 19958 37 A. 2 : Response fact ors for reaction products of the ethanol Guerbet reaction Figure A 2 - 1 TCD response factors for ethanol, 1 - butanol, ethyl acetate, and C 6 + alcohols. Figure A 2 - 2 TCD response factors for acetaldehyde, butyraldehyde, C 4 + esters, 4 - heptanone, and 1 - decanol. 38 Figure A 2 - 3 TCD response factors for 1 - butanol and C 6 + alcohols with butyl hexanoate internal sta ndard. Figure A 2 - 4 FID response factors for ethyl acetate, diethyl ether, 1 - butanol, and C 6 + alcohols with butyl hexanoate internal standard. 39 Figure A 2 - 5 FID response factors for 4 - heptanone, acetaldehyde, butyraldehyde, and C 6 + aldehydes. 40 A. 3 : CO2 chemisorption plot Figure A 2 - 6 CO 2 - - alumina - supported catalysts . A. 4 : NH3 chemisorption plot Figure A 2 - 7 NH 3 - - alumina - supported catalysts. 41 A. 5 : Example BET surface area plot Figure A 2 - 8 Sample BET surface area plot for 8Ni/9La 2 O 3 - - alumina catalyst. 42 REFERENCES 43 R EFERENCES [1] T. Tsuchida, S. Sakuma, T. Takeguchi, W. Ueda, Industrial & Engineering Chemistry Research, 45 (2006) 863 4 - 8642. [2] R. Cascone, Chemical Engineering Progress, 104 (2008) S4 - S9. [3] E.S. Olson, R.K. Sharma, T.R. Aulich, Applied Biochemistry and Biotechnology, 113 (2004) 913 - 932. [4] R.T. Clark, US Patent 3,972,952 1976. [5] M.J.L. Gines, E. Iglesia, Journal o f Catalysis, 176 (1998) 155 - 172. [6] W. Ueda, T. Kuwabara, T. Ohshida, Y. Morikawa, Journal of the Chemical Society - Chemical Communications, (1990) 1558 - 1559. [7] J.P. Wibaut, US Patent 1,910,582 1933. [8] M. Guerbet, Comptes Rendus Hebdomadaires Des Seanc es De L Academie Des Sciences, 133 (1901) 300 - 302. [9] A.S. Ndou, N. Plint, N.J. Coville, Applied Catalysis a - General, 251 (2003) 337 - 345. [10] T. Tsuchida, K. Atsumi, S. Sakuma, T. Inui, US Patent 6,323,383 2001. [11] T. Tsuchida, J. Kubo, T. Yoshioka, S. Sakuma, T. Takeguchi, W. Ueda, Journal of Catalysis, 259 (2008) 183 - 189. [12] L. Wang, Y. Ma, Y. Wang, S. Liu, Y. Deng, Catalysis Communications, 12 (2011) 1458 - 1462. [13] C. Yang, Z.Y. Meng, Journal of Catalysis, 142 (1993) 37 - 44. [14] K.W. Yang, X.Z. Ji ang, W.C. Zhang, Chinese Chemical Letters, 15 (2004) 1497 - 1500. [15] M.W. Farrar, US Patent 2,971,033 1961. [16] H.S. Ghaziaskar, C. Xu, Rsc Advances, 3 (2013) 4271 - 4280. [17] K. Kourtakis, M.B.D. Amore, L.E. Manzer, US Patent 7,700,810 2010. [18] J.T. Koz lowski, R.J. Davis, Journal of Energy Chemistry, 22 (2013) 58 - 64. [19] M. Leon, E. Diaz, S. Ordonez, Catalysis Today, 164 (2011) 436 - 442. [20] I.C. Marcu, D. Tichit, F. Fajula, N. Tanchoux, Catalysis Today, 147 (2009) 231 - 238. 44 [21] A.S. Ndou, N.J. Coville, Applied Catalysis A - General, 275 (2004) 103 - 110. [22] S. Ogo, A. Onda, Y. Iwasa, K. Hara, A. Fukuoka, K. Yanagisawa, Journal of Catalysis, 296 (2012) 24 - 30. [23] S. Ogo, A. Onda, K. Yanagisawa, Applied Catalysis a - General, 402 (2011) 188 - 195. [24] W.F. Ot to, US Patent 1,992,480 1935. [25] T. Riittonen, K. Eranen, P. Maki - Arvela, A. Shchukarev, A. - R. Rautio, K. Kordas, N. Kumar, T. Salmi, J. - P. Mikkola, Renewable Energy, 74 (2015) 369 - 378. [26] T. Riittonen, T. Salmi, J.P. Mikkola, J. Warna, Topics in Catal ysis, 57 (2014) 1425 - 1429. [27] T. Tsuchida, T. Yoshioka, S. Sakuma, T. Takeguchi, W. Ueda, Industrial & Engineering Chemistry Research, 47 (2008) 1443 - 1452. [28] P.T. Barger, US Patent 5,559,275 1996. [29] C.A. Radlowski, G.P. Hagen, US Patent 5,095,156 1 992. [30] T. Riittonen, E. Toukoniitty, D.K. Madnani, A. - R. Leino, K. Kordas, M. Szabo, A. Sapi, K. Arve, J. Warna, J. - P. Mikkola, Catalysts, 2 (2012) 68 - 84. [31] D. Wass, R. Wingad, G. Dowson, J. Lee, Abstracts of Papers of the American Chemical Society, 245 (2013). [32] J.I.D. Cosimo, C.R. Apesteguia, M.J.L. Gines, E. Iglesia, Journal of Catalysis, 190 (2000) 261 - 275. [33] C.L. Yaws, Knovel, 2012. [34] M.C. Sanchez - Sanchez, R.M. Navarro, J.L.G. Fierro, Catalysis Today, 129 (2007) 336 - 345. [35] R. Yang, X. Li, J. Wu, X. Zhang, Z. Zhang, Journal of Physical Chemistry C, 113 (2009) 17787 - 17794. [36] K. Kourtakis, R. Orzer, US 2010/0160692 A1 2010. [37] K. Kourtakis, R. Ozer, M.B.D. Amore, US Patent 2010/0160693 A12010. [38] K. Inui, T. Kurabayashi, S. Sato, J ournal of Catalysis, 212 (2002) 207 - 215. 45 3 VLE Modeling 3.1 Introduction The phase equilibria for near critical ethanol Guerbet reactions with or without gas byproduct reactions have not been characterized in the literature. Alumina - supported nickel catalyst s have shown to produce significant amounts of methane and carbon dioxide. The goal of the ethanol Guerbet reaction is to convert ethanol to 1 - butanol and higher alcohols, therefore it is important to understand the roles of CH 4 and CO 2 . Ethanol can be cl eaved to CH4 and CO2 according to: ( 3 - 1 ) The CH4/CO 2 ratio has been found to be 3 - 5 in my ethanol Guerbet system with nickel catalysts. Reactions are typic ally done at 230°C and autogeneous pressure. Pressures can be anywhere from 5 MPa to 14 MPa. At these conditions, there can be significant quantities of CH4 and CO2 in the liquid phase, while there can be significant quantities of ethanol, water, and buta nol in the vapor phase. In order to do accurate computations of higher alcohol formation kinetics, selectivities, and ethanol conversion, the vapor liquid equilibria (VLE) of the ethanol Guerbet system needs to be quantified. Currently , no one has charact erized and modeled the ethanol Guerbet system. Because of the polar nature of alcohols and water, the system is not ideal. The reaction mixture is also near supercritical, since the critical temperature of ethanol is 241°C and the critical pressure of etha nol is 6.3 MPa [ 1 ] . Consequently, an equation of state model is needed in order to predict higher alcohol VLE at these conditions. Equations of state (EOS) account for the volume of the molecules as well as the interactions between molecules. EOS models are also 46 more suited for predicting P - V - T behavior of pure components and mixtures at near - critical to supercritical conditions. After initial screening of models, th e SR - Polar model in Aspen plus was chosen to model the VLE in the ethanol Guerbet system. Pure component vapor pressure data from the NIST data engine in Aspen was used for checking vapor pressure predictions. Binary data from NIST was also regressed to ob tain binary constants used in the SR - Polar model. Volume translation constants were also applied in the SR - Polar EOS for better prediction of liquid densities. The final model was then used to model the 6 - component system comprised of CH 4 , CO 2 , ethanol, 1 - butanol, 1 - hexanol, and water. Other minor components are grouped with ethanol or 1 - hexanol in the model according to their volatility. 3.2 Thermodynamic Modeling of Reaction 3.2.1 SR Polar EOS The Peng - Robinson - Wong - Sandler (PRWS), predictive Soave - Redlich - Kwo ng (PSRK), and Schwartzentruber - Renon (SR - Polar) equations of state were chosen for initial model c , P c ) [ 2 ] . For the PRWS, three alpha functions were tested, standard PR alpha, Boston - Mathias, and Schwartzentruber. The Figure 3 - 1 ) at 2.1%. Results with other alpha functions for PRW S are shown in Table 3 - 1 . It was decided to use the Schwartz entruber - Renon - (SR) - Polar EOS because it offers the advantage of a temperature dependent molar volume translation parameter. The SR - Polar EOS is also re commended for highly non - ideal systems at high temperatures and pressures. This is needed since predictions of liquid phase density for pure components with an equation of state deviate from experimental data at the near critical region [ 2 ] . 47 Figure 3 - 1 Experimental vapor pressure data is compared with predicted vapor pressure by Peng Robinson and SR Polar [ 3 ] . Table 3 - 1 Comparison of error in vapor pressure calculated by PRWS, PSRK, and SR - Polar. PRWS - 1 denotes standard PR, PRWS - 2 denotes Boston Mathias, and PRWS - 3 denotes Schwartszentruber [ 3 ] . PRWS - 1 PRWS - 2 PRWS - 3 PSRK SR - Polar Ethanol 1.3 1.1 12.5 0.9 0.9 1 - Butanol 15.5 0.8 0.9 1.0 0.8 1 - Hexanol 35.9 6.2 5.7 5.9 7.2 Water 4.9 0.4 0.4 0.4 0.5 Average Abs % error 14.4 2.1 4.9 2.1 2.4 Initial screening of several equations of state led to the selection of the Schwartzentruber - Renon (SR) - Polar equation of state (EOS) to characterize vapor and liquid phases during batch reactor operation. Accurate prediction of liquid and vapor phase densities is important in the current application, but is generally challenging with an equation of state in the near - critical region [ 2 ] . The SR - Polar EOS was chosen because it offers the advantage of a temperature 48 dependent molar volume translation parameter to best predict densities, and is recommended for non - ideal systems at high temperatures and pressures. The SR Polar EOS was developed by Schwartzentruber and Renon in 1989. [ 4 ] It is a Soave - Redlich - Kwong (SRK) type equation of state. The SR - Polar EOS, including volume translation, is ( 3 - 2 ) The s pecies included in the SR - Polar EOS model of the reaction system are ethanol, 1 - butanol, 1 - hexanol, water, CH 4 , and CO 2 , with the ratio of CH 4 :CO 2 determined by GC analysis of vented gas at the end of reaction. In reality, the reaction liquid phase contai ns minor amounts of other species including acetaldehyde, ethyl acetate, diethyl ether, and longer chain alcohols, aldehydes, and esters. For phase equilibrium modeling purposes, these components are combined with one Table 3 - 2 Grouping of observed species into modeled components. Component Modeled Component Ethanol Ethanol 1 - Butanol 1 - Butanol 1 - Hexanol 1 - Hexanol H 2 O H 2 O 2 - Ethyl - 1 - Butanol 1 - Hexanol 1 - Octanol 1 - Hexanol 2 - Ethyl - 1 - Hexanol 1 - Hexanol Diet hyl Ether Ethanol Ethyl Acetate Ethanol Acetaldehyde Ethanol Butyraldehyde Ethanol of the species mentioned above according to volatility. The groupings are given in Table 2 below; unidentified peaks in sample chromatograms were assigned the response factor for 1 - hexanol for calculation purposes, and any unaccounted carbon was assigned as 1 - hexanol for 49 modeling and then subsequently removed to calculate yields based on the model results. Once compositions of the modeled components were established, th eir mole fractions were recalculated according to experimental results to give the modeled quantities of all species observed in Table 3 - 2 . 3.2.2 Parameter Estimation The SR - Polar EOS utilizes quadratic mixing rules for the attractive p arameter a and the repulsive parameter b : ( 3 - 3 ) ( 3 - 4 ) ( 3 - 5 ) ( 3 - 6 ) - ( 3 - 5 - - - 50 Table 3 - 3 - Figure 3 - 2 Table 3 - 3 Binary parameters for the ethanol Guerbet sy stem. Binary ka,ij0 ka,ij1 ka,ij2 lij0 lij2 Abs. Avg. Err. % (T. range) Ref Et OH /CO 2 * - 0.100 40.3 (304 K - 453 K) [ 5 - 7 ] Et OH /1 - Bu OH 0.047 - 15.249 1.0 (323 K - 403 K) [ 8 ] E tOH /Water - 0.004 - 33.871 3.3 (298 K - 473 K) [ 9 , 10 ] 1 - Bu OH /CO2 0.078 24.9 (313 K - 430 K) [ 11 - 13 ] 1 - Bu OH /Water 0.131 - 81.051 0.211 58.836 1.5 (323 K - 403 K) [ 8 ] Et OH /CH 4 * - 2 .362 0.0047 38.4 (398 K - 498 K) [ 14 ] Water/CO 2 - 0.261 0.0006 20.6 (383 K - 523 K) [ 15 ] * The ethanol/CH 4 and ethanol/CO 2 binary parameters were adjusted to our experimental data. The avg. error % was 18% with fitted Aspen NIST data for ethanol/CH 4 , but increased to 38% with parameter adjustment. The avg. error % was 6.5% with fitted Aspen NIST data for ethanol/CO 2 , but increased to 40% with parameter adjustment. Figure 3 - 2 Predicted hase equilibria for ethanol - CH4 is shown with the regressed, temperature dependent ka,EtOH - CH4 and with the ka,EtOH - CH4 adjusted to fit our validation experiments [ 14 ] . 51 The pure component para meters for the SR - Polar EOS are defined by: ( 3 - 7 ) ( 3 - 8 ) The value of i at subcritical temperature is described by the extended Mathias equation ( 3 - 9 ); in the supercritical region the equation to determine i is known as th e Boston - Mathias extrapolat ion ( 3 - 10 ). [ 16 ] For alcohols and water, T r i is less than 1 and i is defined as: ( 3 - 9 ) For CH 4 and CO 2 , T r i is greater than 1 and i is defined as: ( 3 - 10 ) with ( 3 - 11 ) ( 3 - 12 ) ( 3 - 13 ) The parameter i is the acentric factor. For the polar components present in the ethanol Guerbet system, p 1,i , p 2,i , and p 3,i are estimated by the Antoine equation. [ 16 ] The SR - Polar EOS offers a temperature - dependent volume translation parameter c for the purpose of ac curately predicting densities with a linear mixing rule: ( 3 - 14 ) 52 Volume translation does not affect VLE composition calculations but does affect fugacity values. [ 16 ] For alcohols and water, T ri < 1; therefore the pure component volume translation parameter c i for these components is defined by: (for T r <1) ( 3 - 15 ) Table 3 - 4 Volume translation constants are listed for alcohols, water, methane, and CO2. The (*) indicates translation constant was fit to binary with ethanol and not pur e component data [ 3 ] . Component c 0i (m 3 /kmol) c 1i (m 3 /kmol) c 2i (m 3 /kmol) Avg. Abs Error (%) Ethanol 7.00E - 03 2.50E - 03 4.50E - 02 3.8 1 - Butanol 7.00E - 03 2.60E - 03 3.00E - 03 0.6 1 - Hexanol 1.10E - 02 3.00E - 03 0.8 Water 5.00E - 03 7.97E - 04 1.5 Methane* - 1.20E - 01 CO2* - 3.00E - 02 Values of c 0,i , c 1,i , and c 2,i were regressed from pure liquid density data in the AspenPlus database. For CH 4 and CO 2 , T ri > 1 and c 0,i was regressed from binary ethanol/CH 4 and ethanol/CO 2 liquid densities at conditions close to those of reaction (with c 1,i and c 2,i set to zero). Values of all regressed volume translation parameters are given in Table S3 of Supplementary Information. Figur e 3 - 3 SR - Polar translated density predictions are compared with experimental data [ 3 , 14 ] . 