Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Colloidal electrodesposition of Pt-Ru and Pd nanostructires on three-dimensional substrates : application… Cheng, Tsz Hang Tommy 2009

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2009_spring_cheng_tszhang.pdf [ 9.2MB ]
Metadata
JSON: 24-1.0058650.json
JSON-LD: 24-1.0058650-ld.json
RDF/XML (Pretty): 24-1.0058650-rdf.xml
RDF/JSON: 24-1.0058650-rdf.json
Turtle: 24-1.0058650-turtle.txt
N-Triples: 24-1.0058650-rdf-ntriples.txt
Original Record: 24-1.0058650-source.json
Full Text
24-1.0058650-fulltext.txt
Citation
24-1.0058650.ris

Full Text

COLLOIDAL ELECTRODEPOSITION OF Pt-Ru AND Pd NANOSTRUCTURES ON THREE-DIMENSIONAL SUBSTRATES: APPLICATION FOR DIRECT METHANOL AND DIRECT FORMIC ACID FUEL CELL ANODES by TSZ HANG TOMMY CHENG B.A.Sc., The University of British Columbia, 2004 A THESIS SUBMITTED 1N PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Chemical and Biological Engineering) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) April 2009 © Tsz Hang Tommy Cheng, 2009 ABSTRACT Direct liquid fuel cells (DLFC’s) including the direct methanol (DMFC) and formic acid systems (DFAFC) are promising alternatives to hydrogen fuel cells. However, DLFC’s are challenged by slow anodic kinetics, fuel crossover, and CO2 disengagement. In the present work, new colloidal catalyst preparation methods were studied with the aim to electrodeposit Pt-Ru and Pd nanoparticles onto three-dimensional (3-D) substrates: reticulated vitreous carbon, graphite felt (GF), and titanium mesh. The electrodes were investigated as 3-D anodes to mitigate the above-mentioned issues to improve power output, while reducing the catalyst load. The novel catalyst preparation technique involved electrodeposition from various colloidal media including Triton X- 100 and Triton X- 102-based microemulsion and micellar solutions. Compared to conventional aqueous-based media, using colloidal media resulted in catalysts with more desirable Pt-Ru atomic ratio for CH3O oxidation and higher specific surface area. In the absence of surfactants, large catalyst particles (>500nm) were deposited on mostly the exterior surfaces of the 3-D substrates. Employing colloidal media led to more uniform deposition of nanoparticles (5—4Onm, depending on conditions) throughout the substrate thickness. In DMFC experiments at 333K with 1M CH3O and 0.5M H2S04,the Pt-RuJGF anode (lOg m2) generated a peak power density of 741W m2 compared to 703W m2 obtained with a commercial catalyst-coated membrane (CCM) with four-times higher Pt- Ru load. In DFAFC experiments at 333K with 1M HCOOH and 0.5M H2S04,the peak power density using the Pt-Ru/GF anode reached 860W m2, compared to 526W m2 with CCM (Pt-Ru load: lOg m2). However, increasing the HCOOH concentration to 3 and 1 OM led to lower protonic conductivity and resulted in lower peak power density of 727 and 468W m2, respectively. The electrodeposition of Pd on GF was investigated for a comparative study with Pt-Ru/GF for HCOOH electro-oxidation. PdJGF was found to have better HCOOH oxidation kinetics and poorer stability than Pt-RuJGF. In DFAFC experiments at 333K with 1M HCOOH and 0.5MH2S04,the peak power density using Pd/GF anode (57g m2) reached 852W m2, compared to 392W m2 with 40g m2 commercial Pd CCM. The novel catalyst preparation methods along with the 3-D electrode concept were demonstrated to be beneficial for improving DLFC performance at reduced catalyst load. III TABLE OF CONTENTS ABSTRACT.ii TABLE OF CONTENTS iv LIST OF TABLES vii LIST OF FIGURES ix LIST OF SYMBOLS xxiv LIST OF ABBREVIATIONS xxvi ACKNOWLEDGEMENTS xxviii CO-AUTHORSHIP STATEMENT xxix 1 INTRODUCTION I 1.1 General Considerations 1 1.2 Theoretical Background 10 1.2.1 Electrochemical Thermodynamics 10 1.2.2 Half-Cell Potential and Reference Electrode 14 1.2.3 Kinetics of Electrode Processes 15 1.2.4 Mass Transfer, Ionic Conductivity, and Voltage Balance 22 1.2.5 Half-Cell Electrochemical Experimental Methods 26 1.2.5.1 Cyclic Voltammetry 26 1.2.5.2 Chronopotentiometry 28 1.2.5.3 Chronoamperometry 30 1.2.5.4 Catalyst Surface Area Estimation by Electrochemical Techniques 32 1.2.6 Conventional Fuel Cell Electrode Design 39 1.3 Literature Review 42 1.3.1 Methanol Electro-Oxidation 42 1.3.2 Formic Acid Electro-Oxidation 50 1.3.3 Preparation of Electrocatalysts and Nanostructured Materials 60 1.3.3.1 Sot-Gel Method 60 1.3.3.2 Polyol Method 65 1.3.3.3 Colloidal Organosol Method 69 1.3.3.4 Electrodeposition 72 1.3.3.5 Chemical Reduction in Reverse Microemulsion 77 1.3.4 Novel Electrode Design for Direct Liquid Fuel Cells 79 1.4 Objectives 84 1.5 References 87 2 RESULTS AND DISCUSSION: ELECTRODEPOSITION OF Pt-Ru ON RVC FROM REVERSE EMULSIONS AND MICROEMULSIONS* 99 2.1 Introduction 99 2.2 Experimental Procedure 101 2.2.1 RVC Pretreatment 101 iv 2.2.2 Emulsion and Microemulsion Preparation for Electrodeposition 102 2.2.3 Electrochemical Measurements 104 2.2.4 Electrical Conductivity Measurements 106 2.2.5 Surface and Analytical Characterization of Catalysts 106 2.3 Results and Discussion 106 2.3.1 Electrodeposition from Reverse emulsion 106 2.3.1.1 The Effect of Temperature 106 2.3.1.2 The Effect ofDeposition Current Density 109 2.3.1.3 Effect ofRu Precursor and Pt:Ru Ratio 112 2.3.2 Effect of Isopropanol on Solution Phase Behavior: Microemulsification 114 2.3.3 Electrodeposition from Reverse Microemulsion 115 2.3.4 Effect of RVC Pretreatment and Postdeposition Heat Treatment 118 2.4 Conclusion 127 2.5 References 128 3 DIRECT METHANOL FUEL CELLS WITH RETICULATED