Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Development of a continuous reactor for the electro-chemical reduction of carbon dioxide Li, Hui 2006

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

Item Metadata

Download

Media
831-ubc_2006-200155.pdf [ 17.14MB ]
Metadata
JSON: 831-1.0058927.json
JSON-LD: 831-1.0058927-ld.json
RDF/XML (Pretty): 831-1.0058927-rdf.xml
RDF/JSON: 831-1.0058927-rdf.json
Turtle: 831-1.0058927-turtle.txt
N-Triples: 831-1.0058927-rdf-ntriples.txt
Original Record: 831-1.0058927-source.json
Full Text
831-1.0058927-fulltext.txt
Citation
831-1.0058927.ris

Full Text

D E V E L O P M E N T OF A CONTINUOUS R E A C T O R FOR THE E L E C T R O - C H E M I C A L R E D U C T I O N OF C A R B O N DIOXIDE By HUI LI M . A . Sc., Tsinghua University, Beijing, China, 1990 B. A . Sc., Tsinghua University, Beijing, China, 1987 A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE F A C U L T Y OF G R A D U A T E STUDIES (Chemical and Biological Engineering) THE UNIVERSITY OF BRITISH C O L U M B I A May 2006 © Hui L i , 2006 Abstract A laboratory bench-scale continuous reactor has been developed for the electro-reduction of CO2 to formate salts in aqueous solution. This novel reactor operated in the "trickle-bed" mode with co-current flow of reactant gas and catholyte liquid through a flow-by 3-D cathode, to overcome the twin problems of reactant supply and mass transfer limitation encountered in previous work on the electro-reduction of CO2. Two single-cell electro-chemical reactors with cation membrane separators, Reactor A (150 mm long by 30 mm wide) and Reactor B (680 mm long by 50 mm wide), were constructed and used in experiments to study the catalyst (cathode) materials, the effects of process variables on reactor performance and the reactor scale up with respect to a speculative industrial process for reduction of CO2. In Reactor A a variety of tin and lead catalysts (cathodes) were investigated, including tin deposited graphite felt, copper mesh, tin coated copper mesh, lead shot and granules, tin shot and tin granules. Among these catalysts tin coated copper mesh and tin granules were the most intensively studied. The results here showed the primary and secondary reactions were respectively the reduction of C 0 2 to formate (HCOO) and of water to hydrogen, while up to 5% of the current went to production of CO, CH4 and C2H4. Factorial, fractional factorial and parametric experimental designs were employed with both tinned copper mesh and tin granule cathodes to study the separate and combined effects of the following ten process variables on reactor performance: catalyst life (operating time), cathode specific surface area, cathode thickness, catholyte flow rate, current density, CO2 concentration in the feed gas, electrolyte conductivity, electrolyte species, gas flow rate and temperature. For these experiments on Reactor A the current ranged from 1A to 14 A with corresponding superficial current density from 0.22 to 3.11 kA m"2. The formate current efficiency and specific energy consumption ranged respectively from 10 to 96 % and 130 to 1300 kWh kmol"1 formate, while the product concentration of K H C 0 2 was up to 0.08 M in ii single pass flow. In addition, the combinations of two cations ( K + and Na +) and four anions (HCO3", C 0 3 ", Cl" and H C O O ) at various concentrations were studied as the catholyte, the best of which was found to be a solution of 0.5 M K H C O 3 with up to 2 M KC1 as the supporting electrolyte (when necessary). The anolyte used in the present work was 1 to 2 M T 7 / M T K O H In Reactor B tin granules were used as the main cathode material, due to the relatively long lifetime (300 mins vs 30 mins) and high selectivity for formate of tin granules compared with other cathodes in the present work. The effects on formate current efficiency of current density, fluid flow rates, back-pressure, formate concentration in the catholyte feed and cathode pretreatment techniques have been studied. The current in Reactor B ranged from 20 to 101 A with corresponding superficial current density from 0.62 to 3.14 kAm" 2. The results obtained are encouraging, e.g. the formate current efficiency reached 63 % at current density of 3.14 kA/m 2 , a reactor voltage of 3.9 V , specific energy 332 kWh kmol"1 formate and product concentration of up to 1.03 M K H C O 2 in single pass flow. However the problems of cathode deterioration over time (3-7 hours) and formate cross-over must be solved before this process could be commercially viable. A crude reactor model has been established as a guide to the development and scale-up of the process of electro-chemical reduction of CO2. In addition, a conceptual process design and subsequent economic projections were made for the on-site production of sodium formate by reduction of C 0 2 from an 80 M W fossil fuel based power generation plant (600 tonne/day CO2 emissions). For such a CO2 converting plant to be commercially viable the electro-chemical reactors should operate at superficial current densities above about 1 kA m"2 and preferably in the range 2 to 3 kA m" . With a current density of 2.2 k A m the total installed capital cost is estimated at 5.5 x i o 8 $US (2005) +/- 20% and the return on investment is about 24 %/year, based on a carbon credit of 25$US per tonne CO2 and assuming all products ( N a H C 0 2 , NaHC03, H 2 , O2) are sold at prevailing bulk chemical prices. Concerning the energy requirement, electro-chemical reduction would only be feasible as a means to control CO2 emissions in regions where non-fossil based energy, such as solar, wind or nuclear, is abundantly available. iii The present work has broken new ground with respect to the design of electro-chemical reactors for CO2 reduction and demonstrated the feasibility of carrying out this process on an industrial scale - with the caveat that the effective cathode life must be extended from three to several thousand hours and an improved cation membrane separator is needed to prevent formate cross-over. iv Table of Contents Abstract List of Tables List of Figures Abbreviations Acknowledgements 1. Introduction 1 1.1 Control of greenhouse gas CO2 1.2 Conversion and utilization of CO2 1.3 Electro-reduction of C 0 2 (ERC) 1.4 Objectives of the present work 2. Background and Literature Review 1 2.1 Thermo-chemical conversion of CO2 1 2.2 Electro-chemical reduction of CO2 (ERC): Stoichiometry, thermodynamics, mechanism and kinetics 1 2.2.1 Mechanisms of electro-chemical reduction of CO2 (ERC) 12 2.2.2 Kinetics .....13 2.3 Development of ERC 14 2.3.1 Electrodes (cathodes) for ERC 14 2.3.2 ERC in aqueous and non-aqueous media 17 2.3.3 Production of formate/formic acid in aqueous electrolytes 18 2.3.3.1 Electrolyte 2.3.3.2 Pressure, temperature and current density ,22 ,22 2.4 Solubility and speciation of CO2 2.4.1 Chemical reactions and equilibrium constants 2.4.2 Solubility of CO2 in aqueous bicarbonate-carbonate solutions 2.4.3 Solution pH 2.4.4 pH-potential diagram of the C 0 2 / H C 0 3 7 C 0 3 2 " system 2.4.5 C 0 2 absorption in H C 0 3 7 C 0 3 2 " solution 27 ,28 ,26 ,23 ,23 30 2.5 Theory of three dimensional (3-D) electrodes ,35 2.6 Theory of electrochemical trickle-bed reactor 39 3. Strategies and Scope of the Present Work 41 4. Experimental Methods, Apparatus and Materials 44 4.1 Process flow diagram 44 4.2 CO2 Electro-reduction apparatus 47 4.2.1 Reactor A (small reactor) 50 4.2.2 Reactor B (big reactor) 52 4.3 Reactor materials 57 4.3.1 Cathode 57 4.3.2 Anode 60 4.3.2.1 Anode current feeder"and anode mesh 61 4.3.2.2 Spacers 61 4.3.3 Cation exchange membranes 62 4.4 Analytical Methods 63 4.4.1 C 0 2 , CO and H 2 analysis 63 4.4.2 Hydrocarbon analysis 64 4.4.3 Formate analysis 65 4.4.4 Carbonate and bicarbonate analysis 65 5. Experimental Design 66 5.1 Factorial design 67 5.1.1 Main effects 67 5.1.2 Interaction effects 68 5.1.3 Curvature effect 69 5.2 Fractional factorial design 69 5.3 Data interpretation and modeling 70 6. Reactor Modeling 71 6.1 Modeling equations and assumptions 73 6.1.1 Electrode potential and over-potential 73 6.1.2 Concentration of CO2 (aq) and electrolyte pH 75 6.1.3 Partial current densities and current efficiency 77 6.1.4 Electrolyte conductivity and electro-active thickness of the 3-D cathode 79 6.1.5 Material balance 80 vi 6.1.6 Energy balance 82 6.1.7 Pressure gradient and liquid hold-up 82 6.1.8 Transport properties 83 7. Experimental Results and Discussions 89 7.1 Cathode selection and preparation 89 7.1.1 Nanostructured copper on graphite felt by liquid crystal templated electro-deposition 89 7.1.2 Cu-Sn alloy on graphite felt by conventional electro- and electroless deposition 91 7.1.3 Copper mesh 93 7.1.4 Tinned graphite felt by electroless deposition 94 7.2 Anode materials 96 7.2.1 Anode plate (feeder) 97 7.2.2 Anode mesh 97 7.2.3 Anode spacer 98 7.3 Tin-coated copper (tec) mesh cathode in Reactor A 100 7.3.1 Current, operating time (age) and CO2 feed fraction 102 7.3.1.1 Factorial experiments 103 7.3.1.2 Parametric experiments 107 7.3.1.3 Mass transfer constraint I l l 7.3.2 Temperature, (current and CO2 pressure) 112 7.3.3 Eletrolyte species 116 7.3.3.1 Factorial experiments on K + , Na + , CI" and C 0 3 2 " 116 7.3.3.2 K2CO3 vs. K H C O 3 in the catholyte 118 7.3.4 Catholyte conductivity and cathode thickness 121 7.3.5 Specific surface area of the cathode 123 7.3.6 Concentration of the supporting electrolyte (KC1) and cathode thickness 123 7.3.6.1 Factorial runs on KC1 concentration and cathode thickness (number of layers) 124 7.3.6.2 Parametric runs on KC1 concentration and cathode thickness (number of layers) 125 vii 7.3.7 Observations from tec cathode 126 7.4 ERC results with Pb electrodes in Reactor A 129 7.5 ERC results on tin shot and tin granules electrodes in Reactor A 133 7.5.1 Characterization of tin shot and tin granules 133 7.5.2 Exploratory runs on tin shot and Batch 1 tin granules in Reactor A 137 7.5.3 Reactivation of the deteriorating tin granules in Reactor A 138 7.5.3.1 Chemical treatment and recycle of tin granules 139 7.5.3.2 Polarity reversal (PR) for recovering the used tin granules (from Batch 1) 141 7.5.4 Exploratory runs on tin granules from Batch 2 in Reactor A 142 7.5.5 Fractional factorial experiments on the screened Batch 2 tin granules 143 7.5.6 Full factorial experiment on Batch 2 tin granules 148 7.6 ERC results on tin granules in Reactor B (scale-up) 150 7.6.1 Scale-up issues 150 7.6.2 Parametric runs 152 7.6.3 Corrosion problems of tin granules 156 7.6.4 "Recycle" runs and formate cross-over (transport through the membrane) ...163 7.6.5 Back-pressure experiments 168 7.6.6 Comparison between measured and modeled formate CEs 170 8. Conceptual Process Design and Economic Projection 172 8.1 Introduction 172 8.2 Conceptual process design 172 8.3 Operational conditions 175 8.4 Gross economic potential (GEP) 176 8.5 Net Economic Potential (NEP) and Return On Investment (ROI) 179 9. Conclusions, Contributions and Recommendations 188 9.1 Conclusions 188 9.1.1 Electro-catalysts 188 9.1.2 Carbon balance and selectivity 189 9.1.3. Overall cell reactions ..' 189 9.1.4 Process variables 190 viii 9.1.5 Scale-up 191 9.1.6 Modeling 192 9.1.7 Process synthesis & economics 193 9.2 Novelty and contributions of the present work 193 9.3 Recommendations for future work 194 References 196 Nomenclature 205 Appendices A: Overall (full-cell) reactions and performance indicators 210 B: Diffusion coefficient of CO2 in water 216 C: Product analysis and chemicals used for ERC 217 D: Activation of graphite felt surfaces 219 E: Recipes for electro- and electroless deposition 220 F: Determination of factorial points on temperature effects 221 G: Sample for modeling spreadsheet 223 H: Design and cost estimations for the auxiliary equipment 228 ix List of Tables 1.1 Composition of typical flue gas from combustion of fossil fuel 3 2.1 GO2 electro-reduction reactions 12 2.2 Kinetic information of CO2 electro-reduction to C H 4 on a Cu foil cathode at ambient pressure. Catholyte = 0.5 M KHC03 at pH = 7.6 14 2.3 Classified metal electrodes according to the reduction products 15 2.4 Representative results for ERC at solid metal electrodes 15 2.5 Representative results of prior work (since 1970) on electro-reduction of CO2 to formate/formic acid 21 2.6 Equilibrium constants and reaction rate constants for reactions in a C02(g)/C02(aq)/H2C03/HC03-/C032- system at 298 K 25 2.7 Solubility of C 0 2 in water (mol/L at 101 kPa C 0 2 partial pressure) 27 2.8 Estimated and measured compositions of the electrolyte after mixing with CO2 31 2.9 Estimated mass transfer and chemical reaction rates at different pH 34 4.1 Cathode materials studied 58 4.2 Characterizations of graphite felt (Type Grade GF, Metaullics Systems Inc.) 59 4.3 Characterization of copper mesh (ARGUS, US) 60 4.4 Pb plate, Pb shot, Pb granules, Sn shot and Sn granules 60 4.5 Characterization of anode materials 61 4.6 Properties of spacers studied 62 4.7 Diaphragm properties (micro-porous polyethylene) Source: D S M Tech of Heerlen, Netherlands . 62 5.1 Sample factorial matrix of a 2 2 design 69 6.1 Parameter values for best fit of modeled and measured formate CE 86 6.2 Modeling values for Reactor B at increment 1 (N=12) 87 x 7.1 Preliminary experimental results on tin-coated graphite felt 95 7.2 Experiments on anode materials in Reactor A 97 7.3 Variables studied with tin coated copper mesh cathodes in Reactor A 101 7.4 Typical changes along reactor length at 6A 101 7.5 Factorial variables and levels on current, cathode age and CO2 fraction 102 7.6 Other process conditions for runs in Table 7.5 103 7.7 Current efficiencies and carbon balance for 16 runs of the factorial set 105 7.8 Summary of the calculated factorial effects 106 7.9 E D X analyses for tin-coated copper cathode mesh 30# I l l 7.10 Effect of current on current efficiencies I l l 7.11 Factorial variables and levels on current, temperature and CO2 fraction 113 7.12 Factorial effects of current, temperature and CO2 fraction on the formate CE 114 7.13 Effect of temperature coupled with mass transfer 115 7.14 Factorial variables and levels on electrolyte species 116 7.15 Factorial effects of electrolyte species on formate current efficiency 117 7.16 Experiments for the comparison between HCO3" and CO3 ~ ; 119 7.17 Effect of K H C O 3 concentration on formate CE 119 7.18 Factorial variables and levels on conductivity and cathode thickness 121 7.19 Factorial effects on conductivity and cathode thickness (the number of layers) 122 7.20 Effect of the cathode specific surface 123 7.21 Factorial variables and levels on KC1 concentration and cathode thickness 124 7.22 Factorial effects of KC1 concentration and the number of cathode layers 125 7.23 Effect of KC1 concentration 125 xi 7.24 Effect of number of layers of tec 60# 126 7.25 Performance indicators for a special case with tec cathode 128 7.26 Iron (Fe) analysis on Pb catalysts and in the liquid products of ERC 130 7.27 Characterization of tin shot and granules 136 7.28 Impurity analysis in tin shot and tin granules 139 7.29 Experimental results on the recycle of the tin granules (from Batch 1) 140 7.30 ERC results on screened Batch 2 tin granules at 60 minutes of operation 142 7.31 Levels of variables for the 2 ( k"1 )+l fractional factorial design with tin granules from Batch 2 in Reactor A 144 7.32 Design matrix and formate CE's (responses) for the fractional factorial runs with tin granule cathode in Reactor A 145 7.33 Main and two-factor interaction effects for the fractional factorial results with tin granule cathode in Reactor A 146 7.34 Levels of variables for the 2 3 fractional factorial design with Batch 2 tin granules in Reactor A 148 7.35 Factorial effects of current, KC1 concentration and catholyte flow rate on the formate CE with Batch 2 tin granules in Reactor A 149 7.36 Parametric experimental results in Reactor B with Batch 2 tin granules 153 7.37 Experimental results of ERC with different pretreatment techniques 162 7.38 Preliminary experimental results on "recycle" runs in Reactor B 165 7.39 Current efficiency and carbon balance check for "recycle" runs in Reactor B 166 7.40 Experimental results on current efficiency check in Reactor B w.r.t. formate cross-over 167 7.41 Preliminary experimental results on the effects of back-pressure on formate CE in Reactor B 169 8.1 Modeled results for the reactor in the conceptual process design (bipolar mode) 176 xii 8.2 Molar flow rates and values of reactants and products [Chemical Market Reporter, 2005] * ...178 8.3 Cost of electro-chemical unit 182 8.4 Installed capital and operating costs for all units in Figure 8.1 (2005) at various current densities 183 8.5 Cost projection for the process shown in Figure 8.1 at operating current density of 2.2 k A m~2 and carbon credit of 25 $US /tonne C 0 2 184 8.6 NEP and ROI at various C 0 2 credits 185 9.1 Ranges of major process variables and figures of merit 192 B - l Diffusion coefficient of C 0 2 in water 216 E - l " Recipe 1: Electroless deposition of Cu, Sn and Cu/Sn alloy 220 E-2 Recipe 2: Electrodeposition of copper and Cu/Sn alloy 220 F - l 1000x mole fraction of C 0 2 in liquid phase 221 F-2 Correlated Henry's constants 221 F-3 Factorial points and center-point 222 H - l Chemical Engineering Plant Indices 228 H-2 Design diameters and heights for G/L separators 232 H-3 Material and Lang factors for auxiliary equipment 236 xiii List of Figures in 1.1 Diagram of the carbon cycle. The black numbers indicate the carbon stored various reservoirs, the blue numbers indicate the carbon exchange between reservoirs each year. GtC = gigatons of carbon, [http://enwikipedia.org/wiki/ Carbon- cycle] 1 1.2 Changing C 0 2 concentration in the atmosphere over the past 50 years. The measurements are taken from Mauna Loa, Hawaii. Source: Carbon Dioxide Information Analysis Centre (CDIAC), US. Department of Energy (DOE), ppm = ppm(vol)] 2 1.3 Technical areas for CO2 control 3 2.1 Mechanistic pathways commonly proposed for E R C 12 2.2 pH of KHCO3/K2CO3 solution at 298 K 28 2.3 pH-Potential diagram of CO2 and its related compounds at 298 K . pH-potential relations for water are shown in broken lines 29 2.4 Configuration for 3-D electrodes (a) flow-through regime (b) flow-by regime 35 2.5 Potential and current density distributions in a flow-by 3-D cathode 38 2.6 Configuration of trickle-bed electro-chemical reactors [Oloman, 1979] 40 3.1 Map of the chapters and sections of the work done in the present thesis 43 4.1 Process flow diagram. A = ammeter, P = pressure gauge, T = thermometer, V = voltmeter, W = wet gas flow meter, PC = pressure control 46 4.2 Reactor A (small) and Reactor B (big) 48 4.3 Cell configuration 49 4.4 Front view of the cathode for Reactor A 51 4.5 Components of Reactor A with 1 layer of tec 30# being the 3-D cathode 52 4.6 Front view and dimensions of the cathode for Reactor B 54 4.7 Pictures of Reactor B assembly 55 4.8 Reactor B ready for operation 56 xiv 4.9 A silicone glued spacer 62 4.10 Orsat analyzer for CO2 and CO analysis 64 6.1 Modeling scheme for Reactor B 72 6.2 Calculation procedures in each height increment 85 6.3 Local profiles of CE, pH, Pco2, [C02(aq)J, and temperature for the modeling run of Reactor B at 101 A 88 7.1 Electrodeposition of Cu onto graphite felt: a: with surfactant 40 wt % Triton X-100; b: without surfactant. Two pictures are both in a 30 um scale 91 7.2 Cu-Sn alloy catalysts obtained through electro-deposition 92 7.3 Electroless deposited graphite felts 93 7.4 Effect of cathode age with tin-coated graphite felt cathode .....95 7.5 Experimental reactor configurations 99 7.6 Silicone glued tin coated copper 100 7.7 Factorial results for HCOO" production. CE % and ([HCOO] mM) 104 7.8 Effect of current on formate CE. yco2: 100 vol%; cathode age: lOmin 107 7.9 Effect of CO2 concentration on formate CE. Current: 6A; cathode age: lOmin 108 7.10 Effects of cathode age on formate CE 109 7.11 S E M pictures of (a) new tec 30# mesh (b) used tec 30# mesh with 100 min age 110 7.12 CD of C 0 2 reduction vs. cell voltage with 100% CO2+10 min cathode age 112 7.13 Formate current efficiencies for factorial runs in Table 7.11 114 7.14 Formate current efficiency and reactor voltage for factorial runs in Table 7.13 117 7.15 Formate EC vs concentration of K H C O 3 at 6A, 100% C 0 2 and 1 layer tcc30# 120 7.16 Formate CE and reactor voltage for factorial runs in Table 7.18 122 7.17 Formate current .efficiency and reactor voltage for factorial runs in Table 7.21 125 xv 7.19 Effects of chelating agents on the formate CE with Pb plate cathode at 6 A (1.33 kAm"2) in Reactor A 131 7.20 Experimental results on Pb grid and Pb reticulate carbon cathodes at 6 A (1.33 kA m'2) in Reactor A 132 7.21 Size distributions of tin granules 135 7.22 S E M images of fresh tin granules 136 7.23 Formate CE vs cathode age in Reactor A 138 7.24 Formate CE vs cathode age in Reactor A . a: fresh tin granules; b, c and d: used tin granules being recycled once, twice and three times, respectively 141 7.25 Polarity reversal in Reactor A 142 7.26 Exploratory runs with screened Batch 2 granules in Reactor A 143 7.27 Current efficiency for factorial runs in Table 7.33 148 7.28 Comparison of experimental results in reactors A and B with a tin granule cathode ..153 7.29 Specific energy and formate concentration as a function of superficial current density in Reactor B 154 7.30 Effect of 11 wt% HNO3 pretreatment on Batch 1 tin granules. Left: before the pretreatment; right: after the pretreatment (both pictures are in 1 mm magnifications) 156 7.31 Rusty spots after 180 min operation of ERC in Reactor B 158 7.32 Potential- pH diagram of tin in water at 298 K 159 7.33 Modeled vs measured formate CE 170 7.34 Modeled and measured C E vs. superficial current density 171 8.1 Flowsheet for ERC with N a H C 0 3 and HCOONa as the catholyte 174 8.2 Simplified flowsheet of the reactants and products of ERC system 176 8.3 The net economic potential (NEP) and return of investment (ROI) as a function of superficial current density for ERC process shown in Figure 8.1 185 xvi 8.4 Amortized total capital cost (CIC) and total operating cost (COP) as a function of superficial current density for the ERC process of Figure 8.1 F - l Factorial matrix on the effect of temperature H - l Inlet and outlet temperatures of cooling water in heat exchangers CI to C3 H-2 Evaporation unit -xvii Abbreviations CR thermo-chemical reaction 2-D two dimensional 3-D three dimensional DFA direct formic acid (fuel cell) E D X energy dispersive X-ray analysis ERC electro-reduction of carbon dioxide FID flame ionization detector GC gas chromatography GDE gas diffusion electrode GSV gas space velocity HE . hydrogen evolution HPLC high performance liquid chromatography MT mass transfer PCD partial current density PVC polyvinyl chloride RTP room temperature and pressure SEM scanning electron microscope analysis SHE standard hydrogen electrode SS stainless steel sheet SS stainless steel mesh STP standard temperature and pressure tec tin-coated copper xviii Acknowledgements It is a pleasure and honor to express my gratitude to Professor Colin W. Oloman, my research supervisor, for guiding me through the years of my doctoral work in UBC. His strong academic knowledge, broad engineering experience and enthusiasm to teach helped me in building up the fundamental tools to complete the Thesis work; his interests and commitment to the present work, together with his perseverance and continuous encouragement strengthened me to go through the difficult times of the research; his cheerful and friendly nature made the research work enjoyable. Professor Oloman should also be thanked for his desire and great patience in helping me improve my English. I also thank Professor Elliott Burnell, Professor C. Jim Lim, and Professor Elod Gyenge, for providing precious suggestions. It was their suggestion in the Progress Report Meeting that helped directing the present work in searching and investigating more stable cathode catalysts. The staff of Chemical Engineering and Biological Engineering Department should also be thanked for their help in a lot of aspects of my study. I thank NSERC for providing me the PGS scholarship. My husband Mark and my son Gary both deserve my deepest love and gratitude, for years of unconditional and sacrificial support. Apology should also be made to them, for all those meals that I should but did not prepare for them and all those late evenings when they were waiting for me to come back from school at the bus station! xix 1 Introduction Chapter 1 Introduction 1.1 Control of greenhouse gas CO2 Carbon dioxide (CO2) plays an important role in the Earth's carbon cycle, and is a necessary ingredient in the life cycle of animals and plants. Figure 1.1, the Earth's carbon cycle, shows the storage and yearly changes of carbon between the atmosphere, hydrosphere and geosphere in gigatons of carbon (1 Gt = 109 metric tons). Humanity adds about 5.5 gigatons of carbon in the form of carbon dioxide per year, most of which goes directly into the atmosphere. Carbon dioxide is also a greenhouse gas, and the concentration of C 0 2 in the atmosphere has increased significantly since the beginning of the industrial revolution, especially over the past 50 years, as shown in Figure 1.2. Figure 1.1 Diagram of the carbon cycle. The black numbers indicate the carbon stored in various reservoirs, the blue numbers indicate the carbon exchange between reservoirs each year. GtC = gigatons of carbon. [http://enwikipedia.org/wiki/Carbon-cycle]. 1 1 Introduction M A U N A L O A O B S E R V A T O R Y , HAWAII MONTHLY AVERAGE CARBON DIOXIDE CONCENTRATION 390 385 380 375 370 365 360 355 350 345 340 335 330 325 320 315 310 MLO-H5 l T T r r r r r r r p T r r r r n ~ m T n ^ ^ /l/l 0 h 11 11 11 11 11 11 I i 11 11 11 11 11 I i 111 11 11 11 11 11 11 i I 11 i I i I i I 11 i I 11 11 I 11 11 11 11 i I i 1958 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 90 92 94 96 98 00 02 04 YEAR 19-May-05 Figure 1.2 Changing C 0 2 concentration in the atmosphere over the past 50 years. The measurements are taken from Mauna Loa, Hawaii. [Source: Carbon Dioxide Information Analysis Centre (CDIAC), US. Department of Energy (DOE), ppm = ppm(vol)]. The main anthropogenic sources of CO2 emissions are related to the decomposition of carbonate minerals and combustion of fossil fuels. Fossil fuels are consumed in the residential, commercial, industrial and transportation sectors of the economy - the last two of which are the largest contributors to CO2 emissions. Within the industrial sector, there are two principle routes of CO2 formation: i . manufacturing processes where CO2 is a by-product, such as in the production of cement, glass and lime. i i . energy generation (either as process heat or as electricity) by direct combustion of fossil fuels. 2 1 Introduction Table 1.1 gives some information of the flue gas from a typical combustion process. Table 1.1 Composition of typical flue gas from combustion of fossil fuel Composition, vol % co2 H 2 0 N 2 o2 N 0 2 10 20 69 1 0.001 CO2 accumulation in the atmosphere is widely blamed as a main contributor to global warming and a cause of extreme weather [Tryk, 2001]. This situation has stimulated worldwide research directed at reducing/controlling CO2 emissions. Figure 1.3 shows the key issues and technical areas related to C 0 2 control [Chunshan, 2002]. r—~ -\ C 0 2 Control Figure 1.3 Technical areas for CO2 control. 3 1 Introduction Energy choice: lowers CO2 emissions by selecting less CGvintensive forms of fossil fuel (such as natural gas versus coal) or non-carbon based energy sources such as nuclear, solar, wind and hydro. Energy efficiency: lowers CO2 emissions by improving the energy efficiency of the existing fossil fuel-based transportation and power generation systems. Sequestration: long-term burial of C 0 2 in reservoirs with large capacity, such as geologic formations and the ocean. Conversion and utilization: changes CO2 to chemically different forms and/or uses CO2 in both physical and chemical processes. It has to be mentioned that unfortunately most of the research interests of mitigating CO2 exclude "energy conservation" from the above list. When considering an alternative energy system, the very first thing that must be addressed is conservation. No matter how your energy is produced, it is cheaper to not use it or to use it wisely. 1.2 Conversion and utilization of CO2 Carbon dioxide can be used directly in applications such as carbonated beverages, firefighting, secondary oil recovery and supercritical extraction. In addition, the chemical conversion of CO2 is of great interest because it can mitigate CO2 emissions and at the same time utilize the cheap and abundant CO2 as a carbon source for the manufacture of fuels or other useful chemicals, thus adding valueto the CO2 disposal. For example, Bandi et al. from the Center for Solar Energy and Hydrogen Research (Germany) argued that producing carbon-based liquid energy carriers from CO2 is more attractive than producing liquid hydrogen, due to the easier and less expensive handling of the carbon-based fuels and their higher energy density relative to hydrogen [Bandi, 1995]; the N R C (Ottawa, 2004) proposed the electro-chemical synthesis of fuels from captured CO2 using renewable energy (wind, solar, etc.) and the storage of the fuels within the existing infrastructure for transportation applications. Therefore CO2 conversion can be positioned as a step, for CO2 recycling and resource conservation. 4 1 Introduction Possible processes that may be used for CO2 conversion include the radio-chemical, photo-chemical, bio-chemical, thermo-chemical, electro-chemical and photo-electro-chemical methods listed below. Most of these processes are still research subjects in the laboratory and few have reached large-scale production [Scibioh, 2005; Chunshan, 2002]. (1) Radio-chemical [Boyd, 1976] C02 + H20 r-radiMon > HCOOH,HCHO, (1.1) Radio-chemical conversion uses radiation to excite CO2 for reaction. The use of non-fossil fuel energy, i.e nuclear energy might be the attractive point of this method. This is a relatively new research area (first report was in 1960, with less than 10 papers published by 2006). (2) Thermo-chemical Thermo-chemical conversion covers a vast range of reactions of CO2 to organic or inorganic products [Mignard, 2003; Audus, 1997; Qin, 1995]. Only a few examples are given here, among which the production of carbon monoxide (1.3), methanol (1.4) and urea (1.6) have been industrialized. The hydrogenation (1.4) and methane reformation (1.5) of CO2 are the most popular research areas due to their favorable products. More details of these processes are given below in Section 2.1. 2Mg + C02 -> 2MgO + C (1.2) co, + c -> 2co (1.3) C02(g) + xH2(g) -> CH3OH(g),CH„C2H5OH (1.4) C02 +CH4 -^2CO + 2H2 (1.5) C02 (g) + 2NH3 -> H2NCONH2 +H20 (1.6) (3) Photo-chemical C02+H20 >CO,HCHO,HCOOH (1.7) (4) Electro-chemical 5 1 Introduction C02 + xe~ + xH+ — C O , HCOOH, CH,, {COOH)2 (1.8) The electro-chemical conversion of C 0 2 has attracted a lot of. research effort, which is introduced below in section 1.3 and elaborated in Chapter 2 as the topic of the present work. (5) Bio-chemical C02+4H2 bacteria > C H ^ 2 H 2 0 (1.9) This bio-methanation process has been carried out in a laboratory fixed-bed reactor (ca. 50 cm ) to generate CH4 at 80 % of the theoretical conversion with a space-time yield of ca. 5 m 3 STP m"3 h"1 [Jee, 1988]. (6) Photo-electro-chemical C02+2e-+2H+ h v + e V - s " d )CO + H20 (1.10) Photo-electro-chemical conversion is the electro-chemical conversion that uses light (such as sunlight) as the energy input and a semi-conductor (e.g. SiC, GaP, ZnTe) as the cathode [Hashimoto, 1993; Hinogami, 1998 and Tryk, 2001]. With the photo-electro-chemical approach, the solar harvesting process is incorporated into the electro-chemical system. If this can be done without sacrificing the performance of the electro-chemical and light-harvesting systems, there could be benefits in terms of cost and space saving. 1.3 Electro-reduction of C 0 2 (ERC) The electro-reduction of C 0 2 is of great interest in the fields of theoretical and applied electrochemistry, as reflected in the number of papers on ERC published since 1982, i.e. 441 [Schibioh, 2005]. However, the key argument against ERC is that, i f it came from the combustion of fossil fuels, the electric energy used in such a process would increase C 0 2 emissions. Therefore, ERC will have a future only when it can be carried out with high efficiency on an industrial scale and renewable energy is adopted as the source of electricity. 6 1 Introduction The electro-reduction of carbon dioxide has been studied since the late 19 th century [Ayers, 1994]. The majority of this previous work has focused on the electro-catalysis and mechanistic aspects of CO2 reduction, with experiments carried out in the batch mode in small (e.g. 1 x 10"4 m 2) electro-chemical cells (or half-cells) under conditions that are unlikely to sustain a practical process [Li and Oloman, 2005]. There are several issues in the research of ERC that should be dealt with in the development of .a process that will be interesting at the industrial scale (with reproducibility, long-term stability, high efficiency, low cost and so on). (a) Mass transfer constraint For industrial electro-chemical processes, the superficial current density and current efficiency should be respectively at least 1 kA m"2 and 50 %. However, the relatively low solubility of CO2 in aqueous solutions (ca. 70 mM at STP), coupled with the C02(aq)/HC03 _ /CO32" equilibria, creates a mass transfer constraint on the reduction of CO2 that limits the primary current density to a maximum value of the order 0.1 kA m"2 (i.e. 10 mA cm"2) under the typical laboratory reaction conditions (one phase flow, 2-D cathode) with 100 kPa (abs) CO2 pressure at 298 K. Several devices have been suggested to relieve the C 0 2 mass transfer constraint, including operation at super-atmospheric pressure and/or sub-ambient temperature, using a gas-diffusion cathode (GDE) or using a fixed-bed cathode while providing a "3-phase interface" for the reaction by sparging the cathode chamber with CO2 gas [Mahmood, 1987; Todoroki and Hara, 1995; Mizuno, 1995; Koleli, 2004; Yano, 2002; Akahori, 2004]. (b) Reactor capacity Apart from the constraint of the CO2 mass transfer limiting current density, a practical reactor for ERC must be able to handle volumetric gas feed rates in the order of 1 m 3 STP per hour per kA. With the possible exception of gas diffusion systems (GDE's) most electro-chemical reactors, and particularly those used by previous investigators of ERC, are not capable of handling high gas loads at the gas space-velocity (GSV) required of an industrial process (e.g. 100 to 1000 h"1 at STP). 7 1 Introduction GSV = gas space-velocity = volumetric gas feed flow / reactor (or catalyst) volume. (c) Competition of hydrogen evolution Most of the electro-reduction reactions of CO2 are in a potential range in which hydrogen evolution (HE) occurs. So there is always a competition between CO2 reduction and HE, especially in aqueous solution, which can result in a low current efficiency for ERC. There are several ways of suppressing HE, including choosing cathode materials with high over-potential for HE plus high affinity for CO2 and operating under higher CO2 pressure or at a three-phase interface to increase the availablity of CO2 at the cathode [Tryk, 2001]. (d) Lack of engineering research The research of ERC is still at a stage of fundamental investigations on mechanisms and kinetics using tiny electrodes (e.g. 1 x 10"4 m2) with little consideration of the possibility for practical application. Research on the engineering aspects of ERC should be initiated to bridge the gap between the previous laboratory work and industrial reality. Such engineering research includes the design and scale-up of continuous electro-chemical reactors, together with the conception, design and economic projections for complete ERC processes. At the time of writing the thesis (Feb. 2006) only two articles have appeared that describe the electro-reduction of CO2 in a continuous reactor. Akahori et al. [2004], who are apparently the first to report continuous operation, used a lead wire bundle cathode in a flow-by reactor with a cation membrane separator. This reactor obtained a formate current efficiency near 100 % with single-phase flow of a CO2 saturated catholyte solution at 1.4 ml • 1 2 min" and current about 2 mA (0.02 kA m"). The second source is a recent communication [Li and Oloman, 2005] that described the early phase of the present work. 1.4 Objectives of the present work The general objective of the present work is to develop a new system for ERC that could provide both acceptable efficiency and the possibility for practical application in a continuous industrial process for the production of hydrocarbons (such as methane and 8 1 Introduction ethylene), methanol or formate/formic acid. Specifically, the present work proposes a 3-D cathode with 2-phase flow (liquid/gas) in a continuous trickle-bed reactor for ERC to tackle the CO2 mass transfer constraint and increase the current density from the common range of 2 2 0.01 - 0.2 kA m" in the literature to above 1 kA m" , while treating a high volumetric flow of reactant gas at a practical space-velocity. Reactor scale-up is the second primary goal of this study, along with reactor modeling, process synthesis and economic projections aimed to bring ERC research closer to practical application. 9 2 Background and Literature Review Chapter 2 Background and Literature Review 2.1 Thermo-chemical conversion of CO2 The chemistry of thermo-chemical reduction of CO2 is briefly introduced here to show the comparison between thermo-chemical reduction and electro-chemical reduction of C 0 2 . There are two main thermo-chemical processes that use CO2 to produce new chemicals and fuels as a potential means of mitigating CO2 emissions [Eliasson, 1993; Seshan, 1993; Souma, 1993; Qin, 1996; Mignard, 2003] (1) Hydrogenation. C02(g) + 3H2(g)^CH3OH(l) + H20(l) A G ° 2 9 8 = - 9 . 2 x l 0 3 kJ kmol"1 (2.1) AH °298=-131 .2x l0 3 kJkmol" 1 C02(g) + 4H2(g)->CH4(g) + 2H20(l) A G 0 2 9 8 = -130.8x10 3 kJ kmof 1 (2.2) A H ° 2 9 8 ^ - 2 5 2 . 9 x 1 0 3 kJ kmor 1 The exothermic hydrogenation of C 0 2 to produce methanol or methane, as shown in reaction 2.1 or 2.2, is thermodynamically feasible at standard conditions and 298 K but requires a high pressure (1-5 MPa) and temperature (500-800 K) to maintain useful conversion and reaction rate. The major problem with the hydrogenation of CO2 is not the process itself but rather the availability of H2. If the purpose of this process is to limit the emissions of CO2 to the atmosphere then the hydrogen must be produced without emissions of CO2, by using renewable energy sources like nuclear, hydro or solar. (2) Methane reforming. Methane reforming with CO2 to produce synthesis gas is represented by reaction 2.3 and some of the parallel or consecutive reactions are shown by reactions 2.4 and 2.5: 10 2 Background and Literature Review CH4->C + 2H2 2C0 -> C + C02 C02 +CH, ->2CO + 2H2 AG°298 = +288.0x 10 3 kJ kmol"1 AH°298 = +261.0xl0 3 kJ kmol"1 AG°298 = -120.0xl0 3 kJkmol" 1 A H ° 2 9 8 = -172.4xl0 3 kJ kmol"1 AG°298 = +50.0x10 3 kJkmol" 1 AHO298 = +74.9xl0 3 kJkmol" 1 (2.3) (2.4) (2.5) The high standard free energy change of reaction 2.3 dictates a high operating temperature to obtain a favored equilibrium and useful conversion for the reforming process, i.e. 900 - 1200 K . Also, the thermodynamically highly favored formation of coke (reactions 2.4 and 2.5) deactivates the catalyst [Seshan, 1994]. Therefore, the challenges for this process are to achieve it under an input energy as low as possible to avoid secondary generation of CO2, and to develop catalysts that are not prone to coking (noble metals) and more stable over time. 2.2 Electro-chemical reduction of C 0 2 (ERC): Stoichiometry, thermodynamics, mechanism and kinetics Electro-chemical reduction of C 0 2 is an attractive way to convert CO2 because of the following advantages over thermo-chemical methods: (1) water is the proton source; (2) high equilibrium conversion at ambient temperature; (3) pure oxygen is produced as a by-product [Yamamoto and Tryk, 2002]. Carbon dioxide can be electro-chemically reduced to a wide range of end products. Table 2.1 shows the most common products from electro-chemical reduction of CO2 together with their corresponding standard redox potentials in acid solution. However, the extent to which any of these reactions occurs depends (regardless of their E° values) on the electrode kinetics and conditions of the particular system [Chaplin, 2003]. 11 2 Background and Literature Review Table 2.1 C 0 2 electro-reduction reactions [Chaplin, 2003]. Reaction E°/V(SHE) 298K at PH=0 2 C 0 2 + 2 H + + 2e" -> H 2 C 2 0 4 -0.475 C 0 2 + 2 H + + 2e" -> HCOOH -0.199 C 0 2 + 2 H + + 2e~ -> CO + H 2 0 -0.109 C 0 2 + 4 H + + 4e" -»• HCHO + H 2 0 -0.071 C 0 2 + 6 H + + 6e" C H 3 O H + H 2 0 +0.030 C 0 2 + 8H + + 8e" -»• C H 4 + 2 H 2 0 +0.169 2.2.1 Mechanisms of electro-chemical reduction of C 0 2 (ERC) A number of research groups have dedicated their efforts to fundamental studies of the mechanisms of ERC on a variety of electrode surfaces [Udupa, 1971; Sammels, 1993; Ryu, 1972; Paik, 1969]. The reaction pathways and resulting product distributions can be very complex, because they not only are related to the energies of adsorption of a whole range of possible species, including reactants, intermediates and products, but also depend on the electro-catalyst, the electrolyte and the cathode potential. The detailed mechanistic pathways for each product are not clear at present, and in many cases several different schemes have been proposed. Figure 2.1 shows the most commonly proposed pathways for ERC to formate (formic acid), methane and ethylene [Hori, 1989; Jitaru, 1997; Chaplin, 2003; Schibioh, 2005]. C 0 2 (g)-A' , C0 2(aq). A AS C 0 2 (ad) - ± ± - + . C 0 2 " (ad) -HCOO " ( a d ) - ^ - * HCOO" + e" r CO (ads) H + r + 4 e " > : C H 2 H Q + 2 e " > C H 4 : C H 2 I - * C 2 H 4 Figure 2.1 Mechanistic pathways commonly proposed for ERC 12 2 Background and Literature Review Step A, the adsorption of CO2 onto the electrode surface was suggested by some researchers [Sullivan, 1993; Kilnam, 1993; Vassiliev, 1985 and Teeter, 1954] to be a prerequisite to CO2 reduction. In some aqueous systems an intermediate hydration step (step A'i) is required before adsorption (step A'2) occurs. In this case, the low solubility of CO2 in water, i.e. 0.070 M at STP conditions, will limit the availability CC>2(aq) and thus limit the C02(ad) concentration at the cathode sites, especially when CO2 reacts competitively to give HCO3" and /or CO32" in alkaline solution. The use of 2-phase flow (G/L) should overcome this limitation by allowing the formation of C02(ad) directly from the gaseous state (step A) [Chaplin, 2003], and catalysts that have high affinity to CO2 or low activation energy for adsorbing CO2 should help raise the concentration of C02(ad). In addition, ways of increasing CO2 solubility can be useful in aqueous systems, such as operating under high pressure and/or low temperature, using additives to enhance the solubility of CO2. Step B is the one electron reduction of C02(ad) to the intermediate radical GO2", whose standard redox potential is about -1.9 V (SHE) [Tryk, 2001]. It is seen from Table 2.1 that the overall standard redox potentials for ERC are in the same range as that for the reduction of protons to H2 (E°=0.0 V(SHE)), which indicates that ERC is not thermodynamically much more difficult than hydrogen evolution. However, ERC is kinetically suppressed because of the high energy intermediate radical GO2" and the involvement of multi-electron transfer, i.e. 2 electrons for HCOO" (step B —•> C —> D) and 8 electrons for C H 4 (step B —> E —» F —> G). 