3.2.3 Applying SR - Polar EOS to Batch Experiments The SR - Polar EOS is applied to batch Guerbet chemistry experiments to determine compositions and quantities of both vapor and liquid phases. This is done by interfacing the 53 AspenPlus V8.2 process simulator with Microsoft Excel 2013 via the AspenPlus Properties Add - in in Excel. AspenPlus flash operations using TVFlash, particularly to identify bubble and dew point pressures at reaction temperature, are performed in Excel as described below after the Properties Backup (.aprbkp) file with SR - Polar binary constants and parameters is loaded into the Excel workbook. When a liquid phase sample taken during reaction is vented to atmospheric pressure, disso lved gases flash out; consequently, the GC analysis of the remaining material reflects only the composition of the condensable portion of the liquid phase at reaction conditions. Starting with this composition, a mixture of CO 2 + CH 4 with a fixed molar rat io (x CH4 /x CO2 ) is added to the GC - determined liqu i = 1) until the bubble pressure of the combined mixture equals the measured reactor pressure at the point the sample was taken. Once this has been done, the compositions, molecular weights, and densit ies of liquid and vapor phases at reaction conditions are defined. Because the total mass of the reaction mixture ( m T ) is tracked during reaction, and the total mixture volume is constrained by the reactor volume ( V R ), the number of moles of vapor n V (and consequently moles of liquid n L ) in the reactor can be determined by combining Eq. ( 3 - 16 ) and ( 3 - 17 ) to give Eq. ( 3 - 18 ). ( 3 - 16 ) ( 3 - 17 ) ( 3 - 18 ) 54 Once vapor and liquid quantities and compositions are defined, the overall composition ( z i ) of the reaction mixture is calculated for the time of sampling: ( 3 - 19 ) where n tot is the sum of n V and n L . Once this is done, the ratio of liquid - phase CH 4 to CO 2 mole fractions (x CH4 /x CO2 ) chosen for the TVFlash calcu lation is then varied and that calculation is repeated until the calculated overall gas composition ratio (z CH 4 /z CO2 ) matches the overall experimental (z CH4 /z CO2 ) ratio determined by GC analysis. A flowsheet for applying the SR - Polar model is shown in Figure 3 - 5 . The calculation of ethanol conversion and product selectivities and yields are then finally calculated by the following equations: ) x 100% ( 3 - 20 ) x 100% ( 3 - 21 ) ( 3 - 22 ) 55 Figure 3 - 4 re also apply to multi - component mixtures. 56 Figure 3 - 5 Flowsheet for SR - Polar model application to reaction data. 3.2.4 Model Validation and Parameter Adjustment The phase equilibrium model using the SR - Polar E OS for the batch reactor was validated by conducting control experiments in which known quantities of the modeled components were placed into the Parr reactor and the reactor pressure was measured at reaction temperature. Two validation experiments were co nducted at 230 o C: one simulating high ethanol conversion (49%) and the other low ethanol conversion (22%). For these experiments and all subsequent 57 applications of the SR - Polar EOS, a volume of 328 ml for the Parr batch reactor was used. A known liquid m ixture of ethanol, 1 - butanol, 1 - hexanol, and water was first added to the reactor, which was sealed and then weighed. The reactor atmosphere was purged with CO 2 and pressurized to give the desired amount of CO 2 , and then weighed to verify the mass of CO 2 a dded. The gas inlet was then purged with CH 4 , connected to the reactor, and pressurized to a pressure higher than the reactor pressure to prevent backfilling. Methane was then added by monitoring pressure change, and then the reactor was finally weighed ag ain to determine the mass of CH 4 added. The reactor was heated to reaction conditions, reactor pressure was recorded, and liquid samples were taken to determine composition. The SR - Polar EOS was then applied as described in the above section to calculate quantity, composition, and density of liquid and vapor phases in the reactor. Experimental compositions and total calculated gas quantities for the validation experiment are shown in Table 3 - 5 . The SR - polar EOS was found to unde rpredict total gas quantity at the simulated higher conversion; closer inspection of the results revealed that CH 4 solubility in the liquid phase was underpredicted when using the binary parameters regressed from experimental data. The binary VLE parameter s were therefore further checked by performing experiments with just ethanol - CH 4 and ethanol - CO 2 binary mixtures in the batch reactor. Known quantities of the binary mixtures were heated to 215°C and 225°C, and pressure was recorded. The bubble pressure c alculation was performed using TVFlash in AspenPlus for each of the binaries by adjusting k a , EtOH - CH4 0 and k a , EtOH - CO2 0 until the predicated gas error was minimized. At temperatures higher than 225°C, the binary bubble pressure calculation in AspenPlus did not converge because of the proximity to the ethanol critical point. For experimental temperatures up to 225 o C, the best agreement between experimental and predicted 58 quantities of ethanol and CH 4 in the reactor was obtained by adjusting the ethanol/CH 4 bi nary interaction parameter from k a , EtOH - CH4 = 0.0030T - 1.1945 to - = 0.0047T - 2.3619 (T in K). Similarly, the best agreement with the ethanol/CO 2 binary experiment was obtained by adjusting the constant - - . These co rrections resulted in the difference between predicted and actual CO 2 + CH 4 mass decreasing from 0.9 g to 0.6 g (with the total liquid species mass correspondingly adjusted in the other direction) for the high conversion experiment, while the difference be tween predicted and actual total CO 2 + CH 4 in the reactor for the low conversion experiment decreased from 0.4 g to 0.2 g ( Table 3 - 5 ). With this correction, calculated overall reactor composition, ethanol conversion, and higher a lcohol selectivities were within three percentage points of the experimental values in the validation experiments. The adjusted values of k a,EtOH - CH4 and k a,EtOH - CO2 were therefore used in all subsequent applications of the SR - Polar EOS to reaction studie s. Table 3 - 5 Comparison of predicted and experimental gas quantities in SR - Polar validation experiments at 230°C with adjusted ethanol/CH 4 and ethanol/CO 2 binary parameters . Overall mole fraction (z i ) H igh Conversion Low Conversion Component Experimental Predicted Experimental Predicted Ethanol 0.465 0.469 0.770 0.788 1 - Butanol 0.098 0.097 0.077 0.067 1 - Hexanol 0.045 0.041 0.013 0.010 H 2 O 0.202 0.215 0.101 0.099 CH 4 0.133 0.126 0.023 0.022 CO 2 0.0 56 0.053 0.016 0.014 Total Mass of Species (g) 107.0 107.0 84.2 84.2 Observed Pressure (psia) 1565 1565 800 800 Liquid Species Mass (g) 95.1 95.7 82.2 82.4 CH 4 + CO 2 Mass (g) 11.9 11.3 2.0 1.8 Ethanol Conversion 0.485 0.476 0.219 0.194 59 3.2.5 Liquid and V apor Densities The SR - polar EOS gives insight into phase behavior not normally observed or accounted for in batch reaction studies. Most reactions here were carried out at 230 o C, close to the critical temperature of ethanol (238 o C). Under these conditions , the liquid phase is significantly expanded and the vapor phase density is high enough that a significant fraction of the reaction mixture is in the vapor phase. At 230°C, pure ethanol has a liquid phase density of 0.44 g/ml ( Figure 3 - 6 ), slightly more than half its value at 25 o C, and the reaction mixture at partial conversion (right - most picture in Figure 3 - 6 ) is similarly expanded. This liquid phase expansion Figure 3 - 6 Liquid and vapor phase densities for pure ethanol at 25°C (left), pure ethanol at 230°C (center), and for the reaction mixture at 41% ethanol conversion (right). has safety implications , in that if the reactor is initially filled with t oo much ethanol, the liquid phase will expand and fill the reactor head space as reaction temperature is approached, leading to reactor overpressure and possibly reactor failure. Further, solid catalyst is not as easily suspended in the low density liquid solution at reaction conditions, thus leading to possible 60 inaccuracy in rate characterization unless vigorous mixing is ensured. The expanded liquid reaction phase also require s activity - based kinetic models, because of its highly non - idea nature. 3.2.6 Comparis on of SR - Polar EOS analysis with conventional liquid phase analysis Ethanol conversion and alcohol selectivities obtained with the SR - Polar model are compared here to results obtained by conventional analysis of the flashed liquid sample , as is typically d one in batch reaction studies. In conventional liquid phase - only analysis, reactor contents are assumed to be all in the liquid phase, so with pure EtOH as the starting material, EtOH conversion (%) = 100 (wt% EtOH in sample) and selectivity to n - butano l (%) = (2MW EtOH /MW BuOH )(wt% BuOH in sample)/(EtOH conversion (%)). A similar calculation ap plies for other alcohols. Figure 3 - 7 and Figure 3 - 8 give the comparison between these conventional liquid - only values and those calculated using the SR - Polar model. The liquid - only thus misleading. First, EtOH conversion ( Figure 3 - 7 ) is ess entially the same for both analyses, because the EtOH mole fractions in liquid and vapor phases in the SR - Polar model are coincidentally similar at reaction conditions. Butanol selectivity ( Figure 3 - 7 ) and hexanol selectivity ( Figure 3 - 8 ) appear three to seven percentage points higher than the SR - Polar model when calculated using conventional liquid phase analysis, because carbon going to gases (CO 2 + CH 4 ) is not considered and because these heavier specie s are almost entirely present in the liquid phase. Most importantly, the calculated carbon recoveries for both methods are above 95% ( Figure 3 - 8 ) for the conventional liquid phase analysis, this is a direct consequence of the erroneous assumption that all materials are in the liquid phase and is again a misleading result. 61 Figure 3 - 7 Comparison of EtOH conversion and 1 - butanol selectivity for conventional liquid phase analysis and SR - Polar EOS analysis. Figure 3 - 8 Comparison of C6 alcohol selectivity and carbon recovery for conventional liquid phase analysis and SR - Polar EOS analysis. In condensed phase reaction systems where the liquid phase is significantly expanded (e.g. often near the critical point of the mixture), and where species of widely different relative volatility are formed, it is important to do a careful analysis of the reaction mixture to obtain representative results. In this system, 10 - 25% of the reactor contents exist in the vapor state at reaction conditions, and a substantial quantity of gases are formed that are not observed when only the liquid phase is analyzed. The SR - Polar model provides a tool with which a more realistic picture of the products formed in reaction can be characterized. Along with a comparison of carbon recoveries between the liquid - phase only method and using the SR - Polar model, hydrogen and oxygen recoveries were also compared for t he two methods. The hydrogen recovery decreases from 101% to 93% at end of run for the liquid - phase 62 only model ( Table 3 - 6 ). Using the SR - Polar model raised the hydrogen recovery to 98% and above for all the sampl es. Contrary to the carbon and hydrogen recoveries, oxygen recovery was 2 - 3 percentage points above 100% for most of the run and was 99% at end of run for the liquid - phase only method. Applying the SR - Polar model had no real effect on oxygen recovery. I t actually demonstrated an oxygen recovery at end of run that was one percentage point lower than the liquid - phase only method. Table 3 - 6 Hydrogen and oxygen recoveries are compared for the liquid - phase only method and SR - Polar model. Hydrogen Recov. (%) Oxygen Recov. (%) Time (min) Liq. Phase Only SR - Polar Model Liq. Phase Only SR - Polar Model 0 101.0 99.8 102.7 101.4 20 99.8 99.6 101.7 101.4 40 99.5 99.5 101.7 101.6 60 98.9 99.6 101.1 101.6 120 98.2 9 9.0 99.9 100.3 240 97.8 98.8 100.0 100.4 300 97.1 98.3 99.3 99.8 600 97.4 99.2 100.4 100.4 1301 93.3 98.4 98.6 97.7 3.2.7 Experiment al Repeatability Two runs with identical 8 wt% Ni/9 wt% La 2 O 3 catalyst were performed to demonstrate experimental repeatabi lity. The ethanol conversion and carbon recovery profiles for the two runs are similar ( Figure 3 - 9 ). After 10 hours of run time , ethanol co nversion is 32.6% ± 4.6 percentage points and carbon recovery is 97.8% ± 0.01 percentage points . The 1 - butanol selectivity after 10 hours is 52.2% ± 1.2 percentage points and C 6+ alcohol selectivity was 14.5% ± 1.3 percentage points ( Figure 3 - 10 ). The experimental pressure is virtuall y the same for the two runs, but begins to deviate slightly after 10 hours ( Figure 3 - 11 ). At end of run, reactor pressure reached 11.4 MPag for run 1 and reached 14.1 MPag for run 2. Though the end of run 63 reactor pressure is different, the difference did not noticeably affect the gas selectivity (CH 4 +CO 2 ). Figure 3 - 9 Ethanol conversion and carbon recovery are shown for two runs with identical catalyst at 230°C and 0.04 g cat/g EtOH loading. Figure 3 - 10 Selectivity to 1 - butanol and C 6+ alcohols is shown for two runs with identical catalyst at 230°C and 0.04 g cat/g EtOH loading. Figure 3 - 11 Experimental reactor pressure and gas selectivities are shown for two runs with identical catalyst at 230°C and 0.04 g cat/g EtOH loading. 