VITREOUS CARBON, UNCOMPRESSED GRAPHITE FELT, AND TITANIUM MESH ANODES* 130 3.1 Introduction 130 3.2 Experimental Section 133 3.2.1 Pretreatment of Three-Dimensional Catalyst Supports 133 3.2.2 Electrodeposition Procedure 134 3.2.3 Electrochemical Measurements 135 3.2.4 Surface and Analytical Characterization of the Catalysts 136 3.2.5 Membrane Electrode Assembly and Fuel Cell Experiments 136 3.3 Results and Discussion 137 3.3.1 Electrodeposition from Micellar Solution vs. Microemulsion: RVC support 137 3.3.2 Pt-Ru Electrodeposition on RVC from Micellar Media: Effect of Perforated Counter Electrodes and Deposition Current Density 143 3.3.3 Effect of Catalyst Support 148 3.3.4 Direct Methanol Fuel Cell Experiments 153 3.4 Conclusion 159 3.5 References 160 4 EFFICIENT ANODES WITH LOW Pt-Ru LOAD FOR DIRECT METHANOL AND FORMIC ACID FUEL CELLS* 162 4.1 Introduction 162 4.2 Experimental Section 164 4.2.1 Anode Matrix: Graphite Felt Catalyst Support 164 4.2.2 Electrodeposition Procedure 164 4.2.3 Electrochemical Measurements 165 4.2.4 Conductivity Measurements 166 4.2.5 Surface and Analytical Characterization of the Catalysts 166 4.2.6 Membrane Electrode Assembly and Fuel Cell Experiments 166 4.3 Results and Discussion 167 4.3.1 Effect of Non-Ionic Surfactant Type and Micellar Media on the Pt-Ru Electrodeposition.167 4.3.2 Effect of Metal Precursor Concentration on the Electrodeposition from Triton X- 102 Micellar Media and the Associated Electrocatalytic Activity for Methanol Oxidation 175 4.3.3 Catalyst Distribution and Penetration in the Three-Dimensional Graphite Felt Support 181 4.3.4 Direct Methanol Fuel Cell Experiments 184 4.3.5 Direct Formic Acid Fuel Cell Experiments 188 V 4.4 Conclusion .192 4.5 References 193 5 COMPARISON OF Pd AND Pt-Ru SUPPORTED ON GRAPHITE FELT FOR THE DIRECT FORMIC ACID FUEL CELL* 197 5.1 Introduction 197 5.2 Experimental Section 199 5.2.1 Graphite Felt Pretreatment 199 5.2.2 Pd Electrodeposition Procedure 199 5.2.3 Electrochemical Measurements 200 5.2.4 Surface and Analytical Characterization of the Catalysts 201 5.2.5 Membrane Electrode Assembly and Fuel Cell Experiments 202 5.3 Results and Discussion 203 5.3.1 Pd electrodeposition on GF from Triton X-1 02 Micellar Solution: Effect of Metal Precursor Concentration 203 5.3.2 Pd electrodeposition on GF from Triton X-l02 Micellar Solution: Effect of GF Pretreatment and Triton X-102 Concentration 205 5.3.2.1 Effect ofShipley Solution Pretreatment on GF Substrate 205 5.3.2.2 The Synergistic Effect ofGF Pretreatment and Triton X-102 Concentration on the Mass Load, Surface Area, and Morphology ofElectrodeposited Pd: Factorial Experimental Design 208 5.3.2.3 Effect of Triton X-102 Concentration on Pd Electrodeposition 214 5.3.2.4 Crystallography ofElectrodeposited Pd 215 5.3.2.5 Effect ofGF Pretreatment and Triton X-102 Concentration on the Intrinsic Catalytic Activity ofElectrodeposited Pd 217 5.3.3 Catalytic Activity and Long-Term Stability of Pd vs. Pt-Ru 221 5.3.4 DFAFC Performance of Pd vs. Pt-Ru 227 5.4 Conclusion 230 5.5 References 231 6 CONCLUSIONS AND RECOMMENDATIONS 233 6.1 Conclusions 233 6.2 Recommendations 236 6.3 References 239 APPENDIX A - GENERAL BACKGROUND OF METHANOL AND FORMIC ACID 241 APPENDIX B - SEM IMAGES OF Pt-Ru/RVC and Pt-Ru/GF 244 APPENDIX C - THEORETICAL SURFACE AREA CALCULATIONS FROM XRD 246 APPENDIX D - ADDITIONAL SUPPORTING DATA 248 APPENDIX E - DFAFC PERFORMANCE WITH 3 AND 10 M HCOOH 254 APPENDIX F - CYCLIC VOLTAMMOGRAMS OF FORMIC ACID ELECTRO-OXIDATION USING ELECTRODEPOSITED Pd/GF 256 vi LIST OF TABLES Table 1-1: Comparison of modern fuel cell types[6].2 Table 1-2: Volumetric energy density of various fuels [18] 6 Table 1-3: Equilibrium potentials of methanol oxidation, formic acid oxidation, DMFC, and DFAFC at various conditions 13 Table 1-4: Reference electrode potentials [401 15 Table 1-5: Published data of exchange current density and Tafel slope of methanol and formic acid electro-oxidation 21 Table 1-6: Published data of oxidation products of methanol electro-oxidation 43 Table 2-1: RVC pretreatment methods 102 Table 2-2: Mass loading and bulk Pt:Ru ratio of samples prepared by the microemulsion method 118 Table 2-3: Mass loading, Pt:Ru atomic ratio, and active surface area of samples prepared by electrodeposition in microemulsion and aqueous solution 126 Table 3-1: Substrate pretreatment methods 134 Table 3-2: Electrodeposition of Pt-Ru on RVC from colloidal media: comparison between micellar and microemulsion methods. Electrodeposition conditions: 10 A m2, 240 mm, 341 K, flat plate counter electrodes 139 Table 3-3: Effect of perforated counter electrodes on the electrodeposition of Pt-Ru from micellar solution on various substrates. Temperature: 341 K 144 Table 3-4: DMFC performance comparison between published data and results obtained in the present work. Temperature: 333 K. Anode feed: 1 M CH3O - 0.5 M H2S04 (present work), 1 M CH3O in water (literature). Membrane: Nafion® 117 158 Table 4-1: Electrodeposition of Pt-Ru on GF from micellar solution: comparison between Triton X-lOO/isopropanol and Triton X-l02 169 Table 4-2: Effect of the micellar deposition media on the Pt crystallographic for the GF supported catalysts 171 Table 4-3: Apparent Tafel slope and exchange current density of methanol electro oxidation at 298 K on Pt-Ru!GF prepared from 12.5 vol% Triton X-102 with 0.75 mMH2PtC16and (NH4)2RuCl6.Solution: 1 M CH3O and 0.5 M H2S04 181 VII Table 4-4: DMFC performance comparison between published data and results obtained in the present work. Temperature: 