2.2.2 Kinetics The kinetic information/data on ERC is sparse and difficult to summarize because of the variety of approaches and experimental conditions. Therefore the kinetic information is given here in a "list form". It has to be noted that the reaction kinetics depend on many interacting factors, so results obtained by one researcher under his conditions might not apply under other conditions. 13 2 Background and Literature Review (1) Frese [1993] calculated the exchange current density for step B of Figure 2.1 (the formation of -COY radical) at 100 kPa(abs) and 298 K in C 0 2 saturated aqueous solution (-0.036M) to be 4.8 xlO" 7 kA m"2. (2) Kim and Frese [1988] studied the kinetics with a copper foil electrode at 273 and 295 K using two separate electrode pretreatments to remove the oxide layer, i.e. 10 wt % HC1 and 10 wt% HNO3. The kinetic information is listed in Table 2.2. Table 2.2 Kinetic information of C 0 2 electro-reduction to C H 4 on a Cu foil cathode at ambient pressure. Catho lyte = 0.5 M 1 SCHCO3 at p] H = 7.6 Temp. K Pretreatment Tafle slope V/decade jo k A m " 2 Highest CE* % CEat-1.46V(SHE) % 273 10wt%HCl 0.539 N A N A 47 295 10 wt%HCl 0.110 1.94xl0 - 7 50 20 295 10wt%HNO 3 0.170 N A 32 . . N A C E is the "current efficiency" which is one of the performance indicators for electro-chemical processes. Definitions of various performance indicators are given in Appendix A. (3) Kapusta et al. [Kapusta 1988] reported the exchange current densities for ERC to formate at 293 K in (0.95 M KC1+0.05 M NaHC0 3 ) on In, Hg and Sn respectively as: lxlO" 7 , 1x10" 1 0 and l x l O - 8 k A m " 2 2.3 Development of ERC Procedures for electro-reduction of C 0 2 can be classified according to both the nature of the electrodes (cathodes) and the media used for the catholyte (aqueous and non-aqueous). 2.3.1 Electrodes (cathodes) for ERC ERC has been carried out on a variety of electrode surfaces, including most of the metals in the periodic table [Azuma, 1990], a number of alloys [Watanabe, 1991 and 1993; Ishimaru, 2000], metal oxides [Frese, 1991] and chemically modified electrodes [Ogura, 1998]. Among those electrodes metals, in the form of high purity (e.g. 99.9 wt%) foil, plate, 14 2 Background and Literature Review rotating disc, a bundle of wires, a bed of particles, mesh or GDE, are the most intensively studied. Metal electrodes are divided into four groups according to the nature of the main products [Hori, 1994; Azuma, 1990] (see Table 2.3) and representative results on solid metal electrodes are listed in Table 2.4. Table 2.3 Classified metal electrodes according to the reduction products. Group Metal Products A Cu Hydrocarbons and alcohols B Au, Ag, Zn, Pd, and Ga Carbon monoxide C Pb, Hg, In, Sn, B i , Cd, and Tl Formate/formic acid D N i , Pt, Fe, Co, Rh, Ir, and W Multi-products; only exhibit useful catalytic activities at high pressure and low temperature Table 2.4 Representative results for ERC at solid metal electrodes. Electrode Ag In Cu Ag Cu Sn Electrolyte KHCO-3 KHCO*3 K H C 0 3 K O H T B A B F 4 K2CO3 Solvent H 2 0 H 2 0 H 2 0 C H 3 O H C H 3 O H H 2 0 T(K) 298 298 273 248 298 298 P (kPa)(abs) 3000 6000 100 100 4000 100 Product CO HCOO" C H 4 CO CO HCOO" CE (%) 76 108 65 90 87 75 P C D ' " (kA m"2) 1.51 2.15 0.10 0.18 2.88 0.20 ECathode(V SHE) N A N A -1.39 -5.76 •1.50 •1.66 Reference Hara 1995 Todoroki 1995 Hori 1986 Kaneco 1998 Saeki 1995 Ito 1975 Note: TBABF 4 (tetrabutylammonium tetrafluoroborate). *Formic acid (HCOOH, pK a = 3.8 at 298 K) was claimed as the product here - but since pH > 6 the actual product is the formate anion (HCOO"). ** 108 % CE was obtained by summation of the partial CE's obtained by product analysis. *** PCD is the partial current density which is defined in Appendix A. 15 2 Background and Literature Review Due to the low solubility of CO2 in aqueous solution the current density of ERC is typically limited by the mass transfer of CO2 from the bulk catholyte to the electrode surface. In this context, various types of 3-D electrode with three-phase (G/L/S) operation have been used for ERC. They are introduced in the following paragraphs: Gas diffusion electrode (GDE's) Gas diffusion electrodes, based upon PTFE (polytetrafluoroethylene) as a binder, have been developed to overcome mass transport limitations with gaseous reactants in fuel cells. When employed in ERC, GDE's operated in batch mode, provided much higher superficial current densities than those obtained with 2-D metal plate electrodes [Furuya, 1997; Mahmood, 1987; Hara, 1995]. For example, Hara et al. obtained a current density of ERC of about 1.51 k A m"2 with a CE of 76 % for CO at 3000 kPa (abs) with Ag wire cathode (1.6 xlO" 5 m 2 geometric surface area) in 1995, then achieved a superficial current density of 2.58 kA m 2 and a CE of 86 % for ERC at 2000 kPa (abs) with Pt-GDE cathode (1 xlO" 4 m 2 geometric surface area) in 1997. Mahmood et al. reported nearly 100% current efficiency using Pb, In and Sn impregnated GDE's for ERC to produce formic acid in a pH range of 1-5, a current density of 1.15 kA m"2 at ambient pressure and temperature, with a sodium sulfate catholyte whose pH was adjusted by sulfuric acid. Fixed-bed electrode ERC was carried out by Koleli et al. in 2003 [Koleli, 2003] in an un-divided fixed-bed reactor where Pb and Sn granules (1mm and 3 mm diameter, respectively) were placed at the bottom of the reactor as cathode material. Batch mode operation was employed with CO2 gas bubbling up through the bed under ambient conditions. Only one product, "formic acid", was supposedly produced, with the maximum current efficiency of 90 % at a superficial 2 * • current density of 0.008 kA m" . A subsequent publication [Koleli 2004] describes a similar fixed-bed batch reactor, with pressure and temperature respectively up to 50 bar and 353 K , in which CO2 was reduced to "formic acid" with a current efficiency about 90 % at a superficial current density of 0.02 kA m" . The actual product was probably potassium 16 2 Background and Literature Review formate, since HCOOH (pK a = 3.8 at 298 K) should not exist at significant concentration in the catholytes used (0.5 M K H C 0 3 (pH 6-8) and 0.2 M K 2 C 0 3 (pH 11-12)). Mesh electrode Yano et al. performed ERC in a batch reactor at the 3-phase (G/S/L) interface on Cu and Pt mesh electrodes [Yano, 2002; Ogura and Yano, 2003 and 2004] at ambient pressure and temperature. In this system C 0 2 was supplied to the electrode from the gas phase by bubbling C 0 2 through the mesh, aiming to maintain a high concentration of C 0 2 at the cathode surface. The maximum total C E of 83 % for E R C (10 products) was obtained with Cu mesh at a superficial current density of about 0.05 kA m" , using a catholyte of 0.5 M KC1 + HC1, whose initial pH was 2.2 [Ogura, 2003]. 2.3.2 ERC in aqueous and non-aqueous media Because of the influence of the solvent on the nature of the products, there have been two main routes for C 0 2 electro-reduction based on whether the catholyte is aqueous or non-aqueous. Aqueous media have long been a popular selection for ERC research because the water provides a proton source, and aqueous electrolytes have typically much higher electric conductivity than that of non-aqueous electrolytes. A lot of studies have been done in aqueous media on almost all the electrodes described in Section 2.3.1, with a variety of electrolytes (see Section 2.3.3). The solubility of C 0 2 in aqueous electrolytes is relatively low but can be increased by organic additives. For example Ogura et al. [1999] reported that by adding 10 vol% propylene carbonate into an 0.5 M K G catholyte, the maximum total current efficiency of ERC (multi-products) was increased from 50 to 75 % as a result of enhanced C 0 2 solubility. Non-aqueous media present certain advantages for C 0 2 reduction: higher C 0 2 solubility than that in water, the suppression of H 2 evolution, and the possibility to work at 17 2 Background and Literature Review low temperature, e.g., below the freezing point of water. D M F (N, N-dimethylformamide), DMSO (dimethyl sulfoxide) or A N (acetonitrile) are all solvents for non-aqueous media [Sanchez, 2001]. But the most potentially useful results have been reported in methanol with various solid metal electrodes [Naitoh, 1993; Mizuno, 1995; Kaneco, 1998]. Methanol has been industrially used as a physical absorber of CO2 (the Rectisol method). Until 1998, more than 70 large-scale plants for the Rectisol process had been established [Hochgesand, 1970]. Thus, the study on CO2 reduction in methanol provides valuable information on the direct utilization of the industrial methanol-based CO2 absorber medium as the solvent for the electro-reduction process. Although the main products from CO2 reduction in non-aqueous media have been found to be quite similar to those in aqueous media, i.e., carbon monoxide, oxalic acid and formic acid, the reaction mechanisms can follow pathways different from those in aqueous media [Goodridge, 1984; Ikeda, 1987]. The electrolytes used in non-aqueous system are different from those in aqueous media. K O H and TBABF4 (tetrabutylammonium tetrafluoroborate) were both employed as the electrolyte for non-aqueous media, but the latter was more frequently used. 2.3.3 Reduction of CO2 to formate/formic acid in aqueous electrolytes Formic acid is now manufactured by thermo-chemical processes based on the carbonylation of methanol or sodium hydroxide and by the oxidation of hydrocarbons [Wiley, 2005] all of which have negative environmental consequences. The alternative electro-reduction of CO2 to produce formate (or formic acid) in aqueous solution is of particular interest due to: a. The high selectivity for the reaction obtained on a group of so-called "high hydrogen over-potential" cathodes b. The energy storage ability of formic acid both as a fuel for direct formic acid (DFA) fuel cells and as a source of H2 for hydrogen fuel cells [Williams, 1978; Rice, 2002] c. The potential environmental benefits of reducing CO2 emissions. 18 2 Background and Literature Review In this context the electro-reduction of CO2 is usually carried out in mildly alkaline conditions (pH = 7-9) where the main cathode processes are those of reactions 2.6, 2.7 and 2.8. Formic acid (HCOOH), with an acid dissociation constant pK a = 3.8 at 298 K, is produced at significant concentrations only at pH below about 5 (cf. Table 2.1). E°, V (SHE)@298 K (pH = 14) Cathode: C02 (aq) + H20 + 2e~ —»HCOO~ + OH' -1.02 (2.6) 2H20 + 2e~ -+H2 + 20H~ -0.83 (2.7) C02 + H20 + 2e~ -+CO + 20H~ -0.94 (2.8) From the literature it is known that the primary electro-active species C02(aq) is reduced to formate/formic acid (reaction 2.6) in parallel with the reduction of water to hydrogen (reaction 2.7) and tertiary reactions that convert CO2 to CO (reaction 2.8) plus traces of hydrocarbons [Ryu, 1972; Hori, 1982; Kapusta, 1983]. Reaction 2.6 is considered a kinetically "slow" process, occurring at cathode potentials (depending on the CO2 pressure, current density and catholyte pH) from about -0.8 to -1.8 V(SHE), with over-potential ranging from about -0.4 to -1.4 V . The intrinsic kinetics of reaction 2.6 are said to be independent of pH (2<pH<8), but a pH above 6 can have a strong effect on the mass transfer limiting current for this reaction, as C0 2(aq) is depleted through the C02(aq)/HC037C032" equilibria (see Section 2.4 on speciation). Reaction 2.7 is thermodynamically favored over reaction 2.6 over almost the entire pH range at 298 K (see Figure 2.3). However, reaction 2.6 is kinetically favored over reaction 2.7 on a group of "high hydrogen over-potential" cathodes, which leads to the high selectivity of formate/formic acid observed by many researchers. The most studied of this favored group of cathode materials are Hg, In, Pb and Sn, but it is not clear which of these metals is the best because their performance seems to depend on other variables such as CO2 pressure, catholyte composition, potential and temperature. 19 2 Background and Literature Review Table 2.5 summarizes representative results of (published) work since 1970 on the production of formate/formic acid from ERC. As indicated in Table 2.5 all work prior to 2004 was done in laboratory scale batch reactors [Walsh, 1993] operating at currents from a few mA up to about 2A. Most of this work was concerned with the separate effects of variables in subsets from the group [C c, E, I, Me, P, pH, T] on the reaction kinetics and mechanism, along with the selectivity for generation of formate/formic acid. As will be seen in Section 2.3.3.1 and 2.3.3.2, ambiguous and contradictory observations exist in the literature regarding the effects of process variables on the production of formate/formic acid. 20 Table 2.5 Representative results of prior work (since 1970) on electro-reduction of C Q 2 to formate/formic acid. Source Cathode Mode Cathode Catholyte Conditions Variable 'max C E area (aqueous solution) P 1 pH t k A m " 2 a t 'max ( l O ^ m 2 Bar K - h % Udupa 1971 Hg/Cu [rotary cylinder] Batch 39 Na 2 S0 4 , N a H C 0 3 1 293* 7-9 6 i , t 0.5 33 Ryu 1972 Hg [pool] Batch 7 NaHC0 3 , N a H C 0 2 0.1-0.8 274-333 7* ? i , P,T 0.01 98 Ito 1975 Cd,In,Pb,Sn,Zn [sheet*] Batch 5* [Li,Na,K,Rb] C 0 3 , PO„, S 0 4 1* 298 6.8 0.3* C c , E, I, Me 0.2 75 Russell 1977 Hg [pool] Batch 5 NaHC0 3 , N a H C 0 2 1 298 6.8 ? E 0.01 100 Hori 1982 Hg [pool] Batch 5 [Li, K, Na] H C 0 3 , P 0 4 , CI, CIO4, S 0 4 1 298 2-7 1 C c , E, pH 0.01 100 Kapusta 1983 Hg,In,Sn [sheet*] Batch 1 KC1, KHCO3 1 296 6.5 ? Cc, E, P 0.1 99 Mahmmod 1987 Pb,In,Sn [GDE] Batch 3 N a 2 S 0 4 , H 2 S 0 4 1 293 2 0.5 i, Me, pH 1.15 97 Todoroki 1995 Hg, Pb, In [wire,shot] Batch 0.2 KHCO3 1-60 293 7* ? E , i , Me, P 5.6 100 Mizuno 1995 In,Pb,Sn [coil] Batch 3 KHCO3 10-50* 293-373 7* ? P, Me, T ? 100 Koleli 2004 Pb [fixed bed] Batch 2 K 2 C 0 3 1-50 293-353 ? 0.5 E , Me , P, T 0.02* 88 Akahori 2004 Pb [wire] Cont J 1 K 2 H P 0 4 , H 3 P 0 4 1* 288 6 ? Ecell, T 0.02 100 Batch - zero catholyte liquid flow (some had gas flow) Cont = continuous catholyte liquid flow ? = not specified in source article. * = ambiguous or unspecified, but assumed or calculated from context of source article. Cathode area = geometric (superficial) cathode area Fixed bed = bed of particles C c =catholyte composition C E = formate current efficiency E = cathode potential E c e„ = reactor voltage i = geometric (superficial) current density imax = maximum current density Me = cathode material P = C 0 2 pressure T = temperature T = thickness of the 3-D cathode t = operating time 2 Background and literature review 2.3.3.1 Electrolyte The electrolyte (catholyte) composition affects the solubility of CO2, the electrode double-layer, the specific adsorption of reactants and intermediate species, and thus the product distribution and current efficiency of ERC [Murata, 1991; Hori, 1982; Ito, 1987; Spichiger-Ulmann, 1985 and 1986]. A variety of combinations of alkali metal cations and various anions, such as PO4 3", SO42", CO3 2", HCO3", CI", CIO4", HPO4" have been used as the electrolyte in aqueous media for ERC. As for the effect of cations, Hori et al [1982] report that the formate current efficiency increases in the sequence Li + <Na + <K + on Hg, but the reverse sequence is observed by Ito [1975] on an In cathode. With respect to the electrolyte anions some authors imply a unique role for carbonate/bicarbonate [Ito, 1975; Sanchez, 2001] while others [Hori, 1982; Kapusta, 1983] claim that among carbonate, phosphate, acetate and borate the partial current for formate production is independent of the buffer anion on Hg and In cathodes. Analogous cation and anion effects on reaction selectivity are observed in the electro-reduction of CO2 to CO and hydrocarbons on copper [Murata, 1991; Hori, 1988]. Apart from the effects of the electrolyte species, the effects of the composition and pH of the electrolyte have been studied. Most authors prefer K H C O 3 (ca. 0.5 M) with pH 6-8 [Mizuno, 1995; Todoroki, 1995; Hara, 1995] but others have excellent results in Na 2S04 at pH 2 [Mahmood, 1987] and some claim good current efficiency in a catholyte of K2CO3, whose normal pH would be near 12 [Koleli, 2003, Koleli 2004]. 2.3.3.2 Pressure, temperature and current density Operating ERC at high pressure takes advantage of the increased solubility of CO2 in aqueous solution. Some sources indicate that the rate for reaction 2.6 is nearly independent of CO2 pressure [Ryu, 1972], but most workers show that larger current densities for ERC are achieved with increasing pressure, reflecting a reaction order near 1 w.r.t. C02(aq) [Nakagawa, 1991; Kudo, 1993; Todoroki, 1995; Hara, 1995 and 1997]. 22 2 Background and literature review The effect of temperature is confusing in the literature of ERC, probably due to the fact that increasing temperature lowers the solubility of CO2 while raising the exchange current densities for the cathode reactions. One source [Ryu, 1972] shows a 20-fold increase in current density as the temperature rises from 275 to 333 K, without comment on the selectivity. Another [Koleli, 2004] shows a monotonic increase in formate CE on Pb, with increasing temperature (298 - 353 K), while a third [Mizuno, 1995] indicates a parabolic effect of temperature on a Pb cathode, together with monotonic decrease in formate CE on In and Sn over the range 293 - 373 K. Increasing current density typically lowers the CE of ERC [Udupa, 1971; Ito, 1975] by an effect that may be attributed mainly to CO2 concentration polarization. However some data on Sn [Ito, 1975], as well as results from experiments on Pb at 5000 kPa(abs) CO2 pressure [Koleli, 2004] imply a parabolic dependence of formate CE on current density with a maximum CE at about -1.5 V(SHE). From the above review we see that the literature on ERC contains complimentary and contradictory observations concerning the effects of the process variables. To resolve these issues experimental designs that capture the non-linear behavior and interactions of process variables should be employed in the investigation of electro-chemical processes such as ERC, as outlined in Chapter 5 (experimental design). 2.4 Solubility and speciation of C 0 2 When an aqueous solution is in contact with CO2 gas, various chemical species such as dissolved C02, C02(aq), H2CO3, HCO3" and C0 3 2", can be formed. The transformation between those species is complex since it is affected by the pH of the solution, the partial pressure of CO2 in the gas phase, the temperature and the presence of other electrolytes in the solution 2.4.1 Chemical reactions and equilibrium constants 23 2 Background and literature review Many studies address speciation in the C02(g)/C02(aq)/H 2C0 3/HC037C03 2" system with respect to the nature and of the resultant equilibria [Keene, 1993; Palmer, 1983; Williams, 1978; Cotton 1972, Walker, 1927], so only a brief summary will be given here. When CO2 gas is in contact with an aqueous solution containing HC03"/C0 3 2" (the catholyte solution for the present project) CO2 is absorbed into the liquid phase as loosely hydrated C02(aq) (equation 2.9) that subsequently engages in a relatively slow reaction with water to form carbonic acid, H2CO3. C02(g)^lC02(aq) (2.9) C02(aq) + H2O^Zl H2CO, (K #) (2.10) *-o Since less than 1% of the dissolved CO2 is present as H2CO"3, reactions (2.9) and (2.10) are combined to give the pseudo-equilibrium: C02(g) + H20(l)^H2CO; . (Ko) (2.11) in which [HC03*MC0 2 (aq)]+[H 2 C0 3 ] and K0=[H2CO;]/PCOi (2.12) The dissociation of carbonic acid H2C03 ^ H+ + HCO' (K,) (2.13) is a rapid reaction, and the first ionization to produce HCO3" is best expressed in the form of the combination of (2.11) and (2.13): H2CO] H+ + HCO' (Ki*) (2.14) K i * is often referred to as the "apparent" acid dissociation constant of H2CO3. In basic solutions (pH>7), C02(aq) may react directly with OH" CO, (aq) + OH' HCO; ( K 1 ) (2.15) where the forward reaction is much faster than the reverse reaction. 24 2 Background and literature review The second dissociation reaction: HCO; 7Z! H+ + CO]' (K 2) (2.16) can lead to a further equilibrium in solutions containing CO32": C02{aq) + CO]' +H20 7^ 2HCQ- (2.17) however, unless a buffer containing carbonate ion were being used, rate studies indicate the contribution of such a path is negligible compared with the previous equilibria. Table 2.6 lists the equilibrium constants and some reaction rate constants of the above reactions at 298 K, without considering the ionic strength. Table 2.6 Equilibrium constants and reaction rate constants for reactions in a C 0 2 ( g)/C0 2(aq)/H 2C0 3/HC0 37C03 2" system at 298 1 Reaction Equilibrium constant Forward rate constant Reverse rate constant 2.10 K # = 2.6x10"3 k ^ . O x l O " 2 ^ 1 k.0=23.7 s"1 2.11 K o =2.94xl0- 4 2.13 K!=2.0xl0" 4 2.14 Ki*=4.2xl0" 7 2.15 K ' i = 3.3xl0 7 k+i =8.5xl0 3 M V k.i=2.3xl0- 4s-' 2.16 K 2 = 4.7xl0" n ^ote: The relationships of some of the above constants with temperature are as follows: (1): K 0 = 1.0 xlO 1 8 T"8 7 0 5 1 [Perry, 1973] (2): pK] = 17052/T+215.2 log(T)-0.12675T-545.56 [Benefield, 1982] N o data are available on the effect of temperature on K # , K * i , K ' , and K 2 , or on the various reaction rate constants. It has to be noted that above discussion on speciation of C0 2(g) /C0 2(aq) /H2CO3 /HCO37CO3 2" is essentially an equilibrium speciation concerning only thermo-chemical reactions. However, in the system of the present work where a trickle-bed reactor is used (see Section 2.6), the catholyte composition is also affected by electro-chemical reactions and by the gas to liquid mass transfer capacity of C 0 2 , thus a dynamic speciation is more meaningful. This issue will be discussed in Chapter 6 (modeling). 25 2 Background and literature review 2.4.2 Solubility of CO2 in aqueous bicarbonate-carbonate solutions The solubility of CO2 is of great importance for developing the process of ERC, since it determines the concentration of CO2 in the liquid phase (C02(aq)) , which affects the rate of CO2 mass transfer to the cathode surface and thus the formate current efficiency (see Chapter 6 for modeling). The solubility of gases in water increases with pressure but decreases with temperature and can be approximated by Henry's law: P = H'x (2.18) Where: P = partial pressure of solute gas at equilibrium (kPa) x = mole fraction of solute gas in the solution (-) H ' = Henry's constant = f (T) (kPa) In the case of CO2, which reacts with water (reaction 2.10), the value of "x" in equation 2.18 is usually taken to include the concentration of CO2 and H2CO3, so the reported value of H ' is equivalent to the inverse of the equilibrium constant " K 0 " defined in equation 2.12 and correlated with temperature as: K 0 = 1.0 x l 0 1 8 T 8 7 0 5 1 273 < T < 333 K [Perry, 1973] (2.19) Where: T = temperature (K) The solubility of CO2 in the aqueous electrolytes follows the same pattern as that in water but generally decreases with increasing concentration of inorganic salts [Linke 1958]. For the case of bicarbonate-carbonate solutions, which are used in the present work, the dissolution of CO2 is further complicated through the equilibria of reactions 2.14 to 2.17. A great amount of research work has been done to address the issue of CO2 solubility in aqueous bicarbonate-carbonate solution because of the widespread use of this solution as a liquid absorption medium to separate CO2 from industrial flue gas [Greenewalt, 1926]. Here 26 2 Background and literature review the empirical equation from [Harte, 1933] will be introduced to determine the equilibrium relationship between temperature, partial pressure of carbon dioxide in the gas phase, and composition of the liquid: ± £ = io SPCOi(l-X)(\S5-T) Where: C = total concentration of bicarbonate and carbonate species (N) Pco2 = CO2 partial pressure in the gas phase (atm) T = temperature (°C) X = the fraction of the cations ( K + or Na +) in form of bicarbonate (-) S = the solubility of C 0 2 in water at 101 kPa C 0 2 pressure (M). Listed in Table 2.7. Equation 2.20 predicts the relationship between S, Pco2, C and X at equilibrium, which would be of considerable interest in the process development of ERC, especially in the present system where absorption of C 0 2 occurs in both the feed tubing and the cathode. For example, at given C 0 2 partial pressure, temperature and total concentration of HCO3" and CO3 ", equation 2.20 can be used to determine the equilibrium distribution of HCO3" and 2 + C0 3 z " , i.e. X . Then, [FT], thus pH can be calculated using the equilibrium constant K 2 of reaction 2.16. Table 2.7 Solubi ity of C Q 2 in water (mol/L at 101 kPa C Q 2 partial pressure) fHarte, 1933 T, °C 15 25 35 45 55 65 75 85 100 S 0.0455 0.0336 0.0262 0.0215 0.0175 0.0151 0.0120 0.0090 0.0065 2.4.3 Solution pH As will be seen in reactor modeling (Chapter 6), the pH of the catholyte has great impact on ERC because it determines the fraction of dissolved C 0 2 in the form of C0 2(aq) (believed to be the primary reactive species for ERC) along with the equilibrium electrode 27 2 Background and literature review potentials and relative rates of reactions 2.6 and 2.7, and influences the adsorption of ions onto the cathode which also affects the kinetics of reaction 2.6. The calculation of the equilibrium pH in a CO2/HCO37CO3 2" system needs the equilibrium constants (i.e. dissociation constants) of reactions 2.13 or 2.16, the equilibrium concentrations of HCO3" and CO3 ", and the activity coefficients for HCO3" and CO3 . Comstock et al. [1937] calculated the equilibrium pH of carbonate-bicarbonate solution under a set of conditions, using the activity coefficients from Walker et al [1927]. Figure 2.2 shows the graphic results of their calculation and serves as a convenient tool to estimate the pH of the carbonate-bicarbonate electrolyte solution for a limited set of conditions. Note that at pH < 8.5 essentially 100 % of the anions are present in equilibrium as HC0 3 " . When the pH drops below 8 the equilibrium shifts toward C0 2(aq), as indicated below in Figure 2.3. pH vs X (bicarbonate fraction) 9.5 -9 -8.5 -A 0.1 M • 0.5 M • 0.8 M I ~ ! 1 ' r- -0 20 40 60 80 100 X (bicarbonate fraction),% Figure 2.2 pH of K H C 0 3 / K 2 C 0 3 solution at 298 K 2.4.4 pH-potential diagram of the CO2/HCO3 /CO32 system The pH-potential diagram for the C02-water system at 298 K was constructed by Hori et al. [Hori and Suzuki, 1981 on the basis of the thermodynamic data (Figure 2.3). The activities of substances considered in the system were taken to be unity. 28 2 Background and literature review Potential vs pH > uT 0.5 CO CO > C02(aq) 0 ) 3 -05 -Q. -1 0 2 4 6 8 10 12 14 Figure 2.3 pH-Potential diagram of CO2 and its related compounds at 298 K. pH-potential relations for water are shown in broken lines It has been proven by several researchers that C 0 2 ( a q ) is the primary reactive species in the electro-reduction of C O 2 in aqueous solution [Kilnam, 1993; Vassiliev, 1985 and Teeter, 1954]. Figure 2.3 shows that CO"2(aq) exists at equilibrium in the liquid phase when the pH is less than 8. The competitive reaction of hydrogen evolution (reaction 2.7) has a less negative equilibrium potential than that of CO2 reduction to formate when the pH is lower than 7, which means hydrogen evolution is thermodynamically favored over C O 2 reduction in this region [Hori and Suzuki, 1981]. The reaction thermodynamics predict the reduction of CO2 is favored in a narrow pH range between about 7 and 8. Figure 2.3 also shows that formate is favored in the pH range from about 8 to 12. Reaction 2.6 is not viable in this region at equilibrium because C 0 2 ( a q ) is sequestered as H C O 3 7 C O 3 2 " , however as discussed below (Section 2.4.5) the kinetics of reactions 2.9 to 2.13 can allow a sufficient concentration of C02(aq) for reaction 2.6 to proceed at a significant rate under dynamic conditions in a multi-phase (G/L/S) system. As discussed in Section 2.3.3 and elaborated in the model (Chapter 6), the current efficiency for electro-reduction of CO2 to formate also depends strongly on the electrode kinetics of reactions 2.6 and 2.7. Even when reaction 2.7 is thermo-dynamically favored 29 2 Background and literature review over reaction 2.6 the former process can be kinetically suppressed by the use of a cathode material with a low exchange current density for hydrogen evolution, such as Hg, Pb, In or Sn. The predictions of Figure 2.3 apply only to the condition of thermodynamic equilibrium (i.e. they do not consider the process kinetics). The equilibrium state between CO2 and bi-carbonate/carbonate species can be reached by pre-saturating the liquid phase with CO2 gas and maintained approximately by bubbling CO2 into the catholyte during the electrolysis, as was done in most of the prior work on ERC in the batch mode. That is why the most common pH range for prior work on ERC with a bicarbonate electrolyte is from 6 to 8, and that work apparently has given little consideration to the dynamic conditions of mass transfer (G/L and L/S) coupled to thermo-chemical and electro-chemical reaction that exist in a practical continuous reactor such as that of the present study. 2.4.5 C 0 2 absorption in HCO3/CO32 solution The rate of CO2 absorption into the catholyte is determined by the rates of reactions (2.10) and (2.15) and greatly affected by the physical diffusion rate of CO2 (equation (2.9), mass transfer step) through the boundary layers at the gas/liquid interface, which is strongly influenced by the way CO2 gas is contacted with the liquid. In the experiments of the present study, CO2 gas feed is, first of all, mixed with the catholyte and then the 2-phases (G/L) move together in a plug-flow mode through the reactor feed tubing. Finally in the reactor, the 2-phases flow co-currently upwards through the 3-D (porous) cathode. This flow scheme is detailed in the flowsheet of Figure 4.1 in Section 4.1. In the tubing The mixing of CO2 gas with the catholyte in the tubing is essentially a CO2 absorption process, and whether the equilibrium is reached in the solution before entering the reactor is determined by the rates of the mass transfer and chemical reaction steps. This discussion is of great importance because i f the C 0 2 physical absorption (mass transfer) rate 30 2 Background and literature review is lower than the CO2 chemical reaction rate then it may be best to feed a catholye in equilibrium with the feed C 0 2 gas in order not to consume some C 0 2 feed by absorption. If the C 0 2 absorption rate is higher than the chemical reaction rate then it may be best to feed an electrolyte whose equilibrium C 0 2 partial pressure is above that of the feed gas (i.e. lower H C 0 3 7 C 0 3 2 " ratio and higher pH than for the equilibrium), so that the C0 2(aq) is still available at high pH for ERC and hydrogen evolution (reaction 2.7) is suppressed by the high pH. The system in the present research has a gas/liquid mixing tube of 973 cm 2 in contact area prior to the reactor (see Chapter 4 for Experimental apparatus). By using the data of absorption coefficient [Harte, 1933] and the empirical equation 2.20, the compositions of the bicarbonate-carbonate solution after absorbing C 0 2 gas were estimated. The experimental measurements were also carried out with samples being taken immediately after the reactor instead of from the sampling valve in order to exclude the effect of further absorption of C02(gas) in the outlet tubing (see Figure 4.1). The composition of bicarbonate was determined by the Ba(OH) 2 titration given in Chapter 4. Both calculated results and measured results are listed in Table 2.8. Table 2.8 Estimated and measured compositions of the electrolyte after mixing with C 0 2 in feed tubing of the experimental apparatus (Figure 4.1). Liquid feed C0 3 2 " , N 0.52 1.00 HCO3", N 0.00 0.00 Outlet composition after absorbing C 0 2 C 0 3 2 - , N estimated 0.00 0.52 measured 0.00 0.48 HCO3", N estimated 0.54 0.51 measured 0.55 0.56 Table 2.8 shows that measured and estimated values of the composition of the electrolyte after mixing with C 0 2 are in good agreement and they both indicate that the absorption of C 0 2 in the C 0 3 2 " solution is so fast that feeding 0.52 N (0.26M) C 0 3 2 " actually ends up the same thing as feeding ca 0.52 N H C 0 3 " in the present system, except that feeding C 0 3 " would result in a lower gas flow rate. 31 2 Background and literature review In the reactor In the discussion of absorption of CO2 in the tubing, the overall rate of CO2 absorption is the most important consideration. However, in the reactor where the electro-reduction of CO2 takes place, it may be more useful to consider the mass transfer and thermo-chemical reaction steps separately to find the concentration of the electro-active species C02(aq) in the bulk catholyte. The pH-potential diagram in Section 2.4.4 (Figure 2.3) shows that the C02(aq) is not available at equilibrium when pH is over 8. However, the pH-potential diagram only applies for the equilibrium conditions and it might be possible to have C02(aq) available at pH above 8 if the CO2 mass transfer rate exceeds the chemical reaction rate. This material balance is of great importance because i f the CO2 physical absorption (mass transfer) rate is lower than the CO2 thermo-chemical reaction rate then it may be best to feed an electrolyte in equilibrium with the feed CO2 partial pressure in order not to consume some CO2 feed by absorption. If the rate of CO2 absorption is higher than the thermo-chemical reaction rate then it may be best to feed an electrolyte whose equilibrium CO2 partial pressure is above that of the feed gas (i.e. lower HCO37CO3 2 " ratio and higher pH than for the equilibrium), so that the CChfaq) is available at high pH for ERC and hydrogen evolution (reaction 2.7) is suppressed. The corresponding conditions for maintaining C02(aq) should apply throughout the reactor as CO2 is consumed while the pH, bicarbonate concentration and temperature rise. The calculation of mass transfer (MT) and thermo-chemical reaction (CR) rates is important to this situation. Combining equations (2.9), (2.10) and (2.11) gives the mass transfer step as the following equation shows [Perry and Chilton, 1973]. Equation (2.10) and (2.15) are the two thermo-chemical reaction paths that consume C0 2(aq) [Cotton, 1972]: Mass transfer step: A . C02 (g) + H20(l) C02 (aq) + H20(l) s H2CO\ Chemical reaction steps: B. C02(aq) + H20 \ * H2C03 ^Zl H+ + HCO; ' - 0 C. C02 (aq) + OH~ ZZl HCO; 32 2 Background and literature review For the mass transfer step A, the rate is given by: M T rate = kGILaPCOi K0 (2.21) Where: kc/L = the mass transfer coefficient (m s"1) a = the specific G/L surface area (mzm" J) Pco2 = the CO2 partial pressure in the reactor (kPa) K q = the equilibrium constant for equation (2.11), i.e. 2.94><10"4 M kPa"1 at 298 K [Perry, 1973]. For the thermo-chemical reaction steps, for pH < 8, step B dominates and the chemical reaction rate is: - d ^ C 0 ^ a q ^ = k+0hLs[CO2(aq)] = 3x\0-2hLe[CO2(aq)] (2.22) dt Where: k+o = rate constant of reaction 2.10 (s"1) IIL = liquid hold-up (-) s = voidage of the 3-D cathode (-) [C02(aq)] = concentration of dissolved CO2 in the catholyte ( M ) At pH >10, step C is predominant and the rate of C02(aq) consumption is: d[C02(aq)\ dt k+lh,£[C02(aq)][OH~] = t.SxlO3 hLe[C02(aq)][OH-] (2.23) Where: k+i= rate constant of reaction 2.15 (s"1) In the pH range of 8 to 10, both steps are important, hence the rate should be: d[C02(aq)] dt = (3.0xl0" 2 +8.5x\03[OH-])hLs[CO2(aq)] (2.24) 33 2 Background and literature review Table 2.9 shows the calculated mass transfer rate and total thermo-chemical reaction rates at different pH values where C 0 2 partial pressure and temperature are respectively 145 kPa and 298 K. These considerations are elaborated w.r.t. reactor modeling in Chapter 6. Table 2.9 Estimated mass transfer and thermo-chemical reaction rates at different pH pH 7.5 8.5 9.5 10.5 Mass transfer rate, M s"1 0.025 Chemical reaction rate, M s"1 0.0003 0.0006 0.003 0.03 Comparing the mass transfer and thermo-chemical reaction rates it may be seen that, when a catholyte solution of carbonate or of low ratio of bicarbonate/carbonate is fed to the reactor, i.e. pH is higher than about 10, the chemical reaction is so fast that one can not expect to have both a relatively high concentration of C0 2(aq) and high bulk pH simultaneously. Hence, depending on the C 0 2 pressure, G/L mass transfer capacity, liquid hold-up and cathode voidage, the catholyte feed solution should have a pH lower than about 10 and a buffer capacity to maintain the bulk catholyte pH less than 10 throughout the reactor. The C0 2(aq) concentration and pH at the reactive cathode surface depend on these bulk conditions, coupled with the current density and L/S mass transfer coefficient, as detailed by Gupta et al. [2006]. Considered along with the equilibria of Figure 2.3 these mass transfer and thermo-chemical reaction rates may favor the (current density dependent) reduction of C 0 2 to formate in the bulk pH range about 7 to 10 at 298 K. The changes in temperature that occur in a practical reactor will probably shift this pH window. Kinetic data are not available to calculate the effect of temperature, but applying the rule of thumb (doubling of chemical reaction rates for 10 K increase in T) moves the upper limit of pH down from about 10 to 9 as temperature increases from 298 to 333 K. Apart from the above effects the electro-chemical kinetics of reactions 2.6 and 2.7 play a major role in determining the formate current efficiency over the whole pH range, and particularly at pH below about 6, where C 0 2 sequestration as HCO37CO32" is not an issue. 34 2 Background and literature review 2.5 Theory of three-dimensional (3-D) electrodes The above literature review shows that the current density for ERC on 2-D electrodes is limited by the mass transfer constraint caused by the low solubility of CO2 in aqueous electrolytes. This limitation would hinder the realization of industrial processes of ERC but may be relieved by the use of 3-D electrodes. 3-D electrodes, such as reticulates, packed bed, felts and gas diffusion electrodes, provide an extended surface area and therefore enhanced mass transfer capacity for reactions involving low mass transfer limited current density (caused by low concentration of reactive species) and/or slow kinetics. Two flow regimes are used for 3-D electrodes in continuous electro-chemical reactors, based on the relative direction of electrolyte and electric current [Walsh, 1993; Gupta, 2004] (Figure 2.4): G> I I I I 1 1 1 r Porous separator 3-D cathode - > Feeder I Electrolyte flow (a) t _1 1 1 1 • Catholyte flow (b) Figure 2.4 Configuration for 3-D electrodes (a) flow-through regime (b) flow-by regime 35 2 Background and literature review Flow-through electrodes In this regime, shown in Figure 2.4-(a), the current flow and electrolyte flow are parallel. This arrangement has the advantage over flow-by electrodes of more uniform potential distribution since the conversion of reactant occurs along the bed thickness and thus the concentration of the reactant is roughly uniform within each a cross-section perpendicular to the current flow. However, the bed depth is limited by the potential drop which occurs in the direction of current flow and the accompanying electrolyte flow. The typical electro-active bed depth is of the order of 1 - 10 mm. Hence this arrangement gives a restricted conversion per pass and severely limits scale-up and practical applications. Flow-by electrodes A typical configuration of flow-by electrode (shown in Figure 2.4-(b)) uses the arrangement with the current perpendicular to the electrolyte flow. Variations of this configuration are adopted by the majority of porous, gas diffusion and packed-bed electrodes in practice. Higher conversions per pass (compared with those obtained with flow-through configuration) are often possible for this configuration by simply increasing the height of the bed. Therefore this configuration is more amenable to scale-up. The performance of practical 3-D electrodes is greatly affected by the non-uniform potential distribution along the dimension parallel to the current flow (see Figure 2.5 -a). This non-uniformity results in a non-uniform distribution of current density, i.e. the local current density decreases towards the feeder until a point that the current density (reaction rate) is negligible. The ideal operating condition is to maintain the mass transfer limiting current density over the entire thickness of the bed (Figure 2.5-b). This may be achieved, in principle, by ensuring that the potentials at the cathode feeder (E 2) and the separator (Ei) correspond to the ends of the limiting current plateau (Figure 2.5-c). The thickness of the bed should ideally work at the mass transfer limited current density and this thickness is known as "electro-active bed thickness" Terr- The xeff is estimated as follows [Masliy, 1997]: (1) Electrode conductivity (km) » Electrolyte conductivity (ki) 36 2 Background and literature review • V = -|0.5 2kLeffAE nFkMaCr (2.25) (2) Electrode conductivity (km) ~ Electrolyte conductivity (k|) -,0.5 Teff ~ nFkMaCr (2.26) In equations 2.25 and 2.26: xeff = electro-active thickness of 3-D electrode (m) k|>eff = effective electrolyte conductivity (Sm"1) AE =Ei-E2 in Figure 2.5-b (V) = potential window in Equation 6.29. n = number of electrons exchanged F = Faraday's constant (96500 C mol"1) k M = mass transfer coefficient (m s"1) a = specific surface area (m 2 m"3) C r = reactant concentration (mol m"3) The effective electrolyte conductivity is calculated from the Neale-Nader equation [Oloman, 1979]: 2k, s h, 3-shL e = bed porosity of the 3-D electrode (-) IIL = liquid hold-up (-) k| = electrolyte conductivity (S m"1) The mass transfer superficial limited current density is then expressed by: iL=reffnFkMaCr (2.28) iL = mass transfer limited current density (kA m") 37 2 Background and literature review (a) Potential distribution Separator Feeder 0 • x (b) An idealized strategy 1 / / Main reaction y Secondary reaction r * ! • Ei E 2 - E (c) Current density distribution i t 0 h Figure 2.5 Potential and current density distributions in a flow-by 3-D cathode, (a) Potential distribution; (b) An idealized strategy - the entire cathode operates at the limiting current density; (c) The current density distribution. E m and Es: electric potentials of the electrode matrix and electrolyte solution, respectively; E: potential difference between the matrix and solution (driving force); h: bed thickness; i : local current density. 