64 3.3 Conclusions The condensed phase Guerbet reaction of ethanol to 1 - butanol and C 6 + higher al cohols - Al 2 O 3 - supported nickel/lanthanum catalysts. An 8wt% Ni/9wt% La 2 O 3 - Al 2 O 3 catalyst gave a total higher alcohol yield as high as 38%, and total - Al 2 O 3 with La 2 O 3 greatly reduced both formation of ethyl acetate via the Tischenko reaction of acetaldehyde and diethyl ether via the acid - catalyzed dehydration of ethanol. Lanthanum oxide addition leads to a two - fold increase in total base site density, and reduct ion in strength of the catalyst acid sites. Applying the SR - Polar EOS to batch Guerbet reactions provides a more rigorous analysis of reaction than conventional liquid phase sampling. The SR - Polar EOS accurately predicts higher alcohol vapor - liquid equili bria, liquid and vapor phase densities, and total quantity of gases produced in reaction. Although the ethanol conversion profile calculated using the SR - Polar EOS is virtually the same as with liquid - phase - only sampling, the product yields and selectivit ies are more accurately represented because species partitioning between liquid and vapor phases are more accurately modeled. Finally, the ability of the SR - Polar EOS to predict liquid expansion is advantageous in safely designing future ethanol Guerbet re actions at near critical conditions. 65 APPENDIX 66 A PPENDIX B.1: Regressed binary data Figure B 3 - 1 Isothermal VLE data for ethanol - 1 - butanol at 323K, including SR - Polar predictio n from regressed parameters [ 8 ] . Figure B 3 - 2 Isothermal VLE data for ethanol - 1 - butan ol at 403K, including SR - Polar prediction from regressed parameters [ 8 ] . 67 Figure B 3 - 3 Isothermal VLE data for ethanol - water at 298K, including SR - Polar prediction from regressed parameters [ 10 ] . Figure B 3 - 4 Isothermal VLE data for ethanol - water at 348K, including SR - Polar prediction from regressed parameters [ 10 ] . 68 Figure B 3 - 5 Isothermal VLE data for ethanol - water at 473K, including SR - Polar prediction from regressed parameters [ 9 ] . Figure B 3 - 6 Isothermal VLE data for ethanol - methane at 448K, including SR - Polar prediction from regressed parameters [ 14 ] . 69 Figure B 3 - 7 Isothermal VLE data for ethanol - methane at 498K, including SR - Polar prediction from regressed parameters [ 14 ] . Figure B 3 - 8 Isothermal VLE data for ethanol - carbon dioxide at 304K, including SR - Polar prediction from regressed parameters [ 6 ] . 70 Figure B 3 - 9 Isothermal VLE data for ethanol - carbon dioxide at 353K, including SR - Polar prediction from regressed parameters [ 5 ] . Figure B 3 - 10 Isothermal VLE data for ethanol - carbon dioxide at 453K, including SR - Polar prediction from regressed parameters [ 7 ] . 71 Figure B 3 - 11 Isothermal VLE data for 1 - butanol - carbon dioxide at 313K, including SR - Polar prediction from regressed parameters [ 13 ] . Figure B 3 - 12 Isothermal VLE data for 1 - butanol - carbon dioxid e at 393K, including SR - Polar prediction from regressed parameters [ 12 ] . 72 Figure B 3 - 13 Isothermal VLE data fo r 1 - butano l - carbon dioxide at 43 0K, including SR - Polar prediction from regressed parameters [ 11 ] . Figure B 3 - 14 Isothermal VLE data for 1 - butanol - water at 323K, including SR - Polar prediction from regressed parameters [ 8 ] . 73 Figure B 3 - 15 Isothermal VLE data for 1 - butanol - water at 383K, including SR - Polar prediction from regressed parameters [ 8 ] . Figure B 3 - 16 Isothermal VLE data for 1 - butanol - water at 403K, including SR - Polar prediction from regresse d parameters [ 8 ] . 74 Figure B 3 - 17 Isothermal VLE data for carbon dioxide - water at 473K , including SR - Polar prediction from regressed parameters [ 15 ] . Figure B 3 - 18 Isothermal VLE data for carbon dioxide - water at 523K, including SR - Polar prediction from regressed parameter s [ 15 ] . 75 B.2: Regressed Density translation parameter Figure B 3 - 19 SR - Polar liquid density prediction of ethanol with fitted density translation parameter [ 3 ] . Figure B 3 - 20 SR - Polar liquid density prediction of 1 - butanol with fitted density translation parameter [ 3 ] . 76 Figure B 3 - 21 SR - Polar liquid density prediction of 1 - hexanol with fitted density translation parameter [ 3 ] . Figure B 3 - 22 SR - Polar liquid density prediction of water with fitted density translation parameter [ 3 ] . 77 Figure B 3 - 23 SR - Polar liquid density prediction of ethanol - carbon dioxide with fitted density translation parameter [ 7 ] . 78 REFEREENCES 79 R EFEREENCES [1] NIST Standard Reference Database Number 69. [2] J.R. Elliot, C.T. Lira, Introductor y Chemical Engineering Thermodynamics, 2nd ed., Prentice Hall, Upper Saddle River, NJ, 2009. [3] A.T. Inc, Burlington, MA, 2013. [4] J. Schwartzentruber, H. Renon, Industrial & Engineering Chemistry Research, 28 (1989) 1049 - 1055. [5] Z. Knez, M. Skerget, L . Ilic, C. Luetge, Journal of Supercritical Fluids, 43 (2008) 383 - 389. [6] S. Takishima, K. Saiki, K. Arai, S.J. Saito, Chem. Eng. Jpn., 19 (1986) 48. [7] H.G. Zhu, Y.L. Tian, L. Chen, J.J. Feng, H.F. Fu, Chemical Journal of Chinese Universities - Chinese, 2 3 (2002) 1588 - 1591. [8] S.E. Kharin, V.M. Perelygin, G.P.I. Remizov, V.U. Zaved., Khim. Khim. Tekhnol., 12 (1969) 424 - 428. [9] V. Niesen, A. Palavra, A.J. Kidnay, V.F. Yesavage, Fluid Phase Equilibria, 31 (1986) 283 - 298. [10] A.V. Nikolskaya, Z.F. Khim., 2 0 (1946) 421. [11] O. Elizalde - Solis, L.A. Galicia - Luna, L.E. Camacho - Camacho, Fluid Phase Equilibria, 259 (2007) 23 - 32. [12] G. Silva - Oliver, L.A. Galicia - Luna, Fluid Phase Equilibria, 182 (2001) 145 - 156. [13] Z. Yun, M. Shi, J. Shi, Ranliao Huaxue Xuebao , 24(1) (1996) 87 - 92. [14] E. Brunner, W. Hultenschmidt, Journal of Chemical Thermodynamics, 22 (1990) 73 - 84. [15] S. Takenouchi, G.C. Kennedy, Am. J. Sci., 262 (1964) 1055 - 1074. [16] A.T. Inc., Burlington, MA, 2013. 80 4 Impact of Water and its Removal on E thanol Guerbet Reaction 4.1 Introduction In the context of the ethanol Guerbet reaction, few have investigated ethanol decomposition to CH 4 , CO 2 , CO, or H 2 hydroxyapatite, MgO, or mixed oxide catalysts [ 1 - 4 ] . Wang et al do not report producing non - - Al 2 O 3 catalyst in a vapor phase reaction of ethanol at 200°C and atmospheric pressure [ 5 ] - Al 2 O 3 - supported Ni and other metals at 250°C and autogeneous pressure in a batch reactor [ 6 ] . They reported formi ng a gas product that was 2/3 H 2 and 1/3 CH 4 . As they stated, Ni has been shown to cleave carbon - carbon bonds of ethanol [ 7 ] . The pr esence of water may play a role in these decomposition reactions. Marcu et al first described the impact of water on 1 - butanol production over a Cu - Mg - Al mixed oxide catalyst [ 8 ] . It was found when water was added to the reaction mixture, Lewis strong basic sites O 2 - were converted to weaker Br ø nsted OH - sites. They investigated the effect of water removal by taking the reaction mixture after runn ing for certain period of time and drying it using MgSO 4 overnight. The reaction was resumed with fresh catalyst; an increase in ethanol conversion and 1 - butanol selectivity was found when water was removed. Riittonen et al also studied the effect of wat er; they suspected water could be playing a role in the hydrogen formation by wa y of steam reforming of ethanol . This was tested by adding 3 Å molecular sieves to the reactor prior to reaction. Ethanol conversion was increased from 20% to 30%. No further c haracterization was performed on gas production or mechanism. The resulting high H 2 concentration in their product gas was different from our reactions in that the primary components of our system were CH 4 and CO 2 . The amount of H 2 at the end of our 81 react at high ethanol conversion the water gas shift rea In fact, the ratio of CH 4 to CO 2 is >3, which is in agreement with the ethanol decomposition reaction: 2C 2 H 5 4 + CO 2 ( 4 - 1 ) This is an undesirable reaction due to two reasons; CH 4 and CO 2 are greenhouse gases and t he price of methane is much cheaper than the price of ethanol. It was observed in all reactant in this reaction equation, it was hypothesized increasing water concentration was promoting the decomposition reaction. Water concentration increases with conversion due to being a byproduct of the Guerbet reaction: 2C 2 H 5 - Butanol+ H 2 O ( 4 - 2 ) Steam reforming literature [ 9 - 12 ] has shown that water can react with EtOH to form H 2 and CO 2 : C 2 H 5 OH + 3H 2 2 + 6H 2 ( 4 - 3 ) Interestingly though, this is more thermodynamically favorable at higher temperatures (>700K) [ 12 ] . The effect of water and its removal on the ethanol Guerbet reaction was studied. 4.2 Experimental 4.2.1 Materials Ni(NO 3 ) 2 2 O ( Reagent Grade, Jade Scientific), was used as precursor to reduced metal and La(NO 3 ) 2 2 O (99%, Fluka) w as used as precursor to the calcined oxide . The support used was - Al 2 O 3 spheres (Strem Chemical). Anhydrous ethanol (Koptec, 200 proof ) 82 was used as the initial reactor charge. The catalyst composition used in this study was 8 wt% Ni/9 wt% La 2 O 3 - - Al 2 O 3 (8Ni/9La - Al). The nickel catalyst modified with La 2 O 3 was prepared by first depositing La(NO 3 ) 3 by incipient wetness impregnation of th - Al 2 O 3 support followed by drying at 130°C for 18 hours, and calcining at 600°C for 20 hours in 35 ml/min N 2 flow. This assured there was La 2 O 3 on the - Al 2 O 3 surface before the impregnation of the nickel . Nickel was then added by the same wetness tec hnique using Ni(NO 3 ) 2 and dried at 130°C for 18 hours. The nickel was reduced at 450 °C and 1 atm in a tubular flow reactor for 20 hours in 35 ml (STP) H 2 /min. 4.2.2 Catalytic reactions with water addition The effect of water on CH 4 and CO 2 yields was confirme d by performing reactions with different initial water concentration in a 300 ml Parr autoclave reactor. The water concentrations studied were 0, 5 and 10 weight %. Ethanol and/or mixtures with water were placed in the reactor with catalyst and the reac tor was purged with 1 atm of nitrogen and sealed. Reactor pressure was autogen eous and liquid samples were taken at specific time intervals. Reactor pressure was measured with a head pressure gauge at each sampling interval. The reaction temperature was 2 30°C at 900 rpm stir rate and a 0.04 g catatlyst/ g ethanol loading. 4.2.3 Catalytic r eactions with water removal A system was required that could effectively remove water at high temperatures (230°C) and high pressures (700 - 2000 PSIG), while not interfering wi th the 8Ni/8La catalyst. When a desiccant is placed directly into the reactor, there is a chance the desiccant will catalyze side reactions, such as ethanol dehydration to diethyl ether if an acidic desiccant like calcium sulfate or a molecular sieve is u sed. Dessicants also work more effectively at lower temperatures. A 83 water drying loop outside the reactor was devised to eliminate any potential for this issue to occur ( Figure 4 - 1 ). Figure 4 - 1 Recirculation loop with drying bed is shown with attachments to a 300 ml Parr reactor. The loop includes a chiller that cools the mixture to ~40°C and a heater that heats it back up to ~220°C. T he water drying loop consisted of a cooling section that dropped the reaction temperature of 230°C down to 35°C - , . The reaction mixture had to be cooled down to this temperature since molecular sieves 3A have the highest adsorpt ion rate and capacity at lower temperatures. [ 13 ] The cooled mixture was then fed into a magnetically driven, single piston reciprocating pump. A pump was n eeded that could perform recirculation at autogeneous pressures up to 2000 PSIG. The pump was designed by Seifried et al. and was published in Review of Scientific Instruments. [ 14 ] It was well suited for this application since all joints were 316 stainless steel Swagelok. A pump with rubber o - rings and seals may not have stood up to the pressures and temperatures of this reaction system. 84 The pump (assembled by hand) , shown in Figure 4 - 2 , consisted of 1/4" OD stainless tubing with associatively sized union tees and check valves. The four check valves allowed liquid to flow in only one direction. The main piston chamber was comprised of a magnetic 416 stainless steel precision ground rod that was 5/16 OD and 13 cm in length . The piston sat inside a 1/2" OD x 20 cm SS316 smooth bore seamless tube . The piston was driven by the push - pull forces of two solenoids. The piston solenoids were controlled by a n in - house built square wave generator. 3 . The drying bed was filled with roughly 55 g of ~ 2mm molecul ar sieve extrudates. Runs were also performed with ~ 2mm beads. The bead shape versus extrudate shape had negligible impact on packing density. The reaction mixture was then heated close to reaction temperature (~220°C) with heat tape before being return ed to the reactor. Samples were taken at intervals as done previously above. 