333 K; Anode feed: 1 M CH3O - 0.5 M H2S04 (present work), 1 M CH3O in water (literature). Membrane: Nafion® 117 186 Table 4-5: Ionic conductivity of formic acid and sulfuric acid solutions at 333 K 191 Table 5-1: Pd and Sn mass load on GF from Shipley type pretreatment (6 mM PdC12,0.3 M SnC12,4 M HCI) 206 Table 5-2: Pd, Sn, and oxygen surface content of GF before and after Shipley type solution pretreatment 208 Table 5-3: Apparent Tafel slope and exchange current density of formic acid electro oxidation on Pt-Ru/GF and Pd/GF 225 Table D- 1: Mass load and Pt:Ru atomic ratio of Pt-Ru/GF deposited at 20 A m2 in the presence of 12.5 vol% Triton X-102, 0.75 mMH2PtC16,and 0.75 mM (NH4)RuC16 at 341 K for 120 mm 253 Table D-2: Mass load of Pd!GF deposited at 20 A m2 in the presence of 12.5 vol% Triton X-102 and 4.5 mM PdC12 at 341 K for 120 mm after 48 hours of Shipley solution pretreatment 253 VIII LIST OF FIGURES Figure 1-1: Structure of Nafion®.4 Figure 1-2: Mechanisms of proton transport in PEM. (a) Proton hopping; (b) Grotthuss mechanism 4 Figure 1-3: Schematic of DLFC 7 Figure 1-4: Schematic of a three-electrode setup 15 Figure 1-5: Energy of an electrochemical reaction at increasing potential, E> E° 16 Figure 1-6: Graphical representation of Tafel plot 20 Figure 1-7: Effect of gas volume fraction on the effective ionic conductivity 25 Figure 1-8: Concept of cyclic voltammetry 27 Figure 1-9: Concept of chronopotentiometry 29 Figure 1-10: Concept of chronoamperometry 30 Figure 1-11: Concentration profiles of a reactant species at various times into a CA experiment 31 Figure 1-12: Typical Cottrell plot with diffusion control 32 Figure 1-13: Hydrogen adsorption charge 34 Figure 1-14: Typical platinum CV in 0.5 M H2S04recorded at 0.0 10 V s1 showing the hydrogen desorption and CO oxidation peaks [601. Reproduced by permission of The Electrochemical Society 35 Figure 1-15: STM images recorded during the reaction of adsorbed oxygen species with co-adsorbed CO [63]. From J. Wintterlin, S. Volkening, T.V.W. Janssens, T. Zambelli, G. Erti, Science 278 (1997) 1931. Reprinted with permission from AAAS. 36 Figure 1-16: Cu UPD CV in 0.5 M H2S04recorded at 0.010 V s1. (a) Pt; (b) Ru; (c) Pt Ru. Reprinted with permission from [64]. Copyright 2002 American Chemical Society 38 Figure 1-17: Cu UPD stripping charge 39 Figure 1-18: Conventional GDE MEA design comparison 40 ix Figure 1-19: Cross-section SEM image of a full MEA with catalyzed carbon cloths (CCDL) hotpressed onto Nafion® 1135 membrane [70]. Reprinted with permission from Elsevier. Copyright Elsevier (2004) 41 Figure 1-20: Schematic of methanol electro-oxidation pathways 42 Figure 1-21: Linear sweep voltammograms of methanol oxidation on Pt-Ru catalysts with different compositions in 1 M CH3O and 0.5 M H2S04at 298 K. Scan rate: 0.00 1 V s_i [91]. With kind permission from Springer Science + Business Media: J. Appi. Electrochem., “Electrooxidation of Methanol at Platinum-Ruthenium Catalysts Prepared from Colloidal Precursors: Atomic Composition and Temperature Effects”, Vol. 33, 2003, 419-429, L. Dubau, C. Coutanceau, E. Gamier, J.M. Keger, C. Lamy, Fig. 7 44 Figure 1-22: Effect of catalyst substrate withPt80Ru2.Voltammograms measured in 0.5 M H2S04and 1 M CH3O . Temperature: 296 K. Scan rate: 0.0 10 V s [80]. Reprinted with permission from Elsevier. Copyright Elsevier (2004) 46 Figure 1-23: Linear sweep voltammograms of catalysts on Au substrate with different compositions measured in 0.5 M H2S04and 1 M CH3O . Temperature: 296 K. Scan rate: 0.0 10 V s_i [80]. Reprinted with permission from Elsevier. Copyright Elsevier (2004) 47 Figure 1-24: Mass-normalized activity as a function of methanol concentration and anode potential forPt50-Ru,Pt65-Ru20s10,andPt47-Ru290s01r.Cathode: 40 g m2 Pt black with dry air fed at 1 atm and 400 mL mind; Anode: 40 g m2Pt50Ru, Pt65Ru20s10orPt47Ru290s01rwith methanol-water solution fed at 12.5 mL min1. Reprinted with permission from [100]. Copyright 1998 American Chemical Society. 49 Figure 1-25: Dual-path mechanism of the electro-oxidation of formic acid 50 Figure 1-26: Pt and Pd - FTIR spectra recorded in 0.1 M HC1O4 containing 50 mM HCOOH [109]. Reproduced by permission of the PCCP Owner Societies 51 Figure 1-27: Typical cyclic voltammogram of formic acid electro-oxidation on Pt in 0.5 M H2S04with 0.1 M HCOOH at 315 K. Scan rate: 0.100 V s1. Reprinted with permission from [112]. Copyright 2005 American Chemical Society 52 x Figure 1-28: Current oscillation observed in formic acid electro-oxidation on Pt at 1.1 V vs. SHE in 0.5 M H2S04and 1 M HCOOH [113]. Copyright Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission 53 Figure 1-29: Cyclic voltammogram of formic acid electro-oxidation in 0.5 M H2S04and 1 M HCOOH at 0.0 10 V s_i [117]. (a) Pt modified by 10.6 M Bi; (b) Pt. Reprinted with the permission from J. Lee, J. Christoph, P. Strasser, M. Eiswirth, G. Erti, J. Chem. Phys., Vol#. 