38 2 Background and literature review 2.6 Theory of trickle-bed electrodes [Oloman, 1979] A trickle-bed electrode, which is analogous to a thermo-chemical trickle bed reactor, uses co-current flow of a liquid electrolyte and reactant gas (sometimes inert gas) through fixed beds of electro-active catalyst particles or porous materials. A major difference between a thermo-chemical and an electro-chemical trickle bed reactors is that in the latter there is a limitation of effective electrode thickness to a maximum of about 10 mm, which is caused by the decreasing potential difference between the electrode and the electrolyte in the direction of current (see Figure 2.5 above). This constraint requires that trickle-bed electrodes should have a different practical form from that of thermo-chemical trickle-bed reactors: i.e., a long sandwich of electro-catalyst particles between a counter electrode and feeder electrode as shown in Figure 2.6. Depending on the electro-chemical process to be carried out, the reactor may be divided into separate cathode and anode chambers by an ion specific membrane (Figure 2.6-a) or the electrode bed may be prevented from touching the counter-electrode by a porous insulator such as a diaphragm (Figure 2.6-b). It should also be noted that the fluid flow in a "trickle-bed" electro-chemical reactor is typically in the slug flow to pulsing gas-continuous regime. This 2-phase flow can be up or down but upward flow provides higher liquid hold-up, improved effective electrolyte conductivity and better mass transfer than downward flow [Hodgson and Oloman, 1999]. Trickle-bed electrodes have been extensively studied and used commercially for production of hydrogen peroxide by electro-reduction of oxygen [Oloman, 1979 and 1996; Mclntyre, 1983 and 1995]. For some applications, trickle-bed electrodes are considered more practical than gas diffusion electrodes, being easier to make, less mechanically sensitive and unaffected by problems of flooding [Yamada, 1999]. A trickle-bed cathode promises to be a good choice for ERC due to the advantages of 2-phase flow with high gas space-velocity, plus a high mass transfer capacity and ease of scale-up. 39 2 Background and literature review Gas + electrolyte Feeder electrode Gas + electrolyte 3 - L (a) Gas + electrolyte Counter electrode Membrane Feeder electrode — Electro-catalyst bed Gas + electrolyte <-(b) Counter electrode Porous insulation Figure 2.6 Configuration of trickle-bed electro-chemical reactors [Oloman, 1979] 2-phase (G/L) flow may be down or up. 40 3 Strategies and Scope of the Present Work Chapter 3 Strategies and Scope of the Present Work From the above literature review it is evident that there is a lot of R & D to be done on ERC to bring it to practical application. Such work embraces stepping from the small batch . cells of the prior work to a continuous reactor that operates over several thousand hours at practical current density and efficiency (e.g. CD > 1 kA m"2 and CE > 50 %), followed by scaling up to a multi-cell "industrial" reactor in an economically viable continuous electro-chemical process. As mentioned in Chapter 1, the general objective of the present work is to complete the first steps in developing a reactor for ERC that could provide acceptable performance and the possibility for practical application in a continuous process. The main strategies adopted to meet the objective are: i Use of a three-dimensional (3-D) cathode to deal with the twin constraints of slow intrinsic kinetics and C 0 2 mass transfer. ii Operation in the continuous mode. iii Combining i and ii with the use of 2-phase (gas/liquid) flow in a trickle-bed electro-chemical reactor that provides a high mass transfer capacity and low liquid hold-up, to maintain a high concentration of the primary electro-active species (C02(aq)) in the bulk catholyte, while having the capacity to treat high gas loads with practical space-velocity. Specifically, the work in this project includes: selecting and preparing 3-D cathodes that can operate in a trickle-bed for ERC at high superficial current density with high selectivity for a desired product such as hydrocarbons, methanol or formate/formic acid (a.k.a. high space-time yield) and long life. 41 3 Strategies and Scope of the Present Work designing, constructing and testing the reactor to get satisfactory performance. investigating the effects of process variables on the reactor performance, modeling the reactor, scaling up the reactor. conducting a preliminary process synthesis and economic projection for a (speculative) industrial ERC process. Figure 3.1 outlines the scope and hierarchy of the original work carried out for this project. Essentially this work consists of experimentation, reactor modeling and process synthesis/economic projection - with the corresponding chapters shown in the figure. The experimentation is the major endeavor of this thesis and consists of three principal parts, each of which has several elements, as shown in Figure 3.1. The dotted lines indicate transference of results between stages of the hierarchy. 42 Thesis work Experimental work (Chapter 4 and 7) Modeling (Chapter 6) Economic projection (Chapter 8) ERC in Reactor B (big) (Section 7.6) Cathode selection and preparation (Section 7.1) Anode and spacer selection (Section 7.2) Sn-coated Cu mesh cathode (Section 7.3) Pb cathode (Section 7.4) Sn cathode (Section 7.5) SS screen, plastic,etc I Mesh 30" and 60* Cu deposited felt (surfactant, etc) Cu mesh Sn/Cu mesh (electrodeposite) Sn/Cu mesh (electro-less) Pb shot/granule Sn shot/granule i 1 | 1 i Pb shot and I i Sn shot and i [ granules i J granules J L- I L j era 3 Si. s-"-a >! s Figure 3.1 Map of the chapters and sections of the work done in the present thesis 4 Experimental Methods, Apparatus and Materials Chapter 4 Experimental Methods, Apparatus and Materials 4.1 Process flow diagram Figure 4.1 shows the process flow diagram used in the present study for the electro-reduction of carbon dioxide. Pure CO2 or the mixture of CO2 (gas) and N2 (gas) was combined with the catholyte (liquid) at a T junction (mixer), from which the gas and liquid proceeded in slug flow to enter the cathode chamber from the bottom. Thus, the electro-chemical reactor was operated with co-current upward multi-phase (G/L) flow on the cathode side. The anolyte, which was an aqueous K O H solution, also flowed upward through the anode chamber and was recycled to the anolyte storage tank. A l l gases and liquids passed through individual rotameters. Liquid flow was controlled at the pumps, while gas flows were controlled by manual valves to assure the appropriate gas and liquid loads in the reactor. The reactor inlet and outlet pressures and temperatures were measured by visual gauges at the points indicated in the flowsheet. In runs during which the catholyte product temperatures were controlled, pre-cooling or pre-heating of both anolyte and catholyte was employed to keep the temperature at a desired level. Liquid product was withdrawn from the sampling point and analyzed for formate concentration. Gas product from the gas/liquid separator (a packed bed of graphite felt) was controlled by a 3-way valve either to an Orsat gas analyzer for CO2 and CO analysis, to a wet gas flow meter for flow rate measurement, or to a Tedlar sampling bag for subsequent hydrocarbon analysis with gas chromatograph. Galvanostatic electrolysis of CO2 was carried out with a DC power supply connecting across the anode and cathode. A voltmeter was also connected to the unit to measure the reactor voltage. A l l voltages included anode potential, cathode potential and IR drop. The individual electrode potentials were not measured. An automatic pressure control valve was used in the anolyte product line to balance the pressure in the anode chamber against that in the cathode chamber. Such a pressure 44 4 Experimental Methods, Apparatus and Materials balance is required to prevent catholyte by-passing the 3-D cathode and/or the bursting of the membrane that can occur when the cathode pressure exceeds the anode pressure. Most experimental runs were conducted with the cathode outlet at the atmospheric pressure. For some runs in Reactor B a manual back pressure control valve and pressure gauge were installed in the catholyte product line to maintain superatmospheric pressure in the catholyte outlet. 45 DC supply Tedlar bag Liquid product Heat exchanger Heat exchanger T-junction (mixing) Catholyte tank TO 3 ' TO s TO a-o tlx is 1 O a TO S' 5" Figure 4.1 Process flow diagram. A = ammeter, P = pressure gauge,T = thermometer, V = voltmeter, W = wet gas flow meter, PC = pressure control. 4 Experimental Methods, Apparatus and Materials 4.2 CO2 Electro-reduction apparatus Experiments were performed first in Reactor A (small reactor) to carry out fundamental studies of ERC, and then in a seven-fold big Reactor B (big reactor) to see the effects of scale up. Figure 4.2 shows the picture of the appearance of Reactors A and B in comparison with a soft drink can. Both reactors have the configuration shown in Figure 4.3. The reactors consist of a cathode feeder plate and a 3-D cathode, a Nation cation exchange membrane separator, anode spacer/membrane support, an anode feeder plate and gaskets. The cathode mesh, anode mesh and the anode spacer are sealed on their margins by silicone glue, and then the cell assembly is sandwiched between insulated mild steel plates and uniformly compressed with SS bolts (1/4 inch in Reactor A and 3/8 inch in Reactor B) to give a balanced fluid distribution. 47 4 Experimental Methods, Apparatus and Materials 1 2 3 4 5 7 8 9 10 Figure 4.3 Cell configuration. 1 and 2: cell bodies; 2, 7 and 9: gaskets; 3: anode feeder; 4: anode spacer; 5: membrane; 6: 3-D cathode (tin-coated copper mesh, tin shot/granules and Pb shot/granules); 8: cathode feeder. 4 Experimental Methods, Apparatus and Materials 4.2.1 Reactor A (small reactor) Figure 4.4 and Figure 4.5 show respectively a sectioned elevation and exploded view of the single-cell Reactor A. The "flow-by" cathode of this reactor had dimensions of 30 mm width and 150 mm height (geometric surface). The thickness of the cathode depended on which 3-D cathode material was used. For tin-coated copper mesh cathode as shown in Figure 4.5 (see 4.3.1 for details on cathode materials), single or multiple layers of mesh were placed between the membrane and cathode feeder so the thickness of the cathode was the total thickness of these all layers, which ranged from 0.38 to 1.83 mm; for graphite felts and metal granules or shot, the cathode materials were embedded in two layers of Neoprene gasket with the back of the cathode in contact with the cathode feeder, therefore the thickness of the cathode was that of the gasket, i.e. 3.2 mm. The geometric (a.k.a. superficial) cathode area perpendicular to the electric current was 30 mm by 150 mm - 4.5*10~3 m 2 . In Reactor A the applied current ranged from 1 to 14 A with corresponding superficial current density from 0.22 to 3.11 kAm" 2 . 50 4 Experimental Methods, Apparatus and Materials Figure 4.4 Front view of the cathode for Reactor A 51 4 Experimental Methods, Apparatus and Materials Figure 4.5 Components of Reactor A with 1 layer of tec 30* being the 3-D cathode. 4.2.2 Reactor B (big reactor) In Reactor B, a few initial runs were done with tin-coated copper mesh cathode. Due to the fast deterioration of Sn/Cu mesh this material was replaced by tin granules cathode for the bulk of the work in Reactor B. Figure 4.6 presents the dimensioned front view and corresponding dimensions of Reactor B with a tin granule fixed-bed cathode. To minimize the by-pass of the catholyte at the edges of the cathode bed, the gasket was purposely made with five triangles on each side to direct the flow toward the centre of the cathode. Subtracting the areas taken by those triangles, the superficial cathode area was 322 cm", which was about seven times that of Reactor A (45 cm2). The applied current in Reactor B ranged from 20 to 103 A with corresponding superficial current density 0.62 to 3.20 kA/m . Figure 4.7 shows the assembly of Reactor B with a tin granule fixed-bed cathode, according to the following procedures: (1) A sanded tin plate (99.99 wt% Sn, 3mm thick) cathode feeder was put onto the neoprene gasket (Figure 4.7, a); (2) The pretreated tin granules (see section 7.6.3 for details on the pretreatment) were spread uniformly into a Durabla gasket (3 mm thick) on the tin plate, and layers of Netlon screen were inserted into the entrance and exit regions of the catholyte flow to distribute the fluid and support the membrane (Figure 4.7, b); (3) The wet Nation 117 membrane was put on top of the tin granule bed, and then, the P V C screen spacer (Figure 4.7, c), anode SS mesh, and anode feeder (SS plate) were placed on top of one another in that sequence; (4) Lastly, a cell body was put into place, and 24 3/8 inch bolts were employed to compress the sandwiched cell 52 4 Experimental Methods, Apparatus and Materials uniformly using a torque wrench at about 27 N m. Caution had to be taken when putting on the membrane since a small amount of tin granules left on the gasket could cause leaking of the catholyte. 53 4 Experimental Methods, Apparatus and Materials Figure 4.6 Front view and dimensions of the cathode for Reactor B 54 4 Experimental Methods, Apparatus and Materials 4 Experimental Methods, Apparatus and Materials Figure 4.8 shows the fully assembled Reactor B ready for operation. Figure 4.8 Reactor B ready for operation 56 4 Experimental Methods, Apparatus and Materials 4.3 Reactor materials A substantial part of this project involved experiments aimed at the preparation, testing and selection of the materials used for the primary reactor components. 4.3.1 Cathode The cathode (electro-catalyst) is the most important component of the reactor for the electro-chemical reduction of carbon dioxide (ERC). The nature, form, structure and purity of the cathode determine both the intrinsic kinetics of the cathode reactions and the mass transfer capacity of the 3-D electrode. Carbon dioxide can be electro-chemically reduced on almost all groups of metals in the periodic table to give a variety of products with different levels of selectivity. As mentioned in Section 2.3.1 (literature research section), copper has been one of the most common catalysts studied in the research of ERC because of its desirable reaction products, i.e. methane and ethylene, while metals in the group B i , Cd, Hg, In, Pb, Sn and Zn provide high selectivity for formate/formic acid. Therefore, in the present work Cu and Sn were the targeted cathode materials with the prospect of producing hydrocarbons, methanol or formate/formic acid. The main strategy of this project was to use a trickle-bed electro-chemical reactor with 3-D electro-catalyst (cathode) for ERC. However in the initial stage of the project the configuration of the 3-D cathode was uncertain. Thus the first work in the project was concerned with development of an appropriate cathode, by the procedures outlined below. In the present work the cathode materials listed in Table 4.1 have been prepared in the lab or directly bought from manufacturers, however only some of them were utilized for the electro-chemical reduction of CO2. . Tables 4.2 to 4.4 give the specifications of the materials either used to make cathodes or used directly as the cathode. 57 4 Experimental Methods, Apparatus and Materials Table 4.1 Cathode materials studied Catalyst Self prepared From vendor ERC experiments Sections referred to Nano-stuctured copper deposited on graphite felt V 7.1.1 Cu/Sn alloy deposited on graphite felt V 7.1.2 Copper mesh V V 7.1.3 Sn deposited graphite felt V V 7.1.4 Sn coated copper mesh (tec) V 7.3 Pb plate, shot, granules, grid and Pb-C reticulate - V V 7.4 Sn shot and granules V V 7.5 and 7.6 58 4 Experimental Methods, Apparatus and Materials Table 4.2 Characterizations of graphite felt (Type Grade GF, Metaullics Systems Inc.) Property Value Source Initial porosity, e0 0.95 [Oloman et al., 1991] Mean fiber diameter (um), df 20 idem Fiber density (kg m" ) 1500 idem Graphitization (°C hr"1) 2,400/2 idem Carbon content (%) 99 idem Uncompressed thickness (mm), x 0 6.4 measured Compressed thickness (mm), x 3.0 measured Compressed porosity, e 0.89 s = \ Compressed specific surface area (m"), s 22,000 S = 4(1 -s) Electronic conductivity of compressed matrix (S kaps =10 + 2800 m"1), k 49 1--1.55 aps [Oloman et al., 1991] Note: graphite felt was used to make nano-structured copper deposited on graphite felt, Cu/Sn alloy deposited on graphite felt, and Sn deposited on graphite felt. 59 4 Experimental Methods, Apparatus and Materials Table 4.3 Characterization of copper mesh (ARGUS, US) Property Value Mesh 30" Mesh 60" Purity (%) 99.9 99.9 Mesh count per inch 30 60 Thickness (mm) 0.61 0.38 Wire diameter (mm) 0.305 0.191 Opening (mm) 0.541 0.234 Open area (%) 40.8 30.5 Specific surface area, m 2 /m 3 7,000 14,000 Note: copper mesh was used both directly as the cathode and as the base material for tin- coated copper mesh (tec) (see Section 7.3 for details of the tec cathode). Table 4.4 Pb plate, Pb shot, Pb granules, Sn shot and Sn granules. Cathode material Description Purity W t % Vendor Pb plate 3 mm thick N A Unknown (from old lab storage) Pb shot 0.5 mm in diameter 99.9 -Aldrich Pb granules Mesh 30" 99.96 Fisher Sn shot* SN-133 (1 .5- 3 mm diameter) 99.99 A E E (USA)** Sn granules* SN-131 (mesh 30") 99.9 A E E (USA) Note: all these materials were used directly as cathodes; * See details for the characterization of Sn of Sn shot and granules in Section 7.5. ** AEE = Atlantic Equipment Engineers. 4.3.2 Anode Although the selection of cathode material and configuration is the most critical aspect of the process of ERC, the selection of the anode material and configuration can have great impact on the project as well, especially when the purpose of the study is to develop a practical process, where both the anode and cathode affect the current distribution, cell 60 4 Experimental Methods, Apparatus and Materials resistance, reaction temperature and so on. As shown in Figure 4.3, the anode side consists of an anode current feeder and a porous spacer/membrane support that can be inert (plastic screen) or electro-active (SS mesh). 4.3.2.1 Anode current feeder and anode mesh The anolyte used in this work was aqueous 1 M K O H for Reactor A and 2 M K O H for Reactor B with the oxygen evolution being the anode reaction (see anode stoichiometric equation [3] in Appendix A). Due to the highly corrosive nature of the anolyte, the anode feeder should be chemically resistant. Platinized titanium sheet and 316 stainless steel sheet were both studied in Reactor A as the anode current feeder, and their properties are listed in Table 4.5. The stainless steel (ss) mesh types of Table 4.5 were also tested to provide lower reactor voltage and presumably better current distribution. Table 4.5 Characterization of anode materials Anode material Mesh count per inch Thickness mm Wire diameter mm Opening mm Open area % Platinized titanium sheet (Pt) _ 2.0 _ — — Stainless steel (316) sheet(SS) _ 1.0 — — — Stainless steel (316) mesh (ss) 10 1.28 0.64 1.65x1.65 72 Stainless steel (316) mesh (ss) 30 0.51 0.25 0.59x0.59 49 Stainless steel (316) mesh (ss) 40 0.43 0.22 0.42x0.42 44 4.3.2.2 Spacers One or two layers of spacers were employed in the reactor between the cation membrane and ss mesh anode to provide dimensional support for the membrane, prevent inter-electrode contact (hence, electric shorting) and improve the current distribution. To 61 4 Experimental Methods, Apparatus and Materials prevent leaking of the anolyte, the spacer has to be sealed with silicon glue on the margins as shown in Figure 4.9. Silicone glued margins Figure 4.9 A silicone glued spacer Table 4.6 shows the spacers studied in the present work and their properties, and Table 4.7 lists the properties of the diaphragm that was studied. Table 4.6 Properties of spacers studied Spacer Mesh count per inch Thickness mm Wire diameter mm Opening mm Open area % Plastic screen 20 0.75 0.35 1.20x1.20 77 Netlon greenhouse shading (polyolefin) 10 1.20 0.60 1.80x1.20 75 Fly screen (PVC) 20 0.40 0.20 1.50x1.50 88 Table 4.7 Diaphragm properties (micro-porous polyethylene) Source: D S M Tech of Heerlen, Netherlands. Diaphragm type Mean thickness (um) Basis weight (g Mean Pore size (urn) Porosity (%) E075-9H01A 35 1.1 <0.06 74 62 4 Experimental Methods, Apparatus and Materials 4.3.3 Cation exchange membrane The anolyte used in the current project was exclusively K O H , 1 M in Reactor A and 2 M in Reactor B. This strong alkaline solution kept the pH of anode chamber about 13-14. In the cathode chamber, a pH range of 7 to 10 is considered necessary for the electro-chemical reduction of C 0 2 to dominate the secondary reactions (see discussions in Section 2.4.5). The cation membrane was employed to separate the anode and cathode chambers, maintain different electrolyte media on the two sides and act as a barrier to the transport of the product formate ions from the catholyte to the anolyte. Two types of cation membranes have been employed: Nafion 350 and Nafion 117. Both are copolymers of tetrafluoroethylene and perfluoro-(4-methyl-3, 6-diox-7-octene-l-sulfonic acid). They were not much different in terms of the formate CE, but Nafion 350 was a reinforced membrane that had to be masked with silicon glue on the margins of the rough side to prevent leaking. The membranes were soaked for 24 hours in 1 M KC1 solution before being placed into the reactor. In some cases when the re-used membranes were contaminated with rusty spots as a result of the cathode corrosion (especially with tin granules) 1 M HC1 solution was used to soak the membrane until the rusty spots came off the membrane. 4.4 Analytical Methods 4.4.1 C 0 2 , CO and H 2 analysis CO2 and CO were analyzed online with an Orsat, which was connected to the gas phase outlet of the separator through a 3-way valve (Figure 4.1). The Orsat analyzer shown in Figure 4.10 measured CO2 and CO by the decrease in gas volume when absorbed into solutions of chemicals with which they react. The absorbents for C 0 2 and CO were 40 ~ 50 wt % K O H and ammoniacal cuprous chloride (Cu (NTT^Cl), respectively [Vogel, 1978]. 63 4 Experimental Methods, Apparatus and Materials Figure 4.10 Orsat analyzer for C 0 2 and CO analysis 4.4.2 Hydrocarbon analysis A gas chromatograph (Varian 3600) with a FID detector and Carboxen-1010 PLOT capillary column was used to analyze low molecular weight hydrocarbons (basically C H 4 and C 2 H 4 ) in the gas products sampled in Tedlar bags. The analysis conditions for a mixture of methane and ethylene are: 64 4 Experimental Methods, Apparatus and Materials (1) Injection volume: 50 uL (RTP (2) Column temperature: 333 K (2min), 15 °C/min to 498 K (3) FID temperature: 523 K (4) Carrier gas: Helium, 3 mL/min (RTP) Then the hydrogen content of the product gas was found to be the difference of 100% and the sum of C 0 2 , CO, C H 4 and C 2 H 4 values. 4.4.3 Formate analysis Formate was analyzed by the alkaline permanganate oxidation technique detailed in Appendix C. 4.4.4 Carbonate and bicarbonate analysis In some of the experiments the concentrations of carbonate and bicarbonate were needed to check the carbon balance for the system. The total concentration of the two was determined by a sequential titration with standard hydrochloric acid. The concentration of bicarbonate can also be determined separately by back-titration procedures with Ba(OH) 2 and HC1 standard solutions. Details for bicarbonate and carbonate analysis are given in Appendix C. 65 J Experimental Design Chapter 5 Experimental Design The development of a CO2 electro-chemical reduction reactor, including scale-up, demands a quantitative assessment of the relationships between various reaction engineering values and the adjustable process variables. For a given reactor configuration the important reaction engineering values are "figures of merit" such as the current efficiency, specific energy consumption, reactant conversion and product yield that determine the capital and operating cost of the process in which the reactor is used. The adjustable variables include current density, electro-catalyst (form, material, specific surface), electrolyte (species, composition, pH), fluid flow rates, pressure and temperature. The relationships between these variables are on the whole complex and influenced by several factors that are not easily characterized, such as the multi-phase fluid flow regime, current density distribution, electro-catalyst and separator properties, electrolyte impurities and the deterioration in reactor performance over time. This situation leads to the need for extensive experimental research. Since such experiments are time-consuming and costly, good experimental design is required to provide a maximum of information with a minimum number of experiments. Electro-chemical processes often exhibit non-linear behavior, with interactions between process variables that complicate their assessment by experimental methods. Inappropriate experimental design can easily lead to misleading or even wrong conclusions. In the prior work on ERC, studies were almost exclusively based on uni-variate parametric experiments that did not engage interactions among the studied variables and that often implicitly discounted other variables that affected the process. For example, when researchers were investigating the effect of temperature on current efficiency, they discounted the effects of conductivity, C 0 2 solubility and fluid dynamics, which all affected current efficiency and were affected by temperature. That explains (in part) why conflicting observations are common in the research of ERC (Section 2.3.3 to 2.3.5). To explore a wide range of process variables and avoid the pitfalls of uni-variate experimentation the experimental design of the present work used both factorial experiments, 66 5 Experimental Design in which two or more variables were changed simultaneously, and parametric experiments, in which a single variable was changed while all others were (more or less) fixed. Factorial experimental design is an effective but simple method of planning experiments; it provides maximum information with a limited number of experiments by determining both the individual and synergistic effects of various factors on the target functions (responses). Therefore factorial experimental design was employed through all the phases of the present work. In some cases where too many (more than 4) variables were investigated simultaneously fractional factorial design was adopted to obtain preliminary information. As a rule, factorial experiments were used to span multi-dimensional experimental spaces and parametric experiments were carried out when appropriate, to obtain further information on variables that showed statistically significant effects. 5.1 Factorial design [Murphy, 1977; Box, 1961] Factorial experimental design involves nf factors, each at low (-) and high (+) levels. The levels can be either quantitative or qualitative and are usually determined by preliminary trial experiments. This design requires 2"f experiments, plus some replicate experiments to determine the confidence intervals. In addition, a few runs could be added at the centre points between the high (+) and low (-) levels of all factors to detect the possible non-linear system behavior. The notation for the so-called center-points is "0". Three kinds of effects are obtained through the calculations of experimental data: main, interaction and curvature effects. 5.1.1 Main effects The main effect is defined as the change in response caused by a change in the level of the factor. For a factorial design that consists of Nf runs: Main effect of factor X J : 67 5 Experimental Design ^ at high X , - Y Y t at low X, (5.1) -N 2 • f The confidence interval for the main effect is: Cl=±tss f (5.2) In which: CI = confidence interval of main and interaction effect Nf = the total number of factorial runs ts = the student's statistic at the desired confidence level (e.g. 90% or 95%) s = the pooled variance of the response based on fm total degrees of freedom fm = total number of replicate runs minus one Yj = response for run No i 5.1.2 Interaction effects The interaction effect between two factors reflects the fact that the effect of a given factor on the response depends on the level of another factor. The interaction effect between factor X i and X 2 is calculated as the average response difference between the effect of X i at the "high" level of X 2 and the effect of X i at the "low" level of X 2 . For example, in a 2 2 design where the two factors are, respectively, X i and X 2 , and responses are, respectively, Y i to Y4 as shown in Table 5.1, the interaction effect between X i and X 2 is calculated as equation (5.3) and the confidence interval is calculated using the same equation as that for the main effect (equation 5.2). [ ( Y l + Y 4 ) - ( Y 2 + Y 3 ) ] (5.3) 2 Where: 68 5 Experimental Design Xj = factor i X [ X 2 = interaction between factors X] and X 2 Table 5.1 Sample factorial matrix of a 2 2 design Run x 2 X i x 2 Responses 1 - - + Y , 2 + - - Y 2 3 - + - Y 3 4 + + + Y 4 5.1.3 Curvature effect The curvature effect is estimated as the difference between the average of the center-point responses and the average of the factorial points (equation 5.4). Thus, a significant curvature effect suggests the existence of the non-linear behavior of the system studied. Curvature effect = (£Y\ at center-pointsVCf- (£Yj at factorial points)/Nf (5.4) The confidence interval for the curvature effect is calculated as: ci-±,-sJ^ rJ (5-5) Where: CI C = confidence interval of curvature effects Cf = the total number of center-points 5.2 Fractional factorial design When the number of factors nf is greater than 4 fractional factorial designs (e.g. half or one-third) are usually carried out to avoid unnecessary experiments that lead to 69 5 Experimental Design redundancy in terms of an excess number of higher order interactions. In the present work, a two level, 5 variable 25"1 V resolution fractional factorial (half) design was used in Section 7.5.5. With this design, 16 runs, instead of 32 runs, were carried out. The main effects and two factor interactions were calculated using the same equations for full factorial design but the three factor and four factor interactions were confounded. 5.3 Data interpretation and modeling Based on the above methods, the general experimental plan for this project (set out in Figure 3.1) involves the progressive exploration of sets of process variables - leading from the cathode material to the small Reactor A and thence to the scaled-up Reactor B. The experimental data obtained at each stage are interpreted in terms of the theoretical process dynamics and used to build a crude process model, as the basis for further experimental work, scale-up and process development. 70 6 Reactor (cathode) Modeling Chapter 6 Reactor (cathode) Modeling The model described here was developed by Colin Oloman in an Excel spreadsheet and subsequently checked and elaborated by Hui L i , then used to interpret results from Reactor A and provide instruction in scaling up to Reactor B. The model was originally set up and the parameters estimated from experimental data for the .case of the cathode in Reactor A (section 7.3 below) as a differential continuous reactor in isothermal operation at steady-state [Li and Oloman, 2005]. The model was then extended to deal with Reactor B as a plug flow continuous reactor in adiabatic operation at steady-state, with significant gradients in catholyte composition, pressure and temperature along the reactor height. Although it is substantial relative to any in the ERC literature the model is still primitive so far as it does not calculate the potential and current distributions along or across the 3-D cathode, but instead assumes that the current is distributed uniformly along the cathode height and there is uniform current density on the real surface across the electro-active cathode thickness. Both of these assumptions deviate from reality but are considered adequate for the present purposes. Figure 6.1 shows the conceptual approach to modeling Reactor B. The model divides the plug flow reactor into N increments of height and solves the material and energy balances on each of these control volumes as a differential adiabatic reactor. Beginning at the reactor inlet (bottom) the model integrates from the inlet to the outlet along the height of the reactor by forward differencing. The model calculates the values of variables at the outlet of the reactor as shown in Figure 6.1 and provides details on local CO2 partial pressure, pH, [CO2 (aq)], [HCO3"], [HCOO"], formate CE, effective bed thickness, etc. along the height of the reactor. Such local information is instructive for determining operating conditions in scaling up a reactor. 71 6 Reactor (cathode) Modeling Reactor output: CE,=? [HCOO-] o u t l e,=? [HCO3 ] o u t l e t = ? p H o u t l e t = ? yc02(outlet)=? Pc02=? C 0 2 conversion=? Gas: C0 2 +H 2 +H 2 0 Liquid: K H C 0 3 + H C O O K + H 2 0 I Increment output: pH 0 U t , [HCO 3-] 0 U„ [HCOO-] I [C0 2(aq)] o u t, P o u t , yco2out ± AH J I Increment input: ! T i n ; P H i n , [ H C 0 3 - ] i n , [ H C O O - ] i n , I [C0 2(aq)] i n, P i n , y C 0 2 i n 1 Gas: Liquid: C 0 2 K H C 0 3 + H 2 0 Reactor input: Total current, I t o t a i Height of reactor, H t o t a i Number of increments, N Cell voltage, E c en Catholyte flow rate, L c Anolyte flow rate, L a Pressure, P i n i e t Bicarbonate [HC0 3 ' ] i n i e t Carbon dioxide yC02(iniet) • Gas flow rate, G Figure 6.1 Modeling scheme for Reactor B 72 6 Reactor (cathode) Modeling In each height increment the model involves relationships for the stoichiometry, equilibria and kinetics of the thermo-chemical and electrochemical reactions, mass transfer of CO2, pressure drop, electrolyte conductivity and so on, that are used to solve the material and energy balances for the cathode. The following sections describe the equations and assumptions used in the modeling. 6.1 Modeling equations and assumptions. 6.1.1 Electrode potential and over-potential Due to the utilization of a tin electro-catalyst in the present study formate and hydrogen were the dominating cathode products. Therefore the model of the cathode processes is based on the following two competitive cathodic reactions: E° V(SHE)@298 K 1. C0 2(aq) + H 2 0 + 2e~^HCOO" + OH" -1.02 - (6.1) 2. 2H 2 0 + 2 e " ^ H 2 + 2 0 H " -0.83 (6.2) The equilibrium potential of each reaction comes from the'Nernst equation: Erl-E!-**M[HC00~n0H~i) («) IF ( / V 1 0 1 ) RT Er2=E°2 -~\n{[OH-f(PH2 /101)} (6.4) 2F The equilibrium potentials are corrected for temperature by: AS, nF £ r ( r -298) (6.5) E r , i , E R > 2 = equilibrium electrode potential of reactions 6.1, 6.2 ( V ( S H E ) ) E ° i , E ° 2 = standard equilibrium electrode potential of reaction 6.1,6.2 ( V ( S H E ) ) E ° T - standard equilibrium electrode potential at temperature T ( V ( S H E ) ) 73 6 Reactor (cathode) Modeling E°298 = standard equilibrium electrode potential at 298 K (V(SHE)) F = Faraday's number = 96500 (kC kmol"1) n = electron stoichiometry coefficient (-) AS 0 = entropy of reaction (kJ kmof'.K" 1) T = temperature (K) Assuming the intrinsic kinetics of both reactions 6.1 and 6.2 fit the Tafel form with reaction 6.1 under a C 0 2 mass transfer constraint, the individual over-potentials are related to the partial real current densities by: ni = E - E r > i = ai - bilog(ji) + bilog(l-ji/jiL) On tin and on copper (6.6) rj 2 = E - E r j 2 = a 2 S n - b2 Snlog(J2) On tin (6.7) n 2 = E - Er>2 = a 2 C u - b2culog(J2) On copper (6.8) Where: T|2 = over-potential for reaction 6.1, 6.2 (V) ai, a 2 = Tafel constant for reaction 6.1, 6.2 (V) bi, b 2= Tafel slope for reaction 6.1, 6.2 (V decade"1) ji,J2 = partial real current density for reaction 6.1, 6.2 (kA m" ) jiL = CO2 mass transfer limited current density for reaction 6.1 (kA m" ) The Tafel equations 6.6 to 6.8 were set up with Sn and Cu for the tinned-copper mesh cathode in Reactor A , in which the tin coverage decreased with operating time (Section 7.3) [Li and Oloman, 2005]. Modeling Reactor B with a tin granule cathode (Section 7.6) assumed a fixed tin coverage of 100% and so did not use the kinetics on copper. The Tafel slopes of both reactions 6.1 and 6.2 were assumed as 0.118 V/decade at 298 K [Vasiliev, 1985; Pickett, 1977], with a charge transfer coefficient of 0.5, and corrected for temperature by: b = 2.303RT/(ocF) (6.9) in which: a = electro-chemical charge transfer coefficient = 0.5 (assumed) 74 6 Reactor (cathode) Modeling The Tafel constants and their temperature dependence involve several parameters that were fitted in the model from experimental data. First, the exchange current densities are calculated from equations 6.10 and 6.11, in which the effect of temperature on the rate constants is made explicit by not combining the pre-exponential rate factors (ki or k 2) with the activation energy terms: E a FE i0, = 2 F £ J C 0 2 ( ^ ) ] e x p ( - ^ ) e x p ( - ^ — ^ ) (6.10) KI KI E a FE i 0 > 2 = 2Fk2 e x p ( - - ^ ) e x p ( - ^ — ( 6 . 1 1 ) Kl KI Where: io,i, io,2 = exchange current densities (kA m"2) k i , k 2 = electro-chemical pre-exponential rate factors (m s"1) E a , i , E a > 2 = activation energies (kJ kmol"1) Subscripts 1, 2 refer to reactions 6.1 and 6.2 Then the Tafel constants are obtained as: RT fl,=(-^-)ln(i0l) (6.12) axF a2=(-^-)\n(ioa) (6.13) a2F with a i = a 2 = 0.5 (assumed) The electro-chemical parameters k|, k 2, E^i and E a , 2 were fitted in the model from the experimental data - with the constraint that the values of i 0 ; 2 and a 2 should approximate those found in the literature e.g. respectively about 1><10"7 kA m"2 and -0.8 V on Sn at 298 K [Conway, 1952; Pickett, 1977]. The activation energies were also kept within the typical range of electrode reactions, i.e. about 3><104 to 1><105 kJ kmol"1 [Conway, 1952]. 75 6 Reactor (cathode) Modeling 6.1.2 Concentration of CC>2(aq) and electrolyte pH The speciation of CO2 in aqueous HCO37CO3 2 " solution, as outlined in Section 2.4, is the basis for calculating the catholyte composition. The primary reactant C02(aq) reaches the cathode surface by the absorption of CO2 from the gas phase (G/L interface) into the flowing catholyte (reaction 2.9) and mass transfer of C02(aq) to the liquid/solid (L/S) interface. Previous researchers of ERC, using batch reactors, have typically assumed equilibrium between the gas phase and the bulk catholyte liquid (see Sections 2.2 to 2.3), and some authors have taken this assumption as a basis for modeling speciation at the cathode surface [Hori, 1989; Gupta, 2005]. However the G/L equilibrium assumption may not be appropriate for a trickle-bed continuous reactor at "practical" superficial current densities. In such a reactor the concentration of C02(aq) in the bulk catholyte is a dynamic variable that depends on the relative rates of G/L mass transfer, thermo-chemical reaction, L/S mass transfer and electro-chemical reaction. Section 2.4 outlines the principles of this non-equilibrium condition with respect to the G/L mass transfer and thermo-chemical reaction rates. For the working cathode in this condition the steady-state concentration of C02(aq) in the bulk catholyte can be estimated by a material balance on CO2 over a control volume of the cathode, as follows: 0 = (CO2 in from gas - CO2 out to gas) - (CO2 consumed by thermo and electro-chemical reaction) 0 - (kG/LaPco2Ko - k G / L a [C02(aq)]) - (hLe(k+o[C02(aq)] - k. 0[HCO 3-][H +] + k+i[C0 2(aq)][OH-]) + i ia (T e f f Ix) /2F) From which: kGILaPcoK0 +k_Q[HCO-][H+]-ila(^-) t C ( 9 2 ( ^ ) ] = : 7—7, . r n „ - , T (6-14) Where: a = specific surface area of cathode = specific G/L interface area (assumed, m 2 m"3) 76 6 Reactor (cathode) Modeling s = cathode porosity (dimensionless) riL = liquid hold-up (dimensionless-) xeff = electro-active thickness of 3 - D cathode (m) x = actual thickness of 3 - D cathode (m) ii = superficial current density of reaction 6.1 (kA m"2) ko/L,a = G / L mass transfer capacity = 1 / (1/kLa + l/kGa) (s"1) kL = liquid side mass transfer coefficient at G / L interface (m s"1) kc = gas side mass transfer coefficient at G / L interface (m s"1) K 0 , k+o, k.o, k+i, k.i = equilibrium and reaction rate constants defined in Table 2.6. The values of the pseudo Henry's constant (K 0) and the first dissociation constant of H2CO3* (Ki) are corrected for temperature as indicated under Table 2.6 in Chapter 2. Since no data on their temperature dependence were found in the literature the remaining constants in Table 2.6 are held at 298 K . Furthermore, no correction is applied to the constants of Table 2.6 for the effects of supporting electrolytes (e.g. K H C O 2 , KC1) on the CO2 equilibria and reaction kinetics. The mass transfer values are obtained from the literature correlations summarized in [Oloman, 1979]: k La = 0 . 0 1 7 3 ( A P L G U L ) 0 5 (6.15) k G a = 2 + 0 . 6 9 ( A P L G U G ) 0 6 7 (6.16) Where: APLG = pressure gradient in 2-phase flow (see below) (kg m"2 s"2) U L = superficial liquid velocity (m s"1) U G = superficial gas velocity (m s"1) 6.1.3 Partial current densities and current efficiency The liquid to solid (L-S) mass transfer coefficient (kM) for C0 2(aq) to the 3 - D cathode is estimated as that due to forced convection in 2-phase flow (kp), enhanced by a 77 6 Reactor (cathode) Modeling factor (ko) from the cogeneration of hydrogen gas, whose partial current density Q2) gives the Reynolds' number (Ret,) for gas bubble evolution at the cathode surface: [Coppola, 1989] (6.17) [Stephan, 1979] (6.18) (6.19) (6.20) Where: d = average granule diameter from screen analysis (m) db = bubble diameter (m) Dco2 = diffusion coefficient for CO2 in catholyte (m s"2) d e = effective granule diameter = d p / [1 + (x + W)d p / (3xW(l-s))] (m) dp = equivalent spherical diameter of cathode granules = d(p (m) kF = mass transfer coefficient for forced convection (m s"1) kn = mass transfer coefficient for hydrogen gas generation (m s"1) kM = combined L/S mass transfer coefficient (m s"1) P = pressure (kPa(abs)) Sc = Schmidt number = |VpiDco2 (dimensionless) Reb = Reynolds' number for H 2 bubble disengagement from cathode (dimensionless) Reo = Reynolds' number for gas flow = UodePG/pG (dimensionless) ReL = Reynolds' number for liquid flow = UidepJ\iL (dimensionless) W = width of 3-D cathode (m) UL, |^G = viscosity of liquid and gas (kg m"1 s"1) PL, PG - density of liquid and gas (kg m"3) s = voidage of 3-D cathode (dimensionless) 9 = granule shape factor (estimated by microscopic examination of granules) Then the CO2 mass transfer limited real current density of reaction 6.1 is calculated by: k F = (Dco2/de)(2.85Sc0-33 R e L 0 1 2 5 R e G a 2 3 8 ) k H = (D C O2/d b)(0.93Re b a 5 S c 0 4 8 7 ) Re b=(2Fj 2RT/P)(d b P L/p. L) k M = k F [ l+(k H /k F ) 2 ] 0 5 78 6 Reactor (cathode) Modeling j 1 L = 2FkM[C0 2(aq)] (6.21) With this value in equation 6.6 the two Tafel equations (6.6 and 6.7) are solved simultaneously for the partial real current densities j i and j 2 , using the assumption that the total real current density is given by: j = j i + j 2 (6.22) The formate current efficiency is then obtained as: C E , = j 1 / 0 i + j 2 ) (6.23) CEi = current efficiency for reaction 6.1 In these relations the superficial current density is related to the real current density on the cathode surface through the specific electro-active area of the 3-D cathode: i=jaieff (6.24) Where: i = superficial current density = I/WH (kA m"2) I = total current (kA) H = height of 3-D cathode (m) W = width of 3-D cathode (m) j = real current density (kA m"2) a = specific area of 3-D cathode = 6(l-E)/d p (m 2 m"3) Teff = electro-active thickness of 3-D cathode (m) 6.1.4 Electrolyte conductivity and electro-active thickness of the 3-D cathode The electrolyte conductivity was the sum of the conductivities of the primary electrolyte (KHCO3) and supporting electrolyte (KG) , which were estimated as polynomial 79 6 Reactor (cathode) Modeling functions of concentration [CRC Handbook, 2004] and corrected for temperature as shown below [Oloman, 1996]: kKHco, =(-0.8906C^ C O j +7 .5937C m c P j + 0.1677)[1 + 0.02(7 - 293)] (6.25) kKa = (-0.7468C£ c / +11.102CV-, - 0.0112)[1 +0.02(7-293)] (6.26) ~ k-KHco, + kKCt (6.27) The effective electrolyte conductivity was calculated from the Neale-Nader equation [C. Oloman, 1979]: 2k, s h, K , - J Z 7 V L (6-28) Where: kaa = ionic conductivity of K C l solution (S m"1) & K H C O 3 = ionic conductivity of K H C O 3 solution (S m"1) k\ = the electrolyte conductivity of solution containing K H C O 3 and K C l (S m"1) ^i,eff ~ the effective electrolyte conductivity (S m"1) C M = the concentration of species i , i.e. K H C O 3 or K C l (M) Then the effective electro-active thickness under pure mass transfer control (xeff) is estimated by: 2k,e/rAE 0.5 (6.29) 2FkMa[C02(aq)]\ The potential window (AE) is set equal to the absolute value of the over-potential of reaction 6.1, and averages 0.29 V . 