85 Figure 4 - 2 Schematic is shown for the recirculating pump, taken from Seifried [ 14 ] .Circulating pump body parts: (a) pump cylinder, (b) solenoids (S1 and S2), (c) reducing union, (d) union cross, (e) plug, (f) check valve, (g) union - tee, (h) needle valve, (i) tubing, (j) piston, and (k) compression spr ing. The water removal studies involved three different runs. The first run was performed with no loop and pure ethanol. This was one of two control runs. The second control run used the recirculating loop with glass beads used in the dryer instead of molecular sieves. This control was to show if the recirculation loop itself had any influence on the reaction. The third run utilized the drying loo p, starting with pure ethanol. The reaction temperature was 230°C at 900 rpm stir rate and a 0.04 g cataly st / g ethanol loading. 4.2.4 Modeling reaction system with water removal Since water was being removed during the fractions increased accordingly. This increase had to be accounted for in the conversion and selectivity calc ulations, because they are based on the total amount of moles left in the reactor system at time, t i . The theoretical quantity of water formed up to t i is calculated by: 86 ( 4 - 4 ) The accumulated water up to t i is calculated by: ( 4 - 5 ) In a reaction with no water removal, the theoretical water formed can be checked with the water accumulated by the ratio: ( 4 - 6 ) Where n H2Ocheck,ti For the water removal runs, a material balance can first be performed on the entire reactor system, including t he water removal loop . At each time interval samples and rinses are taken out of the reactor, so this is accounted for by: ( 4 - 7 ) Where, is the system mas at t i is the sample mass is the waste mass from rinsing - P olar EOS so thi s mass is subtracted from the system mass to get the reactor mass at t i : ( 4 - 8 ) Where, n loop is th e moles of reactant in the loop and M L is the average molec ular weight of the liquid phase. 87 water taken out in this step of the calculation. As was done with no recirculation loop in the VLE studies, the mass of the re actor is combined with the reactor volume constraint to determine moles in the liquid and vapor phases: ( 4 - 9 ) ( 4 - 10 ) ( 4 - 11 ) With n L and n V determined, water formed in the interval from t i - 1 to t i can be determined by: ( 4 - 12 ) Water accumulated is calculated from: ( 4 - 13 ) Water removed is found by subtracting water accumulated from water formed: ( 4 - 14 ) The total quantity of water removed at t i is found by summing all interval amounts of water removed back to t 0 : ( 4 - 15 ) The system mass is a function of water removed; after the first result of water removed at t i is determined, it is subtracted from the initial approximation of the sys tem mass. Water formation, accumulation, and removal are recalculated based on the new values for n L and n V . When there is no difference between calculated and predicted water removed, the true amount of 88 water removed at t i has been found. Product selec tivities are based on total accumulated product at t i , which includes samples and rinses: ( 4 - 16 ) Where a is the ratio of carbon atoms in the product divided by carbon atoms in ethanol. E thanol co nversion is calculated similarly by: ( 4 - 17 ) 4.3 Results and Discussio n 4.3.1 Water addition runs It was found when starting with 0, 5 , and 10 weight % water, the gas selectivity (CH 4 + CO 2 ) profiles were similar wit h time ( Figure 4 - 3 ) . Gas selectivity for the 10% water run was approximately 4 - 5 points higher than the other two. The gas selectivity profiles vs conversion were virtually identical; gas selectivity steadily increases to 15% at 50% ethanol conversion. It w as observed gas selectivity gradually increased with increasing water concentration ( Figure 4 - 4 ). For the 10% water run, gas selectivity hovered at 10% up to 13% water concentration , then increased 10 points to 20% selectivity wh en water concentration reached 17%. The theoretical amount of water generated for the control runs was calculated using equation ( 4 - 4 ) . The ratio of the theoretical water generated to total water measured was c alculated ( Figure 4 - 5 ). From 0 - 20% ethanol conversion, the 5% and 10% water runs are 0.90 to 0.95. The pure ethanol run deviated more from unity and was around 0.80. Since all three runs had ratios less than 1, this meant more w ater was measured than what could be accounted for. It was reasoned additional could be generated from the hydrogen reacting with ethanol: 89 ( 4 - 18 ) This agreed with experimental data because CH 4 /CO 2 was typically between 3 - 5. The excess methane was then used to calculate additional water generated and incl uded into the accounted water ratio. The results are plotted on the right plot in Figure 4 - 5 . The correction helped shift the ratio closer to 0.9 and above for all runs. Figure 4 - 3 Gas selectivity (CH 4 + CO 2 ) is plotted vs time on the left and vs ethanol conversion on the right. Figure 4 - 4 Gas selectivity is plotted vs water concentration on the left and 1 - butanol, C 4 - C 8 alcohol selectivity is plotted vs conversion on the right. 90 Figure 4 - 5 Accounted water ratio is plotted vs ethanol conversion on the left and the accounted water ratio including a correction for water formed from f ormed CH 4 is plotted vs ethanol conversion on the right. 91 4.3.2 Water removal runs Gas selectivites were found to be lower for the run with water removal than the run with no water removal ( Figure 4 - 6 ) . Interestingly, the run with glas s beads (no drying) subsituted for the molecular sieves had an identical gas selectivty profile. For the run without water removal, gas selectivty starts at 25% at beginning of the run, jumps down to 10% at 40 minutes, and gradually increases to 15% at en d of run. For the runs with water removal, gas selectivity is constant at 5% for most of run and rises to 8% at end of run. Higher alcohol selectivites are slightly higher for the run with water removal versus the run with no water removal. The higher al cohol selectivity for the run with glass beads (no drying) . These results support water removal as a potential route to minimizing gas formation. Figure 4 - 6 Gas selectivites are plotted vs time (top left) and vs ethanol conversion (top right). Higher alcohol selectivites are plotted vs time (bottom left) and vs ethanol conversion (bottom right). 92 G as and higher alcohol selectivities are plotted vs water concentration in Figure 4 - 7 . currently unexplained why the run with glass beads in the drying bed had a lower gas selectivity profile than the run with molecular sieves. The data show with water content reaching a maximum of 12 wt%, which was close to the run with no water removal loop at 10 wt%. On the contrary, higher alcohol selectivity was noticeably higher for the water removal run than the ru n with glass beads at water concentrations less than 5 wt%. For all five runs shown in Figure 4 - 7 , higher alcohol selectivity trends downward with increasing water concentration. Carbon recovers were greater than 95% for the run without the water removal loop and for the run with water removal ( Table 4 - 1 ). The carbon recoveries for the run with glass beads were much lower, with the lowest being 88% at end of run. Figure 4 - 7 Gas (left) and higher alcohol (right) selectivities are plotted vs water concentration (wt%). Table 4 - 1 Ethanol conversion and carbon recoveries are shown for runs with and without the water removal loop. No Loop With Loop (Glass Beads) With Loop (Dryer) Time (min) Conv. (%) C. Rec. (%) Time (min) Conv. (%) C. Rec. (%) Time (min) Conv. (%) C. Rec. (%) 0 2 100 0 3 99 0 2 99 20 5 99 20 6 99 20 7 98 40 8 99 40 8 99 40 10 97 60 9 98 60 11 99 60 12 96 93 Table 4 - 120 14 97 120 15 98 120 18 96 180 17 98 180 18 98 180 22 95 240 20 97 240 21 98 240 24 95 300 23 97 327 25 98 300 27 95 626 31 96 639 33 98 622 39 92 1346 48 96 1369 46 97 1376 52 91 4.4 Conclusions It was hypothes ized removing water would minimize ethanol decomposition to CH 4 and CO 2 . The data supports this, with gas selectivity at 25% for the run starting with 10 wt% initial water concentration. For the run with water removal, gas selectivity stays at 5% at water concentrations up to 4 wt % and 40 % ethanol conversion. It is unclear why minimizing water content decreases ethanol decomposition, because water is not a reactant. Therfore, increasing water concentration must be affecting the catalyst surface. Water may be saturating condensation sites on the alumina, thereby promoting nickel metal sites for decomposition. It is also possible water is irreversibly converting - Al 2 O 3 to boehmite , which lacks the catalytic abilities of - Al 2 O 3 . 94 REFEREENCES 95 REFEREENCES [1] M. Leon, E. Diaz, S. Ordonez, Catalysis Today, 164 (2011) 436 - 442. [2] I. - C. Marc u, N. Tanchoux, F. Fajula, D. Tichit, Catalysis Letters, 143 (2013) 23 - 30. [3] A.S. Ndou, N. Plint, N.J. Coville, Applied Catalysis a - General, 251 (2003) 337 - 345. [4] T. Tsuchida, S. Sakuma, T. Takeguchi, W. Ueda, Industrial & Engineering Chemistry Researc h, 45 (2006) 8634 - 8642. [5] K.W. Yang, X.Z. Jiang, W.C. Zhang, Chinese Chemical Letters, 15 (2004) 1497 - 1500. [6] T. Riittonen, E. Toukoniitty, D.K. Madnani, A. - R. Leino, K. Kordas, M. Szabo, A. Sapi, K. Arve, J. Warna, J. - P. Mikkola, Catalysts, 2 (2012) 6 8 - 84. [7] T. Kratochwil, M. Wittmann, J. Kuppers, Journal of Electron Spectroscopy and Related Phenomena, 64 - 5 (1993) 609 - 617. [8] I.C. Marcu, D. Tichit, F. Fajula, N. Tanchoux, Catalysis Today, 147 (2009) 231 - 238. [9] C.K.S. Choong, Z. Zhong, L. Huang, Z. Wang, T.P. Ang, A. Borgna, J. Lin, L. Hong, L. Chen, Applied Catalysis a - General, 407 (2011) 145 - 154. [10] M.C. Sanchez - Sanchez, R.M. Navarro, J.L.G. Fierro, Catalysis Today, 129 (2007) 336 - 345. [11] J. Sun, X.P. Qiu, F. Wu, W.T. Zhu, International Journa l of Hydrogen Energy, 30 (2005) 437 - 445. [12] I. Fishtik, A. Alexander, R. Datta, D. Geana, International Journal of Hydrogen Energy, 25 (2000) 31 - 45. [13] S. Systems. [14] B. Seifried, F. Temelli, Review of Scientific Instruments, 80 (2009). 96 5 Investigat ion of the Ethanol Guerbet Reaction Mechanism 5.1 Introduction In the condensed phase Guerbet reaction of ethanol to 1 - butanol and higher alcohols, La 2 O 3 - modified - Al 2 O 3 supported Ni has proven to be an active catalyst. Unfortunately, this activity is not only for dehydrogenation/hydrogenation and aldol condensation reactions. Competing reactions occur, such as the Tischenko reaction of acetaldehyde to ethyl acetat e, etherification of ethanol to diethyl ether, ethanol decomposition to CH4 and CO2, and steam reforming reactions of ethanol. Modifying - Al 2 O 3 with La 2 O 3 has already been shown to decrease Tischenko and etherification reactions. There is yet need for und erstanding the ethanol Guerbet reaction so that selectivity to unwanted side products is minimized. Much can be learned from breaking down the steps of the Guerbet reaction: Little has been presented in the literature on the relative reaction rates of these steps or if any of these reactions may be in chemical equilibrium, which may impact the 1 - butanol formatio n rate. Enthalpies of reaction and Gibbs free energies of reaction have been calculated for Guerbet Table 5 - 1 Gas phase enthalpies and free energies of formation (298 K) and equilibr ium constants at 503 K [ 1 ] . o r (kJ/mol) r (kJ/mol) Ke V Reaction (298K) (298K) (503K) 2 60.67 34.71 8.21E - 02 2 Acetaldehyde Crotonaldehyde + H 2 O - 13.42 - 6.57 2.27 Crotonaldehyde + H 2 1 - Butyraldehyde - 103.00 - 72.22 7.74E+05 1 - Butyraldehyde + H 2 1 - Butanol - 67.60 - 34.01 31.83 2 Acetaldehyde Ethyl Acetate - 112.09 - 62.12 4.42E+03 2 O - 38.88 - 13.89 10.82 4 + CO 2 - 163.26 - 210.47 2.46E+26 97 reactions as we ll as observed side reactions ( Table 5 - 1 ). It was determined that ethanol dehydrogenation to acetaldehyde in the vapor phase at 503 K has an equilibrium constant of 0.1 . This reaction has the lowest equilibrium constant of all reactions in the Guerbet reaction system. The only o ther reaction with a low equilibrium constant is the aldol condensation of acetaldehyde , which has an equilibrium constant of 2.3 . All oth er reactions are irreversible. Reactions of ethanol and acetaldehyde with H 2 were probed to examine the kinetics an d equilibrium of ethanol dehydrogenation in the liquid phase. Kinetics of acetaldehyde condensation and butyraldehyde hydrogenatio n were also examined. Pure ethanol reactions were performed with H 2 to determine if H 2 would slow down ethanol reaction rate s. 5.2 Experimental 5.2.1 Materials Ni(NO 3 ) 2 2 O (Reagent Grade, Jade Scientific), was used as precursor to reduced metal and La(NO 3 ) 2 2 O (99%, Fluka) w as used as precursor to the calcined oxide . The support used was - Al 2 O 3 spheres (Strem Chemical). Anhydrous ethanol (Koptec, 20 0 proof) , acetaldehyde (Fluka, 99%), butyraldehyde (Fluka, 99%), and isobutyraldehyde (Aldrich, 98%) were used for ethanol and ethanol/mixed aldehyde reactions . Reactions of 80 mol% ethanol/20 mol% acetaldehyde were performed at 150°C and 200°C with 5.2 M Pa initial charge of ultra pure H 2 (Airgas, 99.999%). Reaction of 51 mol% ethanol/49 mol% butyraldehyde was performed at 230°C. The catalyst composition used in this study was 8 wt% Ni/9 wt% La 2 O 3 - - Al 2 O 3 (8Ni/9La - Al). The nickel catalyst modified with La 2 O 3 was prepared by first depositing La(NO 3 ) 3 by - Al 2 O 3 support followed by drying at 130°C for 18 hours, and calcining at 600°C for 20 hours in 35 ml/min N 2 flow. This assured there was La 2 O 3 on the - Al 2 O 3 surface before the impregnation of the nickel . Nickel was then added by the same 98 wetness technique using Ni(NO 3 ) 2 and dried at 130°C for 18 hours. The nickel was reduced at 450 °C and 1 atm in a tubular flow reactor for 20 hours in 35 ml (STP) H 2 /min. 5.2.2 Ethanol/ acetaldehyde /H 2 reactions All reactions were performed in a 300 ml Parr autoclave reactor. A mixture of 80 mol% ethanol and 20 mol % acetaldehyde was made prior to starting the reaction. The ethan ol/acetaldehyde mixture was placed in the reactor with catalyst at 0.04 g cat/g mixture loading . The reactor was purged with H 2 , sealed, and then pressurized to 5.2 MPa with H 2 . Reactor pressure was autogen e ous and liquid samples were taken at specific ti me intervals. Reactor pressure was measured with a head pressure gauge at each sampling interval. 