115,3, 1485-1492, 2001. Copyright 2001, American Institute of Physics 53 Figure 1-30: Chronopontentiometry of formic acid electro-oxidation on Pt, various Pd-Pt, and Pd (60 g m2) in 0.1 M H2S04+ 10 M HCOOH at 293 K and 1000 A m2 [121]. Reproduced by permission of The Electrochemical Society 55 Figure 1-31: Chronoamperometric curves of Pt/C,Pt05d5/Cand Pd/C in 0.25 M HC1O4 + 0.25 M HCOOH at 0.3 V [122]. Reprinted with permission from Elsevier. Copyright Elsevier (2006) 55 Figure 1-32: Polarization curves of DFAFC at 303 K with Pt, Pt-Pd, and Pt-Ru anode catalysts (40 g m2). Cathode: 70 g m2 Pt black with pure humidified oxygen fed at 2 atm and 100 mE min’; Fuel: 5 M HCOOH fed at 0.5 mL mm4 [1231. Reprinted with permission from Elsevier. Copyright Elsevier (2003) 56 Figure 1-33: Polarization curves of formic acid electro-oxidation at 303 K with Pt-Ru and Pt-Au anode catalysts (30 g m2)with 6 M HCOOH fed at 5 mL mini. Scan rate: 0.010 V s_i [124]. Reprinted with permission from Elsevier. Copyright Elsevier (2006) 58 Figure 1-34: Cyclic voltammograms of formic acid electro-oxidation in 1 M HCOOH and 0.5 MH2S04.Scan rate: 0.050 V s1. (a) Pt and (b) Pt modified with Bi. Reprinted with permission from [128]. Copyright 2006 American Chemical Society 59 Figure 1-35: Cyclic voltammograms of formic acid electro-oxidation on Pt and Pt-Pb in 0.25 M HCOOH and 0.5 M H2S04 [130]. Scan rate: 0.050 V s’. Reprinted with permission from Elsevier. Copyright Elsevier (2006) 60 Figure 1-36: Formation of hydrogel 61 Figure 1-37: SEM image of Pt powder synthesized by the sol-gel method by Kim et al. [134]. Reproduced by permission of The Electrochemical Society 62 xi Figure 1-3 8: TEM image of Pt nanoparticles synthesized by the sol-gel method with tetraethoxysilane [135]. With kind permission from Springer Science + Business Media: Glass Phys. Chem., “Formation of Catalytic Layers from Tetraethoxysilane Based Sols for Use in Polymer Fuel Cells”, Vol. 30, 2004, 98-100, O.A. Shilova, V.V. Shilov, N.D. Koshel, EN. Kozlova, Fig. 2 63 Figure 1-39: SEM images of Pt-WO films synthesized by the sol-gel method. (a) Ethanol condensation; (b) Water condensation [136]. Reprinted with permission from Elsevier. Copyright Elsevier (2005) 64 Figure 1-40: SEM images of Bi nanoparticles prepared by boiling bismuth hydroxide for 4 hours in different liquid polyols [141]. (a) 1,3 Propane diol; (b) Diethylene glycol; (c) Glycerol; (d) Tetraethylene glycol. Reprinted with permission from Materials Research Society and D. Goia 66 Figure 1-41: SEM images of Bi nanoparticles prepared by boiling for 4 hours in propylene glycol with different bismuth precursors [1411. (a) Bismuth acetate; (b) Bismuth nitrate; (c) Bismuth subcarbonate. Reprinted with permission from Materials Research Society and D. Goia 67 Figure 1-42: TEM image of Pt nanoparticles synthesized by the microwave-assisted polyol method. Reprinted with permission from [139]. Copyright 1999 American Chemical Society 68 Figure 1-43: SEM image of Pt-Bi nanoparticles prepared by the microwave-assisted polyol method. Reprinted with permission from [129]. Copyright 2005 American Chemical Society 68 Figure 1-44: N(C8H17)C1stabilized metal core 69 Figure 1-45: High resolution TEM image of Pt-Ru nanoparticles synthesized by the colloidal organosol method. Reprinted with permission from [148]. Copyright 1997 American Chemical Society 70 Figure 1-46: TEM images of Pt-Cr/C nanoparticles synthesized by the colloidal organosol method [149]. (a)Pt80-Cr2/ ; (b)Pt50-Cr/ . Reprinted with permission from Elsevier. Copyright Elsevier (2005) 71 XII Figure 1-47: SEM images of electrodeposited Pt catalyst [152]. (a) Deposition potential of-0.8 V vs. SHE; (b-d) Deposition potential of— 1.8 V vs. SHE. Reprinted with permission from Elsevier. Copyright Elsevier (2005) 73 Figure 1-48: SEM images of Zn electrodeposited at various pulse current densities [153]. (a) 4000 A m2; (b) 8000 A m2; (c) 12000 A m2; (d) 16000 A m2. Reprinted with permission from Elsevier. Copyright Elsevier (2003) 73 Figure 1-49: Schematic of a 3-D structure of a hexagonal liquid crystalline phase (a) and the expected nanostructure of a material produced in its presence (b). Reprinted with permission from [158]. Copyright 1999 American Chemical Society 74 Figure 1-50: SEM image of nanostructured Pt deposited at -0.76 V vs. SHE at 338 K in a hexagonal liquid crystalline phase [158]. Reprinted with permission from [158]. Copyright 1999 American Chemical Society 75 Figure 1-51: SEM image of nanostructured Ni deposited on polished Ni at 50 A m2 at 298 K for 1 hour in a Triton X- 1 00/poly-acrylic acidlwater hexagonal liquid crystalline phase [160]. Reprinted with permission from Elsevier. Copyright Elsevier (2004) 75 Figure 1-52: High resolution SEM micrographs of Pt-Ru (a) and Pt-Ru-Mo (b) deposits on the pressed graphite fiber surface [96]. Reprinted with permission from Elsevier. Copyright Elsevier (2007) 76 Figure 1-53: Schematic of metal nanoparticle formation in reverse microemulsion 77 Figure 1-54: TEM image of Pt-Co nanoparticles prepared by microemulsion [164]. Reproduced by permission of The Royal Society of Chemistry 78 Figure 1-55: SEM images of silica-templated carbon support with different pore size. (a) 25 nm. (b) 68 nm; (c) 245 nm; (d) 512 nm. Reprinted with permission from [167]. Copyright 2004 American Chemical Society 79 Figure 1-56: Single DMFC cell performance comparing the commercial E-Tek catalyst with catalysts prepared on custom-made carbon supports. The numbers in the legend designate the pore diameter of the porous carbon replicas in nm. Temperature: 303 K. Anode: 30 g Pt-Ru m2 with 2 M CH3O fed at 1 mL min’. Cathode: 50 g Pt m2 with dry air fed at 500 mL min1.Reprinted with permission from [167]. Copyright 2004 American Chemical Society 80 XIII Figure 1-57: Schematic representation of the multi-layer anode design for DMFC as suggested by Wilkinson et al. [168] 81 Figure 1-58: Comparison of the performance of the DMFC at 363 K with different anodes under different methanol concentrations [169]. Anode: 40 g m2 Pt-Ru black or Pt-Ru/Ti withCH3OH-water fed at 10 mL min1.Cathode: 35 g m2 Pt/C with dry air fed at 200 mL min1.Reproduced by permission of the PCCP Owner Societies. 