6.1.5 Material balance The flow of CO2 is reduced along the height of the reactor by conversion in reaction 6.1 plus reaction with OH" produced in reactions 6.1 and 6.2. i.e. 80 6 Reactor (cathode) Modeling C 0 2 + O H - ^ H C 0 3 - (6.30) At the same time the flows of HC0 2 " , HC0 3 " and H 2 are increased according to the corresponding reaction stoichiometries. Cations (e.g. K + ) cross the membrane to match the generation of HC0 2 " , HC0 3 " . The flow and composition of both the liquid and gas steams at each height increment of the cathode is obtained using material balance and phase-split calculations, as follows: A(C02) = ~ I_ F (6-31) AOHC00-) = A . ( 6 3 2 ) A ( / / C 0 3 - ) = ( A . + Z l ) ( 6 3 3 ) ^ H l ) = J2F ( 6 ' 3 4 ) Where: A M = incremental change in flow of species M (kmol s"1) 11 = incremental partial current for reaction 6.1 (kA) 12 = incremental partial current for reaction 6.2 (kA) I = Ii+I 2(kA) The liquid phase composition is obtained as: [M] = concentration of species M (kmol m"3) M = flow of species M (kmol s"1) L c = liquid catholyte flow (fixed at cathode inlet) (m 3 s"1) For the G/L phase split the water vapor pressure is taken from Antoine's equation 81 6 Reactor (cathode) Modeling 3985 44 PHO= exp[l6.5362 ] (6 35) Hl° y i (T- 38.9974)J , 1 ; Where: Pmo = water vapor pressure (kPa) The difference between total pressure and water vapor pressure is then the sum of the partial pressures of C 0 2 and H 2 , and the material balances and ideal gas law (not elaborated here) are used to calculate the gas phase composition and volumetric gas flow rates of C 0 2 , H 2 and H 2 0 . The catholyte pH comes from the incremental liquid and gas compositions by: [H+] = KQK\PCO, r2[Hco~] <6-36> pH = -log[FT] Where: Y = activity coefficient in K H C 0 3 solution = 0.7 [Walker, 1927] 6.16 Energy balance Assuming adiabatic operation, the temperature rise in each height increment is calculated with a short-cut energy balance [Oloman, 1996]: A T = M(Ecell-E°cell) (La+Lc)Cp<L K } In which: Al = incremental current (A) L a = anolyte flow (kg s"1) L c = catholyte flow (kg s"1) C P ; L = liquid heat capacity, assumed = 4.18 (kJkg" 1K" 1) Eceii = operating reactor voltage (V) E°Ceii = reversible full cell voltage, approximated as 1 V . 82 6 Reactor (cathode) Modeling 6.1.7 Pressure gradient and liquid hold-up The pressure gradient and liquid hold-up are calculated from literature correlations summarized in [Oloman, 1979]. R = (150 + 1 .75Re i ) / / L £ / / . ( l -* ) 2 < V Pgas =(150 + 1.75ReG)/uGUG(l — e ) 2 d\e l 2 -0.425 ' < 1.30 + 1.85 > P gas _ hL =3.86Re° 5 4 5 G a - 0 4 2 k£ J Where: LG Ml (6.38) (6.39) (6.40) (6.41) (6.42) Piiq = pressure gradient for single-phase liquid flow (kg m" s") Pgas = pressure gradient for single-phase gas flow (kg m"2 s"2) A P L G = pressure gradient in 2-phase flow (kg m"2 s"2) 6.1.8 Transport properties The transport properties in the catholyte are estimated from correlations that account for the changing composition and temperature along the reactor, as follows: Liquid viscosity [Yaws, 1977; CRC handbook, 2004] 1828 (-10.73+^^+1,97xl0"2 7 -1.47xl0~5 7'2) // 7 =0.001x10 T m m c ^ r u ^ - i l (O.O254[#C03T + 0.l04l[/JCO3~] +1.0024) Gas viscosity: [Yaws 1977] Hc =0.0001 x 0.00 l(yCOiMco2 + y^oM^o + yH2MH2) In which the individual viscosities of C 0 2 , H 2 and H 2 0 were calculated respectively by (6.43) (6.44) 83 6 Reactor (cathode) Modeling HCOi = 25.45 + 0.4557- 8.65 x l 0 ~ 5 7 2 (6.45) HHi =21.87 + 0.2227- 3.75 x l 0 ~ 5 7 2 (6.46) juH2o =-31.89 + 0.4157-8.27x10~ 67 2 (6.47) Liquid density Fixed at 1030 kg m"3 Gas density p G = M m P / R T (6.48) In which: M m = mean molar mass of gas phase = £(MjVj) (kg kmol"1) Diffusion coefficient of CO2 in catholyte [Oloman 1996]: (6.49) D = ^ ( 2 9 3 ) — = 1,7444 x l O ' 1 2 — 293 ixL M l Where: Dco2 = diffusion coefficient of C0 2(aq) (m 2 s"1) D°co2(293) = diffusion coefficient of C02(aq) in water at 293 K (m 2 s"1) P293 = viscosity of water at 293 K (kg m ' V 1 ) u L = viscosity of catholyte (kg m ' V 1 ) Figure 6.2 shows a flowchart of the modeling calculations for each height increment. For these calculations Excel is used in the "iteration" mode and thus converges on several simultaneous solutions that are implicit in Figure 6.2 [e.g. calculation of j j L from j 2 ] . 84 6 Reactor (cathode) Modeling Model input: I=Itotai/N, A = A H W , E c e „ , L c , L a , G , P, T Incremental input:, T i n , P i n , P C 02,in, [HC03"]i„, [C0 2(aq)] i n, [HCOQ-Ji, AT (6.37), T o u t=T i n+AT, T a v e=(T i n+T o u t)/2 i> (6.35) 1 A P L G , P , h L (6.38 to 6.42) f K Q , K (Table 2.1) ^ k La, k Ga, kG/La (6.15,6.16) (6.43 to 6.48) [H +], [OK], [C02(aq)] (6.36, 6.14 to 6.16) 1 k F , ku, k M (6.17 to 6.19) JIL(6.21) k M (6.20) kl,eff, T e ff (6.25 to 6.29) 1 E r , i , E r , 2 (6.3 to 6.5); i 0 > 1 , i 0 j 2 (6.10, 6.11); a,,a2(6.12, 6.13); bu b 2 (6.9) i i , i 2 (6.24) Ii=Aii, I 2 =Ai 2 Solve 6.6, 6.7 and 6.22 for ji and j 2 A H 2 , A C 0 2 , A[HC0 3"], A[HCOCT] (6.31 to 6.34) CEj (6.23) pco2> ^//2 (gas law and material balance) pH (6.36) Incremental output: P o u t , Pc02,out, C E , , T o u t [C0 2(aq)] o u t, [HCOOlout, [HC0 3 - ] o u t , p H 0 U t Proceed to the next height increment Figure 6.2 Calculation procedures in each height increment 85 6 Reactor (cathode) Modeling A complete spreadsheet for an example calculation of Reactor B operating at 101 A , with 12 height increments, is shown in Appendix G. Modeling calculations of this type were carried out to match experiments over a wide range of conditions and the set of adjustable parameters (ki, k 2, E a > i and Ea,2) was fitted by the "least squares" technique, using the Excel Solver to give the "best" fit of modeled vs. measured formate C E shown in Figure 7.33. Table 6.1 lists values of the four adjustable parameters for best fit of the modeled and measured formate CE. Table 6.2 summarizes the corresponding input and calculated values of some diagnostic variables at the first height increment (i.e. reactor inlet) for the modeling run of Reactor B at 101 A , and Figure 6.3 shows the computed graphical profiles of critical process values along the reactor height for the modeling run of Reactor B at 101 A. . Table 6.1 Parameter values for best fit of modeled and measured formate CE Parameter k, m s"1 k 2 m s"1 E a , , kJ kmol"1 Ea,2 kJ kmol"1 Value 5 . 5 x l 0 - 4 4.5x10"" 6 . 0 x l 0 4 3 . 2 x l 0 4 86 6 Reactor (cathode) Modeling Table 6.2 Modeling values for Reactor B at increment 1 (N=12) Input values Calculated values Variable Units Value Variable Units Value a m"' 30xlO J [C02(aq)l _ kmol m" 0.14 D°C02 m V 1.96xl0"9 CE, % 89 d m 0.23xl0"3 E V(SHE) -1.07 d b m 50x10"6 En V(SHE) -0.74 AE V 0.29 E r 2 V(SHE) -0.42 E°cell V 1.0 h L - 0.40 Ecell V 3.9 'o.l kA m"2 6.3 xiO" 4 E a , i 60.0x10' >o,2 kAm" 2 7.2 xlO"* Ea,2 31.8xl03 j l kAm"2 0.26 G ml STP min"1 2000 J2 kAm" 2 0.03 H m 0.68 K 0 M kPa"1 3.3xl0"4 rHco 3 - i M 0.5 K , M 4.3 xl0" v Itotal A 101 k F m s"1 1.98xl0"s i kA ml 3.0 k G m s"' 4.8xl0"6 k, m s"1 5.5xl0"4 k M m s"' 2.04x10"5 k2 m s" 4.5x10"" k M a s"1 0.59 [KCl] M 2.0 pH - 7.1 L a ml min"1 40 U G m s"1 0.06 L c ml min"1 20 u L m s"1 0.002 P kPa(abs) 415 A P L G kPa m"1 512 PC02 kPa(abs) 412 T K 288 K e Sm"1 3.4 W m 0.05 111 V -0.34 e - 0.48 Tl2 V -0.66 Y - 0.7 Xeff m 0.36 xlO"3 Pi kg m"3 1030 kg m"1 s"1 1.5xl0"5 X m 3.2xl0" J kg m"' s"1 l . l x lO" 3 <P - 0.45 Pg kg m"3 7.6 N - 12 87 6 Reactor (cathode) Modeling Modeled condition vs reactor height 450 n 1 0 2 4 6 8 10 12 14 Figure 6.3 Local profiles of CE, pH, Pco2, and temperature for the modeling run of Reactor B at 101 A. The local profiles of CE, pH, Pco2, [CO?(aq)] and temperature shown in Figure 6.3 provide detailed information on those critical process variables, which is instructive in learning what is going on locally and determining the operating conditions that can improve the performance of the reactor. 88 7 Experimental Results and Discussion Chapter 7 Experimental Results and Discussion 7.1 Cathode selection and preparation As discussed in Section 4.3.1 (see Table 4.1), 7 categories of cathode material were studied in the present work. This section (7.1) reports the work on 4 cathode materials that were prepared but not used or extensively studied as ERC catalysts in the present work, specifically: - Nanostructured copper on graphite felt - Cu-Sn alloy on graphite felt - Copper mesh - Tinned graphite felt The other 3 groups of cathode material, i.e. - Tin coated copper mesh (tec), - Lead -Tin shot and granules, were each more extensively studied for ERC in the present work and are discussed in separate sections below, i.e. 7.3, 7.4 and 7.5. . 7.1.1 Nanostructured copper on graphite felt by liquid crystal templated electro-deposition The concept of liquid crystalline templating is based on the formation of structured liquid crystalline phases at high concentration of some non-ionic surfactants in aqueous solution [Attard, 1997 and Elliott, 1999]. These liquid crystal structures can be used to direct the deposition of nano-particulate metal catalysts onto suitable substrates. The first objective of the present study was to prepare a nano-structured copper electro-catalyst on graphite felt by depositing copper from a liquid crystal templating solution. Such a cathode material was expected to have a high specific surface area, with consequent high mass transfer capacity for 89 7 Experimental Results and Discussion electro-reduction of CO2. About 5 months were spent in trying to make uniform, penetrating and reproducible copper deposit onto graphite felt. Triton X-100 (octylphenol ethylene oxide condensate) and Brij series (Brij 56 (polyoxyethylene cetyl ether) and Brij 76 (polyoxyethylene stearyl ether)) surfactants were used to investigate the surfactant effects. The recipe and procedures for making surfactant-aided coppered felt cathode are as follows [Elliott, 1999 and 2002; Guerin, 2001]: Plating solution: (A) 60 wt % acidic conventional aqueous Cu plating solution containing 0.2 M C u 2 + and 0.3 M H2SO4; (B) 40 wt% aqueous Triton X-100 or Brij 56 or Brij 76; Preparation: (1) Add A to B; (2) Mix with glass rod for 30 min until "ice cream" state forms; (3) Heat to 317 to 328 K for 30 min with water bath; (4) Soak the graphite felt glued onto a platinum electrode into the above plating solution at the end of the heating process; (5) Stand the electrode and the solution at room temperature for 24 hours. Plating procedures: Electrodeposit copper onto the graphite felt for 5 to 30 minutes under current densities between 0.01 and 0.125 kA m" at room temperature. The high viscosity of the concentrated surfactant solution (semi-solid) trapped a lot of bubbles in the solution, which may have jeopardized the possible liquid crystal structure and surface coverage. But the surfactant did have some effect on the size and distribution of electrodeposited particles (see Figure 7.1). However, the extreme difficulty of dealing with the semi-solid liquid crystal plating solution and the failure in obtaining uniform and penetrating deposition in the present work made us doubt the possibility of large-scale application of this kind of cathode. Therefore, no ERC was attempted with this "nano-structure" copper graphite felt in the present work. 90 7 Experimental Results and Discussion SE WD23.1.HB 20.0kV xl .5k 30un a b Figure 7.1 Electro-deposition of Cu onto graphite felt: a: with surfactant 40 wt % Triton X -100; b: without surfactant. Two pictures are both in a 30 urn scale. 7.1.2 Cu-Sn alloy on graphite felt by conventional electro- and electroless deposition In the process of preparing coppered graphite felt cathode (above), it was found that the copper layer became dark brown (oxidized) very easily, therefore attempts were made to fabricate a Cu-Sn alloy deposited graphite felt cathode which would not only show higher oxidization resistance, but also was expected to provide good catalytic activity for ERC, as indicated by Watanabe's work using a Cu-Sn alloy as the catalyst for ERC to produce HCOO"andCO [Watanabe, 1991, 1991 and 1993]. Investigations into making Cu-Sn alloy electro-deposited felt were carried out with the help of the S E M images and E D X analysis to see how particle size, surface morphology and alloy composition were affected by the depositing conditions, including temperature, current density, ratio of Cu and Sn ion concentrations in the plating solution and plating time. An alkaline pyrophosphate plating bath as shown in Recipe 2 of Appendix E [Watanabe, 1993] was used. It was found that the deposited alloy composition (Cu/Sn ratio) directly related to the surface roughness and particle size, i.e. the lower the ratio, the rougher the surface, and the smaller the particles; and that higher plating current density and lower temperature tended to result in a lower deposit Cu/Sn ratio. But when the deposit Cu/Sn ratio 91 7 Experimental Results and Discussion was lowered to a certain point (~1), two phases of Cu and Sn, instead of Cu-Sn alloy formed. Uniform and rough (high surface area) depositions were achieved at temperatures between 298 and 318 K, current density of 50 to 100 A/m 2 and 0.33 Sn to Cu molar ratio in the plating solution (see Figure 7.2). But the deposit uniformity was only observed at the surface layer of the felt (< 0.5 mm), which means the penetration was bad. This bad penetration might be a result of the screen shell effect (so called "black core") caused by the electronic conductivity difference between metal and felt [Gan, 1994 and Wan, 1997]. Figure 7.2 Cu-Sn alloy catalysts obtained through electro-deposition. A l l the pictures are in 5 pm scale 92 7 Experimental Results and Discussion Electroless deposition of Cu-Sn alloy onto graphite felt was also investigated with an expectation of better penetration. Graphite felt was activated using both sensitizing and nucleation procedures [Plessey, 1980] (see Appendix D for activation procedures), and then a recipe that contained both Sn salt and Cu salt precursors [Bestetti, 2002] (Recipe 1 of Appendix E) was used to plate Cu-Sn alloy onto graphite felt. In comparison with electrodepostion, the surface structure through electroless deposition was not as rough as those obtained through electrodeposition (see Figure 7.3), but the penetration into the felt was better than that in electrodeposition. In the present work no ERC was attempted with either the Cu-Sn electro- or electroless deposited graphite felts due to the bad penetration in the former case and the unstable Sn-Cu deposited layer (weak bonding) in the latter case, along with the knowledge that any uncovered carbon would be a source of hydrogen production. b Figure 7.3 Electroless deposited graphite felts, a: electroless deposition of Cu (100 um); b: electroless deposition of Cu-Sn (50 um). 7.1.3 Copper mesh Copper mesh 30 # from ARGUS (USA) with a specific surface area of 7000 m 2 m"3 (see properties of copper mesh in Table 4.3) was used to carry out ERC in Reactor A. The 93 7 Experimental Results and Discussion detected gas products from the cathode were mostly hydrogen with a low concentration of hydrocarbons (<5 %vol). No liquid product (formate) was detected. Preliminary experimental results showed that current efficiencies with copper mesh were low, with many products (low M W hydrocarbons). Our research partner, N R C Ottawa, who have been studying ERC in a batch reactor with copper foil cathode, also reported that current efficiencies for methane and ethylene of up to 25% could be achieved only with extremely careful pretreatment of both copper foil and electrolyte. Quick deactivation of the copper cathode was also reported by NRC, at 30 min. Such low current efficiency and rapid deterioration would hinder the practical industrial application of ERC. Subsequently the focus of the catalyst (cathode) materials in the present research turned to the group of Pb, Hg and especially Sn, which were reported [Todoroki, 1995; Mahmood, 1987; Mizuno, 1995] to exhibit high selectivity of the single liquid phase product - formate. 7.1.4 Tinned graphite felt by electroless deposition Tin coated graphite felt cathode was tried for ERC in the present work with the purpose of taking advantage of the high specific surface area ( l x l O 4 to 2 x l 0 4 m 2 m"3) and high porosity (ca. 0.9) of graphite felt (see Table 4.2 for the properties of graphite felt). After being activated as described in Appendix D, the graphite felt was electrolessly deposited with copper first, then tin was electrolessly plated onto the coppered felt. The recipes for electroless copper [Plessey, 1980] and tin plating are given in Appendix E. Electroless deposition made it possible to deposit Cu and/or Sn on a big sheet of graphite felt (e.g. 30 mm by 150 mm) with fairly uniform and well-penetrated depositions of Cu and/or Sn. The activity of this cathode for ERC was tested twice with Reactor A under replicate conditions. The formate current efficiency (CE) obtained, along with other 94 7 Experimental Results and Discussion parameters at 10 min of operating time are listed in Table 7.1 and the effect of cathode age is given in Figure 7.4. Table 7.1 Preliminary experimental results on tin-coated graphite felt Run No Current Ecel l Pcathode Formate CE pH A V kPa (abs) % 1 6 4.14 145 7.51 64 2 6 3.84 150 7.66 67 Operating conditions: Reactor A . catholyte =0.45 M KHC0 3 +1 M KC1; catholyte flow rate = 36 ml/min; gas flow rate = 650 ml STP/min; temperature =293 ~ 297 K. Formate CE vs time 10 20 30 Time (cathode age), min 40 50 Figure 7.4 Effect of cathode age with tin-coated graphite felt cathode. Conditions as under Table 7.1. It is seen from Table 7.1 and Figure 7.4 that tin-coated graphite felt showed a high activity for the production of formate, i.e. 67 % at 10 minutes of operation, but suffered a quick drop of CE, i.e. about 10 % decrease within 20 min. This drop in CE was probably caused by the loss of tin from the surface due to the weak binding of the electroless deposit (see Figure 7.3) as observed by visual examination of the cathode at the end of the experiments. Also, a strong rotten-egg smell from H2S occurred during the start-up. The H 2 S 95 7 Experimental Results and Discussion might come from the electro-reduction of residues of the thiourea used as a reducing agent in the electroless deposition. 7.2 Anode materials In the selection of anode materials, ERC experiments were carried out in Reactor A under the following conditions. Cathode: tin-coated copper mesh (tec) 30*; - Current: 6 A (1.33 kA m 2); Catholyte flow rate: 20 ml/min; - C 0 2 gas flow rate: 180 ml STP/min at 100% C 0 2 ; - Catholyte: 0.45 M K H C 0 3 ; Cathode age: 10 min; - Cathode feed: 115 ~ 125 kPa (abs); - Temperature: 297 ~ 301 K. Table 7.2 lists all the experimental results of ERC on the selection of anode materials, including many combinations of anode plate (feeder), anode ss mesh and anode plastic spacer. 96 7 Experimental Results and Discussion Table 7.2 Experiments on anode materials in Reactor A Run No Anode Membrane Spacer Ecell V CE % 1 Pt Nafion 350 2 layer Netlon screens 8.42 30 2 Pt + 2 layers ss40# Nafion 350 No spacer 3.83 35 3 Pt + 2 layers ss40* Nafion 117 No spacer short circuit 4 Pt Nafion 350 2 layer plastic screens 4.87 42 5 Pt+ 1 layer ss 10* Nafion 117 Diaphragm 4.13-6.22* 41 6 Pt + 1 layer ss30# Nafion 350 No spacer 3.54 30 7 Pt+ 1 layer ss 10* Nafion 350 No spacer 3.73 30 8 Pt+ 1 layer ss 10* Nafion 117 2 layer fly screens 3.88 38 9 SS+ 1 layer ss 10* Nafion 117 1 layer fly screen short circuit 10 SS+ 1 layer ss 10* Nafion 117 2 layer fly screens 4.39 40 * The reactor voltage for run No 5 kept increasing after the start, therefore a range of cell voltage is given. Pt =Platinized titanium sheet; SS = stainless steel sheet; sslO*, ss30* and ss40* are stainless steel meshes of mesh count 10, 30 and 40 per inch, respectively. 7.2.1 Anode plate (feeder) As discussed in Section 4.3.2 both the platinized titanium plate (Pt) and 316 stainless steel plate (SS) were tried for ERC in Reactor A. Comparing run No 8 and No 10 in Table 7.2, it is seen that SS yielded a slightly higher formate current efficiency than Pt, i.e. 42 % vs 38 %, but SS resulted in a higher cell voltage, which might cost more energy. In consideration of requirements for a practical industrial application, such as low capital cost, good electrical conductivity, and good physical and mechanical stability, 316 stainless steel sheet was decided to be a better choice than platinized titanium sheet. Therefore, in Reactor B a 316 stainless steel sheet anode feeder was employed. 7.2.2 Anode mesh Stainless steel mesh was also employed together with the stainless steel sheet to provide better current distribution and lower the reactor voltage (see Table 4.5 for the 97 7 Experimental Results and Discussion characterization of anode mesh). Comparing run No 2, 6 and 7 in Table 7.2 suggests that a two-layer stainless steel mesh 40 # provided the highest formate C E among the anode meshes studied, including a single-layer stainless steel mesh 10# and a single-layer stainless steel it a mesh 30 when a Nafion 350 membrane (reinforced), was used. However, ss 40 caused electric short circuit when a Nafion 117 membrane was used, probably due to the filament of the ss 40 # being too fine and protruding through the membrane to contact the cathode. Balancing on all those factors, 1 layer of ss 10# was chosen for the anode mesh for most of the ERC experiments in the both Reactor A and Reactor B. 7.2.3 Anode spacer In the early stages of experimental work on ERC (experiments in Section 7.3.1), two layers of plastic screen were used as the spacer. Experimental results (also see run No 4 in Table 7.2) show that this kind of plastic screen worked quite well for short term operations. However, the plastic screen had movable filaments and tended to deform after being compressed a couple of times, resulting in the spreading out of the silicon glue which not only partially blocked the flow path of the electrolyte and the active surface area of the cathode mesh, but also increased the reactor (anode side) pressure drop. So investigations on other spacers for better long term performance were carried out. Other spacers tested were 10# Netlon greenhouse shading (polyolefin), micro-porous polyethylene diaphragm and fly screen (PVC) (see Table 4.6 and 4.7 for the properties of spacers). The experimental results listed in Table 7.2 show that the run with Netlon spacer (run No 1) not only yielded a low formate CE but also resulted in a very high cell voltage; using a micro-porous diaphragm alone as the spacer resulted in an ever increasing cell voltage (run No 5); and using one layer of fly screen (PVC) led to the short circuit of the cell (run No 9). Thus, 2 layers of fly screen provided the best performance, i.e. high formate CE and long term stability. Figure 7.5 shows the specific reactor configurations used in the listed corresponding sections of experiments. 98 7 Experimental Results and Discussion -Configuration 1 (used in experiments of section 7.3.1): Anode side Separator Cathode side I \ I ~i Pt sheet + plastic screen + Nafion 117 + tec mesh(s) + tec sheet -Configuration 2 (used in experiments of section 7.3.2 to 7.3.6): Anode side Separator Cathode side i I I 1 ss sheet + ss mesh + P V C screen + Nafion 117 + tec mesh(s) + tec sheet —Configuration 3 (used in experiments of section 7.4): Anode side Separator Cathode side : i - r n ss sheet + ss mesh + P V C screen + Nafion 117 + Pb shot/granules + Pb plate —Configuration 4 (used in experiments of section 7.5): Anode side Separator Cathode side I 1 I 1 ss sheet + ss mesh + P V C screen + Nafion 117 + Sn shot/granules + Sn foil on Cu plate -Configuration 5 (used in experiments of section 7.6): Anode side Separator Cathode side I I i I ss sheet + ss mesh + P V C screen + Nafion 117 + Sn granules + Sn plate Figure 7.5 Experimental reactor configurations 99 7 Experimental Results and Discussion 7.3 Tin-coated copper (tec) mesh cathode in Reactor A 3-D tin-coated copper mesh (tec) was made by electroless deposition of tin onto copper mesh. The copper mesh was first etched in 7 wt % nitric acid and washed in distilled water then immersed for 3 minutes at 46 °C in an electro-less acidic tin plating bath consisting of 0.02 M stannous sulphate (SnSCM) and 0.22 M sulfuric acid in water with thiourea as the reductant (see recipe 1 of Appendix E for electroless deposition of tin). Two types of copper mesh, mesh count 30# and 60#, have been used as the mesh substrate to provide different specific surface areas (The specifications of the two meshes are given in Table 4.3). Before plating, the mesh had to be masked on the margins by silicon glue to avoid leaking of electrolyte while assembled (see Figure 7.6). Silicone glued margins Catholyte flow path Figure 7.6 Silicone glued tin-coated copper mesh Tin coated copper (tec) mesh cathode showed much higher activity for ERC than a plain copper mesh in preliminary runs, i.e. 50 % (ERC to formate) vs 5 % (ERC to CO, methane and ethylene) under the same conditions. Therefore, most of the basic experimental work on ERC was done with tec cathodes in Reactor A to investigate several major process variables, either in factorial or in parametric experiments. The process variables studied in this section are listed in Table 7.3. Details of the individual experimental designs are given under Sections 7.3.1 to 7.3.6. The trickle-bed reactor used here operated in nearly plug flow [Walsh, 1993; Oloman, 1979] with consequent changes in process conditions along the reactor height. Table 7.4 summarizes typical values of these changes in the cathode at 6 A, which was the most 100 7 Experimental Results and Discussion commonly used current in the present section (Section 7.3). There are also gradients in local cathode potential and current density, but these variations were not measured and are not given here. Gradients of process conditions in Reactor B at 101 A were calculated by the model and are shown in Appendix H. . It should be noted that, unless otherwise indicated all cathode conditions (except pH and temperature) of ERC in the present work are those at the reactor inlet. The pH and temperature are the catholyte outlet values. Table 7.3 Variables studied with tin coated copper mesh cathodes in Reactor A . Variables Units Range Current Amp 1-10 Cathode age (operating time) min 10-180 CO2 fraction in the feed ' Vol % 16-100 Temperature K 288-328 Catholyte composition - • (K + , Na+),(C1", C0 3 " z , HCO3-) Catholyte ionic conductivity Sm" 1 11-194 3D cathode thickness mm 0.4 to 1.9 (1 - 4 mesh layers) 3D cathode specific surface mz m"3 7000 - 14000 (30" - 60") Table 7.4 Typical changes along reactor length at 6A Condition [in cathode] Unit In Out Total pressure kPa(abs) 120 101 Temperature K 286 288 Formate cone. M 0 0.08 Liquid flow ml min"1 20 20 Gas flow ml min" 165 150 pH - 7.5 8.5 CO2 pressure kPa(abs) 118 70 HCO3" concentration M 0.45 0.60 Cathode residence time s L*=5, G**=0.5 *L =liquid, **G =gas. 101 7 Experimental Results and Discussion 7.3.1 Current, CO2 concentration in the feed gas and cathode age This section presents the results of experiments in Reactor A using tin-coated copper mesh (tec) 30# with the 3 variables: current, operating time (i.e. cathode age) and CO2 concentration in the feed gas, first in the 2 3+l factorial design outlined in Table 7.5 and then in parametric experiments that examine the effect of each of these variables while the others are held constant. In all of these runs other conditions expected to affect the process were kept constant or in the ranges shown in Table 7.6. Table 7.5 Factorial variables and levels on current, cathode age and C Q 2 fraction. Run No. I: Current, A t: Time, min y: yco2, % 1 + 6 + 30 + 100 2 + 6 + 30 33 3 + 6 10 - + 100 4 + 6 10 33 5 2 + 30 + 100 6 2 + 30 33 7 2 10 + 100 8 2 10 33 cp cp 4 cp 20 cp 67 Note: the cathode age (in min) means the operating time for a freshly prepared cathode. The same for all the "cathode age" data mentioned in this thesis. 102 7 Experimental Results and Discussion Table 7.6 Other process conditions for runs in Table 7.5. Condition Units Value Reactor configuration - Configuration 1 and 2 in Figure 7.5 Cathode bed material - 1 layer, 30 s mesh tinned copper Cathode feeder material - tinned copper plate Total pressure in cathode kPa(abs) 120-160 Total pressure in anode kPa(abs) 104-160 Catholyte temperature K 296-303 Catholyte feed composition - 0.45 M K H C 0 3 Anolyte temperature K 290-303 Anolyte feed composition - 1.0 M K O H Anolyte feed flow m V 5.0x10"' (i.e. 30 ml min"1) Catholyte feed flow m V 3.3x10"' (i.e. 20 ml min"1) Cathode gas feed flow m j s"1 STP 3.0xl0"b (i.e. 180 ml STP min"1) Catholyte pH - 7-8 7.3.1.1 Factorial experiments The factorial runs were carried out in random order with three replicates of the centre-point. Replicates were also made for two corner-points and these data were pooled to calculate the confidence level of the results. Including centre-points and replicates 16 experimental runs were carried out in this factorial set. In all of these runs formate and hydrogen were the dominant cathode products, accounting for 95 to 99 % of the current. Methane and ethylene were detected but negligible (<0.06 vol %) and the current efficiency for carbon monoxide ranged from 0 to 5 %. The principal results of these runs with respect to formate production are shown both in the factorial cube of Figure 7.7 and Table 7.7. 103 7 Experimental Results and Discussion 55 (17) 100 C0 2 cone, Vol% 71 (22) 33 54 (17) 60 (19) 47 (29) Current, A 45 (42) 50 (46) 24 (23) 30 24 (22) 10 Time, min. Figure 7.7 Factorial results for HCOO- production. CE % and ([HCOO"] mM) Table 7.7 also gives the check on the integrity of these data by the cathode current efficiency and overall carbon balances (see Appendix A for full-cell reactions and carbon balance). The data in Table 7.7 give an average carbon closure of 101 % with a standard deviation of 3 %. By the Students' t test (95 % confidence level) the average carbon closure is not statistically different from 100 % [Murphy, 1977] so it appears the analyses give an adequate account of the reaction products. The total CEs in Table 7.7 were calculated by summing the CEs for all the measured products. Anomalously high total CEs for two runs were probably caused by the errors in the Orsat analysis for CO2, which is problematic for CO2 concentrations over about 80 vol %. 104 7 Experimental Results and Discussion Table 7.7 Current efficiencies and carbon balance for 16 runs of the factorial set Run No* Ecel l [HCOO-] CE, % Carbon closure V m M HCOO - H 2 CO Total % 1 5.63 41.7 45 61 2 108 100 2 5.67 26.3 25 80 3 108 102 2-rep 5.61 21.7 23 67 4 94 104 2-rep 5.57 21.0 23 75 0.0 98 101 3 5.80 46.3 50 52 3 105 97 4 5.90 25.3 26 76 0 102 102 4-rep 5.83 21.0 23 86 0 109 105 4-rep 5.53 22.3 24 67 4 95 106 4-rep 5.55 20.3 23 80 3 106 105 5 3.85 17.0 55 64 0 119** 99 6 3.90 16.7 54 43 0 97 102 7 3.88 22.0 71 63 0 134** 97 8 4.07 18.7 60 32 4 96 104 cp 4.70 28.0 45 58 5 108 99 cp-rep 5.00 31.0 50 44 3 97 97 cp-rep 4.84 28.7 46 53 4 103 101 Mean (standard deviation) 105(10) 101(3) T h e conditions for Run No 1-8 and cp correspond to those listed in Table 7.5 and other conditions are in Table 7.6. **A high "Total C E " in two runs was suspected to be the result of errors in the Orsat analysis at high C 0 2 levels (>80 vol %) which led to an overestimate of the hydrogen content of the product gas. The formate CE was not affected by this Orsat error. The analysis of variance summarized in Table 7.8 shows significant effects (at the 95 % confidence level) on the formate CE for each of the main factors (I, t and y) plus the (Iy) interaction, with an insignificant curvature over the experimental space. 105 7 Experimental Results and Discussion Table 7.8 Summary of the calculated factorial effects Effects HCOO - CE % 95% confidence intervals [HCOO] m M 95% confidence intervals Main I -24.4 14.7 effects t -6.9 -2.7 y 14.6 ±3.1 11.6 ±3.8 Interaction it 0.9 0.8 effects ty -2.2 -2.1 iy 8.2 9.8 Curvature 0.7 ±2.8 -3.3 ±3.6 Table 7.8 shows that current had the strongest main effect on both the formate CE and [HCOO"], but in opposite directions, i.e., the effect of increasing the current from 2A to 6A, averaged over all levels of C 0 2 concentration and cathode age, was to lower the CE by 24% while raising the HCOO" product concentration by 15 mM. The main effects of cathode age on both the CE and [HCOO"], respectively -6.9% and -2.7 mM, indicated deactivation of the cathode. The "poisoning" of copper cathodes for CO2 reduction has been widely reported [Chaplin, 2003; Jitaru, 1997] and deactivation of tin cathodes has also been observed [Koleli, 2003; Kapusta, 1983]. In the present work the SEM images and E D X spectra of new and used cathodes in Figure 7.11 show a progressive loss of tin from the cathode surface that probably accounted for the effect of cathode age. The positive effect of C02(g) concentration in the gas feed on both CE and [HCOO"] (14.6 % and 11.6 mM, respectively) reflects the fact that C02(aq) was the electro-active species and an increase in its level promoted both the intrinsic kinetic and mass transfer processes of the primary cathode reaction (reaction 6.1). This effect was reinforced in the positive (Iy) interaction which showed how a higher current (a.k.a. superficial current density) was supported by an increased CO2 concentration. 106 7 Experimental Results and Discussion The effects of the interaction between current and cathode age (It), and that between CO2 concentration and cathode age (ty) were statistically insignificant. Nevertheless the weak effect of the interaction between the CO2 concentration and cathode age suggests that the cathode deactivation rate was decreased by a lower CO2 concentration and/or that the primary cathode reaction came under a mass transfer constraint when the CO2 feed concentration dropped to 33 vol %. 7.3.1.2 Parametric experiments Current Figure 7.8 shows the effect of current on the formate current efficiency with fixed cathode age (10 min) and CO2 feed concentration (100 vol %). The formate CE values increased monotonically with decreasing current, reaching 86% at 1A, which corresponded 9 9 to a superficial current density of 0.22 kA m" (22 mA cm - ). The HCOO" concentration reached the maximum of 45 m M at 6A (1.33 kA m"2 ) where the formate CE was 50 %, and « • 9 corresponded to a partial current density for CO2 reduction to formate of 0.67 kA m" , with a reactor voltage of 5.80 Volt and specific energy of 621 kWh kmol"1 of formate. CE vs current 100 -i o 20 0 -I , , , 1 0 2 4 ' 6 8 10 Current, A Figure 7.8 Effect of current on formate CE. yC02: 100 vol%; cathode age: lOmin. 107 7 Experimental Results and Discussion C 0 2 feed concentration Figure 7.9 shows the effect of CO2 feed concentration in the gas phase on the formate CE under fixed current (6A) and cathode age (lOmin). Here the C E increased monotonically with CO2 concentration (a.k.a. CO2 partial pressure), as expected in this process where CO"2(aq) was the primary electro-active species. CE vs C02 feed concentration 100 -1 C02 feed gas concentration, vo l% Figure 7.9 Effect of CO2 concentration on formate CE. Current: 6A; cathode age:10min. Operating time Deactivation of the cathode was observed during the factorial runs, and more experiments were conducted to investigate this effect. Figure 7.10 shows the "deactivation" patterns under different operating conditions while Figure 7.11 and Table 7.9 present results of SEM and E D X measurements on the new and used cathodes. 108 7 Experimental Results and Discussion CE vs cathode age 80 -i * : 60 LU O a> 40 0 20 40 60 80 100 120 140 160 180 Cathode age, min Figure 7.10 Effects of cathode age on formate CE. A 2A and 100% C 0 2 gas feed; • 6A and 100% C 0 2 gas feed; *6A and 33% C 0 2 gas feed. In Figure 7.10 the current efficiency for formate droped more sharply with time under 100 vol % C 0 2 than under 33 vol % C 0 2 . This effect of C 0 2 concentration and time is reflected in the factorial data of Figure 7.7 and might be due to an increase in the rate of cathode deactivation with C 0 2 pressure, and/or to the fact that at 33 vol % C 0 2 the primary reaction was under a mass transfer constraint and thus relatively insensitive to the conditions of the cathode surface that dictated the intrinsic reaction kinetics. In Figure 7.11, the top two photographs show the new tinned copper mesh at 1 mm and 5 um magnification, respectively, and the bottom two the used tinned copper mesh with 100 minutes of operating time at 1 mm and 5 um magnification, respectively. The SEM images at high magnification, combined with the E D X analysis (below) show that the fraction of white spots, which were copper, increased with time. This observation matches the E D X analyses in Table 7.9, which show the Sn/Cu ratio on the surface dropping with time, at a rate that apparently increased with the C 0 2 concentration in the feed gas. Tin was being lost and copper was being exposed on the cathode surface while the reactor was in operation. This loss of tin may be caused by erosion due to hydrogen generation on the cathode surface and/or the formation of tin hydride (SnFLj) or tin-formate complexes. The loss of tin is almost certainly the cause of the "deactivation" of the cathode with respect to 109 7 Experimental Results and Discussion formate generation, since the hydrogen over-potential of copper is about 0.3 V less negative than that of tin. SE 1mm 'WM4..7.-CT 20. OW xT.Ok Sum 1mm magnification 5um magnification (b) Used tinned copper mesh with an age of 100 min Figure 7.11 S E M pictures of (a) new tec 30# mesh (b) used tec 30# mesh with 100 min age 110 7 Experimental Results and Discussion Table 7.9 E D X analyses for tin-coated copper cathode mesh 30# No Mesh description Catalytic activity Sn/Cu ratio on the surface 1 New highest 1/1.05 2 30 min under 33 % C 0 2 at 6A high 1/1.74 3 30 min under 100% C 0 2 at6A middle 1/2.10 4 100 min under 100% C 0 2 at 6A low 1/5.10 7.3.1.3 Mass transfer constraint. The first four rows of Table 7.10 lists the same experimental data as Figure 7.8, i.e. the effect of current on current efficiency. To obtain some information on how the current density of C 0 2 reduction changed with cell voltage, the partial current of the products HCOO - and CO are calculated to obtain the total current density of C 0 2 reduction (the last three rows of Table 7.10). The calculated CD of C 0 2 reduction vs cell voltage is presented in Figure 7.12. Table 7.10 Effect of current on current efficiencies. Current, A 1 2 4 6 8 ECeii, V(abs) 3.12 3.85 4.68 5.80 6.21 CE of HCOO", % 85.6 70.8 61.6 49.6 36.4 CE o f C O , % 0 0 11.6 2.9 3.5 I HCOO-, A 0.86 1.42 2.46 2.98 2.91 I co, A 0 0 0.46 0.17 0.28 CD c o 2 , kA m"2 0.19 0.32 0.65 0.70 0.71 Note: I HCOO- =CEHcoo-xCurrent (A) I c o =CECoxCurrent (A) CD C O 2=10- 3x(I HCOO- +Ico)/Area (kA m") Area=0.0045 m2 (45cm2) 111 7 Experimental Results and Discussion E ,0 < E o" tr LU O Q O 80 70-60 50 40 30 20 10 0 CD vs Ecel Mass transfer limiting current density 2 3 4 5 Reactor voltage (Ecell), V Figure 7.12 CD of C 0 2 reduction vs. cell voltage with 100% CO2+10 min cathode age. It is seen from Figure 7.12 that C 0 2 reduction with 1 layer of tec mesh 30# reached the C 0 2 mass transfer limit at about 6A (i.e. 0.7 kA m"2) or a reactor voltage higher than 5.80 V, with 100 % C 0 2 feed under the experimental conditions (listed in Table 7.6), in which the anode resistance was relatively constant so the inflection in the reactor voltage curve was caused by change in the cathode potential. There are several ways to push up the C 0 2 mass transfer limiting superficial current density, such as using a cathode with higher specific surface, increasing the gas and/or liquid loads, operating at higher C 0 2 partial pressure and enhancing the solubility of C 0 2 , etc. 7.3.2 Temperature, (current and C 0 2 pressure) Temperature is one of the most important variables with respect to engineering industrial scale electro-chemical reactors operating at current densities of 1-5 k A m"2, in which the energy balance can drive the electrolyte temperature up to the boiling point. 112 7 Experimental Results and Discussion Increasing temperature has three primary effects on reactor performance in CO2 reduction (see Chapter 6): (i) . Increase of the exchange current densities of competing reactions by a ratio that depends on the relative activation energies. (ii) . Decrease in the solubility of C 0 2 in the catholyte. (iii) . Increase in the mass transfer coefficient of CO2 to the cathode surface. Secondary effects of increasing temperature include: changes in the pressure gradient, liquid hold-up, electrolyte ionic conductivity, reactor voltage, current and potential distribution, etc. Temperature is thus engaged in complex interactions with current density and CO2 pressure that determine the formate current efficiency and specific energy consumption. The two level, three factor (2+1) factorial experimental arrangement, summarized in Table 7.11-was employed to examine some of these effects. To separate the effect of temperature on the intrinsic kinetics from that on the solubility of CO2 the experiment was designed so the solubility of CO2 in water at the high temperature and high CO2 pressure (328 K , 100%) matched that at the low temperature and low CO2 pressure (288 K, 40%) (see Appendix F for the determination of the factorial points). Table 7.11 Factorial variables and levels on current, temperature and C Q 2 fraction. Variable Symbol Units Level High Center Low Current I A 6 4 2 Temperature T K 328 308 288 CO2 cone. y % 100 70 40 Operating conditions: Reactor A; cathode: 1 layer of tec 30#; cathode age = 10 min; gas flow =180 ml STP min"1; catholyte flow = 20 ml min"1; catholyte = 0.5 M K H C 0 3 ; pressure: = 133 kPa(abs); pH = 7.5-9.5. Figure 7.13 shows the experimental results of Table 7.11 in a factorial cube, with the statistical analysis of effects in Table 7.12. 113 7 Experimental Results and Discussion 85% 100 yco2, % 96% 40 52% 50% 65% Current, A 30% 49% 18% 328 19% Temp, K 288 Figure 7.13 Formate current efficiencies for factorial runs in Table 7.11. Other conditions as under Table 7.11. Table 7.12 Factorial effects of current, temperature and C 0 2 \ fraction on the formate CE. Effects Current efficiency % 95 % confidence intervals % Main I -45 ±5 T -12 y 27 Interaction IT 2 Ty -6 iy 4 Curvature 1.0 ±6 Confidence intervals are estimated from pooled replicates. The data in Table 7.12 confirmed the conclusion from Section 7.3.1 that lowering the current and increasing the C 0 2 volume fraction in the feed gas increased the formate CE, and this conclusion was shown at both levels of temperature. The more complex effect of 114 7 Experimental Results and Discussion temperature can be examined by processing some of the experimental data to provide the information in Table 7.13. In Table 7.13, comparing run 1 with run 2 and run 3 with run 4, exposes the combined effects of temperature and mass transfer on CE at 6A and 2A, respectively, without the confounding effect of CO2 solubility. This is because the levels of T and yco2 for both runs at each current were arranged for respectively the same solubility of CO2. Table 7.13 Effect of temperature coupled with mass transfer. Current A Run No Temperature K yco2 % CE % Relative increase of CE % 6 1 328 100 30 (30-19)/19=58% 2 288 40 19 2 3 328 100 85 (85-65)765=31% 4 288 40 65 The fact that CE increases with increasing temperature at both currents suggests that increasing the temperature favored the generation of formate by the intrinsic kinetics, as observed by Koleli and Balun under 5000 kPa(abs) CO2 pressure [Koleli, 2004]. However, the relative increase of CE at 6 A (58%), was higher than that at 2 A (31%). This effect may be explained by the fact that the reaction kinetics dominated the process at 2 A while mass transfer was more important at 6 A. Thus the dual effect of temperature (i.e. on kinetics and on mass transfer) resulted in a higher impact of temperature on formate CE at 6 A than at 2 A. This is in agreement with the conclusion on the mass transfer limiting current drawn in Section 7.3.1.3. The overall factorial results (Table 7.12) show that temperature generally had a negative 12% main effect on formate current efficiency, which indicates that the kinetic benefit of elevated temperature was outweighed by the corresponding drop in CO2 solubility, at the CO2 pressures used here. The negative effect of temperature has major implications for the reactor scale-up. 115 7 Experimental Results and Discussion The only significant interaction was that between T and yco2- Temperature interacted negatively with yco2 (-6%), which suggests that the benefit of increased yco2 was slightly countered by increasing the temperature. 