5.2.3 Ethanol/H 2 reactions The same reactor used for ethanol/acetaldehyde/H 2 reactions was also used for ethanol/H 2 reactions. Two reactions were performed at 230 ° C and 0.04 g cat/g ethanol catalyst loading. In both runs the 8Ni/9La - Al catalyst was placed in the reactor, sealed, and purged with H 2 overpressure. The reactor was heated to 250 ° C for 20 hours to pre - reduce the catalyst. After the reactor cooled to am bient temperature, ethanol was fed to the reactor from a 150 ml charging vessel. For the control run, the H 2 atmosphere was purged and sealed with N 2 at 1 atm before reactor heat - up . For the run with H 2 , the reactor was pressurized to 1.4 MPa H 2 before r eactor heat - up. 5.3 Results and Discussion 5.3.1 Preliminary Initial Rate Kinetics for 1 - Butanol and 1 - Hexanol Th e formation rates of 1 - butanol and 1 - hexanol were characterized at three different temperatures: 215°C, 230°C, and 239°C ( Figure 5 - 1 ). A run was done at 239°C, since the 99 critical point of ethanol is at 241°C and it was desired to stay out of the supercritical regime. Figure 5 - 1 Initial rate data for 1 - butanol and 1 - hexanol at 0.06 g cat/g EtOH loading. Data points from the first hour of each run provided linear rates from which activation energies could be determined. Data w ere plotted for 0.06 and 0.03 g cat/g ethanol catalyst loadings. The activation energies for 1 - butanol and 1 - hexanol formation were calculated to be 52 KJ/mol and Figure 5 - 2 Arrhenius plots are shown for 1 - butanol and 1 - hexanol at 0.03 and 0.06 g cat/g EtOH loading. 63 KJ/mol respectively from an Arrhenius plot for these rates ( Figure 5 - 2 ). The activation energies were taken as averages for the two different catalyst loadings. For comparison, Tsuchida calculated activation energies of 61 KJ/mol and 70 KJ/mol for 1 - butanol and 1 - hex anol formation over hydroxyapatite catalysts [ 2 ] . 100 5.3.2 Effect of Catalyst L oading The effect of catalyst loading on 1 - butanol and 1 - hexanol formation rates was also examined ( Figure 5 - 3 ). As expected, rates increased with increasing catalyst loading. For 1 - butanol, when the loading was doubled from 0.02 to 0.04 g cat/g EtOH, the rate increased by a factor of 1.9. When the loading was quadrupled from 0.02 to 0.08 g cat/g EtOH, the rate increased by a factor of 3.4. The rate increase may not have equaled a factor of 4 due to not all catalyst spheres being suspended in solution at the higher loading. The rate increase factors for 1 - hexanol were more closer to theoretical with 2.1 and 3.9 when doubling and quadrupling catalyst loading. This demonstrates that the reactions are carried out under a catalyzed, kinetic regime at the conditions used. Figure 5 - 3 Initial reaction rates for 1 - butanol and 1 - hexanol formation at different catalyst loadings. 101 Figure 5 - 4 Selectivites for 1 - butanol and 1 - hexanol are plotted at different catalyst loadings. The selectivity profiles were examined at the different loadings ( Figure 5 - 4 ). It was found for both 1 - butanol and 1 - hexanol the selectivity inc reases with increasing catalyst loading. At 40 minutes the 1 - butanol selectivity increased from 20% to 55% with the higher catalyst loading. At 60 minutes, the 1 - hexanol selectivity increased from 4% to 12% with the higher catalyst loading. Ethanol conve rsion was calculated two ways ( Figure 5 - 5 ). Conversion was first calculated based on reacted ethanol carbon determined from the ethanol concentration. Ethanol conversion was then calculated based on carbon equivalence in the liqu id products. The ethanol conversion based on liquid products is lower than the ethanol conversion based on ethanol concentration. The explanation for this is that there are unaccounted gases in the conversion based on liquid products. It can be seen f rom Figure 5 - 5 that the slopes for the different loadings are different for the two plots. The most noticeable difference is ethanol conversion for the 0.02 g cat /g EtOH 102 loading in the left plot. This shows a higher ethanol reac tion rate for the 0.02 g cat/g EtOH Figure 5 - 5 Plotted on left is ethanol conversion vs time, based on carbon in ethanol. Plotted on right is ethanol conversion vs time based on ethanol carbon equivalence in the liquid product. loading than the 0.04 g cat/g EtOH loading. When looking at the conversion based on liquid products, the 0.02 g cat/g EtOH loading has the lowest ethanol reaction rate, as expected. The 0.04 g cat / g EtOH loading is a factor of 1.7 larger than the 0.02 g cat/g EtOH loading on the right plot, which is close to the theoretical factor of 2. As explained in the paragraph above, there were unaccounted gases in the liquid product. This means for the 0.02 g cat/g EtOH Figure 5 - 6 Selectivities for 1 - butanol and 1 - hexanol are plotted vs conversion at different catalyst loadings. Ethanol conversion was determined from reacted ethanol carbon based on GC analysis of the liquid product. 103 T able 5 - 2 Ethanol conversion based on ethanol concentration and carbon in the products is shown for the different loading runs. Carbon recoveries are also shown for the different loadings. All values shown a re percentages. 0.02 g cat/g EtOH 0.04 g cat/g EtOH 0.08 g cat/g EtOH time (min) EtOH Conv. Prod. Conv C. Recov. EtOH Conv. Prod. Conv. C. Recov. EtOH Conv. Prod. Conv. C. Recov. 0 2.8 0.3 97.6 3.5 0.4 96.9 0.2 0.6 100.4 20 4.6 0.9 96.3 4.5 1.9 97.4 5 .2 3.2 98.0 40 6.3 2.3 96.0 4.1 3.8 99.8 8.6 6.8 98.2 60 7.2 3.4 96.2 6.1 5.7 99.7 10.9 9.7 98.9 120 6.6 6.0 99.4 11.0 9.5 98.5 15.7 14.8 99.1 180 11.0 8.2 97.3 16.6 12.5 96.0 22.5 19.6 97.2 240 14.9 10.1 95.2 19.3 15.7 96.3 28.4 23.8 95.6 300 16.9 1 2.0 95.0 23.4 18.5 95.1 42.1 34.3 92.8 360 15.8 14.0 98.2 25.9 20.9 95.2 68.5 49.8 83.2 loading, the nickel catalyst may have been more active for ethanol decomposition. From both with increasing catalyst loading. VLE modeling demonstrated the liquid phase is significantly expanded; therefore it is possible not all catalyst particles were suspended at the higher catalyst loadings. Selectivites to 1 - butanol and 1 - hexanol are plot ted vs ethanol conversion in Figure 5 - 6 . In both plots, selectivity to 1 - butanol and 1 - hexanol is lower at low ethanol conversion (~5%) for all catalyst loadings. In particular, 1 - butanol selectivity is ~ 5 percentage points low er for the 0.02 g cat/ g EtOH loading at ethanol conversion below 15%. This is consistent with the catalyst being more active for gas selectivity at the low loading. Carbon recoveries are also lower for the 0.02 g cat/ g EtOH loading ( T able 5 - 2 ). The higher loading catalysts may have had more t pre - reduced in - situ . The selectivity to 1 - butanol reaches a maximum of 60% at 15% ethanol conversion. In general it appears the catalyst goes throug h a condition period for all catalyst loadings; this explains the jump in higher alcohol selectivity at ethanol conversion less than 5%. 104 5.3.3 Kinetic modeling of ethanol/acetaldehdye/H 2 r eactions The formation of 1 - butanol from ethanol likely proceeds by way o f an acetaldehyde intermediate; this is considered the classical Guerbet route. The other intermediates involved in this mechanism are the aldol condensation product, crotonaldehyde, and the partial hydrogenated product of crotonaldedhye, butyraldehyde. The literature has not really focused on the kinetics of the intermediate reactions. Tsuchida et al proposed reactions steps for 1 - butanol and C 6 + alcohol formation, but only considered overall reactions for their kinetic modeling [ 2 ] . For instance, the rate constant, k 1 , was found for the reaction, 2EtOH - BuOH + H 2 O . A rate constant was also found for ethanol dehydrogenation to acetaldehyde, but no rate constants were found for acetaldehyde condensation or any aldehydic hydrogenations. For the mixed ethanol/acetaldehyde with H 2 runs in this paper, we proposed the following reaction steps: ( 5 - 5 - 1 ) ( 5 - 5 - 2 ) ( 5 - 5 - 3 ) ( 5 - 5 - 4 ) It was stated earlier that the equilibrium constant is ~ 0.1 for ethanol dehydrogenation (Reaction 5 - 1) in the vapor phase . Since acetaldehyde hydrogenation and not ethanol dehydrogenation was observed in the mixed ethanol/ acetaldehyde runs, an equilibrium constant for Reaction 5 - 1 Hydrogena tion of crotonaldehyde ( R eaction 5 - 3 ) was very rapid, so it was assumed its net rate of formation was zero. These reactions were modeled in Pol ymath by the following differential equations: 105 ( 5 - 5 - 5 ) ( 5 - 5 - 6 ) ( 5 - 5 - 7 ) ( 5 - 5 - 8 ) ( 5 - 5 - 9 ) ( 5 - 5 - 10 ) ( 5 - 5 - 11 ) The concentrations in the rate equations have units of molarity and are represented by [ EtOH ] (ethano l ), [ AD ] (acetaldehyde), [ CAD ] (crotonaldehyde), [ BAD ] (butyraldehyde), [ BuOH ] (1 - butanol), and [ H 2 O ] (water). Hydrogen is quantified by partial pressure and has units of atmospheres. The variables V l and V g are the reactor liquid and gas volumes respectively. The ratio of V l to V g is multiplied by RT and used to convert [AD] and [BAD] to pressure. The solubility of H 2 [ 3 ] . Vapor liquid equilibria for the binary ethanol/H 2 at 498 K is plotted in Figure 5 - 7 . Rate constants k - 1 , k 2 , and k 4 were adjusted until the desired fit was achieved of the model to the experimental data. The following paragraphs describe the fit of the kinetic model to the experimental data at 150°C , 175°C and 200°C. Because the reac tion begins with 20 mol % acetaldehyde, 1,1 - diethoxyethane formation from acetaldehyde and ethanol was observed . Diethoxy butane formation from butyraldehyde and ethanol was also observed. Since these products were minor 106 Figure 5 - 7 Vapor liquid equilibrium is shown for ethanol and H 2 at 498 K [ 3 ] . Figure 5 - 8 Modeled ethanol concentration at 150°C , 175°C and 200°C. Initial charge of 80 mol% ethanol, 20 mol% acetaldehyde with 0.04 g cat /g reactant loading. 107 Table 5 - 3 Modeled ethanol concentrations with percent errors are shown at 150°C, 175°C, and 200°C. 150°C 175°C 200°C Modeled [EtOH] % Error Modeled [EtOH] % Error Modeled [EtOH] % Error 14.3 - 2.0 14.1 0.4 14.4 1.2 14.5 1.5 14.3 1.1 14.9 - 0.7 14.6 3.0 14.5 0. 9 15.0 1.5 14.7 1.9 14.7 0.2 15.1 0.2 14.8 3.7 14.7 - 0.6 15.1 - 1.2 15.0 3.6 14.8 - 0.6 15.1 - 1.4 15.3 1.7 14.9 - 1.1 15.3 1.3 14.9 - 1.5 15.4 0.3 The modeled rate equations fit the ethanol data well at 200 ° C, but not as well at 150 ° C ( Table 5 - 3 ) . Modeled data fits the 175 ° C run the best with absolute error less than 1% for most samples. Due to starting with an initial charge of hydrogen, acetaldehyde hydrogenation to ethanol was immediately obse rved in all runs. The ethanol concentration steadily increased from 14.2 M to a maximum of 15.3 M ( Figure 5 - 8 ). There was some scatter for ethanol data points at time points less than 100 minutes ( Figure 5 - 8 ) . The acetaldehyde concentration started at 3.5M for all runs, but much of it reacted during the heat - up stage for both runs. The modeled rate equations fit the acetaldehyde consumption profiles well ( Figure 5 - 9 ). Specifically, the modeled acetaldehyde concentration profiles fit better for the 150 ° C and 175 ° C runs ( Table 5 - 4 ). Therefore, t 0 was for when the reaction reached steady reaction tem perature. Reactions were modeled with the true acetaldehyde concentration at t 0 . Consumption of butyraldehyde was observed in all run s , but its formation was only observed in the 15 0 ° C run between 0 - 50 minutes and observed in the 175 ° C run between 0 - 10 minutes ( Figure 5 - 10 ). For the 150 ° C run, the rate equations modeled the formation of butyral dehyde well , and also modeled its consumption well from 50 minutes on to the end of the run ( Table 5 - 5 ) . 108 Figure 5 - 9 Modeled acetaldehyde concentration at 150°C , 175°C, and 200°C. Initial charge of 80 mol% ethanol, 20 mol% acetaldehyde with 0.04 g cat /g reactant loading . Table 5 - 4 Modeled acetaldehyde concentrations with percent errors are shown at 150°C, 175°C, and 200°C. 150°C 175°C 200°C Modeled [AD] % Error Modeled [AD] % Error Modeled [AD] % Error 2.1 - 7.0 1.2 - 12 0.91 - 22 1.7 - 15 0.7 - 30 0.27 - 63 1.4 - 19 0.5 - 27 0.17 - 49 1.2 - 22 0.2 - 15 0.04 - 61 1.0 - 14 0.6 - 11 0.2 - 17 109 Figure 5 - 10 Modeled butyraldehyde concentration at 150°C , 175°C and 200°C. Initial ch arge of 80 mol% ethanol, 20 mol% acetaldehyde with 0.04 g cat /g reactant loading. Table 5 - 5 Modeled butyraldehyde concentrations with percent errors are shown at 150°C, 175°C, and 200°C. 150°C 175°C 200°C M odeled [BAD] % Error Modeled [BAD] % Error Modeled [BAD] % Error 0. 