82 Figure 1-59: Comparison of the performance of the DFAFC at 333 K [159]. Anode: 20 g m2 Pd/C ( ) or Pd/Ti mesh (thermal decomposition, ) or Pd/Ti mesh (electrodeposition, ) with 1 M HCOOH fed at 1 mL min’. Cathode: 20 g m2 Pt/C with atmospheric air fed at 400 mL min1.Reproduced with permission from [172]. Copyright 2007 Journal ofNew Materials for Electrochemical Systems 83 Figure 1-60: Comparison of the performance of the DFAFC at 333 K [159]. Anode: (i) 20 g m2 Pt-Sn/C or (ii)Pt-Sn!Ti mesh (thermal decomposition) with 1 M HCOOH fed at 1 mL min1.Cathode: 20 g m2 Pt/C with atmospheric air fed at 400 mL min1. Reproduced with permission from [1721. Copyright 2007 Journal ofNew Materials for Electrochemical Systems 84 Figure 1-61: Conventional vs. extended reaction zone DLFC anode design 86 Figure 2-1: Schematic of electrodeposition vessel and electrode assembly 103 Figure 2-2: Anodic scans obtained on blank RVC subjected to Cu bulk deposition and UPD. Temperature: 298 K; Scan rate: 0.050 V s 105 Figure 2-3: Electrical conductivity of 7 vol% Triton X- 100/ 90 vol% cyclohexane / 3 vol% water emulsion at a temperature range of 319 to 345 K 108 Figure 2-4: Voltammograms of methanol oxidation at 298 K on Pt-RuJRVC in 1 M CH3O and 0.5 MH2SO4.Effect of temperature during catalyst electrodeposition via emulsion with a superficial deposition current density and time of 26 A m2 and 60 minutes. Current density is given in A m2 geometric area basis. Scan rate: 0.005 Vs 109 Figure 2-5: Effect of current density on the Pt-Ru morphology electrodeposited on RVC pretreated in HNO3 at 343 K using W/O emulsion with constant deposition charge. (a) Deposition current and time of 10 A m2 and 234 minutes; (b) Deposition current xiv and time of 26 A m2 and 90 minutes; (c) Deposition current and time of 40 A m2 and 59 minutes 111 Figure 2-6: Voltammograms of methanol oxidation on Pt-RuJRVC in 1 M CH3O and 0.5 MH2S04.Effect of electrodeposition current density and time at 343 K. Current density is given in A m2 geometric area basis. Temperature: 298 K. Scan rate: 0.005 Vs’ 112 Figure 2-7: Voltammograms of methanol oxidation on Pt-Ru/RVC in 1 M CH3O and 0.5 M H2S04at 298K. Effect of ruthenium precursor (RuC13 vs. (NH4)2RuC16. Electrodeposition was carried out at 343 K with a current density and time of 26 A m2 and 90 minutes. Scan rate: 0.005 V s1. Current density is given in A m2 geometric area basis 113 Figure 2-8: Partial phase diagram of the cyclohexane, Triton X-100/isopropanol, and aqueous phase with 0.01 MH2PtCI6and (NH4)2RuCI6ternary system at 341 K... 115 Figure 2-9: The effect of microemulsion on the Pt-Ru morphology electrodeposited on RVC pretreated in 35 vol% HNO3 at 341 K with a superficial deposition current density and time of 10 A m2 and 234 minutes 116 Figure 2-10: Voltammograms of electrodeposited RVC in 1 M CH3O and 0.5 MH2S04. Effect of colloidal media composition, microemulsion vs. emulsion on the catalytic activity of Pt-RuJRVC towards methanol oxidation. Temperature: 298 K. Scan rate: 0.005 V s* Inset: Intrinsic catalytic activity based on real surface area 117 Figure 2-11: SEM images of the RVC surface. (a) No pretreatment; (b) Pretreated with electrochemical cycling from 1.44 to 2.09 V vs. SHE at 0.001 V s1 repeated 50 times in 98% H2S04 119 Figure 2-12: XPS spectra. (a) No pretreatment; (b) Pretreated with electrochemical cycling from 1.44 to 2.09 V vs. SHE at 0.00 1 V s1 repeated 50 times in 98%H2S04. 120 Figure 2-13: SEM images of Pt-Ru electrodeposited on RVC pretreated with electrochemical cycling. Electrodeposition was carried out at 341 K using microemulsion with a superficial deposition current density and time of 10 A m2 and 234 minutes. (a) No heat treatment; (b) After heat treatment in N2 at 573 K for 1 hour 122 xv Figure 2-14: Effect of different RVC pretreatment methods for Pt-Ru electrodeposition via microemulsion. Voltammograms of methanol oxidation for Pt-RuIRVC prepared at 341 K with microemulsion. Superficial deposition current density and time of 10 A m2 and 234 minutes. Electrolyte: 1 M CH3O and 0.5 M H2S04. Current density is given in A m2 geometric area basis. Temperature: 298 K. Scan rate: 0.005 V s1. Inset: Effect of postdeposition heat treatment 123 Figure 2-15: Blank scan and Cu UPD stripping curves of Pt-Ru electrodeposited on RVC prepared at 341 K with microemulsion with a superficial deposition current density and time of 10 A m2 and 234 minutes. Electrolyte: 0.5 M H2S04(blank); 0.5 M H2S04+ 0.002 M CuSO4(Cu UPD) Temperature: 298 K; Scan rate: 0.050 V s’. 124 Figure 2-16: Chronopotentiometry data of methanol oxidation on Pt-RuIRVC in 1 M CH3O and 0.5 MH2S04.Temperature: 298 K. Current density: 50 A m2. Samples prepared in microemulsion and aqueous solution. Both samples were pretreated with electrochemical cycling, prepared at 341 K with a superficial deposition current density and time of 10 A m2 and 234 minutes, and heat treated after deposition.. 