7.3.3 Electrolyte species To explore the effect of some common electrolyte species, the cathodic reduction of CO2 was investigated in aqueous solutions of K H C O 3 , K2CO3, NaHCC»3, Na2CO"3, KC1 and NaCl as well as in some combinations of these electrolytes. Two sets of factorial experiments were carried out here, along with one parametric experiment. 7.3.3.1 Factorial experiments on K \ Na + , CI" and CO3 2" The 2 2+l factorial experiment, summarized in Table 7.14, was carried out in an attempt to see the effects of cation species (K + and Na +) and anion species (CI" and CO32") on formate CE. The catholyte concentrations were all held at 0.5 N and one layer of tec 30# at a low current (2A) was employed as cathode for all the runs to minimize the effects of conductivity and CO2 solubility in the catholyte. The results of the factorial runs are given in Figure 7.14 in which the formate CE at the centre-point is the average of two replicate runs. The factorial effects of cations and anions on formate CE are summarized in Table 7.15. Table 7.14 Factorial variables and levels on electrolyte species. Variable Symbol Level High Low Centre Cation x , K + (0.5 M) N a + (0.5 M) Mixture* Anion e : 1 x 2 CI" (0.5 M) CO3 2"(0.25 M) Mixture Mixture = '/4(0.25 M K2CO3+0.25 M Na2CO3+0.5 M KC1+0.5 M NaCl); ionic conductivity = 3.4 to 6.5 S m"1. Operating conditions: Reactor A; current = 2 A; y C 0 2 = 100%; cathode: 1 layer of tec 30*; cathode age = 10 min; gas flow = 180 ml STP min"1; catholyte flow = 20 ml min"1; temperature = 299 K; pressure = 127 kPa (abs); pH = 6.5 ~ 8.0. 116 7 Experimental Results and Discussion E c e„=3.34 V Ece.i=3.43 V KC1(0.5M) 56% NaCI(0.5M) 28% Mixture cp 76% K 2 CO 3 (0.25M) 72% Na 2CO 3(0.25M) 71% Figure 7.14 Formate current efficiency and reactor voltage for factorial runs in Table 7.14. Other conditions as under Table 7.14. Table 7.15 Factorial effects of electrolyte species on formate current efficiency. Effects Current efficiency % 95 % confidence intervals % Main X t : cation -15 ±5 X 2 : anion -30 Interaction X i X 2 -14 Curvature 19 ±4 Confidence intervals are estimated from pooled replicates. The results of Table 7.15 show that, on average, changing the cation species from K + to N a + decreased the CE by 15%. This observation is qualitatively in agreement with previous reports on the effect of alkali metal cations on the reaction selectivity [Hori 1982; Murata, 1991]. On the other hand, changing the anion feed species from CO3 " to CI" resulted in a decrease of CE by 30% on average. The negative effect of chloride relative to carbonate ions here may be due to the inability of CI" to buffer the cathode surface pH [Gupta, 2005]] and/or to the presence of adsorbed tin-complexes [Sillen, 1964; Pourbaix, 1974] that shift the 117 7 Experimental Results and Discussion kinetics in favour of hydrogen evolution. It can also be seen from the comparison between the results of runs with 0.25 M K2CO3 and 0.25 M Na2CC>3 that CO3 2 - was so much superior to CT in terms of formate CE at 2 A that when C 0 3 2 " was used, the cation ( K + or Na +) had no effect. The high curvature shown in Table 7.15 reflects non-linearity in the experimental system, which suggests a good possibility for optimization around the catholyte composition. 7.3.3.2 K2CO3 vs. K H C O 3 in the catholyte The results of Figure 7.14 do not necessarily mean that alkali carbonate is a good choice of catholyte for electro-reduction of CO2. This situation is complicated by the speciation between CO3 ", HCO3" and CO2 (aq) with respect to the pH, which is considered to be a key factor for electrochemical reduction of CO2 [Hori 1982; Sanchez-Sanchez, 2001; Gupta, 2005; and Section 2.4]. Before further discussion, it would be convenient to rewrite the previously discussed reactions: Electro-chemical reactions at the cathode: Primary reaction: 1. C02 (aq) + H20 + 2e' -> HCOO' + OH' (7.1) Secondary reaction: 2. 2H20 + 2e~ -» H2 + 20H' (7.2) Thermo-chemical reactions in the neighborhood of the cathode: 3. CO]' + C02 (aq) + H20^ 2HC0'3 (7.3) 4. HCO'3 + OH' -> CO]' + H20 (7.4) In the present system (Figure 4.1), the' mixture of gas and liquid electrolyte feed passed through a 6 m long, 5 mm internal diameter tubing before entering the reactor. Absorption of CO2 by the catholyte in the feed line would result in conversion of most of the carbonate to bicarbonate by the chemical reaction 7.3. 118 7 Experimental Results and Discussion From the analysis of the liquid species it was found that pumping 0.25 M CO3 2" in the factorial experiments (Table 7.14 and Figure 7.14) resulted in feeding 0.5 M HC0 3 " to the reactor. In other words, a catholyte with more than 0.25 M CO3 2" could give lower formate CE because the unconverted C03 2" in the reactor would result in a competition between reaction 7.1 and reaction 7.3. Experimental results with a higher C 0 3 2 " feed concentration (shown in Table 7.16) support this hypothesis. Table 7.16 Experiment s for the comparison between HCGy and CO3 2-Electrolyte Conductivity, S m"1 pH Ece l l , V CE, % 0.45 M K H C O 3 4.8 8.07 5.48 50 0.45 M K2CO3 5.2 9.18 5.38 -12 Operating conditions: Reactor A ; current = 6 A ; cathode: 1 layer of tec 30#; cathode age = 10 min; gas flow = 180 ml STP min"1; catholyte flow = 20 ml min"1; temperature = 300 K; pressure = 140 kPa(abs); y C 0 2 = 100%. From this theoretical consideration and experimental result HCO3" appears to be a better anion than C 0 3 . Thus K H C O 3 was adopted as the primary catholyte (in agreement with much prior literature) for subsequent study. The effect of K H C O 3 concentration on formate C E was then investigated by a set of parametric runs whose results are given in Table 7.17 and Figure 7.15. Table 7.17 Effect of K H C O 3 concentration on formate CE. K H C O 3 concentration, M Conductivity, S nf 1 E c e l l , V CE, % 0.10 1.1 5.50 42 0.25 2.6 5.42 46 0.45 4.8 5.43 50 0.75 6.7 5.37 33 1.96 19.4 4.97 27 min; gas flow = 180 ml STP min"1; liquid flow = 20 ml min"1; temperature = 299 K ; pressure = 133 kPa(abs); pH = 7.5~8.0. 119 7 Experimental Results and Discussion CEvsConc. of KHC03 60 T 10 _ 0 -I , , , , 0 0.5 1 1.5 2 2.5 Cone, of KHC03, M Figure 7.15 Formate EC vs concentration of K H C 0 3 at 6A, 100% C 0 2 and 1 layer tcc30#. Other conditions as under Table 7.17. Figure 7.15 shows a maximum formate CE at about 0.5 M K H C O 3 with higher bicarbonate concentration (up to 1.96 M) favouring hydrogen evolution. Hori [Hori, 1982; Murata, 1991] attributed the effect of K H C O 3 concentration to the buffering action of HCO3-at the cathode surface (reaction 7.4), which decreases the surface pH as the bulk concentration of bicarbonate is increased and shifts the process selectivity in favour of hydrogen evolution, i.e. reaction 7.2. This decrease in surface pH has been estimated in the modeling work of Gupta et al. [Gupta, 2005] to be about 0.5 pH unit at the prevailing conditions. While it is probably true that the process selectivity is affected by surface speciation, the results of Figure 7.15 might be more satisfactorily explained by the variation in catholyte conductivity (see Table 7.17) together with the decrease of C 0 2 solubility (ca. 30%) and increase of viscosity (ca. 30%) as the K H C O 3 concentration rose from 0.1 to 2 M [Wong, 2005; Harned, 1943; Lide, 2004]. At 6 A the experimental reactor was operating under C 0 2 mass transfer control (see Section 7.2.1.3), so a decrease in the limiting current density for reaction 6.1 by 50% should cause a corresponding drop in the formate current efficiency. 120 7 Experimental Results and Discussion 7.3.4 Catholyte conductivity and cathode thickness Dilute solutions of K H C 0 3 (e.g. 0.5 M) may be appropriate for academic studies of CO2 reduction but they are not adequate for the superficial current densities of 1-5 kA m"2 required by a 3-D cathode in an industrial process. The need for a low reactor voltage and a relatively uniform potential distribution in the trickle-bed electrode dictates a catholyte conductivity of at least 10 S m"1 and preferably as high as 50 S m"1. Theoretical considerations indicate that the effect of catholyte conductivity becomes increasingly important as the thickness and superficial current density of the 3-D cathode are increased. In the present work the effect of electrolyte conductivity was studied in conjunction with the cathode thickness (i.e. number of mesh layers) in the two level, two factor (2 +1) factorial experiment summarized in Table 7.18. Here the feed concentration of K H C O 3 was held at 0.45 M and the catholyte conductivity was increased by adding KC1 as supporting electrolyte, while the gas and liquid flows were adjusted to maintain constant fluid loads on the cathode. Figure 7.16 shows the results of this experiment and Table 7.19 presents a statistical analysis of the data. Table 7.18 Factorial variables and levels on conductivity and cathode thickness Variable Symbol Units Level High Low Centre Conductivity k, S i n 1 19.4 4.8 12.0 Number of layers - 3 1 2 Operating conditions: Reactor A ; current = 6 A ; y C 0 2 = 100%; cathode: one or multi-layers of tec 30#; cathode age = 10 min; K H C 0 3 concentration = 0.45 M ; gas flow = 180 ml STP min - 1 x layers; catholyte flow = 20 ml min - 1 x layers; temperature = 299 K; pressure = 133 kPa(abs); pH = 7.0 ~ 8.0. 121 7 Experimental Results and Discussion Ece„=5.15 V 64% 73% 3 Number of layers 1 63 % Ecd,=4.90 V EcelI=5.43 V 51% 43% 4.8 Conductivity, S m"1 19.4 Figure 7.16 Formate CE and reactor voltage for factorial runs in Table 7.18. Other conditions as under Table 7.18. Table 7.19 Factorial effects on conductivity and cathode thicl oiess (number of layers) Effects Current efficiency % 95 % confidence intervals % Main ki 1 ±7 X 21 Interaction k| x 8 Curvature 5 ±6 Confidence intervals are estimated from pooled replicates. In considering the results in Figure 7.16 and Table 7.19 it should be noted that other factors were involved here that might confound the observed effects. In particular we know from Section 7.3.3.1 that K C l (alone) gave a lower formate CE than K H C O 3 (alone) and further observe that the addition of any supporting electrolyte to raise conductivity reduced the solubility of CO2 in the catholyte [Linke, 1958] and thus the CO2 mass transfer limited current density. It is seen that with one layer of cathode mesh, increasing conductivity had a negative effect on the formate CE, which indicates that the conductivity of 0.45 M K H C O 3 was high 122 7 Experimental Results and Discussion enough to support the 1 layer cathode and the added KC1 played a negative role. However, with 3 layers of cathode mesh the positive effect of adding KC1 surpassed its negative effect (e.g. CE increased from 64% to 73%). This is because multi-layers needed higher ionic conductivity to support electro-activity over the available cathode thickness. The above results for electrolyte conductivity and cathode thickness are a nice example of how interactions and confounding effects can cloud conclusions drawn from uni-variate experiments such as those conducted in almost all of the prior work on the electro-reduction of CO2. In the present study the results of Figure 7.16 give the direction for process optimization around the cathode thickness, catholyte composition and conductivity. 7.3.5 Specific surface area of the cathode Table 7.20 presents results of experiments that show the effect of increasing the specific surface of the cathode by changing from 30# to 60# mesh (see Table 4.3 in Section 4.3.1 for the mesh properties). The increase in specific surface is accompanied by lower voidage and higher pressure drop through the cathode, however Table 7.20 shows the expected rise in formate C E from increasing the mass transfer capacity of the cathode. Table 7.20 Effect of the cathode specific surface. No. of layers 1 3 Mesh size Specific surface, m _ 1 C E % 30# 7000 51 64 60# 14000 63 82 Operating conditions: Reactor A; catholyte = 0.45 M K H C 0 3 for one ayer and [0.45 M KHCO M KC1] for 3 layers; cathode age = 10 min; current = 6 A; yC02 = 100%; catholyte flow = 20 ml min'1 x layers, gas flow = 180 ml STP min"1 x layers; temperature = 298 K; pressure = 138 kPa(abs); pH = 7.5-8.0. 7.3.6 Concentration of the supporting electrolyte (KC1) and cathode thickness Following the results of Sections 7.3.4 and 7.3.5 a series of experiments was used to explore the effects of the concentration of KC1 (as.a supporting electrolyte) on the reactor 123 7 Experimental Results and Discussion performance with a 60 # cathode. First the number of cathode layers and K C l concentration were tested jointly in a two level, two factor (22+l) factorial experiment and subsequently the K C l concentration and number of layers were examined separately in parametric runs. In all cases the gas and liquid flows were adjusted to maintain constant fluid loads on the cathode. 7.3.6.1 Factorial runs on KCl concentration and cathode thickness (number of layers) Table 7.21, Figure 7.17 and Table 7.22 present the design, data and statistical analysis for this set of factorial experiments. Here it is seen that, over the tested range, K C l concentration has 32% negative main effect on formate CE, and the effect occurs with both 1 and 3 cathode layers, which implies that (under the experimental conditions) K C l concentration should not exceed 1M for both 1 and 3 layers. The results confirm the positive effect of increasing the number of cathode layers and point to possible interactions and curvature. Also, both Figures 7.15 and 7.16 show the expected substantial decrease in reactor voltage (8-18 %) from increasing both the number of layers and the catholyte conductivity. Table 7.21 Factorial variables and levels on K C l concentration and cathode thickness. Variable No. of layers K C l cone Symbol C KCl Units M Levels High Low 1 Centre Operating conditions: Reactor A ; cathode: one or multi-layers of tec 60#; cathode age = 10 min; catholyte = 0.45 M K H C 0 3 + K C l ; current = 6 A ; y C 0 2 = 100% ; gas flow = 90 ml STP min"' x'layers; catholyte flow = 10 ml min"1 x layers; temperature = 298 K; pressure = 140 kPa(abs); pH = 7.6 ~ 9.5.' 124 7 Experimental Results and Discussion E c e „=4.97V KC1 cone, M 1 Ecdi=5.37V 62% 67% 65% Ec-.r4.26V 84% E c d l=5.02V 1 3 Number of layers Figure 7.17 Formate current efficiency and reactor voltage for factorial runs in Table 7.21. Other conditions as under Table 7.21. Table 7.22 Factorial "effects of K C l concentration and the number of cathode layers. Effects Main Interaction - K C l T C KCl Curvature Current efficiency % 29 -32 10 Confidence intervals are estimated from pooled replicates. 95 % confidence intervals % ±8 ±7 7.3.6.2 Parametric runs on KCl concentration and cathode thickness (number of layers) Several parametric runs were performed to seek the appropriate K C l concentration and to get information on the effective electro-active bed thickness. Tables 7.22 and 7.23 show results from the parametric runs. Table 7.23 Effect of K C l concentration. K C l concentration M 3.0 1.0 0.5 Pcath(inlet) kPa(abs) 131 141 146 pH (outlet) 7.84 7.66 7.65 Ecel l V 4.26 4.88 4.27 CE % 62 84 87 Operating conditions: Reactor A; current = 6A; cathode: 3 layers of tec 60#; cathode age = 10 min-catholyte = 0.45M K H C 0 3 + KCl ; gas flow = 270 ml STP min"1; catholyte flow = 30 ml min 1; temperature = 299 K; pressure =133 kPa(abs), y C 0 2 = 100%. 125 7 Experimental Results and Discussion 0.5M K C l is high enough to provide ionic conductivity for 3 layers of tec 60 # (thickness =1.1 mm). Table 7.24 El ffect of number of layers of tec 60 . Number of tcc#60 Pcath(inlet) kPa(abs) pH (outlet) Ecell V C E % [HCOO - ]* M 1 146 9.45 5.37 65 0.101 2 146 7.65 4.12 86 0.080 3 141 7.66 4.88 84 0.052 4 136 7.61 4.68 85 0.039 Operating conditions: Reactor A; current = 6 A; cathode: multi-layers of tec 60#; cathode age = 10 min; catholyte = 0.45 M K H C 0 3 +1 M KCl gas flow = 90 ml STP min"1 x layers; catholyte flow = 10 ml min"1 x layers; temperature = 299 K; reactor pressure =141 kPa(abs). * formate concentration in catholyte product. It is seen from Table 7.24 that increasing the number of layers from two has no significant effect on the formate CE, which suggests that the effective bed thickness under these experimental conditions is no more than the thickness of two layers of tec 60#. This value is in agreement with the theoretical calculation of the effective bed thickness, which gives 0.66 mm, compared to the thickness of two layers of tec 60#, i.e. 0.76 mm. The relatively low voidage of the tinned copper mesh, i.e. 0.41 and 0.31 for tec 30# and tec 60#, respectively, limits the liquid hold-up in the reactor, thus constraining the effective electrolyte conductivity and electro-active bed thickness. From these results one can surmise that the utilization of a more porous cathode material with high specific surface, such as a tin felt or foam, would improve the reactor performance. 7.3.7 Observations from tec cathode The above study of eight process variables for the electro-reduction of CO2, supported by the model (see Chapter 6), allows the following observations, within the ranges of the variables studied: (i) Increasing the current from 2 to 6 A lowered the formate current efficiency by 24 %. 126 7 Experimental Results and Discussion (ii) Increasing the CO2 concentration in the feed from 16 to 100 % increased the formate CE by 15%. (iii) Increasing the reaction temperature from 288 to 328 K favored the intrinsic selectivity for formate over hydrogen. However, with the reaction under a CO2 mass-transfer constraint the positive effect of temperature on the intrinsic kinetics was countered by a corresponding decrease in the solubility of CO2 in the catholyte, with the net result that at fixed CO2 pressure the formate current efficiency decreased with increasing temperature. (iv) With respect to the catholyte composition the generation of formate was favored by K + versus N a + cations and by HCO3" versus CI" anions. The presence of CO3 2 " in the catholyte suppressed formate generation- presumably due to the reaction of CO2 with C 0 3 2 " that inactivated CC»2(aq) as HCO3". The negative effect of CO2 speciation was evident at a catholyte pH over about 9. (v) The effect of K C l as a supporting electrolyte in the K H C O 3 catholyte depended on the thickness of the 3-D cathode. Increasing the catholyte conductivity from 4.8 to 19.4 S m"1 in a single mesh layer cathode (0.61 mm thick) decreased the formate CE, but had the opposite effect in a triple mesh layer cathode (1.9 mm thick). This interaction was attributed mainly to countervailing effects of K C l concentration on the CO2 mass transfer limiting current density and the electro-active thickness of the 3-D cathode. (vi) Within the limits of electro-active thickness, increasing the number of tin-coated copper mesh layers in the 3-D cathode with fixed fluid loads raised the formate CE. This effect was presumably due to the drop in CO2 concentration polarization that resulted from a reduction in the real current density on the cathode surface. (vii) Increasing the specific surface area of the cathode from 7000 to 14000 m"1 also increased the formate CE-again due to the reduction in real current density on the cathode. (viii) Apart from the effects on formate CE, increases in the temperature, catholyte conductivity, cathode thickness or specific surface also reduced the reactor voltage and thus affected the specific energy consumption for formate generation (at a given superficial current density). (ix) The current efficiency for formate decreased with time due to the progressive loss of tin from the cathode surface to expose the copper substrate, with a consequent increase in the 127 7 Experimental Results and Discussion exchange current density for hydrogen evolution that promotes the electro-reduction of water relative to that of carbon dioxide. From a reactor engineering perspective the "best" result of the work with the tec cathode was obtained after 10 minutes operation with a bi-layer 60# cathode and catholyte of [0.5 M KHCO3+O.5 M KCl] , operating on a gas of 100 % C 0 2 at 120 kPa (abs) and 299 K, with gas and liquid flow feed flows respectively 180 ml (STP)/min and 20 ml/min. The performance indicators in this case are listed in Table 7.25. Table 7.25 Performance indicators for a special case with tec cathode. Performance indicator Unit Value Formate current efficiency % 86 Formate product concentration M 0.08 Reactor voltage V 4.1 Superficial current density k A m " 2 1.3 Specific electrochemical energy consumption kWh kmof 1 HCOO" 260 Note: the operating conditions are described in the paragraph above this table. 128 7 Experimental Results and Discussion 7.4 Lead cathodes in Reactor A Lead is a metal with high hydrogen over-potential (i.e. low exchange current density for hydrogen evolution) that is reported in the literature to give a high selectivity for a single product - formate/formic acid. In this context, the Pb plate, Pb shot, and Pb granules listed in Table 4.4 were explored as cathodes for the electro-reduction of CO2 in Reactor A , with reactor configuration 3 (see Figure 7.5) under the following conditions: Catholyte flow rate: 20 ml/min - Gas flow rate: 180 ml STP/min (100 vol% C 0 2 ) - Catholyte: 0.45 M KHCO3 + 0.5 M K C l - Cathode feed pressure: 140 ~ 180 kPa (abs) - Temperature: 293 ~ 300 K Experimental results of formate C E vs time with Pb plate, Pb shot and Pb granules are presented in Figure 7.18. It is seen that up to 70 % formate CE was achieved with a Pb plate at a current density of 0.22 kA m"2. Around 30 % formate C E with Pb shot and 19 % formate CE with Pb granules were obtained, respectively in these 3-D cathodes at a superficial current density of 1.33 kA m~ . However, both the Pb plate cathode and Pb granules deteriorated with time to different extents, while the Pb shot cathode sustained its catalytic activity within the time period studied. 129 7 Experimental Results and Discussion 80 70 UJ 50 4 -o 40 I 30 o 2 0 10 CE vs time Pb plate at 2 A (0.44 kA m"2) Pb shot at 6 A (1.33 kAm" 2 ) Pb granules at 6 A (1.33 kA m - 2) 0 30 60 90 120 150 180 210 240 270 300 330 360 Time, min Figure 7.18 The effect of cathode age on formate CE with Pb plate, shot and granules (Reactor A). ICP-AES analysis was carried out by VIZON SCITEC (former BC Research) on Pb plate, shot and granules, and as well as on liquid catholyte products from ERC with each of these three Pb catalysts, to see the difference in the concentrations of impurities, such as Fe and As, which have lower over-potential for hydrogen evolution than Sn, and thus prompt the deterioration. Table 7.26 Iron (Fe) analysis on Pb catalysts and in the liquid products of ERC Samples Iron concentrations Liquid samples Products from Pb plate . N A mg/L Products from Pb shot 0.1 Products from Pb granules 0.8 Solid Samples Pb plate 1.9 mg/kg Pb shot <0.4 Pb granules 1.2 It is seen from the Fe analysis results in Table 7.26 that Pb shot had the lowest Fe content both on the catalyst (cathode) surface and in the product solution, while on the other 130 7 Experimental Results and Discussion hand, Pb plate had the highest Fe content on the catalyst surface. This pattern of the Fe concentration levels among the three Pb cathodes coincided with the deterioration pattern in the catalytic activity of the Pb cathodes shown in Figure 7.18. Therefore, it is evident that the existence of impurity Fe, whose exchange current density for hydrogen evolution is much higher than that of Sn, was at least one reason for the deterioration of the catalytic activity of Pb cathodes. With this information in mind, two chelating agents, EDTA and DTPA, were added separately into the catholyte feed at a level of about 0.04 wt %, aiming to reduce the levels of Fe impurity on the surface of the cathode. However, preliminary experimental results on ERC did not show any positive effects of the two chelating agents (see Figure 7.19), either because the chelate concentrations were too low or because of other contributing factors to the catalyst deterioration, such as surface morphology changes and surface area losses, etc. CE vs time re E o 30 20 -j 1 10 0 -I , , , , , , , , , 1 0 10 20 30 40 50 60 70 80 90 100 Time, min Figure 7.19 Effects of chelating agents on the formate C E with Pb plate cathode at 2 A (0.44 kA m'2) in Reactor A. a: no chelating agent; b: 0.04 wt% DTPA; c: 0.04 wt% EDTA. 131 7 Experimental Results and Discussion Pb grid and Pb on reticulate carbon were also investigated as ERC cathodes with the results shown in Figure 7.20. It is seen from Figure 7.20 that the highest formate CE with Pb grid under the same conditions as aforementioned was only 9 %, while that for Pb on reticulate carbon was 28 %, and both cathodes deteriorated quickly with time. In addition, the Pb on reticulate carbon was too fragile for any potential industrial application. Formate CE vs time 30 0 10 20 30 40 50 60 70 Time, min Figure 7.20 Experimental results on Pb grid and Pb reticulate carbon cathodes at 6 A (1.33 kA m") in Reactor A. a: Pb on reticulate carbon cathode; b: Pb grid cathode. In the present project, the formate CE's with all lead cathodes were much lower than those with tin cathodes. Thus, no further experiments were carried out with lead cathodes in this project. 132 7 Experimental Results and Discussion 7.5 ERC with tin shot and tin granule cathodes in Reactor A Tin-coated copper mesh cathode deteriorated quickly with time (i.e. within 30 minutes, see page 109). Although the deteriorated tin-coated copper mesh can be stripped of tin, re-tinned, and re-used as a cathode, its poor sustainability in catalytic activity would definitely make it unsuitable as a catalyst for the potential industrial application of ERC. After unsuccessful attempts to obtain pure tin mesh, attention was switched to pure tin shot and granules. 7.5.1 Characterization of tin shot and tin granules In the characterization of tin shot and tin granules, voidage was determined by measuring the volume of water replacement in a certain volume of particles (shot or granules), the average particle diameter was calculated by size distribution measurement using equation 7.5, and specific surface area was calculated by either equation 7.6 (spherical particles) or 7.7 (non-spherical particles): p,average a = 6(1-s) (spherical) (7.6) P,average a -6(1-*) (non-spherical) d (7.7) p,average 9 is the shape factor defined as: S sphere S, (7.8) actual 133 7 Experimental Results and Discussion Where: a = specific surface area (m 2nf 3) dp,i = opening size of screen i in screen analysis (m) dp, average = average particle diameter (m) S s p h e r e = surface area of the sphere that has the same volume of the actual particle (m2) Sactuai = actual surface area of the particle (m2) Wj = weight fraction of granules that pass screen i in screen analysis Tin shot SN-133 from Atlantic Equipment Engineers (US) (see Table 4.4) had nearly mono-disperse size distribution and uniform spherical shape. Therefore the shape factor was assumed to be 1. Tin granules SN-131, on the other hand, had much wider size distribution and irregular shape. In addition, these tin granules (price ca. C A D $ 60/kg) were purchased in three separate batches of 1 to 5 kg - each with ostensibly the same specification (SN-131) but inconveniently arrived with different color, size and shape distribution. Figure 7.21 shows the size distributions from Batch 1 (received May 2005) and Batch 2 (received November 2005) and the preponderance of small particles in Batch 2. To avoid the high cathode pressure drop caused by the fine particles in Batch 2, the granules smaller than 80# mesh of Batch 2 were removed. Figure 7.22 presents the S E M pictures of fresh Batch 1 granules and fresh screened (+80#) Batch 2 granules, which show irregular particle shape, especially for Batch 2. 134 7 Experimental Results and Discussion 35 ^ 30 O 20 jS 15 c 0 u CL 10 5 0 Size distribution a D #25-30 #30-35 #35-40 #40-45 #45-50 #50-60 #60-70 #70 Mesh No Size distribution 60 I50 . 40 0) O) 30 (0 C 20 o O 10 o „ Q. 0 #30-35 #35-40 #40-45 #45-50 #50-60 #60-80 #80-100 #100-Mesh No Figure 7.21 Size distributions of tin granules, a: Batch 1; b: Batch 2. 135 7 Experimental Results and Discussion Figure 7.22 S E M images of fresh tin granules: a: Batch 1; b: Batch 2 By measuring dimensions of individual particles from the S E M images assuming symmetry around the main axis, the shape factors were estimated to be 0.89 and 0.59 for Batch 1 and Batch 2, respectively, using equation 7.8. Due to the non-uniform size distribution of the tin granules, shape factors were determined for five particles from each batch and the averages taken as the respective shape factors. Table 7.27 summarizes the properties for tin shot and the Batch 1 and Batch 2 granules employed for ERC studies in the present project. Table 7.27 Characterization of tin shot and granules Tin particles Voidage % Average size mm Specific area m 2 m- 3 Shape factor Shot 40 2.39 1500 1 Batch 1 granules 38 0.379 11000 0.89 Batch 2 granules (+80#) 40 0.252 17000 0.59 136 7 Experimental Results and Discussion 7.5.2 Exploratory runs on tin shot and Batch 1 tin granules in Reactor A Tin shot and Batch 1 tin granules were pretreated with 11 wt% HNO3 for 2 min to clean the surface, rinsed several times with plenty of de-ionized water for 10 min, and then placed into the 3 mm thick gasket bed of the cathode. ERC was carried out for up to 250 minutes (~4 hours) at 6 A with 0.45 M K H C O 3 and 2 M K C l as the primary and supporting electrolyte, respectively, with the results presented in Figure 7.23. It can be seen that the formate CE with tin shot showed an initial increase and then stabilized within the tested period of time (3 hours); the formate CE with tin granules also showed an initial increase within the first 30 min, and then stabilized for about 2 hours (a plateau area) before starting to decrease slightly. The initial gain of formate CE in both cases might be explained by the further clean-up of the tin surface on the cathode through both the flush of fluid flow and the polarization, which apparently made the cathode more active. It can also be seen that tin granules exhibited much higher CE than tin shot, presumably due to the higher specific surface area of granules than that of tin shot, i.e. 11000 vs. 1500 m 2 m"3, and this made tin granules more interesting for further investigations on recycle, pretreatment techniques and scale-up, etc. 137 7 Experimental Results and Discussion Formate CE vs time 70 10 0 50 100 150 200 250 Time, min Figure 7.23 Formate CE vs cathode age in Reactor A . (a): fresh tin shot; (b): fresh tin granules from Batch 1. Operating conditions: current = 6 A; catholyte = 0.45 M K H C 0 3 + 2 M K C l ; gas flow = 364 ml (STP) min" 1; catholyte flow = 20 ml min" 1; temperature = 300 K ; pressure =116 kPa (abs) on tin shot and 140 ~ 170 kPa (abs) on tin granules. An important feature of Figure 7.23 is the relatively long and stable operating times (200 min) of the tin shot and tin granules, compared to those with the tin-coated copper mesh in which the formate current efficiency dropped from about 50% to 20% in 100 min (Figure 7.10). This result is encouraging in terms of the potential industrial development of ERC. 7.5.3 Reactivation of the deteriorating tin granules in Reactor A The tin granule cathode provided higher formate CE than tin shot but the cathode deteriorated faster with time (Figure 7.23). From the impurity metal analysis shown in Table 7.28, it can be seen that the impurity levels of Sb, As and Cu were much higher in tin granules than in tin shot, which suggested that those impurities might be poisoning the tin granule surface as the ERC was going on. 138 7 Experimental Results and Discussion Table 7.28 Impurity analysis in tin shot and tin granules Impurity Content, wt% Tin shot Tin granules Iron 0.01 0.01 Antimony <0.01 0.05 Arsenic <0.01 0.05 Copper O.01 0.04 The higher formate CE in the tin granule cathode in Figure 7.23 is probably due to the 10 fold increase in specific surface of the granules relative to tin shot (Table 7.27). The effect of the cathode specific surface is explained in Chapter 6 on reactor modeling and elaborated in Sections 7.3.5 and 7.5.4. Although the deterioration rate of tin granules was much lower than that of the tin-coated copper mesh cathode (see Figure 7.10), a couple of hours of sustainable catalyst activity were still not satisfactory for potential industrial application, in which stability over several thousand hours is normally necessary. Therefore, ways of re-activating the deteriorating cathode were investigated. 7.5.3.1 Chemical treatment and recycle of tin granules The used granules from Batch 1, after being scraped from the reactor, were treated with 11 wt% HNO3 again, washed in water and re-used to run ERC under the same operating conditions as described for fresh tin granules, aiming to examine whether the tin granules were recyclable (i.e. resuming the formate CE to the high level obtained with fresh granules). This recycle operation was repeated several times on the same batch of granules (from Batch 1) to see the effects of recycle times. The changes of formate CE with time are presented in Figure 7.24 in which (a) represents fresh granules, and (b), (c) and (d) for recycle once, twice and three times, respectively. Table 7.29 lists experimental data on reactor voltage, catholyte pH and pressure which were taken during the period of the constant CE in those recycle runs. Figure 7.24 shows that tin granules, regardless of how many times they had 139 7 Experimental Results and Discussion been recycled, were not much different in terms of the formate CE. In another words, they yielded almost the same formate CEs and followed the same CE changing pattern with time (having a plateau) under the same operating conditions. The slight increase of CE with recycle time was probably caused by the decreasing size of the tin granules as a result of the HNO3 pretreatment, which was also reflected by the gradual increase of reactor pressure drop, i.e. from 55 kPa with fresh granules to 65 kPa with three-times recycled granules (see Table 7.29). This indication implies that the HNO3 pretreatment might cause a serious loss of tin on the surface, and a less aggressive pretreatment agent may be better (see Section 7.6.3 on the pretreatment techniques). The decrease in reactor voltage with recycing listed in Table 7.29 might also be a result of decreasing tin particle size, as the increased specific surface lowers the CO2 concentration over-potential (under mass transfer control) and possibly increases the effective electronic conductivity of the cathode by increasing the inter-particle contact area. Table 7.29 Experimental results on the recycle of the tin granules from Bate h i ) . Recycle times Formate CE % Ecel l V Pcathode(inlet) kPa(abs) pH outlet [HCOO-] M 0 (fresh granules) 63 3.73 156 7.76 0.059 1 61 3.56 156 7.71 0.058 2 64 3.36 161 7.79 0.059 3 66 3.30 166 7.70 0.062 Operating conditions: the same as under Figure 7.23 except for pressures. Operating time: 30 ~ 150 minutes. 140 7 Experimental Results and Discussion Formate CE vs time o <u 40 -n E 30 o 20 10 0 -I , , , , , 1 0 50 100 150 200 250 300 Time, min Figure 7.24 Formate CE vs cathode age in Reactor A. a: fresh tin granules; b, c and d: used tin granules being recycled once, twice and three times, respectively. Operating conditions: as in Table 7.29. 7.5.3.2 Polarity reversal (PR) for recovering the used tin granules (from Batch 1) If it worked to recover the catalyst activity, polarity reversal could be more convenient and less costly than removing, treating and recycling used granules. Polarity reversal was applied for 5 min at 1A on tin granules from Batch 1 when the formate CE had dropped to 49 % at 6 hours, and the formate CE resumed its previous value of 64 % almost immediately. Unfortunately the high formate CE was not long sustained before starting to drop again, as shown in Figure 7.25. This unsustainable CE might be explained by the heavy rusty spots (brownish deposits) on the membrane formed during the polarity reversal. The brownish deposits were found later to be tin oxides, which might have some adverse effects on the performance of the reactor such as lowering the life, blocking the permeability, or increasing the resistance of the membrane, as well as polluting the cathode surface. 141 7 Experimental Results and Discussion Formate CE vs time 70 -j £ 20 10 -0 0 100 200 300 400 500 Time, min Figure 7.25 Polarity Reversal in Reactor A. Operating conditions: polarity reversal current =1 A; polarity reversal time = 5 min; ERC conditions are the same as under Figure 7.23. 7.5.4 Exploratory runs on tin granules from Batch 2 in Reactor A As mentioned in Section 7.5.1 the two purchased Batches (1 and 2) of tin granules were different in appearance, shape and size distribution. After all granules from Batch 1 had been used exploratory runs were done on Batch 2 granules to see whether they performed as those in Batch 1. Screened Batch 2 tin granules were used to run ERC under the same conditions as for Batch 1 (see Section 7.5.1 for the characterization of screened Batch 2 tin granules). Experimental results on reactor voltage, catholyte pH, pressure, formate CE and formate concentration are given in Table 7.30. The changes of formate CE with time are presented in Figure 7.26. Table 7.30 ERC results on screened Batch 2 tin granules at 60 minutes of operation. Run No Formate C E Ecell Pcathode(inlet) pH [ H C O O ] % V kPa(abs) outlet M 1 76 3.19 171 7.67 0.066 2 76 3.22 178 7.71 0.067 Note: 1 and 2 are replicates to each other. Operating conditions: the same as under Figure 7.23 except for pressures. 142 7 Experimental Results and Discussion Ui o £ re E o 80 70 60 50 40 30 £ 20 10 Formate CE vs time 20 40 60 80 100 Time, min 120 140 160 Figure 7.26 Exploratory runs with screened Batch 2 granules in Reactor A. 1 and 2 are replicate to each other. Operating conditions: the same as under Table 7.30. It can be seen from Table 7.30 and Figure 7.26 that screened Batch 2 granules exhibited similar sustainability in formate CE as Batch 1 granules within about 2 hours, but at a substantially higher level, i.e. 76 vs 63 %. The higher CE was presumably due to the higher specific surface area of the screened Batch 2 granules. However, a higher pressure drop occurred in the reactor (cathode) due to the smaller size of the screened Batch 2 granules, which implied a potential problem in the Reactor B (scale-up) where the pressure drop would be multiplied by the ratio of the heights of Reactor B to Reactor A . 7.5.5 Fractional factorial experiments on the screened Batch 2 tin granules The relatively good results shown in Figure 7.26 encouraged a more extensive investigation of ERC with the screened (+80#) Batch 2 tin granules in Reactor A . To explore the large multi-variable experimental space thus opened for the present project a V resolution, 2 ( k" l ) fractional factorial design was selected with current, concentration of supporting electrolyte (KCl), temperature, gas flow rate and catholyte flow rate being the 5 143 7 Experimental Results and Discussion variables. The joint investigation of these 5 primary process variables using fractional factorial design could provide a preliminary insight on how they separately and interactively affect the formate CE and other performance indicators (see Chapter 5 for details on fractional factorial design). Table 7.31 shows the levels of the five variables. Table 7.31 Levels of variables for the 2 ( k _ 1 )+l fractional factorial design with tin granules from Batch 2 in Reactor A . No Variables Levels Low (-) Center(O) High(+) 1 I: Current, Amp 2 5 8 2 T: Temperature, K 293 308 323 o 3 L: Catholyte flow rate, ml min"1 10 15 20 4 G: Gas flow rate, ml STP min - 1 400 1000 1600 5 C: Cone, of K C l , M 0.5 1.3 2 According to the 2(' '+1 design, seventeen fractional factorial runs were performed in random order with 4 replicates of the centre point. Replicate runs were also conducted for 5 corner-points. The experimental matrix and the formate CEs ("responses") obtained are presented in Table 7.32, in which the outlet pH is also listed. It is seen that the outlet pH under these conditions did not vary much, i.e. between 7.54 and 7.81. Therefore the effect of pH on the formate CE could be considered to be insignificant in Reactor A . 144 7 Experimental Results and Discussion Table 7.32 Design matrix and formate CE's (responses) for the fractional factorial runs with a Batch 2 tin granule cathode in Reactor A. No Variable levels Formate CE % pH I T L G C 1 - - - - + 83 7.58 2 + - - - - 68 7.70 3 - + - - - 72 7.64 4 + + - - + 64 7.76 5 - - + - - 87 7.54 6 + - + - + 79 7.63 7 - + + - + 68 7.70 8 + + + - - 73 7.81 9 - - - + - 77 7.56 10 + - - + + 74 7.68 11 - + - + + 71 7.59 12 + + - + - 66 7.81 13 - - + + + 77 7.57 14 + - + + - 75 7.71 15 - + + + - 65 7.71 16 + + + + + 72 7.76 17 0 0 0 0 0 75 7.63 Operating conditions: catholyte: 0.45 M K HCO3+KCI; pressure = 136-241 kPa (a 3S). Table 7.33 summarizes the calculated main effects, two-factor interactions and the curvature effect. The corresponding confidence intervals were also calculated from pooled variances of the replicate runs. 145 7 Experimental Results and Discussion Table 7.33 Main and two-factor interaction effects for the fractional factorial results with screened Batch 2 tin granule cathode in Reactor A. Effects Current efficiency % 95 % confidence intervals % I -3.7 T -8.5 Main L 2.7 G 6.2 C 0.7 IT 3.4 IL 4.2 IG 2.9 ±1.8 IC 1.0 Two-factor TL -1.5 Interaction TG 1.5 TC -0.9 L G -2.2 L C -1.7 GC 2.1 Curvature 2.0 ±2.1 Table 7.33 shows that, current and temperature had significantly negative effects on formate CE, which means increasing operating current and temperature both lowered the formate CE. This was consistent with the conclusions drawn from the tin-coated copper mesh cathode (Section 7.3). On the other hand, the catholyte (liquid) flow rate and gas flow rate both had positive effects on formate CE. As discussed in Chapter 6 (modeling), an increase in both gas and liquid flow rates raises the G/L and L/S C 0 2 mass transfer capacity through equations 6.15, 6.16 and 6.17, while increasing the liquid flow also increases the liquid hold-up through equation 6.41, all of which increase the concentration of the reactive species [C02(aq)] 146 7 Experimental Results and Discussion through equation 6.14 and promotes the formate current efficiency. It has to be noted that, increasing the gas flow rate from the "low" level (400 ml STP min 1 ) to "high" level (1600 ml STP/min) pushed the pressure in Reactor A up about 65 kPa on average under the experimental conditions, and this increase in reactor pressure can have significant effect on the solubility of CO2 and thus on the mass transfer rate of CO2. Therefore, the effect of gas flow rate is augmented by the effect of reactor pressure. Also, the higher pressure drop caused by the increase in gas flow rate would be many times magnified in bigger reactors (for scale-up or potential industrial applications), and thus might increase the capital and operating costs as far as the process economics is concerned. Interestingly, Table 7.33 shows that the concentration of supporting electrolyte K C l did not have statistically significant effect on formate C E within the tested current range of 2 to 8 A. This fits with the conclusion drawn from previous experimental results (Section 7.3.6.1) that electrolyte K C l had two countervailing effects, one was to increase the viscosity of and decrease the CO2 solubility in the catholyte, thus lowering the CO2 mass transfer limiting current density, the other one was to increase the ionic conductivity, and these two effects counteracted each other under the experimental conditions. Regarding the interaction effects, current interacted positively with both catholyte flow rate and gas flow rate. The positive interactions between them again suggest that the increase in mass transfer capacity of CO2 caused by the increases in gas and liquid flow rate as discussed above was more beneficial at higher current density where ERC tended to be more constrained by mass transfer. Additionally, the current was involved in a significant positive interaction with temperature. From the model (Chapter 6) and previous study (Section 7.3.2) we know that temperature affected the formate C E in many ways, such as favoring the intrinsic selectivity for formate over hydrogen, decreasing the solubility of CO2, increasing the conductivity and decreasing the viscosity of the catholyte, and so on. Among these complex and countervailing effects one can surmise that the positive IT interaction is due to an increase in the electro-active thickness of the cathode with temperature (equation 6.29), giving a lower real current density (equation 6.24) that favors the primary cathode reaction (equations 6.6 and 6.7). 