05 249 0.1 1 - 8.2 0.28 - 2.3 0.1 5 158 0.15 - 1 8 0.21 - 12 0.2 0 35 0.1 4 - 2 9 0.16 16 0.2 4 - 8.0 0.08 - 23 0.06 19 0.2 5 - 16 0.2 3 - 8.7 Formation profiles for water and 1 - butanol ar e shown in Figure 5 - 11 . Water appears to form right away from 0 - 50 minutes, but its concentration remains constant at 0.9 M for the duration of the 150 ° C run , until it jumps to 1.2 M at the end . The water concentr ation was predicted the best for the 175 ° C run ( Table 5 - 7 ). The run was modeled at the starting concentration of 2.5 M and since there is no rate equation accounting for water consumption, the model poorly predic ted water concentration ( Table 5 - 7 ). At 200 ° C, the water profile reaches 1.2 110 M sooner, but levels off at 1.2 M. At 150 ° C, 1 - butanol concentration steadily increases to 0.4 M at end of run. The 1 - butanol conc en tra tion at 175 ° C and 200 ° C follows a similar profile to water , with the concentration leveling off at 0.5 M after 50 minutes and 150 minutes of run time respectively . These observations can be explained by the acetaldehyde concentration profile. At 150 ° C, a ° completely consumed by 50 minutes. This is consistent with no condensation reactions taking place after 300 minutes at 150 ° C and no condensation reactions taking place after 50 mi nutes at 200 ° C. Figure 5 - 11 Modeled 1 - butanol and water concentration at 150°C, 175°C, and 200°C. Initial charge of 80 mol% ethanol, 20 mol% acetaldehyde with 0.04 g cat /g reactant loading. Starting h ydrogen partial pressure was calculated for all runs. The hydrogen partial pressure was also modeled ( Figure 5 - 12 ) . The experimental hydrogen partial pressure was calculated based on the quantity of hydrogen react ed. The number of moles of h ydrogen reacted 111 were found by quantifying Guerbet hydrogenation activity . For instance, one mole of butyraldehyde is produced from one mole of H 2 hydrogenation (Reaction 5 - 3) . O ne mole of 1 - butanol is produced from 2 moles of H 2 hydrogenation (Reaction 5 - 3 and Reaction 5 - 4) . The initial hydrogen partial pressure was determined by multiplying th e starting reactor pressure by the ratio of reaction temperature to room temperature. Table 5 - 6 Modeled 1 - butanol concentrations with percent errors are shown at 150°C, 175°C, and 200°C. 150°C 175°C 200°C Modeled [BuOH] % Error Modeled [BuOH] % Error Modeled [BuOH] % Error 0.02 31 0.04 56 0.07 - 38 0.03 240 0.10 146 0.23 10 0.05 361 0.16 4 4 0.29 - 12 0.08 299 0.25 9.2 0.39 - 10 0.11 164 0.29 - 11 0.45 - 11 0.21 63 0.34 - 5.9 0.45 - 15 0.40 39 0.34 - 6.3 0.44 9.3 0.34 - 7.1 0.46 5.4 Table 5 - 7 Modeled water concentrations with percent e rrors are shown at 150°C, 175°C, and 200°C. 150°C 175°C 200°C Modeled [H 2 O] % Error Modeled [H 2 O] % Error Modeled [H 2 O] % Error 0.41 13 0 .53 4.4 0.97 4.5 0.52 33 0.65 - 2.5 1.07 21 0.60 5 0.70 - 25 1.08 8.3 0.67 0 0.73 - 2.5 1.08 6.9 0.71 - 2 0.74 - 22 1. 08 - 1.2 0.79 - 10 0.74 - 5.1 1.08 - 5.3 0.86 - 11 0.74 - 7.0 0.87 - 2 0.74 - 4.9 0.87 - 25 The modeled partial hydrogen pressure fits the experimental data well for the 175 ° C run and the 200 ° C run, particularly between 0 - 50 minutes run time ( Table 5 - 8 ). For the 150 ° C run, - 50 minutes, but fits the data well from 200 minutes to end of 112 run. The model may be overpredicting hydrogen consumption for hydrogenation reactions in the 150 ° C run. Figure 5 - 12 Modeled hydrogen partial pressure at 150°C , 175°C, and 200°C. Initial charge of 80 mol% ethanol, 20 mol% acetaldehyde with 0.04 g cat /g reactant loading. Table 5 - 8 Modeled partial pressures of hydrogen with percent errors are shown at 150°C, 175°C, and 200°C. 150°C 175°C 200°C Modeled PH 2 % Error Modeled PH 2 % Error Modeled PH 2 % Error 54.0 - 2.6 72.5 40 53.1 - 9.4 50.3 - 19 65.8 41 38.8 2.2 47.2 - 23 60.4 55 35.3 - 13 43.6 - 18 54.4 86 30.1 0.3 41.0 - 25 51.4 133 27.7 26 34.6 - 27 48.3 153 27.6 37 24.7 - 23 47.6 187 22.7 - 18 47.4 222 21.7 - 3.0 With the rate constants found at 150 ° C , 175 ° C and 200 ° C, approximate activ ation energies could be calculated with an Arrhenius plot ( Figure 5 - 13 ) . The activation energies and 113 frequency factors are shown in Table 5 - 9 . The activation energy of ac etaldehyde hydrogenation to ethanol is 6 9 KJ/mol. The activation energy of b utyraldehyde to 1 - butanol is 63 KJ/mol. T he activation energy of acetaldehyde condensation to crotonaldehyde is 6 6 KJ/mol. Crotonaldehyde reaction is so fast, the net rate of formation of crotonaldehyde is approximately zero. Figure 5 - 13 Arrhenius plot is shown for the ethanol Guerbet reaction mechanism. Table 5 - 9 Rate constants and activation energies are shown for ethanol/acetaldehyde runs at 150°C and 200°C. 150 ° C 175 ° C 200 ° C E a (KJ/mol) Freq. Factor (Unit) k - 1 (atm - 1 *min - 1 ) 2.0E - 04 4.5E - 04 1.6E - 03 68.6 5.8 E+04 atm - 1 *min - 1 k 2 (L*mol - 1 *m in - 1 ) 7.3E - 03 2.5E - 02 5.2E - 02 65.6 9.7 E+05 L*mol - 1 *min - 1 k 4 (atm - 1 *min - 1 ) 2.0E - 04 7.0E - 04 1.3E - 03 62.6 1.2 E+0 4 atm - 1 *min - 1 1 14 Table 5 - 10 Acetaldehyde conversion and carbon recoveries are shown for 20% ace taldehyde/80 % ethanol runs done at 150°C and 200°C. Ethanol conversion is not reported because ethanol was formed and not consumed. Catalyst loading was 0.04 g cat/g mixture. 150°C 175°C 200°C AD Conv. (%) C. Recov. (%) AD Conv. (%) C. Recov. (%) AD Conv. (%) C. Recov. (%) 37.1 99.7 62.4 91.7 67.4 96.0 43.5 97.0 70.1 91.7 79.3 99.2 50.3 96.1 81.5 91.9 90.7 95.9 58.3 97.5 92.0 91.8 96.9 96.9 67.9 95.5 98.1 92.0 99.7 97.5 80.3 95.1 99.7 92.8 99.7 97.7 92.8 96.7 98.6 95.9 98.6 97.3 5.3.4 Effect of H2 on higher alcohol formation r ate It was found hydrogen had little effect on ethanol conversion and the ethanol concentra tion profile ( Figure 5 - 14 ). From 100 to 300 minutes, ethanol conversion was 3 percentage points higher for the run without hydrogen. It was hypothesized starting with an initial hydrogen charge would slow down ethanol reaction rates due to ethanol dehydrogenation being an equilibrium limited reaction . A possible explanation for ethanol conv ersion rate being the same with and without H 2 is that while the presence of H 2 reversed ethanol dehydrogenation, it was offset by H 2 reacting with ethanol to produce CH 4 as described earlier: ( 5 - 12 ) Evidence for this reaction is found in the water concentration profile ( Figure 5 - 15 ). Though less water is being produced from condensation reactions in the run with H 2 , it is being produced b y reaction ( 5 - 12 ) . Therefore, the water concentration profiles for the two runs are virtually identical. Evidence of reaction ( 5 - 12 ) can also be seen in the carbon recov eries (based on liquid products) between the two runs ( Table 5 - 11 ). For the run with no H 2 , carbon recovery is 98 % 115 or greater at up to 20 % ethanol conversion. For the run with H 2 , carbon recovery is 94 - 97 % at up to 17 % ethan ol conversion. Since carbon recovery is only based on liquid products, this implies there was carbon lost to unaccounted gases. Figure 5 - 14 Ethanol conversion and ethanol concentration are plotted for run s with and without hydrogen. Reaction temperature was 230°C with 0.04 g cat/g ethanol loading. Figure 5 - 15 Water concentration is plotted vs time for the reactions with and without H 2 at 230°C and 0.04 g cat/ g ethanol loading. The 1 - butanol and C 6 alcohol concentration profiles were different between the two runs, most notably for C 6 alcohols ( Figure 5 - 16 ). At 200 minutes, the C 6 concentration was approximately a 116 factor of 4 mor e for the run without hydrogen, versus the one with hydrogen. The 1 - butanol concentration was 1.25 times more for the run without hydrogen than the run with hydrogen at 200 minutes. The difference in 1 - butanol and C 6 alcohol rates could also be seen in s electivity ( Figure 5 - 17 ). Without hydrogen the 1 - butanol selectivity is 60%+ and stays there during the Table 5 - 11 Ethanol conversion (%) and carbon recoveries (%) are shown for run s with and without H 2 . Carbon recoveries are based on liquid products only. No H 2 With H 2 EtOH Conv. (%) C. Recov. (%) EtOH Conv. (%) C. Recov. (%) 1.4 99.9 2.3 97.6 3.7 99.7 2.6 98.1 6.1 99.4 4.6 97.2 8.2 99.5 7.6 95.5 11.2 99.5 9.3 96.4 16.5 98.2 13.5 95.6 19.6 98.3 17.2 94.8 28.2 95.0 27.7 93.2 course of the entire run. With hydrogen present, 1 - end of the run. The C 6 - OH selectivity without hydrogen present was double the run with hydrogen at a m aximum of 16%. Figure 5 - 16 Concentration of 1 - butanol and C 6 alcohols are plotted for runs with and without hydrogen. Reaction temperature was 230°C with 0.04 g cat/g ethanol loading. 117 Figure 5 - 17 Selectivites of 1 - butanol and C 6 alcohols are plotted for runs with and without hydrogen. Reaction temperature was 230°C with 0.04 g cat/g ethanol loading. The ethanol conversion rate was nearly the same w ith and without H 2 due to the side reaction of H 2 with ethanol to CH 4 and water offsetting the negative effect of hydrogen on acetaldehyde formation rate . One would expect a higher CH 4 /CO 2 ratio for the run with H 2 due to reaction ( 5 - 12 ). This is supported by GC analysis of the gas composition at end of the run. The CH 4 /CO 2 ratio for the run with H 2 is 5.7 and for the run without H 2 is 4.5 ( Table 5 - 12 ). Since the butanol and C 6 alcohol rates were lower with H 2 present, this is evidence of the ethanol dehydrogenation equilibrium limiting the reaction rate. Equations ( 5 - 5 - 1 ) and ( 5 - 5 - 2 ) demonstrate ace taldehyde is formed until equilibrium is achieved. The formed acetaldehyde than condenses to form crotonaldehyde. If the acetaldehyde condensation rate was the rate - limiting step in the Table 5 - 12 Molar com position of gas phase at end of run is shown with CH 4 /CO 2 ratio for the runs with and without H 2 . With H 2 No H 2 Run Time (min) 617 592 EtOH Conv. (%) 27.7 28.2 CH4 (mol %) 78.8 67.0 CO2 (mol %) 13.7 14.9 CH4/CO2 5.7 4.5 118 ethanol Guerbet reaction, then excess H 2 - butanol and C 6 alcohol formation. Therefore, not only is ethanol dehydrogenation slower compared to acetaldehyde condensation, ethanol dehydrogenation is also equilibrium limited. The undesired side reactions forming et hyl acetate and acetal (1,1 - diethoxy ethane) were also compared for the runs with and without H 2 ( Figure 5 - 18 ) . It would be expected for ethyl acetate formation to be lower with H 2 , due to less acetaldehyde being available for the Tischenko reaction to occur. Ethyl acetate formation was found to be slightly lower with H 2 , but it was expected to be more noticeably lower. This may be connected with the formation of acetal: ( 5 - 13 ) Acetal formation for the H 2 run was approximately half that of the run without H 2 from beginning of run to 100 minutes. Both runs level o ff at 0.02 M. It is unclear why H 2 lowered acetal formation and not ethyl acetate formation. Ethyl acetate may form from direct dehydrogenation of two ethanol molecules over an 8 wt% Ni/9 Wt% La 2 O 3 - Al 2 O 3 catalyst. Figure 5 - 18 Concentration of ethyl acetate and acetal are plotted for runs with and without hydrogen. Reaction temperature was 230°C with 0.04 g cat/g ethanol loading. 5.3.5 Reaction of neat 1 - butanol Reactions with ethanol/acetaldehyde/H 2 helpe d explain the ethanol Guerbet mechanism, - 119 butanol, other reactions of butyraldehyde are known to occur, due to the formation of C 6 and C 8 alcohols. It is not cl ear if the 1 - butanol formed is ever dehydrogenated back to butyraldehyde. The data supports ethanol dehydrogenation a s an equilibrium react ion, which limits the reaction rate of acetaldehyde to crotonaldehyde, and the subsequent hydrogenation to butyraldeh yde and 1 - butanol . The equilibrium constant for 1 - butanol dehydrogenation in the gas phase is ~0.07 at 503 K. In the liquid phase the equilibrium constant is ~0.03 at 503 K. This implies that a run with pure 1 - butanol would produce butyraldehyde at a sma ll maximum concentration, much like acetaldehyde is produced in a neat ethanol run. D ata in Figure 5 - 19 from our neat 1 - butanol run supports this hypothesis. Butyraldehyde concentration was higher than 2 - ethyl - 1 - hexanol from 0 to 240 minutes, but reaches a maximum at 0.30 M. The Guerbet product from 1 - butanol is the branched alcohol, 2 - ethyl - 1 - hexanol. In Figure 5 - 19 1 - Butanol concentration is plotted 1 - butanol concentration v s. time for the pure 1 - butanol run at 230°C on the left. Butyraldehyde and 2 - ethyl - hexanol concentrations are plotted vs time on right. Table 5 - 13 Carbon recoveries and 1 - butanol conversion are shown for the neat 1 - butanol run at 230°C. Time (min) Conversion (%) Carbon Recovery (%) 15 2.9 98.4 60 5.4 97.1 120 8.2 95.2 180 9.4 94.9 240 11.9 93.2 546 18.1 88.5 1455 30.5 81.8 120 the run with pure 1 - butanol at 230 ° C, 2 - ethyl - 1 - hexanol was the major product ( Figure 5 - 19 ). At 500 minutes, the 2 - ethyl - 1 - hexanol concentration increased to 0.65 M. Similar to the neat ethanol runs, there was decomposition of 1 - butanol to gases. The main product gas was propane, with small amounts of CH 4 and CO 2 . This can be seen in carbon recoveries based on carbon in condensable liquid product ( Table 5 - 13 ). The carbon recovery decreased from 98% at 15 minutes to 82% at 1455 minutes. Much like neat ethanol runs, there wa s decomposition of 1 - butanol to gases. Interestingly, propane was the major gas product at 56% ( Table 5 - 14 ). There was also 23% CO 2 and 13% CH 4 . The mole ratio of propane:CH 4 :CO 2 was 4.2:1:1.7, which indicates pr opane may be decomposing under the same mechanism as ethanol decomposition. The reaction for 1 - butanol decomposition is: ( 5 - 14 ) The mole ratios of pr opane to CH 4 to CO 2 - 14, which would have mole ratios of 2:1:1 for propane:CH 4 :CO 2 . This means additional propane either came from hydrogenation of 1 - butanol or it was a different mechanism. Table 5 - 14 End of run gas analysis is shown for neat 1 - butanol run at 230°C. CH 4 (%) CO 2 (%) Propane (%) 13.4 22.6 56.3 5.3.6 Reaction of ethanol/butyraldehyde A reaction was performed with 51 mol% ethanol/49 mol% butyraldehyde at 230 ° C a nd 0.04 g cat/g ethanol loading to examine the hydrogen scavenging ability of butyraldehyde. Ethanol undergoes dehydrogenation, which leads to butyraldehyde hydrogenation to 1 - butanol. 