126 Figure 3-1: SEM images of 3-D substrates. (a) RVC structure; (b) RVC surface; (c) GF structure; (d) GF surface; (e) Ti mesh structure [211; (f) Ti mesh surface. Reproduced by permission of The Electrochemical Society 133 Figure 3-2: Voltammograms of methanol electro-oxidation using Pt-Ru electrodeposited on RVC: Effect of colloidal electrodeposition media. Electrolyte: 1 M CH3O and 0.5 MH2S04.Temperature: 298 K. Current density is given in A m2 geometric area basis. Scan rate: 0.005 V s’ 139 Figure 3-3: SEM images of electrodeposited Pt-Ru on RVC prepared with microemulsion and micellar solution. (a) Microemulsion; (b) Micellar solution 141 Figure 3-4: Blank scan and Cu UPD stripping curves of electrodeposited RVC prepared at 341 K from micellar media with a superficial deposition current density and time of 10 A cm2 and 240 minutes. Test solution: 0.5 M H2S04(blank); 0.5 M H2S04+ 0.002 M CuSO4 (Cu UPD) Temperature: 298 K; Scan rate: 0.050 V s1 142 Figure 3-5: Schematic of counter electrodes used in colloidal electrodeposition 143 xvi Figure 3-6: SEM images of electrodeposited Pt-Ru on RVC prepared from micellar solution with perforated counter electrodes at 341 K. Effect of deposition current density (a) 10 A m2; (b) 20 A m2; (c) 40 A m2; (d) 60 A m2 145 Figure 3-7: Blank scan and Cu UPD stripping curves of electrodeposited RVC prepared at 341 K from micellar media with perforated counter electrodes with a superficial deposition current density and time of 10 A cm2 and 240 minutes. Test solution: 0.5 M H2S04(blank); 0.5 M H2S04+ 0.002 M CuSO4 (Cu UPD) Temperature: 298 K; Scan rate: 0.050 V s 146 Figure 3-8: Voltammograms of methanol electro-oxidation using Pt-Ru electrodeposited on RVC. Effect of electrodeposition conditions: counter electrode type and superficial current density. 1 M CH3O and 0.5 MH2S04.Temperature: 298 K. Scan rate: 0.005 V s 147 Figure 3-9: SEM images of Pt-Ru electrodeposits prepared from micellar solution with different substrates at 20 A m2, 341 K and perforated counter electrodes. (a) RVC; (b) uncompressed GF; (c) Ti Mesh 149 Figure 3-10: Effect of three-dimensional support on the electrocatalytic activity of Pt-Ru toward methanol electro-oxidation. Solution: 1 M CH3O and 0.5 MH2S04. Temperature: 298 K. Scan rate: 0.005 V s1. Samples prepared at a temperature of 341 K and a deposition current density of 20 A m2 for 120 mm with perforated counter electrodes in micellar solution 150 Figure 3-11: Chronopotentiometric data for methanol electro-oxidation of catalyzed RVC, GF, and Ti mesh in 1 M CH3O and 0.5 M H2S04at 298 K. Current density: 50 A m2. Samples prepared at a temperature of 341 K and a deposition current density of 20 A m2 for 120 mm with perforated counter electrodes in micellar solution 151 Figure 3-12: Effect of three-dimensional anode with electrodeposited Pt-Ru on DMFC performance. Catalysts electrodeposited at 341 K with a superficial current density and time of 20 A m2 and 120 mi respectively. Fuel: 1 M CH3O in 0.5 M H2S04 at 2 mL min1.Cathode: 40 g Pt nf2, dry 02 fed at 2.5 bar and 500 mL min1. Temperature: 333 K 154 xvii Figure 3-13: DMFC power density. Three-dimensional anode support comparison. Conditions idem Figure 3-12. (a) Area specific power density; (b) Mass specific power output 155 Figure 4-1: Voltammograms of methanol electro-oxidation using Pt-Ru electrodeposited on GF: Effect of micellar electrodeposition media. Solution: 1 M CH3O and 0.5 M H2S04.Temperature: 298 K. Scan rate: 0.005 V s1. Inset: enlarged view from 0 to 0.5V 169 Figure 4-2: Blank scan and Cu UPD stripping curves of electrodeposited GF prepared at 341 K from 5 vol% Triton X-102 micellar media with a superficial deposition current density and time of 20 A cm2 and 120 minutes. Test solution: 0.5 M H2S04 (blank); 0.5 M H2S04+ 0.002 M CuSO4 (Cu UPD) Temperature: 298 K; Scan rate: 0.050Vs’ 170 Figure 4-3: XRD spectra of Pt-Ru prepared from 5vol%/20vol% Triton X 100/isopropanol and 12.5 vol% Triton X-102. Precursor concentration — Pt: 0.25 mM, Ru: 0.25 mM 172 Figure 4-4: Effect of Triton X- 102 content. (a) Cathodic voltammograms of Pt electrodeposition on GF with 0.75 mMH2PtC16;(b) Cathodic voltammograms of Ru electrodeposition on GF with 0.75 mM (NH4)RuC1 Temperature: 341 K. Scan rate: 0.005 V s 174 Figure 4-5: Effect of metal precursor concentration in the electrodeposition media on the Pt-Ru catalyst characteristics. Superficial current density 20 A m2, 341 K 175 Figure 4-6: SEM images of Pt-Ru electrodeposited on GF prepared with 12.5 vol% Triton X-102 and different precursor concentrations. Superficial current density 20 A m2, 341 K 176 Figure 4-7: Voltammograms of methanol electro-oxidation using Pt-Ru electrodeposited on GF: Effect of precursor concentration and deposition current. (a) Real surface area basis; (b) mass basis. Solution: I M CH3O and 0.5 MH2S04.Temperature: 298 K. Scan rate: 0.005 V s1 178 Figure 4-8: Chronopotentiometry data of methanol electro-oxidation using Pt-Ru electrodeposited on GF: Effect of precursor concentration and deposition current. xviii Solution: 1 M CH3O and 0.5 MH2S04.Current Density: 50 A m2. (a) Temperature: 298 K; (b) Temperature: 333 K 179 Figure 4-9: XRD spectra of Pt-Ru prepared from 12.5 vol% Triton X-102. Precursor concentration — Pt: 0.75 mM, Ru: 0.75 mM 180 Figure 4-10: Locations of fiber pullout for SEM imaging of the Pt-Ru catalyst dispersion and penetration in the three-dimensional GF support (see Figure 4-11) 182 Figure 4-11: Pt-Ru dispersion and penetration into the three-dimensional support: SEM images of catalyzed GF at locations ito 12 (see Figure 4-10). Sample prepared at 20 A m2 with 12.5 vol% Triton X-102, 0.75 mMH2PtC16,and 0.75 mM (NH4)2RuC16, 341 K 183 Figure 4-12: Schematic of Different Anode Designs — CCM, CCDL, and extended reaction zone anodes 184 Figure 4-13: DMFC performance at 333 K — effect of extended reaction zone vs. conventional anode design. Fuel: 1 M CH3O in 0.5 M H2S04at 2 mL min1. Cathode: 40 g Pt m2, dry 02 fed at 2.5 bar and 500 mL min* (a) Polarization curve; (b) Power