147 7 Experimental Results and Discussion 7.5.6 Full factorial experiment on Batch 2 tin granules To further understand the effects of some major process variables, a full 2 3 factorial experiment was carried out with current (I), concentration of K C l (C) and catholyte flow rate (L) being the three variables, and in this set the current was extended to an extreme high level of 14 A (3.1 kA m"2) to force the process into mass transfer control where C and L might be more important than observed in the previous set of experiments. Table 7.34 shows the levels of the three variables, Figure 7.27 presents the experimental results in the factorial cubes and Table 7.35 summarizes the calculated factorial effects. Table 7.34 Levels of variables for the 2 fractional factorial design with Batch 2 tin granules in Reactor A. No Variables Levels Low (-) High(+) 1 I: Current, Amp 2 14 2 C: Cone, of K C l , M 0.5 2 3 L: Catholyte flow rate, ml min" 10 20 Other operating conditions: catholyte = 0.45 M KHC0 3 + KCl ; gas flow = 400 ml (STP) min"1; temperature = 323 K; pressure =181 ~231 kPa (abs). 42 14 38 Current, A 45 39 93 72 95 20 10 L, ml min"1 0.5 KCl 148 7 Experimental Results and Discussion Figure 7.27 Current efficiency for factorial runs in Table 7.34. Table 7.35 Factorial effects of current, K C l concentration and catholyte flow rate on the formate CE with Batch 2 tin granules in Reactor A . Effects Current efficiency % 95 % confidence intervals % I -42 Main C 0 L 14 ±6 IC 1 Interaction IL -9.8 C L 2 Calculated main and interaction effects listed in Table 7.35 are in agreement with that from fractional factorial runs except that the negative main effect of current density on formate CE was more significant in this set. This can be explained by the fact that in the factorial runs the current was extended to a much higher level (14 A) where ERC was totally mass transfer controlled so that an increase in current density only promoted the secondary reaction, i.e. hydrogen evolution. 149 7 Experimental Results and Discussion 7.6 Tin granules in Reactor B (scale-up) Tin granules yielded higher formate CE's than tin shot with longer catalyst life than tin-coated copper mesh, and therefore were adopted as the primary cathode material for the scale-up investigations in Reactor B . 7.6.1 Scale-up issues The philosophy of scale-up for continuous fhermochemical reactors is to ensure several similarities between the small and big reactors, i.e. geometric similarity, kinematic similarity and thermal similarity. However, an additional similarity has to be maintained in the scale-up of an electrochemical reactor, i.e. electrical similarity that respects the distribution of electrode potential and current density [Goodridge and Scott, 1995; Gupta, 2005]. Geometric similarity in thermochemical reactors is obtained by fixing the dimensional ratios of the small and big reactors. In the electrochemical reactors, however, it is not recommended to increase the inter-electrode gap (i.e. the thickness of the 3-D electrode and/or the counter-electrode chamber) since that increases the Ohmic drop, thus raising the cell voltage and energy costs. Changing the thickness of a 3-D electrode can also change the potential distribution along the direction of current and thus promote undesired secondary reactions (see Section 2.5). Therefore, the long-sandwiched configuration (small inter-electrode gap) used in Reactor A was also employed in Reactor B (scale-up) for the present work to realize the more important potential similarity criterion discussed below. Kinematic similarity concerns the similarity in flow regime and flow velocities (fluid loads) or more generally the Reynolds' numbers that govern the pressure drop, liquid hold-up, and mass transfer capacity of the system [Hodgson and Oloman, 1999]. In the present work, the same bottom-to-top "trickling" 2-phase flow mode as used in the Reactor A was employed in Reactor B . The catholyte (liquid) flow rate of 12 ~ 33 ml min"1 (i.e. 1.3x10" 4 ~ 3.7xl0" 4 m s"1 superficial velocity) in Reactor B was intended to match the liquid flow rate 150 7 Experimental Results and Discussion of 8 ~ 20 ml min - 1 (i.e. 1.4xl0" 4 ~ 3.7><10"4 m s"1 superficial velocity) in Reactor A . Unfortunately, the upper limit of catholyte flow rate in Reactor B (33 ml min 1 ) could not be realized because the resulting high pressure drop (equation 6.40) stopped the flow of CO2 gas to the reactor (the maximum regulator pressure at the CO2 gas cylinder was about 700 kPa(abs) which could deliver CO2 gas to the system only where the cathode inlet pressure was below about 680 kPa(abs)). Therefore, the actual liquid flow rate used in Reactor B was 10 to 22 ml min 1 (i.e. l . l x l O " 3 ~ 2.4xl0" 3 m s"1 velocity). The gas flow rate in Reactor B for the present work was determined more based on the stoichiometric requirements of the process than on the kinematic similarity criteria. As a rule of thumb the minimum gas (CO2) flow was set such that the CO2 conversion at the maximum current did not exceed 75 %, assuming a stoichiometric coefficient of 1 Faraday/mole CO2. For example, with a current of 100 A and maximum C 0 2 conversion of 75 %, the feed C 0 2 rate should be above 1860 ml STP min"1. A gas flow rate (inlet) of 400 ~ 1600 ml STP min"1 (i.e. 5.3><10"2 ~ U x l O " 1 m s"1 superficial velocity) was used in Reactor A , whereas a gas flow rate of 1000 ~ 3500 ml STP min"1 (i.e. 2 .4xl0" 2 ~ 6.2xl0" 2 m s"1 in velocity) was intended for Reactor B. However, the desired upper limit of the gas flow rate in Reactor B (3500 ml STP min"1) could not be realized in the present gas delivery system, therefore the actual gas flow rate used was 1000 to 2500 ml STP min"1 (inlet) (i.e. 2 .4x l0" 2 ~ 5.2x 10" m s" in velocity). It has to be noted that, the greater height of Reactor B (680 mm) caused proportionally higher pressure drop than in Reactor A (150 mm) which might to some extent confound the results of scale-up. Thermal similarity calls for maintaining the same temperatures at the corresponding points of the small and big reactors. The heat exchange between an electrochemical reactor and its surroundings changes when scaled up. From the energy balance described in Chapter 6 (equation 6.37) it is seen that in the scaled-up electrochemical reactors, where the I/L ratio (current over liquid flow rate) is normally higher than in the small reactors, the temperature increase is more significant. Also, due to the lower ratio of external surface area to current the heat loss from a large reactor are proportionally lower than from a small reactor, so higher temperatures are established by Joule heating. 151 7 Experimental Results and Discussion Since the data from Reactor A had shown a negative effect of temperature on the formate CE, pre-cooling of both the anolyte and catholyte were used with Reactor B, to lower the catholyte temperature. Electrical similarity, which is often the most important consideration in the scale-up of electrochemical reactors, implies maintaining the same potential and current density distributions between the small and big reactors. The dimensionless Wagner number (Wa = dr)/dj / (1/ki)), which governs the secondary current distribution [Walsh, 1993], may be invoked here but has limited utility in application to 3-D electrodes under mass transfer constraints [Prentice, 1991]. The criterion of similar potential and current density usually requires a constant inter-electrode gap on scale-up, plus constant specific surface and thickness for 3-D electrodes, as in the present work. However, the current density and potential distributions were also affected by other factors such as, in Reactor B the higher catholyte conductivity caused by the higher temperature, the higher formate concentration due to the higher ratio of current to catholyte flow and so on. Modeling the reactor (Chapter 6) is useful in accounting for these interacting effects and in predicting the performance of the scaled-up reactor. 7.6.2 Parametric runs Parametric runs were carried out in Reactor B within a current range of 20 to 101 A to see the effect of scale-up on the performance of the reactor. Table 7.36 gives the formate CE's achieved along with some other experimental data, Figure 7.28 shows the comparison of formate CE in reactors A and B, both with screened (+80#) Batch 2 tin granules (see Figure 7.21 and Table 7.27 for the characterization of Batch 2 tin granules), and Figure 7.29 summarizes the dependence of specific energy and formate concentration on current density. 152 7 Experimental Results and Discussion Table 7.36 Parametric experimental results in Reactor B with Batch 2 tin granules Current, A CD, kA m" CE, % Operating time, min [HCOO"], M Ecell, V T (out), K Pcathode(in) kPa(abs) pH (out) 20 0.62 91 60 0.28 2.70 293 690 8.10 40 1.24 86 80 0.54 3.40 296 690 8.20 51 1.58 83 30 0.64 3.10 299 620 8.16 88 2.73 67 35 0.90 4.30 308 500 9.29 91 2.83 65 0.92 4.25 305 500 9.46 94 2.92 64 100 0.94 4.10 322 540 9.47 Operating conditions: Reactor B; cathode: recycled Batch 2 tin granules; catholyte= 0.45 M K H C O 3 + 2 M KCl ; catholyte flow rate = 20 ml min"1; gas flow rate =1600 ~ 2200 ml STP min"1; C 0 2 = 100 vol%. Inlet pH =7.50; inlet pressure =101 kPa (abs); inlet temperature = 291 K. Formate CE vs CD 100 -i o 30 20 10 . 0 -I r— , , , r , 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 CD, kA/m2 Figure 7.28 Comparison of experimental results in reactors A and B with a tin granule cathode. Operating conditions in Reactor A: catholyte = 0.45 M KHC0 3 +2 M K C l ; catholyte 153 7 Experimental Results and Discussion flow = 20 ml min 1 ; gas flow = 400 ml STP min"1; temperature (out) = 323 K; pressure (in) = 2 1 0 - 2 1 5 kPa(abs). Note: 3.1 kA m"1 = 14 A Reactor A , 100 A Reactor B. Operating conditions in Reactor B: same as under Table 7.36. - * - S E vs CD -•—[HCOO-] vs CD 400 0.00 0.50 1.00 1.50 2.00 CD, kA/m2 2.50 3.00 3.50 Figure 7.29 Specific energy and formate concentration as a function of superficial current density in Reactor B. Operating conditions are the same as under Table 7.36. Before discussion, it has to be pointed out that the experimental data for each current in Table 7.36 were taken at different operating times and in some cases with cathodes from different recycled lots of screened Batch 2 tin granules. Due to tin corrosion in the recycling sequence (see below) the catalyst particle size might change from run to run, thus resulting in some apparently inconsistent data. For example, the cell voltage at the current 101 A was lower than that at 91 A . The following paragraphs present the observations and discussions of the results shown in Table 7.36, Figure 7.28 and Figure 7.29. (i) "Corrosion" of the tin granules: 154 7 Experimental Results and Discussion As observed for some runs in Reactor A, during most runs in Reactor B a black/grey suspension appeared in the catholyte product within about 5 minutes of applying the current. This suspension was later determined to be caused by the corrosion of tin metal, involving the formation of tin oxide and/or possibly tin hydride, depending on the pH and redox conditions of the media which the tin granules were in contact with. More details of this corrosion problem will be given in Section 7.6.3. (ii) High pressure drop: It is seen from Table 7.36 that the Reactor B inlet pressures were much higher than those in Reactor A , i.e. 595 ± 95 vs -213 kPa(abs). It was also observed that, in most runs the reactor inlet pressure increased by some 200 kPa within about 5 minutes of applying the current. This sudden increase in pressure might relate to the acceleration of the corrosion of tin as the current was applied, which resulted in more suspension in the catholyte and blockage of the flow in the cathode bed and outlet flow channel. (iii) High pH: Compared to a product pH range of 7.50 to 7.80 in Reactor A , a product pH as high as 9.56 was observed in the catholyte of Reactor B. This observation suggests that, in the section of the reactor close to the outlet the pH, especially the local pH at the electrode surface (which would be about 0.5 unit higher than the bulk pH according to Gupta's modeling [Gupta, 2006]), might be too high for the reactive species CO2 (aq) to exist (see Section 2.4 for discussions on the speciation of CO2), therefore the selectivity of the process was shifted in favor of hydrogen evolution (reaction 6.2). Also, the high local pH might have promoted the corrosion of tin granules (see below) which in turn caused the suspensions to form, thus raising the feed pressure. Referring to the model (Chapter 6) it may be concluded that a system that could operate at a higher liquid and gas flow rates and/or with super-atmospheric outlet pressure would be beneficial for maintaining the local pH at the right range, thus sustaining a high formate CE throughout the reactor. (iv) Formate current efficiency (CE). 155 7 Experimental Results and Discussion From the comparison of Reactors A and B shown in Figure 7.28, we see that at superficial CD's above about 1 kA m"2 the formate CE in Reactor B is higher than that in Reactor A. This difference can be explained by the positive effect of the higher pressure in Reactor B on the C 0 2 mass transfer limited current density, since at the higher CD's the selectivity for reaction 6.1 is reduced through the C 0 2 mass transfer constraint (see equations 6.14 and 6.21 in Chapter 6). (v) Specific energy (SE) and formate concentration in the product. Figure 7.29 shows that the specific energy for formate increased with superficial 1 2 current density, reaching about 342 kwh kmol" formate at 101 A (ca. 3.14 kA m" ). The highest formate concentration obtained (one pass) in the catholyte product was 1.03 M , which has never been achieved within such a short residence time (ca. 2 minutes for the liquid phase) in the research history of ERC. 7.6.3 Corrosion of tin granules For most of the work with tin granule cathodes, the granules were pretreated with 11 wt% FTN03 for 2 minutes and then rinsed with de-ionized water, with the objective to clean and roughen the surface (see Figure 7.30 for the effect of 11 wt% FINO3 pretreatment). 156 7 Experimental Results and Discussion Figure 7.30 Effect of 11 wt% HNO3 pretreatment on Batch 1 tin granules. Left: before the pretreatment; Right: after the pretreatment (both pictures are in 1 mm magnifications). During runs at high current (8 to 14 Amp) in Reactor A where HNGvpretreated tin granules were used as the cathode, blackish/grey suspensions showed up in the product solution for about 10 minutes after the current was applied and then cleared. In Reactor B all the runs with HNGvpretreated tin granules produced more blackish/grey suspensions within extended operating periods. Sometimes the suspensions from Reactor B never cleared, until the system had to be shut down due to the rising feed pressure, which eventually (at ca. 700 kPa(abs)) caused the failure of CO2 delivery to the cathode. However, when un-pretreated fresh tin granules were used as the cathode, only a small amount of brownish suspension was observed within a few minutes after the current was applied, and the reactor pressure was stable. These observations indicated that there was a serious loss of tin associated with the HNO3 pretreatment, either in the metal form or in the oxide or hydride form, and that the suspensions blocked the fluid flow path and raised the feed pressure. Apart from the suspensions in the catholyte product solution, the gas and liquid cathode product had a faint garlic-like smell, and rusty spots were observed on both the membrane and the surface of the cathode bed touching the membrane (see Figure 7.31). 157 7 Experimental Results and Discussion Figure 7.31 Rusty spots after 180 min operation of ERC in Reactor B. Top: tin granule bed before ERC; middle: Nafion 117 membrane after ERC; bottom: tin granule bed after ERC. 158 7 Experimental Results and Discussion To understand and solve those "corrosion" problems, one needs some knowledge of tin chemistry. Unfortunately, there is limited information about tin chemistry in the literature. The following potential-pH diagram may help understand the conversion between tin and its compounds that was happening before and after the pretreatment and during the ERC. Potential vs pH -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 PH Figure 7.32 Potential- pH diagram of tin in water at 298 K [Pourbaix, 1974] From the Potential-pH diagram (Figure 7.32) it can be seen that metallic tin exists at equilibrium only within a narrow potential and pH region in water. Tin can undergo reaction, not only by oxidation, with the formation of a layer of oxide SnC>2 or the soluble derivatives Sn 2 + , Sn 4 + , HSnOy and SnC>32_' but also by reduction with the formation of the gaseous 159 7 Experimental Results and Discussion hydride SnH 4 [Pourbaix, 1974; Donaldson, 1967; Hoar, 1937]. In solutions where some ions are present, such as the electrolyte used in the present work containing HCOy, CI" and HCOO - , the region of stable tin might be narrowed due to the formation of complexes of tin with those anions [Sillen, 1964; Pourbaix, 1974]. Also an increase in temperature from 298 K probably narrows the regime of stability of Sn. During the pretreatment of tin granules with 11 wt% H N 0 3 in the present work, a straw-yellow liquid and a whitish suspension were formed. The same colors were described and explained by [Mellor, 1956] as a result of the formation of stannous or stannic nitrate which readily decomposed to a- and P- stannic acids. The formation of these acids explains why if was found in the present work that the water from rinsing HNO3 pretreated tin granules tended to stay at a pH below 5 even after a long period with multiple rinsing (15 minutes). As a result, the HNO3 pretreated tin granules, i f not neutralized with dilute alkaline solution before use in the reactor for ERC, gave a strong garlic-like smell during the reactor start-up, heavy grey/black suspensions in the product solution, and a lot of "rusty" (brown) spots on the membrane and the tin granule bed. The garlic-like smell, mostly occurring during the initial few minutes after the current was on, might be attributed to the formation of the gaseous species - tin hydride (SnH4 - stannine) at the low pH region shown in Figure 7.32, or possibly ASH3 (arsine) or SbH3 (stibine) from As and Sb impurities that total about 0.1 wt% of the tin. As the ERC proceeded, the pH at the surface of tin granule cathode was brought up by the generation of O H - accompanying the reduction of CO2 and hydrogen evolution (reactions 6.1 and 6.2) and the pH-potential regime started to move from the left to the right (Figure 7.32). As a result, the tin compounds formed during the HNO3 pretreatment converted to particulate tin oxides and/or hydroxides (e.g. black SnO, yellow-white Sn02) which gave the observed reddish-brown to black color [Mellor, 1956] and blocked the catholye flow path, thus raising the cathode feed pressure and causing the system to shut down. Nitric acid is a strong oxidizing agent, therefore a thick (e.g. 1 to 5 micron) layer of tin oxides could be expected to form during the pretreatment, which caused the serious problems discussed earlier. With this consideration, other less oxidizing or even reducing 160 7 Experimental Results and Discussion agents were investigated for the pretreatment of the tin granules, including sulfuric acid, hydrochloric acid and potassium hydroxide. Table 7.37 lists the experimental results of ERC with different pretreatment techniques. 161 Table 7.37 Experimental results of ERC with different pretreatment techniques Run No Pretreatment 11 wt% H N 0 3 + rinse 11 wt% HNO3 + rinse 11 wt% H N 0 3 + rinse No pretreatment I M H 2 S O 4 + rinse 1MHC1 +rinse 1.6 M K O H +rinse 1.6M K O H +rinse+1M HC1 +rinse+1.6M K O H + rinse Current, A 8 - 1 4 20-40 103 50-91 50 50 50 2 0 - 5 0 Reactor B B B B B B B pH, in/out 7.20/8.10 7.20/9.37 Running time, min 7.20/9.56 7.20/9.46 7.20/8.06 7.20/8.01 7.20/8.06 7.20/9.50 120 40 17 411 170 90 60 220 CE, % 76 -44 91 -86 Observations on tin corrosions Grey susp for 10 min from current on and then cleared. 65 74 -65 83 Grey/black susp from current on until pressure was too high for the system. Garlic-like smell. Grey/black susp from current on until pressure was too high for the system. Strong garlic-like smell. Grey/brown susp for 6 min from current on then cleared. Grey/brown susp for 30 min from current on then cleared 85 81 69 Brown susp for 8 min from current on then cleared. Slightly brown susp for 7 min from current on then cleared. No color and no smell. Opening conditions: cathode = recycled Batch 2 granules (30* -80" for run No 1 to 7, and 30" - 50* for run No 8); catholyte = 0 45 K H C O 3 + 2 M K C l ; temperature = 291-323 K; pressure = 220 - 700 kPa(abs). M 7 Experimental Results and Discussion Since H2SO4 is a less oxidizing acid than HNO3, a thinner layer of oxide, i f there would be any, may be formed during the pretreatment with H 2 S O 4 than with H N O 3 , and much less suspension would appear during the ERC. This assumption was proved correct by run No 5 in Table 7.37 in that the suspension lasted for 30 minutes but did not cause the system to shut down, as happened with HNO3 pretreatment (run No 2 and 3 in Table 7.37). Further proof of this assumption comes from runs No 6, 7 and 8 in Table 7.37. In run No 6, where the reducing acid HC1 was used as the pre-treating agent, the suspension lasted for only 8 minutes before the product solution cleared; in run No 7 where K O H was the pre-treating agent, only a slightly brownish suspension was observed for 7 minutes; in run No 8 where K O H and HC1 were employed sequentially no suspension was observed. It has to be emphasized here that using a less oxidizing pre-treating agent or even a reducing agent not only solved the tin "corrosion" problem but also fulfilled the pretreatment purpose, i.e. restoring the formate CE to the level that was obtained with fresh tin granules. The latter was reflected by the similar CE's achieved with different pretreatment techniques, i.e. 8 1 - 8 5 % at 50 A. The reason for the low CE of run No 8 (69 %) was that the granules used there were larger than those used in the other runs, i.e. 30# - 50# vs 30# - 80#. 7.6.4 "Recycle" runs and formate cross-over (transport through the membrane). In the conceptual process design of Chapter 8 the catholyte product solution would be recycled (see Figure 8.1) and a steady state concentration of the ERC product formate as high as 5 M is proposed to reduce the energy cost for the separation of sodium formate from NaHC03 (see Chapter 8 for details on the conceptual process design). To investigate whether the high concentration of formate in the catholyte feed would adversely affect the performance of the process, experiments were carried out in Reactor B with formate feed concentration ranging from 0.66 to 3.42 M and designated as "recycle" runs. 163 7 Experimental Results and Discussion Table 7.38 gives the results for some preliminary "recycle" runs. It is seen that as the formate concentration in the feed increased from 0.66 M in run No 1 to 2.79 M in run No 4, the formate CE dropped significantly, i.e. from 75 to 36 % at 50 A . The decrease of formate CE with increasing formate concentration in the feed could be due to any/some/all of the following reasons: (1) formate was electrochemically reduced to other species that might not be detected or detected but at a different stoichiometry than formate with the permanganate titration technique, such as methanol or formaldehyde, so that the current efficiency was underestimated; (2) formate crossed the membrane to the anode side to be partly oxidized to CO2, which reacted with OH" to generate C03 27HC03" ; (3) formate accelerated the deterioration of the tin cathode by forming complexes that changed the surface state of the cathode or adversely affected the adsorption of intermediate species involved in the reduction of CO2. To understand which of the above three reasons account for the dropping CE, an experiment without CO2 was carried out (run No 5) in which N2 gas was used to duplicate the reactor pressure and flow regime. The catholyte product (liquid) analysis showed a negative formate CE (-13 %), which indicated the possibility of both formate reduction and formate cross-over. At the same time, the catholyte product of run No 4 was analyzed by VIZON SCITEC (former BC Research) for methanol with GC (FID detector, DB-Wax column, < 50 ppm) and formaldehyde with H P L C (< 5 ppm), from which no methanol or formaldehyde were detected, thus excluding the possibility of , formate reduction. 164 7 Experimental Results and Discussion Table 7.38 Preliminary experimental results on "recycle" runs with Batch 2 tin granules in Reactor B Run No [HCOO], M pH Gas feed CE inlet outlet out % 1 0.66 1.30 9.25 C 0 2 75 2 1.44 1.80 9.39 c o 2 44 3 2.12 2.44 9.38 c o 2 44 4 2.79 2.95 9.22 C 0 2 36 5 1.55 1.44 9.68 N 2 -13 flow = 20 ml min' 1; gas flow = 1000 ml STP min"1; operating time: 20 ~ 330 min; temperature = 291 ~ 305 K; pressure = 510 ~ 630 kPa (abs). Two experiments were also conducted aiming to check the carbon balance at the cathode side with the high concentration of formate in the feed (run No 1 and 2 in Table 7.39). Results on these two runs show that the formate CE calculated from the difference of formate concentration between the catholyte inlet (feed) and outlet (product) were respectively 42 and 36 % for run No 1 and 2, and the total carbon closures were near 100 % but both with negative deviations. The total CE at the cathode side, however, did not add to 100 % for both cases, which indicated the transport of formate through the membrane. Runs 1' and 2' in Table 7.39 give the data of 1 and 2, respectively, with corrected formate CEs and carbon closures. The corrected CE was calculated by adding the measured formate CE and the difference between 100 % and the total cathode CE, assuming this difference was caused by the transport of formate to the anode side. The corrected carbon closure was then calculated based on the corrected formate CE. The corrected formate CE (i.e. the "real" CE = 68 %) and the new carbon closure for both cases are more plausible than the uncorrected values, thus indicating that formate cross-over was occuring. 165 7 Experimental Results and Discussion Table 7.39 Current efficiency and carbon balance check for "recycle" runs in Reactor B. Run No [HCOO"], M CE, % Carbon closure % inlet outlet HCOO" CO H 2 Total 1 1.55 1.89 42 3 30 75 97 2 1.55, 1.84 36 3 30 69 94 r 1.55 2.09 68 3 ' 30 100 102 2' 1.55 2.10 68 3 30 100 100 Operating conditions: current = 50 A; catholyte = 0.45 M K H C 0 3 + HCOO"; catholyte flow = 20 ml min"1; gas flow = 1000 ml STP min"1; temperature = 291 ~ 301 K; pressure = 580 kPa (abs); pH = -9.2. To confirm the transport of formate through the membrane, experiments were carried out during which the anolyte product was analyzed for both formate and bicarbonate/carbonate to see how much of the formate CE from ERC was lost due to cross-over of formate. Experimental results are listed in Table 7.40, in which the formate CE on the cathode side ("apparent" formate CE) was measured by the difference of formate concentrations between the inlet and outlet of the catholyte, the formate CE on the anode side was calculated by the amount of formate and bicarbonate/carbonate in the anolyte due to formate cross-over. The "real" formate CE was the sum of the cathode side formate CE and the anode side formate CE. Table 7.40 shows that the "apparent" CE's were respectively 26, -26, -14 and -28 % for run No 1 to 4, and the corresponding "real" CE's were respectively 85, 91, 41 and 20 %. The huge differences between the "apparent" CE's and "real" CE's again indicated the significant problem of formate cross-over. 166 Table 7.40 Experimental results on current efficiency check in Reactor B w.r.t formate cross-over Run Current A 40 20 40 50 Cathode side [HCOO"], M Inlet 2.57 3.42 3.42 3.42 Outlet 2.75 3.34 3.34 3.22 CE % 26 -26 -14 -28 Anode side [HCOO"], M Outlet 0.025 0.015 0.036 0.048 [HC0 3"], M Inlet 0.17 0.12 0.12 0.12 Outlet 0.28 0.25 0.21 0.21 CE % 59 117 55 48 Real CE % 85 91 41 20 Tin granules used Screened Batch 2 granules (-30#+ 80#) Screened Batch 3* granules (-20# + 50#) Operating conditions: catholyte = 0.5 M KHCO3 + HCOOK for run No 1 and 1 M HCOOH + HCOOK for run No 2, 3 and 4; Catholyte flow rate = 20 ml min"1; gas flow rate = 1480 ~ 2460 ml STP min"1; temperature = 290 ~ 297 K ; pressure = 471 -681 kPa (abs); pH = 5.22 - 8.19. *Batch 3 tin granules were basically the same as Batch 1 tin granules in particle shape and size distribution. 7 Experimental Results and Discussion 7.6.5 Back- pressure Back-pressure was applied to the cathode in some runs in Reactor B, to operate the process at super-atmospheric outlet pressure, with an expectation to increase the solubility and thus promote the mass transfer of CO2. However, due to the high feed pressure caused by the fine Batch 2 tin granules and consequently the difficulty to deliver C 0 2 to the reactor, there was not much room to explore the effect of back-pressure on the present system. Therefore only a few preliminary experiments were conducted. The results of the back-pressure runs are listed in Table 7.41. Runs 1 and 2 of Table 7.41 were carried out under the same conditions except for gas flow rate, i.e. 2000 and 1260 ml(STP) min"1 respectively for runs 1 and 2, and the back-pressure applied in run 2, i.e. 14 psi(gauge) = 197 kPa(abs). From the fractional factorial experimental results discussed in Section 7.5.4 and factorial experimental results in Section 7.5.5 we learned that gas flow rate had a significant positive effect on formate CE, therefore we would expect a lower formate CE for run 2 than 1 at the lower gas flow rate. However, the application of the 14 psi (gauge) back-pressure in run 2 had a positive effect on formate CE, thus resulting in the same formate CE of run 2 as run 1. In addition, the application of back-pressure under the experimental conditions decreased the reactor voltage from 3.10 to 2.98 V , which would be significant in terms of reducing the energy cost in the potential application. Of course the extra pressure increases the CO2 feed compression cost. Similarly, comparing runs 3, 4 and 5 in Table 7.41 we see that imposing a back-pressure of 30 psi (gauge) = 307 kPa(abs) on the reactor increased the formate CE from 18 % to 38 ~ 44 % at 30 A , and at the same time reduced the reactor voltage from 3.75 to about 3.56 V . 168 Table 7.41 Preliminary experimental results on the effects of back-pressure on formate CE in Reactor B Run No Electrolyte 0.5 M K H C O 3 + 2 M K C l 0.5 M K H C O 3 + 2.77 M H C O O K Current A 50 30 Gas flow rate ml(STP) min" 2000 1260 1380 1000 Back-pressure psi (gauge) 0 14 30 1000 30 Ecel l V 3.10 2.98 3.75 3.55 3.58 Formate CE % 83 83 18 44 38 Tin granules used Screened Batch 2 granules (-30#+ 80#) Screened Batch 3 granules (-20# + 50#) Operating conditions: catholyte flow = 20 ml min"1; temperature = 298 ~ 303 K; reactor inlet pressure = 601 ~ 701 kPa (abs)- pH = 8 10 8.77. Note: 30 psi (gauge)« 300 kPa(abs). 7 Experimental Results and Discussion 7.6.6 Comparison between measured and modeled formate CE's Figure 7.33 is a parity graph showing the experimental formate CE's versus the modeled formate CE's with tin granule cathode over a wide range of conditions in both Reactor A and Reactor B. This figure shows the best fit obtained by using the model equations introduced in Chapter 6 and adjusting the four cathode kinetic parameters (ki, k 2, E a , i and E a,2) elaborated on page 86-88 and tabulated here. Regression of the measured against the modeled values in Figure 7.33 gives a slope of 1.01, a regression coefficient (R2) = 0.64 and maximum deviation of +/- 20 %. Figure 7.34 shows a corresponding parametric plot of the measured and modeled formate CE versus superficial current density in each of Reactors A and B. These correlations give a measure of support to Chapter 6 but shows that more work is needed in the area of reactor modeling. U J o •o 0 ) <D TJ O 100 90 80 70 60 50 40 -| 30 20 10 0 CE meas vs model Duplicate runs excluded • Reactor A (small Reactor B (big) k, = 5.5x10^8"' k2 = 4.5x10" ms"1 E a > l =6.0xl04kJkmol"' E a ' 2 = 3.2x10* kJkmof 20 40 60 Measured CE % 80 100 Figure 7.33 Modeled vs measured formate CE. 170 7 Experimental Results and Discussion Formate CE vs CD 100 -90 -80 -^ 70-UJ 60 -O (D 50 -A ~~ ^ ^ ^ 2 40- • Meas CE in Reactor B A Forr ro GO o o • Meas CE in Reactor A — — Model CE in Reactor B — — Model CE in Reactor A 10 -0 - i i i l l 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 Current Density, kA/m2 Figure 7.34 Modeled and measured CE vs. superficial current density. 7.6.7 Observations on scale-up (1) ERC was successfully carried out in the scaled-up Reactor B using a tin granule fixed-bed cathode at a current range of 20 to 101 A with corresponding superficial current density of 0.62 to 3.14 kA m"2. (2) Serious tin corrosion occurred during the ERC with HNO3 pretreatment of the tin granule cathode, but the corrosion problem seemed to be solved by using HC1 or K O H pretreatment. (3) ERC was also performed with high concentration of formate (2 to 3 M) in the feed. Experimental results indicated serious cross-over of formate through the Nafion 117 membrane, which would challenge the proposed conceptual design in Chapter 8. (4) ERC with back-pressure on the cathode side was carried out, and the preliminary experimental results showed that applying the back-pressure on the cathode side increased the formate CE and reduced the reactor voltage. 171 8 Conceptual Process Design and Economic Projection Chapter 8 Conceptual ERC Process Design and Economic Projection 8.1 Introduction In the previous chapters the focus was on fundamental studies of the electrochemical reduction of CO2 (ERC) related to the electro-catalyst, electrochemical reactor design and reactor performance. However, an electrochemical reactor is part of a process plant, so the potential for industrial application of ERC should be determined not only by the performance of the reactor itself, but more importantly, based on the assessment of the whole process flowsheet, which includes an examination of the technological feasibility and process economics. The following case study relates to the electrochemical conversion of CO2 from the equivalent of a fossil-fuel burning electric power plant in which the CO2 from the combustion gases has been recovered in relatively pure form for subsequent disposal [R. J. Socolow, 2005]. From the perspective of the global carbon balance the electric power for the process would preferably come from a non-fossil energy source such as hydro, nuclear or solar. The basis for this conceptual design and economic analysis is a CO2 electro-reduction plant located on-site of a 80 megawatts Ontario fossil-fuel-burning power generation plant from which about 600 tonnes of CO2 is emitted per day [Ontario Power Generation website: www.opg.com]. 8.2 Conceptual process design From the present study, it is known that potassium ion is superior to sodium ion in terms of current efficiency for formate (see Section 7.3.3). Therefore potassium bicarbonate and potassium chloride (KCl) have been used as the primary catholyte and supporting electrolyte, respectively, in most of the experimental work in this project. However, sodium salts, instead of potassium salts, are employed in the conceptual process design because 172 8 Conceptual Process Design and Economic Projection sodium salts are less expensive than potassium salts. Besides, using K C l or NaCl as the supporting electrolyte in the industrial application would demand extra auxiliary equipment for separating CI" from main product HCOO", which would definitely be a disadvantage. Therefore, sodium bicarbonate and sodium formate are finally chosen as the primary and supporting electrolyte in the conceptual process design. Although there were some problems associated with high concentration of formate in the feed in ERC experiments, i.e. the transport of formate through the membrane (see Section 7.6.4), the process design is based on the assumption that a different membrane is used that prevents formate cross-over. The process design also assumes a cathode life of at least 4000 hours. The separation of formate/formic acid from the catholyte would be a major cost-intensive part of the whole ERC process. Even in the conventional industrial production of formic acid, the difficulties and main cost are associated with the separation of formic acid from water and/or other products. Actually the largest single use of formic acid world wide is in silage additives [Wiley & Sons, 2005], and mixtures of formic acid with formate salts are also used for this purpose. As outlined in Section 2.3 there is also potential for a substantial market in formic acid and/or formate salts as energy carriers (fuels) for fuel cells in a "carbon neutral" energy cycle. Figure 8.1 is the conceptual process flowsheet upon which the current economic evaluation of ERC is based. In this flowsheet, the most complicated and costly process equipment would be the electrochemical reactors, each of which usually consist of a number of individual cells arranged with either bipolar or monopolar connections. In addition to the electrochemical reactors, a number of auxiliary unit operations are employed to carry out the separation of the products, recycle the unreacted CO2 and partially recycle the electrolytes. 173 8 Conceptual Process Design and Economic Projection From C 0 2 emissions G/L Separator (SI) HCOONa NaHCO N a H C 0 3 J 1 Evaporation r C4 Cool Filter G/L Separator (S2) NaOH Liquid product HCOONa Figure 8.1 Conceptual flowsheet for ERC with NaHC0 3 and HCOONa as the catholyte 174 8 Conceptual Process Design and Economic Projection 8.3 Operational conditions In the conceptual process, the individual electrochemical cells will be scaled-up to 2.7 times as high and 5 times as wide as the present Reactor B, i.e. 1.8 m high and 0.25 m wide, which would yield a 0.45 m 2 superficial cathode area per cell. The same current density range as with Reactor B is employed, and also, gas and liquid flow rates are adjusted to set the fluid loads for the same performance as Reactor B. The individual cell voltages are assumed the same as in Reactor B and formate current efficiencies are obtained from the reactor model outlined in Chapter 6. The following operational conditions are employed in the economic evaluation of the process of Figure 8.1. Cell dimension: 1.8 m high and 0.25 m wide Cell components: Cathode: Pure tin mesh (99.9+ wt% Sn) Diaphragm: Nafion 11II some other membrane that prevents HCOO" transport Anode: Stainless steel plate and mesh Superficial current density: Current efficiency: Cell voltage: Operation mode: Pressure in/out: Temperature in/out: 0.44-2.22 kA m"2 92 to 77 % (based on the model) 3 to 5 V (from the data with Reactor B) galvanostatic, one-pass upward G/L flow 520/101 kPa(abs) 288/ 323 K (based on the model) Catholyte: Inlet composition: Outlet composition: 0.5M N a H C 0 3 + 5 M HCOONa at pH -7.5 see Table 8.1 (at pH -9.5) 175 8 Conceptual Process Design and Economic Projection Flow rate: Anolyte: Inlet composition: Outlet composition: Flow rate: CO2 flow rate: 0.35 L min"1 per cell 2 M N a O H ( ~ 8 wt%) see Table 8.1 0.35 L min"1 per cell 21 L STP min"1 per cell Table 8.1 Modeled results for the read .or in the conceptual Drocess design (bipo ar mode) Current, A 200 400 600 800 1000 Current density, kA m"2 0.44 0.89 1.33 1.78 2.22 Formate current efficiency, % 92 92 86 83 77 Volt per cell 3.4 3.6 3.8 4.0 4.2 Electrolyte composition: Catholyte Inlet N a H C 0 3 , M 0.5 0.5 0.5 0.5 0.5 HCOONa, M 5.0 5.0 5.0 5.0 5.0 Outlet N a H C 0 3 , M 0.69 0.88 1.11 1.33 1.59 HCOONa, M 5.16 5.33 5.46 5.59 5.68 Anolyte Inlet NaOH, M 2.0 2.0 2.0 2.0 2.0 Outlet NaOH, M 1.64 1.29 0.93 0.58 0.22 8.4 Gross economic potential (GEP) From the conceptual process design shown in Figure 8.1, the species flowsheet at steady state can be simplified as Figure 8.2. co 2 ERC system ^ HCOONa H 20 o 2 t NaOH —*—+ • Figure 8.2 Simplified flowsheet of the reactants and products of ERC system 176 8 Conceptual Process Design and Economic Projection The gross economic potential associated with the 80 megawatt power station (600 t/day CO2 emissions) is calculated as: GEP= [values of products]-[values of reactants] (8.1) It should be noted that the economic analysis is carried out over a range of current densities, aiming to see the effect of current density on the costs. Based on the modeled current efficiencies and the material balance calculations, the molar flow rates of all species are tabulated in Table 8.2, in which the number of electrochemical reactors, and the GEPs under each current density are also listed, along with bulk chemical prices of reactants and products. The cost calculations are based on the assumption that all products from the process are sold at the existing market prices listed in Table 8.2 [Chemical Market Reporter, 2005] and the value of the carbon credits is added as income for the process. The value of carbon credits for CO2 is now (2006) about $US25/tonne CO2 [www.pointcarbon.com]. With an expectation of increasing demand for reducing CO2 emissions, an ever rising value of carbon credits is predicted. Therefore, a carbon credit of $US100/tonne CO2 is also used for the economic analysis. Table 8.2 presents GEP's corresponding to the present carbon credit ($25US/tonne CO2) and the projected credit of $100 US /tonne CO2 are given under each current density. The gross economic potential is a basic measure of the potential feasibility of the process, and the positive values of GEP for both levels of carbon credit justify the further economic analysis. 