121 Butyraldehyde readily und ergoes condensation reactions to C 6 and C 8 al dehydes ( Figure 5 - 20 ). Butyraldehdye is almost completely consumed by 500 minutes. Unsaturated aldehydes , hexenal and 2 - ethyl - 2 - hexenal, form early during the run and decrease at approximately 100 minutes d ue to hydrogenation reac tions to the saturated aldehydes, hexanal and 2 - ethyl - 2 - hexanal. Figure 5 - 20 Ethanol and butyraldehyde concentration vs time is shown in the top left. 1 - Butanol and 2 - ethyl - hexanol concentration vs time is shown top right. Concentrations of higher aldehydes hexenal (HEN), hexanal (HAN), 2 - ethyl - hexenal (2 - E - HEN), and 2 - ethyl - hexanal (2 - E - HAN) are shown in bottom left . Initial rates for higher aldehydes and 1 - butanol are shown in bottom right. The concen tration profiles of these aldehydes and their initial reaction rates are shown in Figure 5 - 20 . The initial reaction rate of butyraldehdye was found to be - 0.05 0 mol/L - min. The sum of 1 - butanol, C 6 and C 8 aldehyde initial format ion rates was found to be 0.034 mol/L - min. This indicates 1 - butyralde is reacting to other products. There were many unidentified products in the liquid gas chromatogram ( Figure 5 - 21 ). Gaseous products were also formed, but it was unclear whether they were formed from ethanol or butyraldehyde. Loss of carbon to these unidentified products and gases is quantified in Table 5 - 15 . 122 An interesting comparison is made between the concentration profiles of 1 - b utanol and 2 - ethyl - 1 - hexanol in the top right of Figure 5 - 20 . The 1 - butanol concentration reaches a maximum of 1.2 M by 400 minutes. The 1 - butanol concentration remains at 1.2 M from 400 minutes to the end of the run because it is either in thermodynamic equilibrium or its net rate of formation is zero . The net rate of formation of 1 - butanol is zero after 400 minutes, since 2 - ethyl - 1 - hexanol may be forming from butyraldehyde from 1 - butanol dehydrogenation . that the 2 - ethyl - 1 - hexanol concentration is much lower than 1 - butanol up to 400 minutes , but steadily increases from 0.2 M at 400 minutes to 1.3 M at the end of the run . Figure 5 - 21 Gas chromatogra m for ethanol/butyraldehyde reaction at end of run. Unmarked peaks are unidentified peaks. Ethanol 1 - Butanol Butyraldehyde 2 - Eth - 1 - HeOH Int. Std. 2 - Eth - 1 - BuOH 1 - Hexanol 123 Table 5 - 15 Butyraldehyde conversion and ethanol conversions are shown. Carbon recoveries are also shown. Time (min) BAD Conv. (%) EtOH Conv. (%) Carbon Recov. (%) 0 4.6 3.5 107.1 20 22.8 5.2 105.2 40 37.9 8.9 101.9 60 50.1 14.2 97.4 120 64.9 15.8 92.9 180 78.5 23.4 85.1 240 86.7 33.3 78.0 300 90.5 38.9 74.5 627 94.3 49.0 69.8 1401 98.5 73.1 69.6 5.3.7 Re action of ethanol/isobutyraldehyde It was found in the previous sections ethanol dehydrogenation is in equilibrium. Since ethanol is in equilibrium with acetaldehyde and hydrogen, the concentration of acetaldehyde is small. This translates to a lower ac etaldehyde condensation rate. By starting with a mixture of ethanol and isobutyraldehyde, the mixture is deficient in hydrogen. This deficiency in hydrogen was hypothesized to increase higher alcohol formation rates, and therefore ethanol reaction rate. When comparing the slope of the ethanol concentration curve for the neat ethanol run versus the EtOH/isobutyraldehye run, the slope is steeper at 10 M [EtOH] for the EtOH/isobutyraldehyde run ( Figure 5 - 22 ) . Theor etically, if isobutyraldehyde was an inert, the reaction rate would be the same for both runs at 10 M [EtOH]. This implies the H 2 deficiency is driving the ethanol dehydrogenation equilibrium to acetaldehyde and H 2 . Differences were observed in the aceta ldehyde concentration profiles ( Figure 5 - 22 ). Acetaldehyde reaches a maximum of 0.10 M for the neat ethanol run, while it reached as high as 0.27 M for the mixed aldehyde run. Since equilibrium is pushed forward from isobutyraldehyde uptaking H 2 , acetaldehyde is able to reach a higher equilibrium value. This was also observed 124 with butyraldehyde ( Figure 5 - 23 ). For the neat ethanol run, butyraldehyde reached a maximum of 0 .04 M. Butyraldehyde increased to 0.14 M with isobutyraldehyde present, then decreased to 0.02 M at end of run. - butanol rate for the mixed aldehyde run ( Figure 5 - 23 ). This c ould be due to the competition for H 2 from isobutyraldehyde. The isobutyraldehdye concentration started out much larger than butyraldehyde, therefore this may be to blame for the poor 1 - butanol formation rate. Figure 5 - 22 Ethanol and acetaldehyde concentration profiles are shown for a neat ethanol run and for the mixed 70 mol% EtOH/30 mol % isobutyraldehyde run at 230°C and 0.04 g cat/g mixture loading. Figure 5 - 23 Concentration profiles are shown for the neat ethanol run and the mixed 70 mol % EtOH/30 mol % isobutyraldehyde run at 230°C and 0.04 g cat/g mixture loading. Concentration profiles for isobutyraldehyde and isobutanol are shown in Figure 5 - 24 . As expected, isobutanol was formed from the dehydrogenation of ethanol and subsequent 125 hydrogenation of isobutyraldehdye to isobutanol. Isobutyraldehdye was not expected to undergo condensation reactions, but a sign ificant amount of it was converted to un identified pro ducts ( Figure 5 - 25 ). Approxmately 1.5 M of isobutyraldehdye is accounted for in isobutanol . The remaining 3 M is presumed to be in the unidentified peaks at t he end of the sample chromatogram. Figure 5 - 24 Concentration profiles are shown for isobutyraldehyde and isobutanol for the mixed 70 mol % EtOH/30 mol % isobutyraldehyde run at 230°C and 0.04 g cat/g mixtu re loading. 126 Figure 5 - 25 Chromatogram is show n for end of run of 70 mol% EtOH/30 mol% isobutyraldehdye at 230°C and 0.04 g cat/g mixture loading. 5.4 Conclusions Reactions of ethanol with acetaldehyde a nd H 2 were performed to probe the ethanol Guerbet mechanism. It was hypothesized the first step of this mechanism, ethanol dehydrogenation, is in equilibrium. Reactions performed at 150 ° C , 175 ° C and 200 ° C were modeled to determine rate constants and activ ation energies. Runs at 230 ° C with hydrogen and without hydrogen demonstrated the ethanol conversion rate was virtually the same. Higher alcohol production was less for the run with H 2 , confirming the presence of excess H 2 drove ethanol dehydrogenation backwards. Ethanol conversion rates were the same due to the side reaction of H 2 with ethanol to CH 4 and water, which offsets the negative effect of hydrogen on acetaldehyde formation rate. The CH 4 /CO 2 ratio was expected to be higher for the run with H 2 , which was confirmed by gas chromatographic analysis of the gas product. Isobutanol Internal std 1 - Buanol EtOH Water Unidentified iBAD condensation products 127 APPENDIX 128 A PPENDIX C.1: Master reaction List Table C 5 - 1 Part 1 of the master reaction list is shown. Runs with (*) are preliminary continuous runs with ethanol flow rate in ml/min under starting reactor mass column. Run ID Starting Material Start. R. Mass (g) Catalyst 50TLJ122111 Ethanol 100.66 - Al2O3 52TLJ010312 Ethanol 104.04 - Al2O 3 62TLJ01 1212 Ethanol 78.76 - Al2O3 66TLJ012012 Ethanol 76.05 8Ni - - Al2O3 68TLJ012412 Ethanol 80.7 8Ni - 2Cu/3La2O3 - Al2O3 71TLJ020112 Ethanol 91.65 - Al2O3 73TLJ020812 Ethanol 86.76 8Ni/14La 2O3 - - Al2O3 75TLJ021012 Ethanol 75.57 - Al2O3 77TLJ021912 Ethanol 85.9 8Ni/9La2O3 - - Al2O3 Spheres 79TLJ022812 Ethanol 84.67 8Ni/9La2O3 - - Al2O 3 Spheres 80TLJ030512 Ethanol 86.76 8Ni/9La2O3 - - Al2O 3 Spheres 81TLJ030612 Ethanol 72.79 8Ni/10Ce2O3 - - Al2O3 Spheres 82TLJ030812 Ethanol 128.77 8Ni/14La - - Al2O3 Spheres 83TLJ031512 Ethanol 103.86 8Ni/11La2O3 - 86TLJ032212 Ethanol 100.69 8Ni/11La2O3 - 3 91TLJ041012 Ethanol 93.59 8Ni/11La2O3 - 3 92TLJ041212 Ethanol 109.6 8Ni/11La2O3 - 3 112TLJ072412 Ethanol 0.4 ml/min* 8Ni - 2Cu/ 3 114TLJ 072712 Ethanol 0.4 ml/min* 8Ni/ 3 116TLJ080112 Ethanol 0.4 ml/min* 8Ni/11La2O3 - 120TLJ081412 Ethanol 88.6 8Ni/9La2O3 - - Al2O3 122TLJ081612 Ethanol 0.8 ml/min* 8Ni/9La2O3 - - Al2O3 124TLJ082112 Ethanol 0.4 ml/min* 8Ni/9La2O3 - - Al2O3 136TLJ09251 2 Ethanol 110.48 8Ni/9La2O3 - - Al2O 3 141TLJ101012 Ethanol 109.83 8Ni/9La2O3 - - Al2O 3 142TLJ101612 Ethanol 110.22 8Ni/9La2O3 - - Al2O 3 143TLJ102212 Ethanol 110.04 8Ni/9La2O3 - - Al2O 3 146TLJ110712 50 wt%EtOH/50 wt % Water 109.56 8Ni/9La2O3 - - Al2O 3 147TLJ1112 12 50 wt%EtOH/50 wt % BuOH 108.26 8Ni/9La2O3 - - Al2O 3 149TLJ112712 1 - Butanol 113.3 8Ni/9La2O3 - - Al2O 3 2 - 04TLJ021413 Ethanol 151.12 8Ni/9La2O3 - - Al2O 3 2 - 05TLJ021913 Ethanol 160.34 8Ni/9La2O3 - - Al2O 3 2 - 07TLJ022613 Ethanol 160.42 8Ni/9La2O3 - - Al2O 3 2 - 09TL J032813 95 wt % EtOH/ 5wt % Water 111.13 8Ni/9La2O3 - - Al2O 3 2 - 10TLJ041613 93 wt % EtOH/ 7wt % Water 112.99 8Ni/9La2O3 - - Al2O 3 2 - 11TLJ051413 90 wt % EtOH/ 10 wt % Water 111.34 8Ni/9La2O3 - - Al2O 3 129 Table C5 - 02 - 14TLJ052313 1.6 mol EA 98.4 mol E tOH 112.52 8Ni/9La2O3 - - Al2O3 02 - 17TLJ071013 Ethanol 112.19 - Al2O4 02 - 19TLJ071713 Ethanol 112.49 8Ni/9La2O3 - - Al2O3 02 - 21TLJ073113 Ethanol 111.53 8Ni/9La2O3 - - Al2O 3 02 - 22TLJ080613 Ethanol 112.7 8Ni/4La2O3 - - Al2O 3 02 - 24TLJ080813 Ethanol 112.59 8Ni /9La2O3 - - Al2O 3 02 - 26TLJ081513 Ethanol 112.6 8Ni/9La2O3 - - Al2O 3 02 - 27TLJ091013 97 EtOH 3 mol iBAD 113.73 8Ni/9La2O3 - - Al2O 3 02 - 28TLJ091613 70 mol EtOH/30 mol IBAD 110.37 8Ni/9La2O3 - - Al2O 3 02 - 29TLJ092013 80 mol EtOH/ 20 mol AD 111.65 8Ni/9La2O3 - - Al2 O 3 02 - 30TLJ092413 51 mol EtOH 49 mol BAD 110.11 8Ni/9La2O3 - - Al2O 3 02 - 32TLJ100713 Ethanol 110.9 8Ni/9La2O3 - - Al2O 3 02 - 33TLJ100813 Ethanol 111.08 8Ni/9La2O3 - - Al2O 3 02 - 35TLJ101613 Ethanol 112.85 8Ni/9La2O3 - - Al2O 3 02 - 36TLJ101713 Ethanol 112.12 8Ni/9La2 O3 - - Al2O 3 02 - 39TLJ110513 80 mol 2 - EHAN 20 mol AD 113.59 8Ni/9La2O3 - - Al2O 3 02 - 40TLJ110813 80 mol EA 20 mol AD 110.27 8Ni/9La2O3 - - Al2O 3 02 - 67TLJ010715 1 - Butanol 109.28 8Ni/9La2O3 - - Al2O 3 02 - 72TLJ021115 Ethanol 117.82 4Ni - 4Cu/11La2O3 - - Al2O 3 02 - 73TLJ 021815 Ethanol 108.68 4Ni/14La2O3 - - Al2O 3 02 - 77TLJ051815 61 mol EtOH 39 mol H2O 109.24 8Ni/9La2O3 - - Al2O 3 02 - 83TLJ072115 80 mol EtOH/ 20 mol AD 98.22 8Ni/9La2O3 - - Al2O 3 02 - 85TLJ072315 80 mol EtOH/ 20 mol AD 98.31 8Ni/9La2O3 - - Al2O 3 02 - 86TLJ072815 90 mo l EtOH 10 mol EA 99.62 8Ni/9La2O3 - - Al2O 3 02 - 87TLJ072915 60 mol ETOH 20/20 AD/BAD 99.91 8Ni/9La2O3 - - Al2O 3 02 - 89TLJ081015 Ethanol 113.3 8Ni/9La2O3 - - Al2O 3 2 - 90TLJ081215 Ethanol 111.8 8Ni/9La2O3 - - Al2O 3 2 - 99TLJ121615 80 mol EtOH/ 20 mol AD 98.83 8Ni/9La 2O3 - - Al2O 3 Table C 5 - 2 P art 2 of the master reaction list is shown. Runs with (*) are preliminary continuous runs with catalyst bed weight under g cat/g react. column. Run ID Catalyst ID Meas. Disp (%) g cat/g react. T. (°C) Run t. (min) Conv. (%) 50TLJ122111 0.093 230 600 46 52TLJ010312 0.125 210 1200 37 62TLJ011212 0.093 230 600 52 66TLJ012012 0.093 230 600 35 68TLJ012412 0.093 230 600 34 71TLJ020112 0.093 230 600 50 73TLJ 020812 0.093 230 600 49 75TLJ021012 0.093 230 600 16 77TLJ021912 0.093 230 600 55 79TLJ022812 0.093 230 1320 72 80TLJ030512 0.093 200 600 26 130 Table C5 - 81TLJ030612 0.093 230 600 50 82TLJ030812 0.093 230 240 35 8 3TLJ031512 0.04 200 120 7 86TLJ032212 0.033 210 140 8 91TLJ041012 0.033 230 90 13 92TLJ041212 0.033 220 90 11 112TLJ072412 15.03 g * 210 165 20 114TLJ072712 15.03 g * 210 225 23 116TLJ080112 12.58 g * 210 240 20 120TLJ081412 0 .033 210 120 9 122TLJ081612 12.52 g* 210 225 17 124TLJ082112 12.50 g* 230 240 33 136TLJ092512 0.060 230 1608 67 141TLJ101012 0.060 215 1595 51 142TLJ101612 0.060 239 491 50 143TLJ102212 118TLJ080609 0.066 230 1409 67 146TLJ110712 1 18TLJ080610 0.060 230 558 11 147TLJ111212 118TLJ080611 0.060 230 1750 85 149TLJ112712 118TLJ080612 0.060 230 1340 45 2 - 04TLJ021413 155TLJ012303 0.040 230 1346 48 2 - 05TLJ021913 155TLJ012304 0.040 230 1574 52 2 - 07TLJ022613 155TLJ012305 0.040 230 1629 48 2 - 09TLJ032813 155TLJ012306 0.040 230 1297 48 2 - 10TLJ041613 155TLJ012307 0.040 230 2 - 11TLJ051413 155TLJ012308 0.040 230 1410 41 02 - 14TLJ052313 155TLJ012313 2.09 0.04 230 1348 