density 185 Figure 4-14: DFAFC performance — effect of extended reaction zone vs. conventional anode design. Fuel: I M HCOOH with or without 0.5 M H2504at 6 mL min* Cathode: 40 g Pt m2, dry 02 fed at 2.5 bar and 500 mL min’. Temperature: 333 K. (a) Polarization curve; (b) Power density 189 Figure 4-15: DFAFC performance at 333 K — effect of formic acid concentration on peak power density. Extended reaction zone vs. conventional anode design. Fuel: 1, 3, and 10 M HCOOH with or without 0.5 M H2S04at 6 mL min1.Cathode: 40 g Pt m 2 dry 02 fed at 2.5 bar and 500 mL min1 190 Figure 5-1: SEM images of Pd catalyzed GF (unpretreated) prepared with 12.5 vol% Triton X-102 at 341 K with a deposition current density of 20 A m2 for 120 mm. (a) Pd (0.75 mM PdC12); (b) Pd (4.5 mM PdC12) 204 Figure 5-2: Pd and Sn deposition yield as a function of pretreatment time 206 Figure 5-3: The effect of Shipley (6 mM PdC12 and 0.3 M SnC12 in 4 M HC1) pretreatment on the GF surface (middle of the felt). (a) Unpretreated; (b) 24-hr pretreatment; (c) 48-hr pretreatment 207 xix Figure 5-4: Effect of Triton X-102 concentration and Shipley solution pretreatment time on the Pd mass load and surface area obtained by electrodeposition at 341 K with a deposition current density of 20 A m2: Experimental design matrix with 2 variables at 3 levels 209 Figure 5-5: SEM images of electrodeposited Pd/GF with 0% Triton X-102. Effect of Shipley solution pretreatment time. (a, b) no pretreatment (exterior vs. interior); (c, d) 24 hr pretreatment (exterior vs. interior); (e, f) 48 hr pretreatment (exterior vs. interior) 210 Figure 5-6: SEM images of electrodeposited Pd/OF with 12.5% Triton X-102. Effect of Shipley solution pretreatment time. (a, b) no pretreatment (exterior vs. interior); (c, d) 24 hr pretreatment (exterior vs. interior); (e, f) 48 hr pretreatment (exterior vs. interior) 211 Figure 5-7: SEM images of electrodeposited Pd/GF with 25% Triton X-102. Effect of Shipley solution pretreatment time. (a, b) no pretreatment (exterior vs. interior); (c, d) 24 hr pretreatment (exterior vs. interior); (e, 48 hr pretreatment (exterior vs. interior) 212 Figure 5-8: Voltammograms of Pd electrodeposition on GF: Effect of Triton X-102 content and surface pretreatment. PdCl2 concentration: 4.5 mM. Temperature: 341 K. Scan rate: 0.005 V 215 Figure 5-9: XRD spectra of Pci/OF electrodeposited in the presence of 12.5 vol% Triton X- 102 after 48 hours of Shipley solution pretreatment 216 Figure 5-10: Blank scan and Cu UPD stripping curves of Pd/OF electrodeposited in the presence of 12.5 vol% Triton X- 102 after 48 hours of Shipley solution pretreatment. Test solution: 0.5 M H2S04(blank); 0.5 M H2S04+ 0.002 M CuSO4 (Cu UPD) Temperature: 298 K; Scan rate: 0.050 V s1 217 Figure 5-11: Cyclic voltammograms of formic acid electro-oxidation using Pd electrodeposited on OF. (a) no Triton X-102 and 24-hr pretreatment; (b) 25 vol% Triton X-102 and no pretreatment. Solution: 1 M HCOOH and 0.5 MH2S04. Temperature: 298K. Scan rate: 0.005 V s’ 218 xx Figure 5-12: Cyclic voltammograms of formic acid electro-oxidation with Pt/C and Pd/C in 3 M HCOOH and 1 M H2S04at 298 K [25] Scan rate: 0.0 10 V s1. Reprinted with permission from Elsevier. Copyright Elsevier (2006) 219 Figure 5-13: Oscillatory phenomena during formic acid electro-oxidation using Pd electrodeposited on GF prepared with 25 vol% Triton X- 102 with 24-hr Shipley pretreatment. Solution: 1 M HCOOH and 0.5 MH2S04.Temperature: 298 K. Scan rate: 0.005 V s 220 Figure 5-14: Intrinsic formic acid oxidation current density at 0.3 V vs. SHE on electrodeposited Pd/OF catalysts in 1 M HCOOH and 0.5 M H2S04at 298 K 221 Figure 5-15: Superficial formic acid oxidation current density at 0.3 V vs. SHE on electrodeposited Pd/GF catalysts in I M HCOOH and 0.5 MH2S04at 298 K 222 Figure 5-16: Voltammograms of formic acid electro-oxidation using Pt-Ru/OF and Pd/OF (prepared with 12.5 vol% Triton X-102 with 48 hours of Shipley pretreatment). (a) Real surface area basis; (b) Mass basis. Solution: 1 M HCOOH and 0.5 MH2S04.Temperature: 298 K. Scan rate: 0.005 V s 223 Figure 5-17: Chronoamperometry data of formic acid electro-oxidation using Pt-Ru and Pd electrodeposited on OF. (a) PdJGF; (b) Expanded view of Pd/OF at 0.45 V vs. SHE; (c) Pt-RuJOF; (d) Expanded view of Pt-RuJGF at 0.45 V vs. SHE; Solution: 1 M HCOOH and 0.5 MH2S04.Temperature: 298 K 224 Figure 5-18: Tafel plots of formic acid electro-oxidation in 1 M HCOOH and 0.5 M H2S04at 298 K. (a) 10 g m2 (4:1) Pt-Ru/OF; (b) 57 g m2 Pd/GF 225 Figure 5-19: Long-term chronoamperometry data of formic acid electro-oxidation using Pt-Ru and Pd electrodeposited on OF. Solution: 3 M HCOOH and 0.5 M H2S04. Temperature: 298 K. Constant potential: 0.65 V vs. SHE 226 Figure 5-20: DFAFC performance — effect of extended reaction zone vs. conventional anode design and Pd vs. Pt-Ru. The PdJGF was prepared with 12.5 vol% Triton X 102 and 48 hours of Shipley pretreatment. Fuel: 1 M HCOOH with or without 0.5 M H2S04at 6 mL min’. Cathode: 40 g Pt m2, dry 02 fed at 2.5 bar and 500 mL min’. Temperature: 333 K. (a) Polarization curve; (b) Power density 228 Figure 5-21: Mass-specific DFAFC performance — effect of extended reaction zone vs. conventional anode design and Pd vs. Pt-Ru. The Pd/OF was prepared with 12.5 xxi vol% Triton X-102 and 48 hours of Shipley pretreatment. Fuel: 1 M HCOOH with or without 0.5 M H2S04at 6 mL min1.Cathode: 40 g Pt m2, dry 02 fed at 2.5 bar and 500 mL min1.Temperature: 333 K. (a) Polarization curve; (b) Power density. 