177 Table 8.2 Molar flow rates and values of reactants and products [Chemical Market Reporter, 2005] CD, kA m-2 0.44 0.89 1.33 1.78 2.22 Price $US kg"1 2005 Formate C E , % 92 92 86 83 77 Volt per cell 3.4 3.6 3.8 4.0 4.2 Specific energy, kWh kmol"1 198 210 237 258 292 No of reactors* 761 380 254 190 152 Molar flow rates at different current density, 102x kmol h"1 C 0 2 ' 5.7 5.7 5.7 5.7 5.7 -0.025-0.10 H 2 0 0.18 0.18 0.18 0.18 0.18 0.01 NaOH 5.7 5.7 5.7 5.7 5.7 0.37 NaHCC-3 2.8 2.8 2.8 2.8 2.8 0.50 HCOONa 2.6 2.6 2.6 2.6 2.6 1.80 H 2 0.23 0.23 0.23 0.23 0.23 5.00 0 2 1.4 1.4 1.4 1.4 1.4 0.10 GEP, 104 x$US/h 3.88 3.88 3.98 3.81 3.78 $25/tonneC02 4.07 4.07 4.17 4.00 3.97 $100/tonneCO2 * Each reactor consists of 100 cells (see equations 8.5 and 8.6 for the number of reactors and cells). 8 Conceptual Process Design and Economic Projection 8.5 Net Economic Potential (NEP) and Return On Investment (ROI) The net economic potential (NEP) and return on investment (ROI) yield a more practical evaluation of the process by taking into account operating and capital costs for the full flowsheet of Figure 8.1. The net economic potential is the gross economic potential (GEP) minus the operating costs [Oloman, 1996]: NEP = GEP-Operating costs (utilities + labor and maintenance) (8.2) Normally only options with positive NEP's deserve further economic assessment and they can be ranked in order of the return on investment (ROI) which is given by the following equation with 8000 operating hours a year: ™ w n / \ JV£Px8000 ROI (% per year) = x 100 (8.3) Total • installed • capital • cos t The capital and operating cost estimates are obtained based on the preliminary process design presented in Figure 8.1. For the most important component of the flowsheet, i.e. the electrochemical unit, the total installed capital cost Ci ; E, is given by the sum of the installed capital cost of the cells QEC and the installed capital cost of the power supply system Cps [Oloman, 1996]. That is: ~CIEC + ^ F S (8-4) The installed capital cost for electrochemical reactors, Q E C is given as [Oloman, 1996]: CIEC=NR.GR\ARY (8.5) 179 5 Conceptual Process Design and Economic Projection The number of reactors (N R ) is the total number of cells (N C ) divided by the number of cells per reactor (100 cells/reactor assumed in the present design), and the total number of cells is related to the cathode area through Faraday's law, i.e. A R = electrode area per reactor = AC(NC/NR) (m2) CE = current efficiency F = Faraday's constant (96486 kC kmol"1) GR = cost constant ( $ U S m"2) i = superficial current density (kA m"2) m = scale exponent (0.8 to 0.9) no = electron stoichiometry coefficient for formate = 2 N c = total number of cells required (-) NR = number of reactors =Nc/100 (-) P r = formate production rate (kmol s"1) The capital cost of the installed single cell is taken as 2.1 *10 4 $ U S m"2 for the present case by assuming a Lang factor of 3 for an uninstalled cell cost of 7x10 3 $ U S m"2, which includes the cost of membrane (-900 $ U S m"2), tin mesh cathode (500 $ U S m"2) and stainless steel anode. Assuming that all the cell components (e.g. membrane & cathode) are replaced every six months, G R then should be doubling the installed cell cost, i.e. 4.2 x l f j 4 $ U S m"2. The installed capital cost of the power supply system Cps is [Oloman, 1996]: n0FPr iAc(CE) (8.6) Where: Ac = electrode area per cell (m2) C P S = N P P P { B ' + —} (8.7) p Nc-i-Ac-E{ Cell (8.8) 180 8 Conceptual Process Design and Economic Projection P w p=J~ (8-9) p Pp=IPVP (8.10) Where: B ' and C = cost constants (200 and 40,000 $US kW" 1, respectively, for 2005) Cps = installed capital cost of all DC supplies ($US) Ecell = voltage per cell (V) Np = the number of DC power supplies required (-) IP = the current output of the power supply (10 kA assumed) PP = the power output of each power supply (kW) Pw = total power consumption (kW) Vp = the voltage output of the power supply (1000 V assumed) The arrangement of the huge number of electrochemical cells in the present process design, including the electrical and hydraulic connections, is important [Goodridge, 1995]. The electrical connection (either monopolar or bipolar) affects a lot of factors, such as the cost of the power supply, the current by-pass, potential and current distribution and so on. In the present process design, bipolar electrical connection is adopted because of the many advantages associated with it, such as better potential and current distribution, lower resistance losses due to fewer electrical connections, avoiding expensive bus bars to each frame, etc. Hydraulic connection (the electrolyte flow either in parallel or series) determines the conversion per pass and pressure drops. Parallel electrolyte flow is utilized in the present design to assure the same conversion for each unit and maintain the designed pressure drop along the electrolyte flow. The operating cost of DC electrical energy associated with the electrochemical unit, C d ($US h"1), is expressed as [Oloman, 1996]: C P RE (8.10) 181 8 Conceptual Process Design and Economic Projection Where: C a c = cost of A C electricity (0.1 $ U S kWh"1, 2005) RE = rectifier efficiency (96%). The installed capital cost of electrochemical cells (CIEC), the installed capital cost of DC power supply system Cps, and operating cost C e i for different current densities are listed in Table 8.3. Table 8.3 Cost of electrochemical unit Current density, kA m"2 0.44 0.89 1.33 1.78 2.22 C,EC, X 1 0 9 $ U S 1.6 0.78 0.52 0.39 0.31 C P S , x l 0 7 $ U S 1.2 1.3 1.4 1.5 • 1.5 Cd, x l 0 3 $ U S h _ 1 5.4 5.7 6.0 6.3 6.7 The installed capital and operating costs for the auxiliary equipment are tabulated in Table 8.4 together with the capital and operating cost for the electrochemical units. The sizing and cost calculations for auxiliary equipment are shown in the Appendix H. The total installed capital cost (Cic) is therefore the sum of the installed capital cost of the electrochemical unit (CIEC), installed cost of power supply equipment (Cps) and the installed capital cost of auxiliary equipment (CIA). Also a 30% contingency is added to the total capital cost to capture the possible underestimation of the capital cost, which usually happens in any economic analysis. Therefore, Clc=l.3(C!EC+CPS+CIA) (8.11) The annual maintenance and labor cost (CE,ML, $ U S year"1) associated with the electrochemical plant is typically 5 to 10% of the total capital cost [Oloman, 1996]. In the present economic analysis, 10% of the total installed capital cost is chosen for the maintenance and labor cost. 182 Table 8.4 Installed capital and operating costs for all units in Figure 8.1 (2005) at various current densities CD, kA m 2 0.44 0.89 1.33 1.78 2.22 Equipment Installed capital cost $US Operating cost, $ US h 1 Installed capital cost $US Operating cost, $ US h 1 Installed capital cost $US Operating cost, $ US h 1 Installed capital cost $US Operating cost, $ US h"1 Installed capital cost $US Operating cost, $ US h 1 ERC units 1 .6xl0 9 5 . 4 x l 0 3 7 . 9 x l 0 8 5 . 7 x l 0 3 5.3x10 s 6 .0x l0 3 4.0x10 s 6 .3x l0 3 3.3x10 s 6 . 7 x l 0 3 Maint & labour 3 . 4 x l 0 4 1.7xl0 4 1.2xl0 4 8 .7x l0 3 6 . 8 x l 0 3 Compressor (PI) 2 . 1 x l 0 8 5 . 9 x l 0 2 9 . 5 x l 0 7 3 . 0 x l 0 2 7 . 1 x l 0 7 2 . 0 x l 0 2 5 .4x l0 7 1.5xl0 2 3 . 6 x l 0 7 1.2xl0 2 Compressor (P2) 2 . 5 x l 0 8 8 . 0 x l 0 2 1.1x10 s 3 . 4 x l 0 2 7 . 0 x l 0 7 1.9xl0 2 4 . 5 x l 0 7 1.2xl0 2 2 . 9 x l 0 7 7.5x10 Pump (P3) 3 . 3 x l 0 6 6.7x10 1.9xl0 6 3.3x10 1.4xl0 6 2.2x10 9.3x10 s 1.7x10 9.3x10 s 1.3x10 Pump (P4) 3 . 3 x l 0 6 6.7x10 1.9xl0 6 3.3x10 1.4xl0 6 2.2x10 9.3 x10 s 1.7x10 9.3x10 s 1.3x10 Pump (P5) 4 . 7 x l 0 5 1.0 4.7x10 s 0.6 4 . 7 x l 0 5 0.4 4.7x10 s 0.3 4.7x10 s 0.2 Pump (P6) 4 . 7 x l 0 5 1.0 4 . 7 x l 0 5 0.6 4.7x10 s 0.4 4.7x10 s 0.3 4.7x10 s 0.3 Pump (P7) 4.7x10 5 1.0 4 . 7 x l 0 5 0.6 4.7x10 s 0.4 4.7x10 s 0.3 4.7x10 s 0.3 Cooler (CI) 2.3 x l O 5 l . l x l O 2 1.4xl0 5 8.2x10 9 . 8 x l 0 4 5.7x10 8.3 x l O 4 4.9x10 6 . 9 x l 0 4 4.1x10 Cooler (C2) 4 . 9 x l 0 5 4 . 8 x l 0 2 2 . 9 x l 0 5 2 . 7 x l 0 2 2.6x10 s 1.9xl0 2 2.1x10 s 1.5xl0 2 1.8x10 s 1.2xl0 2 Cooler (C3) 5 . 7 x l 0 5 4.4x10 2 5.2x10 s 5 . 5 x l 0 2 4.7x10 s 5 .2x l0 2 4.4x10 s 606x10 2 4.1x10 s 7.2x10 2 Cooler (C4) 9 . 8 x l 0 5 1.4xl0 3 7 . 4 x l 0 5 8 .2x l0 2 6.4x10 s 5 .6x l0 2 5.4x10 s 4 . 7 x l 0 2 5.1x10 s 3 . 9 x l 0 2 Separator (SI) 6 . 0 x l 0 6 - 3 . 8 x l 0 6 - 2.8x10 s - 2.3x10 s - 1.8x10 s -Separator (S2) 3 . 8 x l 0 5 - 3 . 8 x l 0 5 - 3.8x10 s - 3.8x10 s - 3.8x10 s -H2 recovery 4 . 5 x l 0 6 - 4 . 5 x l 0 6 - 4.5x10 s - 4.5x10 s - 4.5x10 s -Evaporation 4 . 5 x l 0 7 2 . 6 x l 0 4 2 . 1 x l 0 7 1.5xl0 4 1.7xl0 7 l . O x l O 4 1.6xl0 7 7 .9x l0 3 1.3xl0 7 6.3 x l O 3 Filtering 2.3 x l O 6 - 1 .9xl0 6 - 1.3x10 s •- 1.3x10 s - 1.3x10 s -Feedtank (Fl) 4.2x10 6 - 2 . 4 x l 0 6 - 1.4x10 s - 1.4x10 s - 9.4x10 s -Feed tank (F2) 4 . 2 x l 0 6 - 2.4x10 s - 1.4x10 s - 1.4x10 s - 9.4x10 s -Total* 2 . 1 x l 0 9 7 . 0 x l 0 4 l . O x l O 9 4 . 0 x l 0 4 7.1x10 s 3.0x10" 5.3x10 s 2 . 5 x l 0 4 4.2x10 s 2 . 1 x l 0 4 00 D * Total capital cost is sum of capital costs of individual units, excluding the 30 % contingency. ERC unit includes reactors and DC power supply. 8 Conceptual Process Design and Economic Projection For clarity, compact cost data, including GEP, NEP and ROI, at a current density of 2.2 kA m"2 are provided in Table 8.5, in which the same auxiliary equipments are grouped together. Table 8.5 Cost projection for the process shown in Figure 8.1 at operating current density of 2.2 kA m" and carbon credit of 25 $US /tonne CO2 Units Utility cost, $ US year"1 Installed capital cost, $US ERC unit 5 .4x l0 7 3 . 3 x l 0 8 Compressors (2) 1.8xl0 6 6 . 5 x l 0 7 Pumps (5) 2 . 2 x l 0 5 3 . 3 x l 0 6 Coolers (4) l . l x l O 7 1.2xl0 6 Separators (2) - 2 . 2 x l 0 6 H2 recovery - 4 . 5 x l 0 6 Evaporation 5 .1x l0 7 1.3xl0 7 Filtering - 1.3xl0 6 Feed tanks (2) - 1.9xl0 6 Total 1.2xl0 8 4 . 2 x l 0 8 Total installed capital cost plus 30% contingency, $US 5 . 5 x l 0 8 Maintenance & labor, $US year"1 5 . 5 x l 0 7 GEP, $US year1 3 . 0 x l 0 8 NEP, $US year1 1.3xl0 8 ROI, % year1 24 Figures 8.3 and 8.4 show the dependence of NEP, ROI, total operating cost (COP) and total installed capital cost amortized over 10 years (CIC), respectively, on superficial current density. Table 8.6 lists the NEP and ROI at a C 0 2 credit range of 25 to 300 $US/tonne with operating current density of 2.2 kA m"2. It has to be noted that the CO2 credit alone would not make the process economically viable, and the product 184 8 Conceptual Process Design and Economic Projection formate as well as the by-products bicarbonate, hydrogen and oxygen have to be sold to get the 24 %/year ROI. Superficial C D , kA/m2 Figure 8.3 The net economic potential (NEP) and return of investment (ROI) as a function of superficial current density for ERC process shown in Figure 8.1. C 0 2 credit is 25$ US/tonne. Table 8.6 NEP and ROI as various C Q 2 credits C 0 2 credit, $US/tonne 25 100 200 300 NEP at 2.22 kA m"2, (108) $US/year 1.32 1.47 1.67 1.87 ROI at 2.22 kA m"2, %/year 24 27 31 34 185 8 Conceptual Process Design and Economic Projection 0.00 0.50 1.00 1.50 .2.00 2.50 CD, kA/m2 Figure 8.4 Amortized total capital cost (CIC) and total operating cost (COP) as a function of superficial current density for the ERC process of Figure 8.1 Over the range of current density in Figure 8.4 both the utility and labor & maintenance components of the operating cost decrease monotonically with increasing current density (see Table 8.4). The decrease in utility cost with increasing CD is contrary to the usual expectation for an electrochemical process. This decrease is due to the fact that in the present process design increasing the C D raises the CO2 conversion and the salt concentration in the catholyte, thus lowering the energy load for recycle and crystallization. These auxiliary energy savings outweigh the increase in the cost of energy used to drive the electrochemical reactors as the CD moves up to about 2.2 kA m"2. The above cost estimates (e.g. Figure 8.3) show that in the present economic climate the electro-reduction of CO2 to formate can be commercially viable only at superficial current densities exceeding about 1 kA m"2, and a decent return on investment would require a current efficiency above 70 % at current densities in the range of 2 to 3 kA m"2. With the possible exception of gas diffusion electrodes the previous work on 186 8 Conceptual Process Design and Economic Projection ERC (summarized in Table 2.5) used reactor and process conditions that could not sustain a commercial ERC process. The trickle-bed electro-chemical reactor, explored in the present work for the first time in electro-reduction of CO2, does have potential to be the basis of a commercially viable ERC process. 187 9 Conclusions and Recommendations Chapter 9 Conclusions, Contributions and Recommendations 9.1 Conclusions The electro-chemical reduction of carbon dioxide (ERC) has been carried out in continuous single-cell laboratory reactors using cation membrane separators and operated in the "trickle-bed" mode, with co-current flow of reactant gas and catholyte liquid through flow-by 3-D cathodes. This work was performed first in a small reactor (Reactor A: 150 mm high by 30 mm wide), then in a 7-fold scaled-up reactor (Reactor B: 680 mm high by 50 mm wide) to demonstrate the feasibility of reduction of CO2 to formate (HCOO") on an industrial scale. Extensive experimental work was carried out in Reactor A to investigate different kinds of 3-D cathodes, as well as anode materials, to establish a carbon balance and subsequently to examine the effects of 10 process variables on reactor performance, using factorial, fractional factorial and parametric experimental designs. The studied process variables include catalyst life (operating time), cathode specific surface area, cathode thickness, catholyte flow rate and gas flow rate, current density, CO2 concentration in the feed gas, electrolyte conductivity, electrolyte species and temperature. In Reactor B, the consequences of scale-up were studied, using currents up to 101 A (equivalent to a superficial current density of 3.14 kA m"2), together with the effects of cathode pretreatment techniques, back-pressure and formate concentration in the catholyte. In addition, a reactor (cathode) model was developed and used to guide the experimental work, plus a process synthesis exercise was conducted for a preliminary design and economic projection of a (speculative) industrial ERC process. 9.1.1 Electro-catalysts 188 9 Conclusions and Recommendations Among the catalysts (cathodes) prepared and used in the present work, tin coated copper mesh and tin granules were the most intensively studied for ERC. The tin coated copper mesh prepared in-house showed useful activity for CO2 reduction and selectivity for the production of formate but exhibited rapid deterioration with time after 10 minutes of operation. This deterioration was due to the progressive loss of the tin from the cathode surface to expose the copper substrate, with a consequent increase in the exchange current density for hydrogen evolution that promoted the electro-reduction of water relative to that of carbon dioxide. Tin granules, after being pretreated with appropriate techniques (Section 7.6.3), exhibited not only high superficial current density for the production of formate but also sustained the formate current efficiency for 200 minutes in operation. However, the tin granules eventually deteriorated after about 200 minutes due to the accumulation of impurities and/or reaction products on the cathode surface and/or in the catholyte solution. Also, when HNO3 pretreated tin granules were used as the cathode catalyst serious tin corrosion occurred during the ERC, with the appearance of black/grey suspensions in the catholyte, that partially blocked the flow path and raised the pressure drop. The cathode corrosion problem was solved by using HC1 and/or K O H pretreatment of the tin granules. 9.1.2 Carbon balance and selectivity The experimental results indicated that the primary and secondary cathode products were respectively formate and hydrogen (together accounting for 95 to 99% of the current), with tertiary products including CO, C H 4 , and C2H4. The integrity of the experimental data was confirmed by a carbon balance over Reactor A , in which the average steady-state "carbon closure" (defined in Appendix A) from 16 factorial runs was 101 %, with a standard deviation of 3 %. 9.1.3. Overall cell reactions 189 9 Conclusions and Recommendations As well as the cathode reactions, the overall (full cell) reactions are important for the material and energy balances used in the process synthesis of a complete ERC plant. The overall reactor process in this work consumed hydroxide from the anolyte and resulted in the transfer of CO2 into the catholyte to give formate, together with bicarbonate to match the hydroxyl generation in the cathode reactions. The material balance for the process thus involved the transport of cations (mostly K + in the present work) across the membrane to match the total conversion of CO2, while a fraction of these cations (equivalent to the formate yield) went to produce the formate salt. As well as the bicarbonate, major by-products of this process were stoichiometric oxygen and substantial amounts of hydrogen. 9.1.4 Process variables The ranges of the process variables and figures of merit for Reactors A and B are given below in Table 9.1. Observations on the effects of the 10 process variables, supported by the model, were obtained from ERC experiments in Reactor A with tin coated copper mesh and tin granule cathodes, as follows: a. Increase in superficial current density and temperature both lowered the formate current efficiency; b. Increase in CO2 concentration in the feed gas, specific area of the cathode and thickness of the cathode (within the limits of electro-active thickness) all resulted in the increase in formate current efficiency; c. The effect of the supporting electrolyte K C l on formate current efficiency, interacting with the effect on CO2 mass transfer and electro-active thickness, was positive or negative depending on the thickness of the cathode; d. With respect to the cations and anions studied, the formate production was favored over hydrogen evolution in the order of K >Na and HC03>C1>C03 "; 190 9 Conclusions and Recommendations e. Increases in catholyte liquid flow rate and gas flow rate both increased the mass transfer capacity of CO2, and thus raised the formate current efficiency by promoting formate production over the secondary hydrogen evolution reaction. More detailed observations on the process variables can be found in Section 7.3.7. 9.1.5 Scale-up ERC was carried out in the scaled-up Reactor B at inlet pressures from 350 to 600 kPa(abs) and outlet (liquid product) temperatures 295 to 325 K , with catholyte and anolyte respectively [0.5 M K H C O 3 + 2 M KCl] and 2 M K O H . Conclusions are listed below, and ranges of major process variables and figures of merit are presented in Table 9.1. a. Formate current efficiencies achieved in the scaled-up Reactor (B) were at the same levels as that in Reactor A , i.e. 91 to 63 % formate CE, corresponding to superficial current density of 0.62 to 3.14 kA m" (20 to 101 A in Reactor B), under the same range of reactor voltages (2.7 to 4.3 V), which reflects the success of scale-up in the present work b. Formate concentration up to 1 M was achieved in a single pass through Reactor B. However when the catholyte feed was spiked with 2-3 M potassium formate the results indicated serious cross-over of formate through the Nafion 117 membrane, with subsequent loss of formate by oxidation at the anode. c. Preliminary experimental results on the effects of back-pressure up to 300 kPa(abs) showed that imposing a back-pressure improved formate current efficiency and reduced the reactor voltage. 191 9 Conclusions and Recommendations Table 9.1 Ranges of major process variables and figures of merit Units Range Process variables Reactor A Reactor B Reactor pressure (inlet) kPa(abs) 120-240 350-600 Reactor temperature (outlet) K 288-328 295-325 Catholyte pH (outlet) 7-8.5 8.1-9.6 Catholyte composition - (K,Na)(HC0 3 , C0 3,C1) (K)(HC0 3.C1, H C 0 2 ) Catholyte liquid flow rate ml min"1 8-20 10-22 Cathode thickness mm 0.6-3.2 3.2 Cathode specific surface 2 -3 m m (2-17)x 103 (11-17) x 103 Gas flow rate ml(STP) min"1 180-1600 1000-2500 Gas CO2 feed concentration Vol % 16-100 100 Operating time min 5-420 5-352 Figures of merit Current density kAm" 2 0.22-3.14 0.62-3.14 Current efficiency for formate % 16-96 18-91 Conversion of CO2 % 7-80 18-71 Formate product concentration M 0.01-0.08 0.28-1.03 Reactor voltage V 2.0-6.0 2.7-4.3 Space-time-yield of HCOO" kmol m" s" (2-23)xl0"4 (9-30)xl0"4 Specific energy for HCOO" kWh kmol"1 130-1300 160-350 9.1.6 Modeling The steady-state reactor (cathode) model was based on the assumption of parallel reactions of CO2 to HCOO" and H2O to H2, with uniform current distribution over the 3-D cathode in a plug-flow reactor. Using four adjustable parameters representing the intrinsic kinetics of the cathode reactions (kj, k2, E a > i , E ^ ) , with 27 experimental points embracing a wide range of conditions in both Reactor A and Reactor B, the modeled values of formate CE 192 9 Conclusions and Recommendations regressed with the measured values to give a slope of 1.01, a regression coefficient (R 2) = 0.64 and maximum deviation of +/- 20 % (Figure 7.33 in Section 7.6.6). 9.1.7 Process synthesis & economics A conceptual process flowsheet was developed (Figure 8.1 in Chapter 8) for a (speculative) ERC process to treat 600 tonne/day CO2 (from an 80 M W fossil fuel burning power station) to produce sodium formate, with by-products sodium bicarbonate, hydrogen and oxygen. The preliminary process design and economic analysis shows that in the present economic climate the electro-reduction of CO2 to formate can be commercially viable only at superficial current densities exceeding about 1 kA m - 2 , and a decent return on investment would require a current efficiency above 70 % at current densities in the range of 2 to 3 kA m"2. The modeling calculations indicate that at 2.2 k A m"2 such a plant would have a capital cost ($US 2005) 5.5 x i o 8 +/- 20%. If all products were sold at prevailing bulk chemical prices, with a carbon credit of 25$US per tonne CO2, the return on investment is estimated as 24 %/year. 9.2 Novelty and contributions of the present work The following aspects of the present work are the "first time"s in the published research history of ERC: a. Operation of a continuous reactor for electro-reduction of CO2 with 2-phase (G/L) flow. b. Exploration of a broad multi-variable process space for ERC by factorial and parametric experimental design. c. Scale-up in the order of 100 fold in reactor size, current, CO2 processing capacity and space-time yield compared to the published prior work on ERC (Tables 2.5 and 9.1). 193 9 Conclusions and Recommendations d. Formate concentration up to 1 M in a single-pass, with a catholyte residence time of about 2 minutes. e. Operation of an ERC reactor at industrially practical pressures (up to 600 kPa(abs)) and superficial current densities (1 to 3 kA m"2) with useful current efficiency (90 to 63 %) and specific energy consumption (160-350 kWh kmol"1 formate). f. Operation of a continuous ERC reactor for unbroken periods up to 7 hours. g. Reactor (cathode) modeling. h. Conceptual process design and economic projection. In sum, the present work has broken new ground with respect to the design of electrochemical reactors for CO2 reduction and demonstrated the feasibility of carrying out this process on an industrial scale - with the caveat that the effective cathode life must be extended from three to several thousand hours and an improved cation membrane separator is needed to prevent formate cross-over. 9.3 Recommendations for future work The major goal of further work should be in the direction of bridging the gap between laboratory work and industrial reality. Specific recommendations are as follows: 1. Acquire a pure tin mesh or felt cathode which would presumably have the advantages of tin coated copper mesh (high selectivity for formate and low pressure drop), tin granules (high specific surface area) and tin shot (long life) that have been seen in the present work. 2. Study the effects of increased pressure (back-pressure) and decreased CO2 feed concentration on reactor performance (Reactor B). 194 9 Conclusions and Recommendations 3. Study the deterioration of the tin catalyst in order to prolong the operating life. Particularly, study the effect of high concentrations of formate (>2 M) and electrolyte impurities (trace elements) on the cathode surface properties and cathode life. This work may be combined with investigations of chelating agents and polarity reversal. 4. Study other metals and alloys (e.g. Pb/Sn alloys) as cathode materials. 5. Investigate the reasons for formate cross-over and search for ways of reducing it, such as using counter-flow of anolyte and catholyte or new membranes that would suppress the transport of formate. 6. Perform ERC with recycles of both anolyte and catholyte. Study the potential crystallization problem of potassium/sodium salts accumulated in the catholyte. Investigate the operability of separating of formate from bicarbonate through crystallization, separating hydrogen from C 0 2 by adsorption, and similar issues associated with the complete ERC process flowsheet. 7. Investigate the control and optimization of catholyte pH through the use of acidic anolytes, along with corresponding ERC process concepts for the production of formic acid. 8. Carry out more work in reactor modeling. 9. Look for alternative anode reactions to make the whole process more economic. Such reactions include, for example, the production of chlorine, persalts or hydrogen peroxide instead of oxygen. 10. Investigate the optimization of the reactor design including the flow regime, reactor geometry (width and height), cathode specific surface and voidage, etc. in combination with process variables such as current density, pressure and temperature. 195 References REFERENCES Akahori, Y., Iwanaga, N . , Kato, Y., and Kamamoto, O., Electrochemistry (Tokyo, Japan), 72 (4), 266 (2004). Akahori, Y., Iwanaga, N . , Hato, Y., and Ishii, M . , Electrochem. (Tokyo, Japan), 72 (4), 266 (2004). Arakawa, H., Aresta, M . , and Armor, J. N . , Chem. Rev., 101, 953 (2001). Attard, G. S., Electrochemistry Communications, 3, 544 (2001). Attard, G. S., J. M . Elliott, Science, 278, 838 (1997). Audus, H., Energy, 22(2/3), 217 (1997). Ayers, W. M., Carbon Dioxide Chemistry: Environmental Issues, Ed by Paul, J., and Pradier C. M . , The Royal Society of Chemistry (1994). Azuma, M., Hashimoto, K. , Hiramoto, M . , Watanabe, M . , Sakata, T., J. Electrochem. Soc, 137, 1772 (1990). Bandi, A, Specht, M . , Weimer, T. and Schaber, K. , Energy Convers. Mgmt., 36(6-9), 899 (1995). Benefield, L., Judkins, J., and Weand, B., Process Chemistry for Water and Wastewater Treatment, Prentice-Hall, Inglewood Cliffs (1982). Bestetti, M., Vicenzo, A. , and Cavallotti, P. L., AESF SUR/FIN Proceedings, 745 (2002). Box, G. E. P. and Hunter, J. S., Technometrics, 3 (3), 311 (1961). Boyd, A. W., and Willeis, C , Int. J. Radiat. Phys. Chem., 8 (1-2), 71 (1976). Chaplin, R. S. P., and Wragg, A . A. , J. Applied Electrochemistry, 33, 1107 (2003). Chunshan, S., Gaffney, A . M . , and Fujimato, K., CO2 Conversion and Utilization, American Chemical Society, Washington DC (2002). Comstock, C. S., and Dodge, B. F., Ind. Eng. Chem., 29, 520 (1937). Conway, B. E., Electrochemical data, Elsevier, Amsterdam (1952). Cook, R. L., MacDuff, R. C , and Sammells, A . F., J. Electrochem. Soc, 137 (2), 607 196 References (1990). Cook, R. L., MacDuff, R. C , Sammells, A . F., J. Electrochem. Soc, 135, 1320 (1988). Cook, R. L., MacDuff, R. C , Sammells, A . F., J. Electrochem. Soc, 136, 1982 (1989). Cook, R. L., MacDuff, R. C , Sammels, A . F., J. Electrochem. Soc, 134, 1873 (1987). Coppola, L., Cavatort, O., Bohm, U. , J. Appl. Electrochem., 19, 100 (1989). Cotton, F. A. , and Geoffrey Wilkinson, F. R. S., Advanced Inorganic Chemistry-A Comprehensive Text, New York, London, Sydney, Toronto (1972). DeWulf, D. W., Jim, T., Bard, A . J., J. Electrochem. Soc 136, 1686 (1989). Donaldson, J. D., Progress in Inorganic Chemistry, Vol 8, InterScience John Wiley (1967). Eliasson, B., Carbon Dioxide Chemistry: Environmental Issues, Ed by Paul, J., and Pradier, C. M . , The Royal Society of Chemistry (1994). Elliott, J. M., Attard, G. S., Chem. Commun, 331(1999). Elliott, J. M., Attard, G S., Chem. Mater., 11, 3602 (1999). Elliott, J. M. , Attard, G. S., Chem. Mater., 14, 524 (2002). Elliott, J. M., Attard, G. S., Journal of New Materials for Electrochemical Systems, 2, 239 (1999). Elliott, J. M., Attard, G. S., Langmuir, 15 (22), 7411(1999). Falkowski, P., Scholes, R. J., Boyle, E., Canadell, J., Canfield, D., Elser, J., Gruber N . , Hibbard, K., Hogberg, P., Linder, S., Mackenzie, F. T., Moore in, B. , Pedersen, T., Rosenthal, Y. , Seitzinger, S., Smetacek, V. , and Steffen, W., Science, 290 (2000). Frese, K. W., and Jr, Electrochemical and Electrocatalytic Reactions of Carbon Dioxide, Elesevier Amsterdam (1993). Frese, K. W., Jr., and Canfield, D., J. Electrochem. Soc, 131, 2518 (1984). Frese, K. W., Jr., and Leach, S., J. Electrchem. Soc, 132,259 (1985). Frese, K. W., Jr., and Summers, D. P., Patent No. US 4609441 (1985). Furuya, N., Yamazaki, T., Shibata, M . , J. Electroanal. Chem., 431, 39 (1997). Gan, Y. X., Materials and Manufacturing Processes, 9 (2), 263 (1994). 197 References Goodridge, F., and Scott, K. , Electrochemical Process Engineering: A Guide to the Design of Electrolytic Plant, Plenum Press, New York and London (1995). Goodridge, F., and Presland, G., J. Appl. Electrochem. 14, 791 (1984). Greenewalt, Ind. Eng. Chem., 18, 1291 (1926). Gupta, N., Gattrell, M . , and Macdougal, B., J. App. Electrochem., 36, 161 (2006). Gupta, N., Scale-up of the Trickle Bed Eletrochemical Reactor for the Generation of Alkaline Peroxide (UBC Thesis) (2004). Hara, K., and Sakata, T., J. Electrochem. Soc, 144(2); 539(1997). Hara, K., Kudo, A. , and Sakata, T., J. Electrochem. Soc, 142(4), L57 (1995). Hara, K., Kudo, A. , Sakata, T., J. Electroanal. Chem., 391, 141 (1995). Hara, K., Kudo, A. , Sakata, T., J. Electroanal. Chem., 421, 1 (1997). Harned, H., and Davies, R., J. Am. Chem. Soc, 65, 2030 (1943). Harte, C. R., Jr., and Baker, E. M . , Ind. Eng. Chem., 25, 1128 (1933). Harte, C. R., Jr., Baker, E. M . , and Purcell, H . H., Ind. Eng. Chem., 25, 528 (1933). Hashimoto, K., and Fujishima, A. , Carbon Dioxide Chemistry: Environmental Issues, The Royal Society of Chemistry (1994). Hinogami, R., Nakamura, Y. , and Yae, S., J. Phys. Chem. B, 102, 9834 (1998). Hoar, T. P., Trans Faraday Soc, 33, 1152 (1937). Hochgesand, G., Industrial and Engineering chemistry, 62(7), 37 (1970). Hodgson, I., Oloman, C , Chem. Eng. Sci., 54, 5777 (1999). Hori, Y., and Murata, A. , Bull . Chem. Soc. Jpn., 64,123 (1991). Hori, Y., Kikuchi, K. , and Susuki, S., Chem. Lett., 11, 1695 (1985). Hori, Y., Kikuchi, K. , Murata, A. , and Suzuki, S., Chem. Lett., 897 (1986). Hori, Y., Murata, A. , and Takahashi, R., J. Chem. Soc, Faraday Trans. 1, 85 (8), 2309 (1989). 198 References Hori, Y . , Murata, A. , and Yoshinami, Y . , J. Chem. Soc., Faraday Trans., 87 (1), 125 (1991). Hori, Y . , Murata, A. , Kikuchi, K. , Chem. Lett., 1665 (1987). Hori, Y . , Murata, A. , Takahashi, R., and Suzuki, S., J. Chem. Soc, Chem. Commun., 17 (1988). Hori, Y . , Wakebe, H. , Tsukamoto, T., Koga, O., Electrochim. Acta., 39 (11/12), 1833 (1994). Ikeda, S., Takagi, T., Ito, K. , Bull. Chem. Soc. Jpn., 60, 2517 (1987). Ito, K., and Ikeda, S., Bull. Chem. Soc. Jpn., 60, 2517 (1987). Jaeyoung, L., and Yongsug, T., Electrochim. Acta, 46, 3015 (2001). Jee, H. S., Nishio, N . , and Nagai, S., J. Ferment. Technol., 66 (2), 235 (1988) Jermann, B., Augustynski, J., Electrochim. Acta., 39, 1893 (1994). Jermann, B.,and Augustynski, J., Electrochimica Acta 39(11/12), 1891 (1994). Jesus-Cardona, H. D., Moral, C. D., Cabrera, C. R., J. Electroanal. Chem., 513, 45 (2001). Jitaru, M., Lowy, D. A. , Toma, M . , Toma, B. C , Oniciu, L., J. Appl. Electrochem., 27, 875 (1997). Kaneco, S., Hiei, N . , Suzuki, T., and Ohta, K., Electrochim. Acta., 48, 51 (2002). Kaneco, S., Iiba, K. , Ohta, K. , Mizuno, T., Saji, A . , Electrochim. Acta, 44, 573 (1998). Kaneco, S., Iiba, K. , Ohta, K. , Mizuno, T., Saji, A. , J. Electroanal. Chem., 441, 215 (1998). Kaneco, S., Ohta, K. , Electrochimita Acta, 48 (1), 51 (2002). Kapusta, S., and Hackerman, N . , J. Electrochem. Soc, 130 (3), 607 (1983). Keene, F. R., Electrochemical and Electrocatalytic Reactions of Carbon Dioxide, Elsevier, Amsterdam, London, New York, Yokyo (1993). Kilnam, H., Nam, M . , Park, D. J., Proceedings of the Fifth International Symposium on Redox Mechanisms and Interfacial Properties: Properties of Molecules of Biological Importance. 93(11), 248 (1993). Kim, J. J., Summers, D. P., and Frese, K. W., Jr., J. Electroanal. Chem., 245, 223 (1988). 199 References Koleli, F., Atilan, T., and Palamut, N . , Fresenius Environmental Bulletin, 11(6), 278(2002). Koleli, F., Atilan, T., and Palamut, N . , J. Appl. Electrochem., 33, 447 (2003). Koleli, F., Balun, D., Applied Catalysis A: General, 274, 237 (2004). Kudo, A. , Nakagawa, S., Tsuneto, A. , Sakata, T., J. Electrochem. Soc., 140,1541(1993). Kyriacou, G. Z., and Anagnostopoulos, A . K. , J. Appl. Electrochem., 23, 483(1993). Kyriacou, G., and Anagnostopoulos, A. , J. Electroanal. Chem., 332, 233 (1992). Ladouceur, M., Belanger, G., and Ashok, K. V . , Canadian Chemical News, June, 17 (2001). Li, H., and Oloman, C , J. Appl. Electrochem., 35, 955 (2005). Lide, D. R., CRC Handbook of Chemistry and Physics, London, New York (2004). Linke, W. F., Solubilities: Inorganic and Metal-Organic Compounds, D.Van Nostrand Princeton (1958). Mahmood, M. N., Masheder, D., Harty, C. J., J. Appl. Electrochem., 17, 1159 (1987). Masliy, A . I., and Poddubny, N . P., J. Appl. Electrochem., 27, 1036 (1997). Mclntyre, J. A . , and Phillips, R. F., Patent No. US 4406758 (1983). Mclntyre, J. A. , and Phillips, R. F., Electrochemical Society Interface, Spring, 29 (1995). Mignard, D., Sahibzada, M . , Duthie, J. M . , and Whittington, H. W., International Journal of Hydrogen Energy, 28, 455 (2003). Mizuno, T., Naitoh, A. , Ohta, K. , J. Electroanal. Chem., 391, 199 (1995). Mizuno, T., Ohta, K. , and Sasaki, A . , Energy Sources, 17, 503 (1995). Momose, Y . , Sato, K. , and Ohno, O., Surface and Interface Analysis, 34, 615 (2002). Montiel, V., Tryk, D. A. , Aldaz, A. , Fujishima, A. , Pure Appl. chem., 73 (12), 1917 (2001). Mukerjee, S., Srinivasan, S., J. Electrochem. Soc, 142, 1409 (1995). Murata, A . and Hori, Y . , Bull. Chem. Soc. Jpn., 64, 123 (1991). Murphy, T. D., Chemical Engineering, 6,168 (1977). 200 References Naitoh, A. , Ohta, K. , Mizuno, T., Yoshida, H., Sakai, M . , Noda, H. , Electrochim. Acta., 38, 2177(1993). Nakagawa, S., Kudo, A. , Azuma M . , Sakata, T., J. Electroanal. Chem., 308, 339 (1991). O'Grady, W. E., Ross, P. N . , Wi l l , F. G., Electrocatalysis, Electrochemical Society, Pennington, NJ(1982). Ogura, K., and Endo, N . , J. Electrochem. Soc, 146 (10), 3736 (1999). Ogura, K., Endo, N . , andNakayama, M . , J. Electrochem. Soc, 145, 3801 (1998). Ogura, K., Yano, H. , A C S Symposium Series ( C 0 2 Conversion and Utilization), 809,344 (2002). Oloman, C , and Watkinson, A . P., J. of Appl. Electrochem., 9, 117(1979). Oloman, C , Ben Lee and Leyten, W., The Canadian Journal of Chemical Engineering, 68, 1004(1990). Oloman, C , Journal of the Electrochem. Soc, 126(11), 1886 (1979). Oloman, C , Electrochemical Processing for the Pulp and Paper Industry, The Electrochemistry Consultancy, Romsey, U K (1996). Paik, W., Andersen, T. N . , and Eyring, H., Electrochim. Acta, 14, 1217 (1969). Palmer, D. A . , and Eldik, R. V. , Chemical Reviews, 83(6), 651 (1983). Perry, R. H., and Chilton, C. H., Chemical Engineers' Handbook, 5 t h ed., McGraw Hi l l , New York (1973). Pickett, D. J., Electrochemical Reactor Design, Elsevier, Amsterdam (1977). Pourbaix, M., Atlas of Electrochemical Equilibria in Aqueous Solutions. 475 (1974). PLESSEY CO LTD, Patent No: GB 2041342 (1980). Prentice, G., Electrochemical Engineering Principles, Prentice Hall, New Jersey (1991). Qin, D., Lapszewicz, J., and Jiang, X . , J. Catalysis, 159,140 (1996). Russell, P. G., Kovac, N . , Srinivasan, S., J. Electrochem. Soc, 124 (9), 1329 (1977). Ryu, J., Andersen, T. N . , and Eyring, H., J. Phys. Chem., 76, 3278 (1972). 201 References Saeki, T., Fujishima, A. , J. Phys. Chem., 99 (20), 8440 (1995). Sammels, A . F. and Cook, R. L., Electrochemical and Electrocatalytic Reactions of Carbon Dioxide, Elsevier, Amsterdam, London, New York, Yokyo (1993). Sanchez -Sanchez, C M . , Montiel, V. , Pure Appl. Chem., 73(12), 1917 (2001). Sanchez-Sanchez, X . M . , Montiel, V. , Tryk, D. A. , Aldaz, A. , and Fujishima, A. , Pure Appl. Chem., 73 (12), 1917(2001). Schibioh, M. A . and Viswanathan, B., Proc. Indian Natn. Sci. Acad., 70 A (3), 407 (2005). Seshan, K., and Lercher, J. A. , Carbon Dioxide Chemistry: Environmental Issues, The Royal Society of Chemistry (1994). Shiratsuchi, R., Aikoh, Y . , Nogami, G., J. Electrochem. Soc, 140, 3479 (1993). Sillen, L. G. , and Martel, A . E., Stability Constants of Metal Ion Complexes, Chem. Soc, London, Special publication 17 (1964). Smetacek, V., and Steffen, W., Science, 290(2000). Smith, B. D., Irish, D. E., kedzierzawski, P., Augustynski, J., J. Electrochem. Soc, 144, 4288 (1997). Socolow, R. J., Scientific American., 293, 49-55 (2005). Souma, Y . , Ando, H. , and fujiwara, M . , Carbon Dioxide Chemistry: Environmental Issues, The Royal Society of Chemistry (1994). Specchia, V., Baldi, G., Chem. Eng. Science, 32, 515 (1977). Spichiger-UImann M., and Augustynski, J., J.Chem. 10, 487 (1986). Spichiger-Ulmann, M., and Augustynski, J., J. Chem. Soc, Faraday Trans., 181, 713 (1985). Stephan, K., and Vogt, H. , Electrochim. Acta, 24, 11 (1979). Sullivan, B. P., Krist, S. K., and Guard, H. E., Electrochemical and Electrocatalytic Reduction of Carbon Dioxde, Elsevier, Amsterdam (1993). Summers, D. P., Frese, K. W., Jr., J. Electrochem. Soc, 135, 264 (1988). Summers, D. P., Frese, K. W., Jr., Langmuir, 4, 51(1988). 202 References Summers, D. P., Leach, S., and K. W. Frese Jr., J. Electroanal. Chem. 205, 219(1986). Teeter, T. E., and Ryselberghe, P. V . , J. Chem. Phys., 22, 759 (1954). Todoroki, ML, Hara, K. , Kudo, A. , Sakata, T., J. Electroanal. Chem. 394, 199 (1995). Tryk, D. A . , and Fujishima, A. , The Electrochemical Society Interface. Spring, 32 (2001). Udupa, K. S., Subramanian, G. S., and Udupa, H. V . K. , Electrochim. Acta, 16, 1593 (1971). Vassiliev, Y . B., Bagotsky, V . S., Osetrova, N . V. , Kahzova, O. A. , Mayorova, N . A. , J. Electroanal. Chem., 189, 295 (1985). Vassiliev, Y . B., Bagotzky, V . S., and Osetrova, V . V. , J. Electroanal. Chem., 189, 271 (1985). Vogel, A . I., A Text Book of Quantitative Inorganic Analysis, Longmans, New York (1978). Walker, A . C , Bray, U . B. and Johnstone, J., J. Am. Chem. Soc., 49, 1235(1927). Walsh, F. C , A First Course in Electrochemical Engineering, the E C C Electrochemical Consultancy, England (1993). Wan, Y . Z., Wang, Y . L. , Journal of Materials Science Letters, 16, 1561 (1997). Wasmus, S., Cattaneo, E., and Vielstich, W., Electrochim. Acta, 35 (4), 771 (1990). Wasmus, S., Cattaneo, E., Vielstich, W., Electrochim. Acta, 35, 771 (1990). Watanabe, M., and Shibata, M . , J.Electroanal.Chem., 305, 319 (1991). Watanabe, M., and Shibata, M . , Proceedings of the Symposium of Environmental Aspects of Electrochemistry and Photoelectrochemistry, 18, 55 (1993). Watanabe, M., Shibata, M.,and Kato, A. , J. Electrochem. Soc, 138 (11), 3383 (1991). Wikipedia Encyclopedia online: http://enwikipedia.org/wiki/carbon cycle (2005). Wiley, J., and Sons, Kirk-Othmer Encyclopedia of Chemical Technology (2005). Williams, R., Crandall, R. S., and Bloom, A. , Appl. Phys. Lett., 33 (5), 381 (1978). Wong, C. S., Tishchenko, P. Y. , and Johnson, W. K. , J. Chem. Eng. Data, 50, 817 (2005). Xuebing, Li, Qunying, Yan, Fenzi Cuihua, 6 (4), 297 (1992). 203 References Yamada, N., Yaguchi, T., Journal of the Electrochem. Soc, 146 (7), 2587 (1999). Yamamoto, T., Tryk, D. A. , Fujishima, A. , and Ohata, H. , Electrochim. Acta, 47, 3327 (2002). Yamamoto, T., Tryk, D. A. , Hashimoto, K. , J. Electrochem. Soc, 147 (9), 3393 (2000). Yano, H., Shirai, F., Nakayama, M . , Ogura, K. , J. Electroanal. Chem., 519, 93 (2002). Yaws, C , Physical properties, McGraw Hill New York (1977). Zhu, Z., Kuang, X . , and Carotenuto, G., J. Materials Science, 32, 1061 (1997). 