02 - 17TLJ071013 02 - 16TLJ070813 9.68 0.04 230 1344 38 02 - 19TLJ071713 02 - 16TLJ070813 0.04 230 1301 43 02 - 21TLJ073113 02 - 16TLJ070813 (450C) 8.04 0.04 230 1340 52 02 - 22TLJ080613 02 - 20TLJ073013 6.44 0.04 230 1376 43 02 - 24TLJ080813 02 - 16TLJ070813 (450C) 8.04 0.04 230 1373 29 02 - 26TLJ081513 02 - 16TLJ070813 ( 450C) 8.04 0.13 200 1351 38 02 - 27TLJ091013 02 - 13TLJ052113 2.5 0.04 230 327 36 02 - 28TLJ091613 02 - 13TLJ052113 2.5 0.04 230 1328 53 02 - 29TLJ092013 02 - 13TLJ052113 2.5 0.04 230 495 21 02 - 30TLJ092413 02 - 13TLJ052113 2.5 0.04 230 1401 73 02 - 32TLJ100713 02 - 13T LJ052113 2.5 0.02 230 360 16 02 - 33TLJ100813 02 - 13TLJ052113 2.5 0.04 230 360 26 02 - 35TLJ101613 02 - 13TLJ052113 2.5 0.04 230 300 21 02 - 36TLJ101713 02 - 13TLJ052113 2.5 0.08 230 1416 68 02 - 39TLJ110513 02 - 13TLJ052113 2.5 0.04 230 189 98 02 - 40TLJ110813 02 - 37 TLJ102913 0.04 230 260 89 02 - 67TLJ010715 02 - 37TLJ102913 0.04 230 1455 02 - 72TLJ021115 02 - 69TLJ012315 0.04 230 1275 21 02 - 73TLJ021815 02 - 69TLJ012315 - noCu 0.04 230 1305 41 02 - 77TLJ051815 02 - 37TLJ102913 0.04 230 530 02 - 83TLJ072115 02 - 37TLJ10 2913 0.04 200 246 100 02 - 85TLJ072315 02 - 37TLJ102913 0.04 150 267 99 02 - 86TLJ072815 02 - 37TLJ102913 0.04 200 277 131 Table C5 - 02 - 87TLJ072915 02 - 37TLJ102913 0.04 120 566 02 - 89TLJ081015 02 - 37TLJ102913 0.04 230 592 28 2 - 90TLJ081215 02 - 37TLJ102913 0.04 230 617 28 2 - 99TLJ121615 02 - 37TLJ102914 0.04 175 240 Table C 5 - 3 Part 3 of the master reaction list . Run ID Hi. 1 - BuOH Sel (%) End. 1 - BuOH Sel (%) Gas Wt. Reaction Notes 50TLJ12211 1 48 22.56 52TLJ010312 55 19.44 62TLJ011212 38 18.7 66TLJ012012 58 5 68TLJ012412 59 4.7 71TLJ020112 47 11.5 73TLJ020812 45 11 75TLJ021012 55 0.7 77TLJ021912 42 8.9 79TLJ022812 32 25.4 80TLJ030512 54 2.37 81TLJ030612 47 6.9 82TLJ030812 54 4.9 83TLJ031512 57 57 0.6 86TLJ032212 74 50 91TLJ041012 45 45 0.5 92TLJ041212 35 27 0.6 112TLJ072412 57 57 Trickle bed run (continuous) 114TLJ072712 42 42 Trickle bed run (continuous) 116TLJ0 80112 47 47 Trickle bed run (continuous) 120TLJ081412 45.8 45.8 122TLJ081612 46 46 Trickle bed run (continuous) 124TLJ082112 30 30 Trickle bed run (continuous) 136TLJ092512 62 40.5 17.5 141TLJ101012 56 50.8 6.5 142TLJ101612 56 49 9.2 14 3TLJ102212 57.7 42.5 18.5 Reuse test for catlayst used in 142TLJ101612 146TLJ110712 62 17 8.4 147TLJ111212 21.2 Conversion for EtOH 149TLJ112712 54 54 14.1 2 - Ethyl - 1 - Hexanol Selectivty shown 2 - 04TLJ021413 50 43 7.9 2 - 05TLJ021913 54 43 With d rying loop (molecular sieves) 2 - 07TLJ022613 50 44 With drying loop (glass beads) 132 Table C5 - 2 - 09TLJ032813 44 39 8.8 2 - 10TLJ041613 2 - 11TLJ051413 42 37 10.8 02 - 14TLJ052313 6.7 02 - 17TLJ071013 60 57 7.7 02 - 19TLJ071713 6 4 58 12.9 02 - 21TLJ073113 63 50 18.9 02 - 22TLJ080613 61 55 6.0 02 - 24TLJ080813 49 43 11.6 4.59g Na2CO3 added 02 - 26TLJ081513 50 41 6.8 Higher catalyst loading run 02 - 27TLJ091013 1.2 02 - 28TLJ091613 8.1 02 - 29TLJ092013 1.8 02 - 30TLJ0 92413 16.1 02 - 32TLJ100713 55 55 0.7 02 - 33TLJ100813 56 51 1.3 02 - 35TLJ101613 53 53 1.3 Catalyst ground into fine particls 02 - 36TLJ101713 61 41 17.8 02 - 39TLJ110513 0.5 02 - 40TLJ110813 0.3 02 - 67TLJ010715 15.9 02 - 72TLJ0211 15 72 72 2.8 02 - 73TLJ021815 56 55 8.0 02 - 77TLJ051815 7.7 02 - 83TLJ072115 0.9 started with 750 PSIG H2 02 - 85TLJ072315 0.4 started with 750 PSIG H2 02 - 86TLJ072815 1.7 started with 750 PSIG H2 02 - 87TLJ072915 0.4 started with 100 0 PSIG H2 02 - 89TLJ081015 67 58 5.0 Control run with pre - reducing catalyst at 250C, purged with nitrogen 2 - 90TLJ081215 57 57 6.9 Control run, started with 200 psi H2 2 - 99TLJ121615 started with 750 PSIG H2 133 R EFERENCES 134 R EFERENCES [1] C.L. Yaws, Knovel, 2012. [2] T. Tsuchida, S. Sakuma, T. Takeguchi, W. Ueda, Industrial & Engineering Chemistry Research, 45 (2006) 8634 - 8642. [3] E. Brunner, W. Hultenschmidt, Journal of Chemical Thermodynamics , 22 (1990) 73 - 84. 135 6 Summary and Recommendations for Future Work 6.1 Catalyst Screening Initial catalyst screening with - alumina supported nickel catalysts demonstrated they were active catalysts for the liquid - phase conversion of ethanol to 1 - butanol and h igher alcohols at 230 °C and autogeneous pressure in a 300 ml Parr reactor . Specifically, the optimum catalyst configuration was found to be 8 wt% Ni - 9 wt% La - Al. Total higher alcohol selectivity reached over 80% at 5 0% ethanol conversion . Adding La 2 O 3 t o the - alumina increased basicity, which decreased the Tischenko reaction of acetaldehyde to ethyl acetate and also decreased the etherification of ethanol to diethyl ether. This was confirmed with CO 2 TPD, which demonstrated a 2 - fold increase of base si tes when adding La 2 O 3 . Nickel is active for carbon - carbon bond cleavage though and the decomposition of ethanol to CH 4 and CO 2 was overserved. 6.2 VLE Modeling In the catalyst screening stud ies, CH 4 and CO 2 was usually observed at the end of each run. Gas s quantify CH 4 and CO 2 in both the liquid and vapor phases. Due to the system being near the critical point of ethanol (241 ° C, 6.3 MPa) and having polar components, it was decided to use an equation of state to model the reactor. The SR - Polar equation of state was chosen because it o ffers accurate prediction of non - ideal mixtures and volume translation for better prediction of density. Accurate prediction of phase density is needed d ue to how expanded the liquid phase is. Applying the SR - Polar EOS to batch Guerbet reactions provides a more rigorous analysis of reaction than conventional liquid phase sampling. The SR - Polar EOS accurately predicts higher alcohol vapor - liquid equilibri a, liquid and vapor phase densities, and total quantity of gases produced in reaction. Although the ethanol conversion profile calculated using the SR - 136 Polar EOS is virtually the same as with liquid - phase - only sampling, the product yields and selectivities are more accurately represented because species partitioning between liquid and vapor phases are more accurately modeled. Finally, the ability of the SR - Polar EOS to predict liquid expansion is advantageous in safely designing future ethanol Guerbet react io ns at near critical conditions. 6.3 Impact of Water It was hypothesized removing water would minimize ethanol decomposition to CH 4 and CO 2 . The data supports this, with gas selectivity at 25% for the run starting with 10 wt% initial water concentration. For the run with water removal, gas selectivity stays at 5% at water concentrations up to 4 wt % and 40 % ethanol conversion. It is unclear why minimizing water content decreases ethanol decomposition, because water is not a reactant. Ther e fore, increasing water concentration must be affecting the catalyst surface. Water may be saturating condensation sites on the alumina, thereby promoting nickel metal sites for decomposition. It is also possible water is irreversibly converting - Al 2 O 3 to boehmite , which lacks the catalytic abilities of - Al 2 O 3 . 6.4 Investigation of the Ethanol Guerbet Reaction Mechanism Preliminary initial rate kinetics determined the reaction of ethanol to 1 - butanol had an activation energy of 52 kJ/mol over a 8Ni/9La - Al catalyst. The act ivation energy for 1 - hexanol formation was found to be 63 kJ/mol. The effect of catalyst loading was investigated and it was found the 1 - butanol and 1 - loading. Reactions of ethanol with ac etaldehyde and H 2 were performed to probe the ethanol Guerbet mechanism. Rate constants were found for acetaldehyde hydrogenation, acetaldehyde condensation, and butyraldehyde hydrogenation. 137 Runs at 230 ° C with hydrogen and without hydrogen demonstrated t he ethanol conversion rate was the same. Carbon recoveries were lower for the run with H 2 , indicating carbon was lost to unaccounted non - condensable products. The primary component of these products was reasoned to be CH 4 from the reaction of H 2 with eth anol, which also produces water. The 1 - butanol and C 6 alcohol formation rates were found to be lower for the run with H 2 then the run without H 2 . 6.5 Recommendations for Future Work 6.5.1 Condensed - phase c ontinuous reactions The batch reactor has exclusively been used in this work for catalyst optimization, VLE modeling, reactor modification, and mechanism probing. Future work should look at performing condensed - phase reactions in a continuous reactor. Higher throughputs can be attained with continuous reactors. Condensed - phase continuous reactors, such as trickle bed reactors, offer the advantage of longer contact times. Preliminary testing was performed with a trickled bed reactor and the run conditions and results summary are shown in Table 6 - 1 and Tabl e 6 - 2 respectively. Table 6 - 1 Reaction conditions are shown for preliminary trickle bed reactions . Run ID Catalyst Temp. (°C) Ca t. Wt. (g) EtOH (ml/min) 112TLJ072412 8Ni - 210 15.03 0.4 114TLJ072712 210 15.03 0.4 116TLJ080112 8Ni/11La2O3 - 210 12.58 0.4 122TLJ081612 8Ni/9La2O3 - 3 210 12.52 0.8 124TLJ082112 8Ni/9La2O3 - 3 230 12.5 0.4 138 Tabl e 6 - 2 Results summary for preliminary trickle bed reactions. Run ID Time (min) EtOH Conv. (%) 1 - BuOH Sel. (%) 112TLJ072412 165 20 57 114TLJ072712 225 23 42 116TLJ080112 240 20 47 122TLJ081612 225 17 46 124TLJ082112 240 33 30 It was found ethanol decomposition to CH 4 and CO 2 was hard to quantify due to the liquid sampling configuration. Ethanol conversion was also low compared to batch runs at 33%. Selectivty to 1 - butanol was also low with a maximum o f 57% for the run with nickel and copper. The data supports copper as a potential metal supplement to nickel to improve higher alcohol selectivity. Copper is not as rugged as nickel and was found to wear off the - alumina easily in batch reactions. It is recommended for both future continuous and batch runs to examine copper as a catalyst modifier. Since - alumina - supported nickel catalysts are able to decompose ethanol to CH 4 and CO 2 , the continuous reactor will need to have a sampling system to allow s ampling both the liquid and vapor phases. Reactions are at high pressures (>5 MPa) so a backpressure regulator will be needed so that gas production can be quantified. When initial continuous testing runs were done with a trickle bed reactor, sample cyl i nders were pressurized with N 2 to match their pressure with reactor pressure. This caused gas samples to be diluted and CH 4 /CO 2 composition Instead of sampling non - condensable product gases this way, redu cing the reactor pressure to atmospheric with a back pressure regulator will enable appropriate gas phase sampling. Gas production was correlated with increasing water concentration in the water studies section. The non - condensable gas composition at end 139 a reactant to reform ethanol to CH 4 and CO 2 . This means wate r must be affecting active site s on the 8 wt% Ni/9 wt% La 2 O 3 catlyst. It is unclear if the water is converting - alumina to boehmite or converting s trong Lewis sites to weaker Br ønsted sites. During a continuous run this will need to be investigated by examining different sections of the catalyst bed. Ethanol conversion will be higher at the downstream end of the bed, therefore the catalyst would be different at the downstream end from the upstream end. For a given run it is recommended to sample catalyst from the top and sample catalyst from the bottom of the reactor . The crystal structure can be checked with x - ray diffraction and acid/base sites can be checked with NH 3 /CO 2 chemisorption. Other molecules could be used for chemisorption to distinguish Lewis base sites from Br ønsted sites. 6.5.2 Minimizing ethanol decomposition to CH 4 and CO 2 The catalyst configuration optimized in this work was 8 wt% Ni supported by - alumina, which was modified with 9 wt % La 2 O 3 . Modifying - alumina with La 2 O 3 added basic sites, and therefore helped decreased undesired, acid - catalyzed side reactions. Nickel metal is highly active for cleaving carbon - carbon bonds though, which was observed in reactions with ethanol. Nickel easily decomposes ethanol to CH 4 and CO 2 . Adding basic sites to the - alumina did not have an impact on this decomposition reaction. Future work should look at ways to modify the a ctivity of n ickel, without hurting its activity towards converting ethanol to 1 - butanol and higher alcohols. Other metals, such as copper, could be added with nickel to reduce activity for cleaving carbon - carbon bonds. Besides modifying the catalyst to minimize eth anol decomposition, modifying process conditions could help minimize ethanol decomposition. It is has already been shown decreasing 140 water content decreases selectivity to gases. Another poten tial route to decreasing selectivity to gases is to utilize carb on monoxide and the water - gas shift reaction (WGSR) : ( 6 - 1 ) The operating temperature (230 ° C) is low enough to favor CO 2 and H 2 formation. If the WGSR produces H 2 , it is likely the excess H 2 will decrease acetaldehyde concentration due to an equilibrium shift to ethanol. As was demonstrated in the ethanol/acetaldehyde/H 2 runs, the excess H 2 decreased selectivity to 1 - butanol and 1 - hexanol. This would be from t he potential reaction of the excess H 2 reacting with ethanol to produce CH 4 and water. If that were to happen, then adding CO to the reaction would not be beneficial. It is worth trying CO as a reducing reagent in a future reaction.