204 Nomenclature Nomenclature A C electrode area per cell m 2 A R electrode area per reactor = AC(NC/NR) m 2 a specific surface area of cathode m"1 dp,i opening size of screen i in screen analysis m a2Cu Tafel constant for reaction 6.2 on copper V a2Sn Tafel constant for reaction 6.2 on tin V bi Tafel slopes for reaction i V/decade C total concentration of bicarbonate and carbonate in eq. 2.6 N Cac cost of A C electricity (0.1 $ kWh - 1 , 2005) $US kW h" C C cathode composition in Table 2.5 -CKCI concentration of K C l in the catholyte M CD current density k A m " 2 CE, ML annual maintenance and labor cost $ year"1 CEJ current efficiency for reaction i -C e l operating cost of DC electrical energy ($US h"1 Cf total number of center-points in factorial experiments -CI confidence interval for main and interaction effects ClA installed capital cost of auxiliary equipment $us C I C total installed capital cost $us CIc confidence interval for curvature effects Q,E installed capital cost for electro-chemical units $us CIEC installed capital cost for electro-chemical reactors $us Cpi.f final concentration of product Pi M Cpi,i initial concentration of product Pj M Cpi,out outlet concentration of product Pi M Cpijn inlet concentration of product Pi M Cp,L liquid heat capacity kJ k g 1 K" 1 Cps installed capital cost of power supply $US c r reactant concentration in equations 2.25 and 2.26 mol m" 205 Nomenclature Dc02 diffusion coefficient of CO2 in catholyte mV D °C02 diffusion coefficient of CO2 in water mV d average granule diameter from screen analysis m d b bubble diameter m d e effective granule diameter = d p / [1 + (T + W)dp / (3xW(l-s))] m d P equivalent spherical diameter of cathode granules = dcp m dp, avera g e average particle diameter m dp,i opening size of screen i in screen analysis m E electrode potential V(SHE) Ea >i activation energy for reaction i kJ kmol"1 Ecell full-cell operating voltage V E e equilibrium electrode potential V(SHE) Eg ~Eg a equilibrium cell voltage V Er,i reversible electrode potential of reaction i V(SHE) E° standard electrode potential of reaction i V(SHE) AE voltage window in operation of 3-D flow-by electrode V F Faraday's number • kCkmol fm degree of freedom in factorial experimental design G gas load in 3-D flow-by electrode kg m"2 s" GEP gross economic potential $US G R cost constant in equation 8.5 $US m"2 h L liquid hold-up in 3-D flow-by electrode -H incremental height of the reactor m FT Henry's constant = f (T) kPa Htotal total height of the reactor I current kA I P the current output of the power supply ( 10 kA assumed) kA i superficial current density k A m " 2 io,i exchange current density for reaction I k A m "2 i i L mass transfer limited superficial current density for reaction 6.1 k A m " 2 ji partial real current density of reaction i k A m "2 206 Nomenclature J 1 L mass transfer limited real current density for reaction 6.1 k A m " 2 K',Ko K i reaction equilibrium constants -. M kPa 1 kaps electronic conductivity of compressed felt matrix S m _ 1 k + 0 , k. o, k+i, k.i rate constants in Table 2.6 k F mass transfer coefficient due to forced convection m s"1 k G gas side mass transfer coefficient at G/L interface m/s k G / L a G/L mass transfer capacity = 1 / ( l /k L a + 1/kca) s-1 k H mass transfer coefficient due to gas (H2) generation m s"1 ki electro-chemical pre-exponential rate factor for reaction i m s"1 k L liquid side mass transfer coefficient at G/L interface m/s k, electrolyte conductivity S i n 1 kl.eff effective electrolyte conductivity Sm" 1 km electrode conductivity S i n 1 k M liquid to solid (L-S) mass transfer coefficient m s"1 L liquid load in 3-D flow-by electrode kg m"2 s"1 L a anolyte flow rate m3 s"1 L c catholyte flow rate m3 s-1 M species in the catholyte, HC0 3 " , HCOO - , C 0 2 , H 2 -Me cathode material in Table2.5 -m scale exponent (0.8 to 0.9) for equation 8.5 -N number of increments in reactor modeling -N c number of cells required in process design -N f total number of factorial runs -N P the number of DC power supplies required -N R number of reactors required in process design -NEP net economic potential $US year no number of electrons involved in ERC to formate = 2 -P total pressure kPa(abs) PCD partial current density k A m " 2 PC02 CO2 partial pressure in the gas phase kPa Pgas pressure gradient for single-phase gas flow kg m"2 s"2 207 Nomenclature PH2 H2 partial pressure in the gas phase kPa Pii g pressure gradient for single-phase liquid flow kg m"2 s"2 PP the power output of each power supply kW Pr formate production rate kmol s"1 Pw total power consumption kW APLG pressure gradient in 2-phase flow kg m"2 s"2 R gas constant kJ kmol"1 K Rceii ohmic resistance Ohm RE rectifier efficiency Reb Reynolds' number for bubble generation at electrode Reo Reynolds' number for gas flow = U o d p G / p - G ReL Reynolds' number for liquid flow = U L d p L / | J . L ROI return on investment % year-1 9 1 r electro-chemical reaction rate kmol m" s" S solubility of CO2 in water in equation 2.20 mol L " 1 s the pooled variance of the response based on fm total degrees of freedom SE specific energy for formate production kWh kmof 1 Sc Schmidt number = P L ' P L D AS 0 entropy of reaction kJ kmol"1 K Spi selectivity for product Pi STY space-time-yield kmol m"3 s"1 T temperature K t operating time min ts the student's statistic at the desired confidence level U Q , U L superficial gas velocity, superficial liquid velocity m s"1 V catholyte volume m 3 Vceii reactor volume m Vp the voltage output of the power supply (1000 V assumed) V Wj weight percent of granules that pass screen i in screen analysis % X fraction of cations ( K + or Na +) in form of bicarbonate in Eq. 2.20 -Xj factorial variable i , such as current, liquid flow rate, etc 208 Nomenclature X m , i main effect of factorial variable X i x conversion of CO2 y volume fraction (i.e. mole fraction) in gas phase ctj electro-chemical charge transfer coefficient s voidage of 3-D flow-by electrode cp granule shape factor r\; over-potential of reaction i V r|COnc concentration over-potential (due to mass transfer constraint) V Akinetic activation over-potential (due to intrinsic electrode kinetics) V y activity coefficient u L , po viscosity of liquid, viscosity of gas kg m"1 P293 viscosity of water at 293 K kg m"1 PL, PG density of liquid, density of gas kg m"3 t thickness of 3-D flow-by electrode m xeff effective electro-active thickness of 3-D flow-by electrode m 209 Appendices Appendix A Overall (full-cell) reactions and performance indicators For a practical process development, it is necessary to examine the reactions, material balance and performance indicators (i.e. figures of merit) for the overall reactor. Overall (full-cell) reactions Considering only the production of formate and its competing reaction of hydrogen evolution at the cathode, plus the oxygen evolution at the anode, the full cell reactions A and B are obtained by combining the respective cathode and anode half-cell reactions (with potassium as the principal cation) and adding the subsequent reaction of C 0 2 with hydroxide in the bulk catholyte. The cation membrane prevents the transport of bicarbonate, formate, and hydroxide from the cathode to the anode. Cathode [ 1 ] C0 2(aq) + H 2 0 + 2e" -> HCOO - +OH" [2] 2 H 2 0 + 2e" -»• H 2 + 20H" Anode [3] 20H" -» l / 20 2 +H 2 0 + 2e" Bulk catholyte [4] C 0 2 + OH" -+ HC0 3 " Full-cell [A] 2KOH + 2 C 0 2 -»• HCOOK + K H C O 3 + l / 20 2 (2 Faraday) Full-cell [B] 2KOH + 2 C 0 2 + H 2 0 2KOH + H 2 + l / 20 2 (2 Faraday) The overall carbon (material) balance was checked in some of the runs (see Section 7.3.1) in terms of the carbon closure, which is defined as: ^ 1 , Carbonml Carbon closure = — x 100% (A-1) Carbon^ Performance indicators (figures of merit) Several figures of merit are used to quantify the performance of the ERC process. To define the performance indicators in a general ERC process in which more than one product 210 Appendices is normally produced, the following general reactions for ERC are considered to occur in parallel: [1] C0 2 (aq) + n , e - ^ P 1 [2] C0 2 (aq) + n 2 e ' ->P 2 Current density (CD) j and partial current density (PCD) jj: For a single reaction, the current density is related to the surface reaction rate by Faraday's Law. The current density j is defined as the current per unit area: In which: 2 1 r = surface reaction rate (kmol m" s") I = current (kA) j = current density (kA m"2) A = electrode area (m2) n = electron stoichiometry coefficient F = Faraday's number (kC kmol'1) ERC often involves more than one electrode reaction and associated products, so the partial current density, jj, defined as the current density for a specific product " i " , is more meaningful than the total current density j which is calculated by j = X(j0-Current efficiency (CE) This is a product yield based upon the electrical charge applied. So far all the research work on ERC, except for that of Akahori et al's [2004], has been carried out on batch processes for which the current efficiency is defined as the ratio of the charge required to form the amount of a product determined by the product analysis to the total charge in the specific time interval. For example, the CE for product Pi in a batch process is calculated as: i = — = nFr A (A-2) CEPi = nxF(CPa-CP^)V (A-3) It 211 Appendices Cpi.f = final concentration of product P i (M) CPIJ = initial concentration of product P i (M) I = current (kA) t = batch operating time (s) V = catholyte volume (m ) For a continuous process such as that of the present study the steady-state C E is given by: CEP =  ! ! • Where: CEpi = current efficiency for product P i (-) C p i ; 0 U t = outlet concentration of product P i (M) C p y n = inlet concentration of product P i (M) L c = catholyte flow rate (m 3 s"1) nj = electron stoichiometry coefficient for P i (-) Overall reactor voltage (Ec en) The overall cell voltage Eceu is given by: Ecell = (Ece -Eae)-\Vc\-\Va\-IRcell (A-5) Where: E e c - E e a = equilibrium cell voltage (V) IRceii = ohmic drop - due to electric conductivity of electrodes, electrolyte & separator (V) I = current (A) Rcei i = ohmic resistance (Ohm) n c and n a = cathode and anode over-potentials (V), which can each be expressed as: M = \ E -  Ee\ = \nklne«c\ + h„nc\ (A"6) E - operating electrode potential (VSHE) E e = equilibrium electrode potential (VSHE) ^kinetic= activation over-potential (due to intrinsic electrode kinetics) (V) "Hconc= concentration over-potential (due to mass transfer constraint) (V) 212 Appendices The reversible cathode potential E e c is associated with the. electrode reaction thermodynamics; the electron transfer over-potential | rjkinetic | may be lowered by choosing catalytic electrode materials and electrolysis conditions; the concentration over-potential | nCOnc | can be lowered by effective mass transport conditions; the ohmic potential drop IRceii results in an energy wastage and can be decreased by decreasing the inter-electrode gap and/or increasing the electrolyte and/or separator conductivity. Selectivity (Spi) The electro-reduction of CO2 usually gives a range of products, with the consequent problems of recovering a desired product and disposal or recycling of byproducts. So selectivity is considered a key performance indicator of CO2 reduction. The selectivity for product Pi , Spi, is defined as: S = pJ—L_— ( A_7) Pi CEP2/n2+CEn/nx Where: CEj = current efficiency of product i (-) ni = electron stoichiometry coefficient for product i Spi = selectivity for product Pi (-) Specific Energy (SE) The specific energy is the energy consumption referred to unit mass (or moles) of CO2 reduced or to unit mass (or moles) of product P i : FE ,,, co^mms = _ 7 7 T ^ ~77^ ; Z7Z (A-8) {CEPJnx+CEpJn2)MCOi nxFEcell Pi-mass (CEP )MP ( A " 9 ) Where: SEi ;mass = specific energy for species i (kJ kg"1) M i = molecular weight of species i (kg kmol1) 213 Appendices It can be seen that the specific energy depends upon the cell voltage and the current efficiency. For example, i f a 90 % current efficiency for formate and a 3 % current efficiency for CO are obtained at a cell voltage of-2.5 V and the following reduction reactions occur at the cathode (such as the case at 20 A in Reactor B in the present work): C 0 2 +H 2 0 +2e"=HCOO" +OFF C02+H20+8e~=CO +20FT the specific energy can be calculated as: 96500 x (-2.5) S E C O « „ , = " " {U I kg) = 11191(kJ I kg) C O 2 , « « B (0.9/2+ 0.03/2) x 44 & 11791 = — (kWh/kg) = 33 (kWh/kgC02) Or for the product K H C 0 2 : SEKHCQ2,mass = - 2 x 9 ^ ° X g ( : 2 , 5 ) ( ^ / * g ) = 6382(AJ/*g) (0.9) x 84 = (kWh I kg) = 1.11 (kWh I kgKHC02) Conversion of C 0 2 (x) The consumption of C 0 2 is that from both the production of formate (reaction 6.1) and hydrogen evolution (reaction 6.2), which amounts to 1 mol C 0 2 per Faraday. Thus the conversion of C 0 2 is given by: C02 feed rate Space-time-yield (STY) Space-time-yield is the amount of product (either in kg or kmol) per unit time that can be obtained in a unit reactor volume. In the present work where formate is the only primary product, the space-time-yield is calculated by: 214 Appendices CE I S T Y = 2FV~ ( A - l l ) ^ r y cell Where: CE = formate current efficiency (-) STY = space-time-yield (kmol m"3 s"1) V c e i i - reactor volume (m3) 215 Appendices Appendix B Diffusion coefficient of C 0 2 in water The diffusion coefficient of C 0 2 (D° C o2) determines the mass transfer rate of C 0 2 in the catholyte and is a function of composition and temperature. Table B-l lists the diffusion coefficient of C 0 2 in water within a temperature range of 283 to 308 K [CRC handbook, 2004]. Table B-l Diffusion coe ficient of C 0 2 in water Temperature, K 283 288 293 298 303 308 D ° C 0 2 , x l0" 9 m 2 s- ' 1.26 1.45 1.67 1.91 2.17 2.45 D°co2 in water is correlated as an exponential function of temperature in equation B - l , with a regression coefficient of 0.9991: . D°co2 = 5.0x 10~6 e x p ( - ^ ^ ) In which: D°co2 = diffusion coefficient of C 0 2 in water (m 2 s"1) T = temperature (K). (B-l) In reality, the diffusion coefficient is also affected by the viscosity of the catholyte solution that is in turn affected by the addition of supporting electrolyte and the accumulation of the product formate. These factors are accounted for in the reactor model (Chapter 6). 216 Appendices Appendix C Product analysis and Chemicals used for ERC Determination of formate concentration Formate concentration was determined by the following back-titration steps: [Vogel, 1978]: (1) Add excess Na2C03 to the formate containing sample solution; (2) Add 0.1 N (or 1 N) potassium permanganate from a burette to the sample. A brown precipitate of MnG"2 was formed with the addition of permanganate, and the end point of this step would be from a suspension of brown precipitate to a mixture of brown precipitate and purple solution; 2MnO; + 2HCOO' + OH~ -> 3CO]~ + 2Mn02(l) + 2H20 (C-l) (3) Add excess 10 wt % H2SO4 to acidify the solution; (4) Add excess 0.1 N sodium oxalate to the above solution to dissolve all the precipitate; M „ 0 2 + C20]- + AH+ -> 2C02 ( t ) + M„2+ + 2H20 (C-2) (5) Back titrate the solution with permanganate to a purple end point. 2MnO; + 5C202- + 16/T -> 1OC02(T) + 2 M 2 + + %H20 (C-3) The following equation is used to calculate the concentration of formate: 1 y x N -V x N j-i A , perm perm oxl oxl A ^ formate = TV ~ ) (S^'^) sample Cformate= the concentration of formate (M) Nperm = the normality of permanganate (N) N o x i - the normality of oxalate (N) V 0xi - the volumes consumed of oxalate (ml) Vperm = the volumes consumed of permanganate (ml) Vsampie = the volume of the sample (ml) Determination of the total concentration of bicarbonate and carbonate 217 Appendices The total concentration of the bicarbonate and carbonate was determined by a sequential titration with standard hydrochloric acid after enough hydroxide was added to convert bicarbonate to carbonate [Vogel, 1978]. The first end point was from pink to colorless with phenolphthalein as the indicator (pH ca. 8.6) and the second end point was from yellow to pink with methyl orange as the indicator (pH ca. 4.0). The total concentration of HCO3" and CO3 " was then calculated by the volume difference of hydrochloric acid at the two end points: C ^ / a * - 4(^)JV"° (C~5) sample Where: Cu™- mr,i- = the total concentration of bicarbonate and carbonate (M) N H c i = the normality of HCI (N) V i = the volume of HCI consumed at the first end point (ml) V2 = the volume of HCI consumed at the second end point (ml) VSampie = the volume of the sample (ml) Determination of bicarbonate The concentration of bicarbonate was also determined separately by the following back-titration procedures: (1) Add enough volume ( V i ) of standard IN Ba(OH) 2 solution to convert all H C 0 3 _ to C0 3 2 " ; (2) Add enough BaCl 2 solution to precipitate all CO3 " to BaCCh; (3) Titrate with standard IN HCI solution (V 2) using phenolphthalein as the indicator (from pink to colorless). The concentration of HCOy can be calculated by the volume difference of V i and V 2 using the following equation: Cm-,=iJf:L1»«a " (C-6) sample Chemicals used for ERC: Standard analytical grade chemicals were used, including: K H C O 3 , K C l , K O H , HCOOH, K 2 C 0 3 . 218 Appendices Appendix D Activation of graphite felt surfaces The two-step activation process involved [Plessey, 1980]: Step 1: Sensitization. Sn 2 + ions are absorbed on the felt surface from the solution of Sn 2 + ions: C + S n 2 + s o l u t i o n ^ C . S n 2 + a d s (D-l) The sensitizing'solution was: SnCl 2 : lOg/L HC1 (37 wt%): 40 mL/L Step 2: Nucleation. During nucleation the following reaction occurs: C.Sn ads+Pd solution ^C.Pdads+Sn " solution . (D-2) The nucleating solution was: PdCl 2 : 0.1-1.0 g/L HC1 (37 wt%): 5~10mL/L The nucleation process produced small Pd catalytic sites dispersed on the surface and thus activated the graphite felt for the subsequent electroless reduction step. 219 Appendices Appendix E Recipes for electro- and electroless deposition Table E- l : Recipe 1: Electroless deposition of Cu, Sn and Cu/Sn alloy [Bestetti, 2002] Copper plating [Plessey, 1980] Tin plating [http://vvrww.finishing.com/020 0-0399/260.html ] Cu/Tin alloy plating [Bestetti, 2002] CuS0 4 .5H 2 0, g/L 55 SnCl 2 , g/L 3.8 Cu 2 0 , g/L 15 NaKC 4 H 4 06, g/L 228 Thiourea, g/L 50 SnS0 4 , g/L 11 NaOH, g/L 50 H 2 S 0 4 , ml/L 12 H 2 S 0 4 , ml/L 56 HCOOH, ml/L 80 Time, min 30 Thiourea, g/L 200 Time, min 10 Temperature, K 303-323 Time, min 30 Temperature, K 298 Temperature, K 298 Table E-2: Recipe 2: Electrodeposition of copper and Cu/Tin alloy [Watanabe, 1993] Acidic Cu plating Alkaline Cu plating Alkaline Cu/Sn plating CuS0 4 .5H 2 0,g/L 150-250 CuS0 4 .5H 2 0, g/L 55 Cu 2 P 2 0 7 . 3H 2 0 , g/L 14 H 2 S 0 4 , g/L 45-110 K 4 P 2 0 7 , g/L 174 Sn 2 P 2 0 7 , g/L 44 Temperature, K 298 N a 2 H P 0 4 , g / L 95 K 4 P 2 0 7 , g/L 304 CD, A m' 2 50-300 K N a C 4 H 4 0 6 . 4 H 2 0 , g/L 22 ( N H 4 ) 2 C 2 0 4 . H 2 0 , g/L 23 Temperature, K 298 Temperature, K 298 CD, A m"2 50-300 CD, A m"2 50-200 220 Appendices Appendix F Determination of factorial points on temperature effects (see Section 7.3.2) Table F-l shows the selected data from section 8-92 of CRC Handbook (84th Edition) on the solubility of carbon dioxide in'water at various temperatures and pressures. Table F-l lOOQx mole fraction of C Q 2 in liquid phase Temperature K Partial pressure of C 0 2 in kPa 5 10 20 30 40 50 100 288 0.041 0.082 0.164 0.245 0.327 0.409 0.814 308 0.024 0.048 0.097 0.145 0.193 0.242 0.481 328 0.016 0.033 0.065 0.098 0.131 0.163 0.325 Relationships between C 0 2 partial pressures and C 0 2 solubility under different temperatures are correlated in the form of Henry's Law to obtain the Henry's constants at different temperatures. The correlated Henry's constants are shown in Table F-2 with R-square value being 1 for all the correlations. Table F-2 Correlated Henry's constants Temperature K Henry's constant kPa/mole fraction R square value 288 122893 1 308 207834 1 328 307660 1 In order to compensate the difference of C 0 2 solubility at the two factorial levels of temperature, 288 and 328 K respectively, the ratio of the two C 0 2 partial pressures corresponding to the two temperatures has to be equal to the ratio of the corresponding Henry's constants, which is 122980/307660=1/2.5. This C 0 2 partial pressure ratio can be realized by setting the same ratio for the C 0 2 volume fraction in the feeds. Table F-3 and Figure F- l show the matrix of the factorial design in which the effect of temperature and effect of solubility on current efficiency can be investigated separately and jointly. 221 Appendices Table F-3 Factorial points and center-point Run No. cp Current, A + + + + cp Temp, K + + + cp 328 328 288 288 329 328 288 288 308 Yco2, % + + + + cp 100 40 100 40 100 40 100 40 60 100 Yco2, % 40 cp Current, A 328 288 Temp, K Figure F- l Factorial matrix on the effect of temperature 222 to to Appendix G Samples for modeling spreadsheet C02 Electro-Reduction Reactor Model Cathode dimensions Total Height = 0.66 Tin granules-Number of height increments Oloman & Li 09 Sept. 2005 Update TO HQ 12 10# SS anode 28-Mar-06 L = BIG REACTOR Big run 38 55 mm/increment W = 101 50 15 22 2000 Volt CE Area = 000275 m2/incr Fluid properties. 500 3.90 63 Thickness = t Voidage 3.18 Density 0.48 kg/m3 Viscosity Tcoeffs A B kg/m.s Particle diam. 0.233 mm Liquid D 1030 -10.73 Shape factor Sphere diam = dp 0.449541181 Gas C02 1828 1.97E-02 -1 47E-05 Fitted from delP 25.45 4.55E-01 -8.65E-05 Cond coeffs y = ax2+bx+c -0.8906 7.5937 0.1677 293K KHC03 -0.7468 11.102 -0.0112 KCl 0.105 mm Effective diam = d' 0.102 mm Cond Spec, surf 1000 S/m 29787 m2/m3 H20 -31.89 4.15E-01 -8.27E-06 Real area H2 0.2605 m2/incr 21.87 2.22E-01 -3.75E-05 N2 30.43 4.99E-01 -1.09E-04 Reaction conditions Run Antoine const H2Q 16.5362 3985.44 -38.9974 Current 101 Stoch C02 1407 Amp measured CD av ml STP/min 3.06 kA/m2 calculated superficial CQ2 conversion 70 % Op. time 10 min Cell volts measured 3.90 V Cath flow in measured 22 Gas flow in / out ml/min measured Empirical pressure drop factor = 2000 ml STP/min Anolyte flow in measured Empirical liguid hold-up factor: 1.00 40 ml/min Anolyte cone. measured Pin / Poutcath M KOH 421 measured T in / Tout cath 101 kPa(abs) TRUE measured 288 308 K 101 modelled out delP CQ2 in /CQ2out measured 320 100 KHCQ3 in / out vol % 350 modelled out measured 0.5 KCl measured M measured pH in / pH out 7.5 - measured 9.12 modelled out CE meas 63 % measured CE modeled 67 % calculated by model CE meas/CEmod 0.94 Parameters fitted by Solver for CEmeas/CE mod = 1 in small reactor Equilibria [0] C02(g) + H20(l) < > C02(aq) + H2C03(aq) = H2C03* (aq) Viscfact 1 Ko = [H2C03*] / pC02 = 1.00E+18 T A -8.7051 ref Perry. C02 sol factor 1 [0*1 C02(aq) + H20 <—> H2C03(aq) 2.60E- ref Ko* = [H2C03aq] / [C02aq] = 03 at 298K? Keene Temperature effect not available [11 H2C03*(aq) <—> H+ + HC03 K1 = [H+1[HC03-] / [H2C03*1 ref Benefield pK1 = 17052 n + 215.2 loq(T) - 0.12675 T - 545.56 Kinetics Homogeneous pH < 8 C02 + H2C -> H2C03 rate = k1[C02aql k1 = 0.062 S-1 298K? ref Keene at 298 K? C02 + ref pH < 8 H2C03 —> H20 rate = k2[H2C031 k2 = 23.7 S-1 298K? Keene C02 + pH > 10 OH- —> HC03- rate = k3[C02aq][OH-l k3 = 8500 l/mol.s 298K? ref C&W Cathode Ee,o 298K k alpha Eact Calculated in program Tin cathode VSHE m/s kJ/kmol a b Tafel coef Rxn 1 C02 + H20 + 2e- > HCOO- + OH- -1.02 5.48E-04 0.5 6.00E+04 V V So J/mol.K 121 70 92 -10 delSo -109 -0.4 -0.12 [HCOO-] 0.01 M on Cu and on Sn 2H20 + H2 + Rxn 2 2e- > 20H- Sn -0.83 4.54E-11 0.5 3.18E+04 ref Vassiliev So J/mol.K 70 130 -10 delSo -30 -0.90 -0.12 Cu 0.60 2.33E-06 0.5 3.18E+04 Pickett p.71 293K Experimental data on time vs Sn coverage Sn coverage = 1 exp 0 t 3D potential window = 0.29 Volt Reversible cell volts .1.00 Volt Gas - Liquid mass transfer kG/L 1.00E-04 m/s guess not used not aG/L 2.98E+04 m2/m3 estimate = aS/L used Liquid - Solid mass transfer Calculated by Sato correlation (cf. Oloman JECS 126(11),1979. p.1890 fa Height increment 1 2 3 4 5 6 7 8 9 10 11 12 Time min 10 10 10 10 10 10 10 10 10 10 10 10 Current Amp / incr. 8.42 8.42 8.42 8.42 8.42 8.42 8.42 8.42 8.42 8.42 8.42 8.42 C02 vol % 100 99.1 98.7 97.8 96.7 95.2 93.1 90.1 86.0 80.3 72.4 61.9 Cell Volts V Estimate 3.9 3.9 3.9 3.9 3.9 3.9 ' 3.9 3.9 3.9 3.9 3.9 3.9 Anol flow ml/min 40 40 40 40 40 40 40 40 40 40 40 40 Press in kPa(abs) 421 391.0 360.9 332.1 304.3 277.5 251.3 225.8 200.6 175.8 151.0 126.2 delP (LM) kPa 29.66 30.11 28.81 27.73 26.85 26.14 25.57 25.15 24.86 24.73 24.81 25.22 Press out kPa(abs) 391 361 332 304 277 251 226 201 176 151 126 101 Press av kPa(abs) 406 376 346 318 291 264 239 213 188 163 139 114 Temp in K 288 294 299 305 311 316 322 328 333 339 345 350 DelT (adiabatic) K 5.7 5.7 5.7 5.7 5.7 5.7 5.7 5.7 5.7 5.7 5.7 5.7 Temp out K 288 294 299 305 311 316 322 328 333 339 345 350 Temp av K 294 294 299 305 311 316 322 328 333 339 345 350 H20 vp kPa(abs) 2.42 2.42 3.41 4.72 6.44 8.69 " 11.57 15.25 19.89 25.67 32.82 41.60 Liq vise kg/m.s 1.1E-03 1.1E-03 9.8E-04 8.9E-04 8.1E-04 7.4E-04 6.9E-04 6.4E-04 6.0E-04 5.7E-04 5.4E-04 5.2E-04 Gas vise kg/m.s 1.5E-05 1.5E-05 1.5E-05 1.5E-05 1.6E-05 1.6E-05 1.6E-05 1.5E-05 1.5E-05 1.5E-05 1.4E-05 1.3E-05 Gas MW kg/kmol 43.9 43.5 43.2 42.7 42.1 41.2 40.0 38.4 36.1 33.1 29.3 24.7 Gas density kg/m3 7.70 6.96 6.26 5.60 4.96 4.35 3.76 3.18 2.62 2.07 1.54 1.07 [HC03]in M 0.5 . 0.63 0.76 0.90 1.04 1.18 1.32 1.47 1.63 1.81 1.99 2.19 [HC03]out M 0.63 0.76 0.90 1.04 1.18 1.32 1.47 1.63 1.81 1.99 2.19 2.40 [HC03]av M 0.57 0.70 0.83 0.97 1.11 1.25 1.40 1.55 1.72 1.90 2.09 2.30 Liq flow ml/m 22 22 22 22 22 22 22 22 22 22 22 22 Gas flow (dry) ml/m STP 2000 1889.0 1778.5 1668.9 1560.5 1453.8 1349.5 1248.4 1151.8 1060.6 975.8 898.0 Gas flow (wet) ml/minSTP 2000 1899.9 1789.6 1684.8 1583.0 1485.2 1393.1 1308.7 1235.2 1177.3 1142.7 1147.3 Liq vel m/s 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 Gas vel m/s 0.06 0.06 0.06 0.06 0.06 0.07 0.07 0.07 0.08 0.09 0.09 0.11 Liq Re 0.23 0.22 0.25 0.27 0.30 0.33 0.35 0.38 0.41 0.43 0.45 0.47 Gas Re 2.93 2.71 2.51 2.31 2.12 1.93 1.75 1.57 1.40 1.22 1.05 0.88 fliq 150 150 150 150 151 151 151 151 151 151 151 151' fgas 155 155 154 154 154 153 153 153 152 152 152 152 Pliq kPa/m 87 89 79 72 65 60 56 52 49 46 44 42 Pgas kPa/m 31 31 33 34 36 37 39 41 43 45 47 51 X _ 0.64 0.64 0.69 0.73 0.77 0.82 0.86 0.90 0.95 0.99 1.03 1.08 delPIg kPa 29.66 30.11 28.81 27.73 26.85 26.14 25.57 25.15 24.86 24.73 24.81 25.22 Liq load kg/m2.s 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3 Gas load kg/m2.s 0.375 0.353 0.330 0.308 0.285 0.262 0.238 0.215 0.191 0.167 0.143 0.121 to to Galileo no. = Ga Liq hu Cond KHCQ3 Cond KCl Eff cond Ko S/m S/m S/m K1 pC02 in C02 consumed H2 produced C02 flow H2 flow mfCQ2out mfC02out M/kPa M kPa(abs) ml/minSTP 5.12E+02 0.41 4.23 19.46 3.37 3.29E-04 4.31 E-07 418 117.2 ml/minSTP ml/min STP ml/minSTP 6.2 5.03E+02 0.41 5.10 6.00E+02 0.41 19.46 3.49 3.29E-04 4.31 E-07 385 117.2 6.8 2000.0 % (wet) pC02 out pCQ2 av [H2CQ3'lav [H2C03*]in [H2C03*lout [CQ2aqlequil.av % (dry) kPa(abs) kPa(abs) M M 6.2 99.1 99.7 387 403 1.33E-01 1.38E-01 1.27E-01 1882.8 13.0 6.61 7.08E+02 0.40 8.28 8.27E+02 0.40 21.63 3.95 2.79E-04 4.63E-07 353 117.2 7.7 1765.5 98.7 99.3 356 371 1.22E-01 1.27E-01 1.32E-01 CQ2 material balance (steady-state) in bulk liquid 1.17E-01 1.22E-01 20.7 23.80 4.42 2.37E-04 4.89E-07 320 117.2 8.8 1648.3 97.8 98.8 325 339 9.44E-02 9.83E-02 9.06E-02 9.42E-02 29.5 10.11 9.56E+02 0.39 25.97 4.90 2.02E-04 5.09E-07 288 117.2 10.5 1531.0 96.7 98.1 294 307 7.28E-02 7.59E-02 6.97E-02 7.26E-02 40.0 12.10 1.09E+03 0.39 28.14 5.38 1.73E-04 5.23E-07 256 117.2 12.9 1413.8 52.9 14.26 30.31 5.87 1.48E-04 5.32E-07 223 117.2 1.24E+03 0.38 16.60 1.39E+03 0.38 32.48 6.37 1.27E-04 5.37E-07 190 117.2 16.2 1296.5 95.2 97.2 264 276 5.57E-02 5.82E-02 5.33E-02 5.56E-02 93.1 96.1 234 245 4.23E-02 4.41 E-02 4.04E-02 4.21 E-02 69.1 90.1 94.5 203 213 3.15E-02 20.6 1179.3 19.11 34.65 6.88 1.10E-04 5.38E-07 155 117.2 1.54E+03 0.37 21.80 36.82 7.39 9.46E-05 1.71E+03 0.36 24.66 38.99 7.89 5.35E-07 120 117.2 26.1 89.7 86.0 92.2 172 181 3.30E-02 3.01 E-02 3.15E-02 2.30E-02 2.41 E-02 2.19E-02 1062.0 115.8 32.5 944.8 80.3 89.1 141 148 1.62E-02 1.70E-02 1.54E-02 2.30E-02 I 1.62E-02 I 148.3 72.4 8.19E-05 5.30E-07 86 1.88E+03 0.36 27.67 41.16 8.36 7.11E-05 5.23E-07 117.2 39.4 827.5 187.7 84.8 109 115 1.09E-02 1.14E-02 1.03E-02 1.08E-02 kgagPCQ2Ko J1as/2F hLk2[HC03-HH+1 hLk3fOH-l [1/kLa+1/kGal-1 hl_k1 [CQ2aqlkin,av 61.9 79.1 52 117.2 46.6 710.3 234.3 48.1 71.7 78 82 6.71 E-03 7.01 E-03 6.40E-03 6.69E-03 49 50 3.59E-03 3.73E-03 3.45E-03 3.58E-03 kmol/m3.s kmol/m3.s kmol/m3.s s-1 s-1 s-1 kmol/m3 8.08E-02 4.46E-03 1.07E-02 7.17E-04 6.10E-01 2.57E-02 1.37E-01 7.49E-02 4.41 E-03 9.63E-03 9.84E-04 6.14E-01 2.57E-02 1.25E-01 5.67E-02 4.34E-03 7.25E-03 1.43E-03 6.01 E-01 2.53E-02 9.50E-02 4.29E-02 4.24E-03 5.46E-03 2.06E-03 5.89E-01 2.50E-02 7.16E-02 3.23E-02 4.09E-03 4.09E-03 2.97E-03 5.80E-01 2.46E-02 5.32E-02 2.42E-02 3.89E-03 3.05E-03 4.33E-03 5.72E-01 2.43E-02 3.88E-02 1.79E-02 3.61 E-03 2.24E-03 6.39E-03 5.66E-01 2.40E-02 2.76E-02 1.29E-02 3.24E-03 1.61 E-03 9.67E-03 5.61 E-01 2.36E-02 1.90E-02 9.06E-03 2.77E-03 1.12E-03 1.52E-02 5.58E-01 2.33E-02 1.24E-02 pH in calc 6.05E-03 2.23E-03 7.36E-04 2.52E-02 5.57E-01 2.30E-02 7.54E-03 pHin meas pHout calc ph x 10 pHout meas E1 atT 3.74E-03 1.63E-03 4.46E-04 4.55E-02 5.58E-01 2.26E-02 4.08E-03 E2 atT Re(b) Sc S"(g) DC02 J<2-V(SHE) V(SHE) m/s 2.02E-03 1.02E-03 2.32E-04 9.54E-02 5.62E-01 2.22E-02 1.80E-03 6.93 7.06 70.6 7.06 7.18 71.8 -0.735 -0.417 9.54E-07 m2/s m/s 4.55E-05 647 1.47E-01 1.62E-09 4.75E-06 7.23 7.33 73.3 -0.740 -0.425 1.07E-06 5.03E-05 669 1.57E-01 1.59E-09 4.99E-06 -0.744 7.38 7.48 74.8 -0.428 1.09E-06 5.73E-05 526 1.49E-01 1.81E-09 5.40E-06 -0.748 -0.431 1.16E-06 6.72E-05 7.54 7.64 7.71 76.4 -0.752 -0.434 1.28E-06 421 1.45E-01 2.05E-09 5.92E-06 8.12E-05 343 1.44E-01 7.80 78.0 -0.757 7.88 7.96 79.6 -0.438 1.46E-06 1.01E-04 284 1.46E-01 2.29E-09 6.58E-06 2.54E-09 7.43E-06 -0.763 -0.442 1.72E-06 1.29E-04 239 1.52E-01 8.06 8.14 81.4 8.25 8.34 83.4 -0.770 -0.448 2.06E-06 1.66E-04 204 2.79E-09 8.49E-06 1.60E-01 3.05E-09 9.74E-06 -0.778 -0.455 2.47E-06 8.47 8.56 85.6 8.73 8.81 9.05 9.12 91.2 -0.787 -0.465 2.12E-04 176 1.68E-01 3.30E-09 1.11E-05 2.89E-06 2.62E-04 155 1.76E-01 3.55E-09 -0.800 -0.477 3.25E-06 3.09E-04 139 1.81E-01 3.79E-09 -0.818 -0.494 3.43E-06 3.40E-04 126 1.80E-01 4.01 E-09 1.25E-05 I 1.37E-05 I 1.45E-flfi TO a • TO TO Co --4 kLa kGa [1/kLa-H/kGa)-1 kSa kMa Kf km jL C02 iL C02 Supr CD Real CD (over teff) io1 a1 bl io2Sn a2Sn b2Sn io2Cu a2Cu b2Cu a2 Sn C cov'age delHCOO [HCOO-lin [HCOOlout j1 (over teff) j2 (over teff) n2 Ecath CE1 calc K(m)a overall Eff thick s-1 s-1 s-1 s-1 s-1 m/s m/s kA/m2 kA/m2 kA/m2 kA/m2 kA/m2 _V _V kA/m2 _V _v kA/m2 V mM mM mM kA/m2 kA/m2 VSHE % s-1 mm 6.10E-01 6.96E+02 6.10E-01 1.79E+01 5.90E-01 1.98E-05 2.04E-05 0.538 50.93 3.061 0.289 6.3E-04 -0.37 -0.12 7.2E-08 -0.83 -0.12 3.7E-03 -0.28 -0.12 -0.83 1.000 106.4 106.4 1.211 30 -26 4.70E+00 0.258 0.031 7.53 0.89 -0.337 -0.656 -1.073 89 6.06E-01 0.36 0.36 6.15E-01 7.12E+02 6.14E-01 1.78E+01 5.94E-01 1.99E-05 2.05E-05 0.496 46.94 3.061 0.274 6.4E-04 -0.37 -0.12 8.5E-08 -0.82 -0.12 4.4E-03 -0.28 -0.12 -0.82 1.000 105.1 106.4 211.5 1.172 30 -24 4.07E+00 0.242 0.032 7.44 0.98 -0.334 -0.650 -1.075 88 6.12E-01 0.38 0.38 6.01 E-01 7.10E+02 6.01 E-01 1.93E+01 5.83E-01 1.96E-05 203E-05 0.372 35.25 3.061 0.225 6.3E-04 -0.38 -0.12 9.8E-08 -0.83 -0.12 5.0E-03 -0.27 -0.12 -0.83 1.000 103.4 211.5 314.9 1.147 38 -24 3.16E+00 0.196 0.029 7.32 1.10 -0.334 -0.651 -1.078 87 6.04E-01 0.46 0.46 5.90E-01 7.11E+02 5.89E-01 2.08E+01 5.73E-01 1.92E-05 2.01E-05 0.278 26.34 3.061 0.186 6.1E-04 -0.39 -0.12 1.1E-07 -0.84 -0.12 5.8E-03 -0.27 -0.12 -0.84 1.000 101.0 314.9 415.9 1.115 47 -23 2.42E+00 0.758 0.028 7.15 1.27 -0.336 -0.653 -1.084 85 6.00E-01 0.55 0.55 5.80E-01 7.15E+02 5.80E-01 2.23E+01 5.65E-01 1.90E-05 2.01 E-05 0.206 19.53 3.061 0.154 5.9E-04 -0.40 -0.12 1.3E-07 -0.85 -0.12 6.6E-03 -0.27 -0.12 -0.85 1.000 97.6 415.9 513.6 1.074 58 -22 1.83E+00 0.727 0.028 6.91 1.51 -0.339 | -0.657 -1.091 82 5.98E-01 0.67 0.67 5.73E-01 7.23E+02 5.72E-01 2.38E+01 5.59E-01 1.88E-05 2.02E-05 0.151 14.32 3.061 0.129 5.5E-04 -0.41 -0.13 1.5E-07 -0.86 -0.13 7.7E-03 -0.27 -0.13 -0.86 1.000 92.8 513.6 606.3 1.022 70 -20 1.35E+00 0.100 0.028 6.56 1.85 -0.343 -0.662 -1.100 78 6.01 E-01 0.80 0.80 5.66E-01 7.34E+02 5.66E-01 2.53E+01 5.54E-01 1.86E-05 2.04E-05 0.109 10.32 3.061 0.107 5.1E-04 -0.42 -0.13 1.7E-07 -0.86 -0.13 9.0E-03 -0.26 -0.13 -0.86 1.000 86.1 606.3 692.4 0.956 83 -19 9.67E-01 0.077 0.030 6.09 2.33 -0.347 -0.668 -1.110 72 6.09E-01 0.96 0.96 5.62E-01 7.52E+02 5.61 E-01 2.68E+01 5.50E-01 1.85E-05 2.09E-05 0.076 7.24 3.061 0.089 4.6E-04 -0.43 -0.13 2.1E-07 -0.87 -0.13 1.1E-02 -0.26 -0.13 -0.87 1.000 77.2 692.4 769.6 0.872 97 -17 6.60E-01 0.058 0.031 5.46 2.96 -0.352 -0.673 -1.121 65 6.22E-01 1.16 1.16 5.59E-01 7.77E+02 5.58E-01 2.83E+01 5.47E-01 1.84E-05 2.15E-05 0.051 4.87 3.061 0.073 4.0E-04 -0.45 -0.13 2.5E-07 -0.87 -0.13 1.3E-02 -0.25 -0.13 -0.87 1.000 66.0 769.6 835.6 0.763 113 -15 4.22E-01 0.040 0.032 4.67 3.74 -0.354 -0.676 -1.131 56 6.39E-01 1.41 1.41 5.57E-01 8.15E+02 5.57E-01 2.98E+01 5.46E-01 1.83E-05 2.22E-05 0.032 3.06 3.061 0.058 3.3E-04 -0.47 -0.13 3.1E-07 -0.88 -0.13 1.6E-02 -0.24 -0.13 -0.88 1.000 53.1 835.6 888.7 0.620 129 -13 2.44E-01 0.026 0.032 3.76 4.66 -0.352 -0.675 -1.139 45 _ 6.61 E-01 1.76 1.76 5.58E-01 8.72E+02 5.58E-01 3.12E+01 5.48E-01 1.84E-05 2.29E-05 0.018 1.71 3.061 0.045 2.5E-04 -0.49 -0.14 4.0E-07 -0.87 -0.14 2.1E-02 -0.23 -0.14 -0.87 1.000 38.9 888.7 927.6 0.425 147 d ° j 1.20E-01 0.075 0.030 2.75 5.66 -0.343 -0.667 -1.144 33 6.83E-01 2.28 2.28 5.63E-01 9.63E+02 5.62E-01 3.27E+01 5.53E-01 1.86E-05 2.35E-05 0.008 0.78 3.061 0.032 1.7E-04 -0.53 -0.14 5.6E-07 -0.87 -0.14 2.9E-02 -0.21 -0.14 -0.87 1.000 24.4 927.6 952.0 0.139 168 -8 4.47E-02 0.007 0.026 1.73 6.69 -0.324 -0.648 -1.142 21 7.01 E-01 3.17 3.17 a S-f t Mean CE 67% Appendices Appendix H Design and cost estimations for the auxiliary equipment Design and cost calculations for the auxiliary equipment are based on the conceptual process flowsheet presented in Figure 8.1. It needs to be noted that the cost estimation of the auxiliary equipment in Table 8.4 is for 5 current densities, as in the cost estimation for the electrochemical units. For this purpose the flow rates of all streams (a total of 22 streams in Figure 8.1) were obtained from the material balance at each of the 5 current densities. The following equipment is involved in Figure 8.1: two feed tanks, seven pumps including two gas compressors, four coolers, two G/L separators, the H 2 recovery unit (pressure swing adsorption), an evaporator and a filter. The materials for auxiliary equipment are exclusively stainless steel therefore a material factor applies for each unit when obtaining the purchase costs. The purchase costs of the process equipment are obtained from 1982 [Ulrich, 1984], and then updated using the Marshall-Swift cost index of 2005 shown in Table H-l . Table H-l Chemical Engineering Plant Cost Indices Year Equipment Index Reference General Equipment 549.5 2005 Heat Exchanger 520.7 [Chem. Eng., 10/2005] Pumps & Compressors 756.1 1982 General Equipment 315 [Ulrich, 1984] (1) Feed tanks (Fl and F2) Feed tanks are employed to hold anolyte and catholyte for the specified flow rates. The total flow rates for both anolyte and catholyte can be obtained through multiplying the number of cells needed at various current densities by the flow rate per cell which is 0.35 3 1 l/min (0.021 m h" ) for both anode side and cathode side. Then assuming a typical residence 228 Appendices time of 2000 s and using a height to diameter ratio of 2 for vertically oriented feed tanks, the dimensions of the anolyte tank and catholyte tank can be calculated. Take the current density of 2.22 kA m"2 (1000 A) as an example. The number of cells needed is 15200 and therefore the total flow rate for both anolyte and catholyte is 319 m 3 h"1, which would require a tank volume of 177 m 3 . The numbers of tanks needed for catholyte and anolyte are both 2, each of which has the dimension of 4 m in diameter and 8 m in height (with the 2 to 1 aspect ratio). Table 8.4 lists the installed capital costs of feed tanks for both catholyte (Fl) and anolyte (F2) under all current densities, and the material and Lang factors are summarized in Table H-3. (2) Compressors (PI and P2) PI CO2 gas from the emission of power plant (feed) and from the H2 recovery unit (recycle) is required to be compressed from a pressure of near atmosphere (100 kPa abs.) to a pressure of 520 kPa (abs). For a compression ratio > 4 the compressor has to be broken into stages to prevent over heating and minimize the shaft power to the compressors. Two stage compressor is employed in the present design with a compression ratio of 2.3 for each stage. The ratio 2.3 is calculated by the geometric progression in the pressures of two stages, i.e. the outlet pressure of the first stage is the geometric mean of the inlet pressure of the first stage (100 kPa abs.) and the outlet pressure of the second stage (520 kPa abs.). The shaft power is calculated from the following equation [Timmerhaus and Flynn, 1989]: where m v is the total molar flow rate of C 0 2 gas into the reactor (mol s"1) including the C 0 2 recycle, R the gas constant (8.314 J mol' 1 K ' 1 ) , T i n the inlet C 0 2 temperature to the compressor (K), r the ratio of specific heats (Cp/Cv), P o u t /P i n the ratio of outlet to inlet (H-l) pressure for the compressor and sc is the compressor efficiency. 229 Appendices For the present work, T i n is 298 K, r is 1.29 (CRC Handbook, 2004), outlet to inlet pressure ratio is roughly 2.3 for all current densities and the compressor efficiency is taken as 80%. Table 8.4 shows the installed capital costs for compressors at various current densities and Table H-3 lists the material and Lang factors. P2 The mixture of H 2 and C 0 2 from the cathode side G/L separator needs to be compressed for H 2 recovery in the HyQuestor unit from a pressure of near atmospheric (100 kPa(abs)). The outlet pressure of the compressor has to be near 1000 kPa(abs) for a satisfactory H 2 recovery (data from QuestAir) [Ed Rode, 2005]. Therefore the compression ratio would be about 10. As a.rule of thumb, i f the pressure ratio is greater than 4:1, multiple stages are employed, and total power is minimized i f each stage has the same compression ratio. For the present case, two-stage compression is adequate with a compression ratio of 3.25 for each stage. The outlet pressure for the first stage or the inlet pressure for the second stage is thus 325 kPa(abs). The shaft power is estimated from equation H - l , in which the inlet temperature is obtained from the modeling and is a function of the current density. For example, at 2.22 kA m" the outlet temperature from the reactor, which is also the inlet temperature of the compressor, would be 321 K, the flow rate of the mixture of H 2 and C 0 2 is 79.8 mol s"1 from material balance and the estimated shaft power for each stage is 373 kW. The installed capital costs for two compressors under different current densities are presented in Table 8.4, and material factor and Lang factor are listed in Table H-3. . (3) Pumps (P3 to P7) Five pumps are installed in the system to transfer the liquid for feed, recycle and filtering (see Fig. 8.1). The capital and operating cost of a pump is a function of the required shaft power W s [Uhlrich, 1984] which is given by the following equation: 230 Appendices bp Where F v is the volumetric flow rate (m3/s), A P the pressure differential (kPa) and e p the pump efficiency which is taken as 60% The pressure differential is specific to each pump. The capital cost and operating cost for P3 to P7 under different current densities are listed in Table 8.4. (4) G/L separators The function of the separator on the cathode side is to separate the gas mixture of H2 and CO2 from the catholyte stream, and that of the separator on the anode side is to release O2 from the anolyte stream. G/L separators are typically vertical columns with packing in the upper part to remove the entrained liquid droplets. The design diameter of separators is based on the following equations [Uhlrich, 1984]: D = [-i~Y (H-3) . (H-4) Pi — P -where Q g is the gas flow rate (m3/h) which, in the cathode side, is a function of current density and current efficiency which are obtained from the model outlined in Chapter 6; U a v e is the gas velocity in the separator (m/s); pi the liquid density (kg/m3) and p g the gas density (kg/m ). The gas densities in the cathode side are calculated for the mole-fraction average of H 2 and C 0 2 . The designed diameters and corresponding column heights are tabulated in Table H-2 with the aspect ratio of 3. 231 Appendices Table H-2 Design diameters and heights for G/L separators CD, k A m " 2 0.44 0.89 1.33 1.78 2.22 Cathode side G/L separator Diameter, m 4.4 2.9 2.1 1.6 1.2 Height, m 13.2 8.7 6.3 4.8 3.6 Anode side G/L separator Diameter, m 0.8 0.8 0.8 0.8 0.8 Height, m 2.4 2.4 2.4 2.4 2.4 From the designed diameter the nominal flow areas are calculated and the purchase costs taken from [Uhlrich, 1984]. The capital costs of separators are shown in Table 8.4, and the material and Lang factors are given in Table H-3. (5) Heat exchangers (CI to C4) To prevent the high temperature of the electrolyte caused by Joule heating in the electrochemical reactors, pre-cooling of both the catholyte and anolyte by 'chilled water' (278 K at a cost of $1 US/m 3) is provided (CI to C3). In addition, another cooler (C4) is needed after the evaporator for crystallization of the NaHCOs. The heat transfer area of the heat exchangers and the flow rate of cooling water are calculated using the following equations: UAtLTmmm=QlCPJ*Tl=QwCP,wATw mean Tp r-p j A I,in ~*W,out >. (H-5) (H-6) where U is the heat transfer coefficient (1.6 kW.m _ 2 K _ 1 [Uhlrich, 1984]), A T m e a n the logarithmic temperature difference (K), C P J and C P ; W the heat capacity of the liquid cooled 232 Appendices and that of cooling water ( k J . k g V ) , Q, and Q w the flow rate of liquid cooled and that of the cooling water (kg/s),AT, and A T W the temperature decrease of the liquid cooled and the temperature increase of the cooling water (K), and A the heat transfer area needed (m2). Hot fluid 288 K 278 K Figure H-l Inlet and outlet temperatures of cooling water in heat exchangers CI to C3 CI, C2 and C3 Shell and tube heat exchangers with counter-current flow mode are selected for coolers CI to C3 as shown in Figure H-l . The temperature decreases of hot' fluid in CI and C3 are both designed to be equal to the temperature increases in the electrochemical reactor obtained from the model; while the temperature of the hot fluid (the additional water to the system) in C2, is 298 and 288 K, respectively, at the inlet and outlet. This way, the feed temperature of the electrochemical reactor is always kept at 288 K. C4 Cooler C4 is provided for the crystallization of NaHC03, which needs a substantial cooling load because of the high temperature in the evaporator. The inlet temperature of the liquid cooled, which is the temperature from the evaporator, is 379 K, and the outlet temperature is set at 298 K. Therefore ATi for C4 would be 8IK. The cooling water flows counter-currently to the hot fluid with inlet and outlet temperatures of respectively 278 and 280 K. The flow rate of the hot fluid and thus of the cooling water is a function of current density and current efficiency. The utility cost of the "chilled water" is taken as 1$US m - 3 which is based on the fuel price of 2005. The capital costs and operating costs for the four coolers are given in Table 8.4 as a function of current density, and material and Lang factors are listed in Table H-3. (6) Filtering 233 Appendices Rotary disk filter is chosen for filtering the N a H C 0 3 crystals [Uhlrich, 1984] because of its being relatively inexpensive to purchase and operate, and the most efficient in use of floor space. The filter is sized and nominal area is calculated by the following equations: D = [^-]°5 (H-7) U, = 0 . 0 3 [ ^ ^ ] 0 5 (H-8) Pi K>rmmai=^f (H-9) where Qj is the liquid flow rate through the filter (m 3 s"1), U | is the velocity of the liquid (m/s), pi and p s are the densities for liquid and solid (2.159 g ml"1 for N a H C 0 3 crystals), respectively, and An0minai is the nominal area for filtration. The capital cost is listed in Table 8.4 as a function of operating current density. (7) Evaporator The evaporator has the highest operating-cost among the auxiliary equipment. Vertical forced-circulation evaporation has the advantage of alleviating scale-forming problems and providing efficient heat transfer, thus is considered to be a good option for the present system. Water vapor out in Evaporator Figure H-2 Evaporation unit In Figure H-2, the water would be evaporated at a temperature of 379 K, which is about 6 K higher than the normal boiling point of water due to the boiling point elevation. 234 Appendices The flow rates of water evaporated Q w i n kg/s and m 3/h are both calculated from the material balance. Then the design diameter is obtained by the following equation: D = (——-) (H-10) where U a V e is the water vapor velocity at atmospheric pressure which is taken as 2.4 m s"1 as suggested by [Ulhrich, 1984]. Heat duty H e Vap (kJ s"1) is estimated from the overall energy balance on the evaporation unit: Hevap -QinCPiin(l06-Tm) + ^AHevap (H- l l ) where Q m is the flow rate of the inlet to evaporator (kg s"1), C P i i n is the heat capacity of the inlet (taken as 4.18 kJ kg"1 K" 1), T m i s the inlet temperature (K), which is a function of the current density, Q w is the water vapor flow rate in kg s"1 and AH e Vap is the latent heat of water (kJ mol"1). The cost of heating steam is then estimated using a price of 0.01 $US kWh"1. The design heat transfer area A (m ) can be estimated from the following: UAATm=Hemp (H-l 2) where A T m is the mean temperature differential assuming 413 K (140 °C) as the steam temperature. U , the overall heat transfer coefficient, is taken as 6 kW m"2 K" 1 for the vertical forced-circulation evaporator as suggested by [Ulhrich, 1984]. The capital cost is then estimated and listed in Table 8.4 as a function of current density, and material and Lang factors are in Table H-3. (8) HyQuestor A "HyQuestor" pressure-swing adsorption unit from QuestAir (Vancover) is projected as the equipment to recover H 2 from the gas product ( C 0 2 + H 2 + H 20(g)) with a 235 Appendices separation efficiency around 90 % at 313 K and 1000 kPa (abs). For a typical-gas feed 4000 STP m3/h, two HyQuestor 628 units would be required and would totally cost ~ 1.2 106 $US (The above estimate comes from Ed Rode, QuestAir August 2005). Table H-3 Material and Lang factors for auxiliary equipment Auxiliary equipment Material factor Lang factor 1. Compressors 5.5 4.5 2. Pumps 2.4 4.5 3. Coolers 3.3 3.0 4. Separators 3.6 4.0 5. H2 recovery NA 3.0 6. Evaporator 6.2 3.0 7. Filter 3.6 4.0 8. Feed tanks 3.0 3.0 236 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0058927/manifest

Comment

Related Items