229 Figure B-i: SEM images of Pt-Ru/RVC and Pt-RuJGF prepared without surfactant at 341 K. (a, b) Pt-Ru/RVC prepared with 0.25 mMH2PtC16and (NH4)2RuC16and deposition current density 10 A m2 (a: exterior vs. b: interior); (c, d) Pt-Ru/GF prepared with 0.75 mMH2PtC16and (NH4)2RuC16and deposition current density 20 A m2 (c: exterior vs. d: interior) 244 Figure B-2: SEM images of Pt-Ru!RVC and Pt-Ru/GF prepared with surfactants at 341 K. (a, b) Pt-Ru/RVC prepared with 0.25 mMH2PtC16and (NH4)2RuC16in 5 vol% Triton X-100 / 20 vol% isopropanol micellar media and a deposition current density 10 A m2 (a: exterior vs. b: interior); (c, d) Pt-Ru/GF prepared with 0.75 mM H2PtC16and (NH4)2RuC16in 12.5 vol% Triton X-102 micellar media and a deposition current density 20 A m2 (C: exterior vs. d: interior) 245 Figure D-i: Voltammograms of methanol electro-oxidation using Pt-Ru electrodeposited on GF: Triton X-102 vs. aqueous deposition media. (a) Real surface area basis; (b) Mass basis. Solution: 1 M CH3O and 0.5 MH2S04.Temperature: 298 K. Scan rate: 0.005 V s1 248 Figure D-2: Chronoamperometry data of methanol electro-oxidation using Pt-Ru/GF (10 g m2, 4:1 atomic ratio, prepared in 12.5 vol% Triton X-102 with 0.75 mMH2PtC16 and (NH)RuC16in i M CH3O and 0.5 MH2S04.Temperature: 298 K 249 Figure D-3: Tafel plots of methanol electro-oxidation in 1 M HCOOH and 0.5 M H2S04 at 298 K with Pt-RuIGF prepared in 12.5 vol% Triton X-102 with 0.75 mMH2PtC16 and (NH4)2RuCI6(10 g m2, 4:1 atomic ratio) 250 Figure D-4: DMFC performance of 10 g m2 Pt-RuJGF. Fuel: 1 M CH3O with 0.5 M H2S04at 2 mL mind. Cathode: 40 g Pt m2, dry 02 fed at 2.5 bar and 500 mL min1. Temperature: 333 K. (a) Polarization curve; (b) Power density 251 xxii Figure D-5: Catalytic activity of Pd/OF deposited at 20 A m2 in the presence of 12.5 vol% Triton X-102 and 4.5 mM PdC12 at 341 K for 120 mm after 48 hours of Shipley solution pretreatment. Solution: 1 M HCOOH and 0.5 MH2S04. Temperature: 298 K. (a) CA data, Potential: 0.25 V vs. SHE; (b) CP data, Current density: 50 A m2 252 Figure E- 1: DFAFC performance — effect of extended reaction zone vs. conventional anode design. Fuel: 3 M HCOOH with or without 0.5 M H2S04at 6 mL mi&’. Cathode: 40 g Pt m2, dry 02 fed at 2.5 bar and 500 mL min1.Temperature: 333 K. (a) Polarization curve; (b) Power density 254 Figure E-2: DFAFC performance — effect of extended reaction zone vs. conventional anode design. Fuel: 10 M HCOOH with or without 0.5 M H2S04at 6 mL min* Cathode: 40 g Pt m2, dry 02 fed at 2.5 bar and 500 mL min1.Temperature: 333 K. (a) Polarization curve; (b) Power density 255 Figure F-i: Cyclic voltammograms of formic acid electro-oxidation using Pd/OF with no pretreatment: Effect of Triton X-102 concentration. Solution: 1 M HCOOH and 0.5 MH2S04.Temperature: 298 K. Scan rate: 0.005 V s 256 Figure F-2: Cyclic voltammograms of formic acid electro-oxidation using Pd! OF with 24-hr Shipley pretreatment: Effect of Triton X-102 concentration. Solution: 1 M HCOOH and 0.5 MH2S04.Temperature: 298 K. Scan rate: 0.005 V s 257 Figure F-3: Cyclic voltammograms of formic acid electro-oxidation using Pd/with 48-hr Shipley pretreatment: Effect of Triton X-102 concentration. Solution: 1 M HCOOH and 0.5 M H2S04.Temperature: 298 K. Scan rate: 0.005 V s 258 xxiii LIST OF SYMBOLS a Tafel Parameter V am Mass-Specific Surface Area m2 g a Area Enhancement Factor m2total m2 geom a Volume-Specific Surface Area m2totalm3geom a Chemical Activity B Full Width at Half Height rad b Tafel Slope V decade1 c Concentration mol m3 d Diameter of Crystallite m E Cell Potential V F Faraday’s Constant 96485 C mol1 g Interaction Parameter - i Current Density per Geometric Area or A m2 or Current Density per Effective Catalytic Area A m2real Km Mass Transport Coefficient m 11 k Reaction Rate Constant m m Catalyst Load g m2geom P Pressure Pa or bar q Heat J s Reaction Coefficient - T Temperature K V Volume m3 w Work Done by System J We Electrical Work Done by System J x Length of Conducting Medium m Transfer Coefficient mol Symmetry Factor 6 Diffusion Layer Thickness m xxiv U Overpotential V K Conductivity S rn1 Wavelength of X-Ray (XRD) rn 0 Equilibrium Coverage - 0 Diffraction Angie rad Scan Rate V s Time to Deplete Surface Reactant Concentration s AG Change in Gibbs Free Energy J mol’ AH Change in Enthalpy J mol’ AS Change in Entropy J mol K’ AU Change in Internal Energy J mol’ MDohm Ohmic Loss V xxv LIST OF ABBREVIATIONS AFC Alkaline Fuel Cell BEV Butler-Volmer-Erdey-Grüz CA Chronoamperometry CCDL Catalyst-Coated Diffusion Layer CCM Catalyst-Coated Membrane CP Chronopotentiometry CV Cyclic Voltammetry DBFC Direct Borohydride Fuel Cell DEFC Direct Ethanol Fuel Cell DFAFC Direct Formic Acid Fuel Cell DHE Dynamic Hydrogen Electrode DMFC Direct Methanol Fuel Cell DT Decal Transfer ECA Effective Catalyst Area FESEM Field-Emission Scanning Electron Microscope FTIR Fourier Transform Infrared Spectroscopy GDE Gas-Diffusion Electrode GF Graphite Felt ICP-AES Inductively Coupled Plasma Atomic Emission Spectroscopy IUPAC International Union of Pure and Applied Chemistry MCFC Molten-Carbonate Fuel Cell MEA Membrane-Electrode Assembly MSE Mercury-Mercurous Sulfate Electrode MTBE Methyl Tertiary-Butyl Ether NASA National Aeronautics and Space Administration OCV Open-Circuit Voltage PAFC Phosphoric Acid Fuel Cell PEMFC Proton Exchange Membrane Fuel Cell xxvi PPC Pores Per Centimeter PTFE Polytetrafluoroethylene RVC Reticulated Vitreous Carbon SEM Scanning Electron Microscope SHE Standard Hydrogen Electrode SG Sol-Gel SOFC Solid Oxide Fuel Cell SPEEK Sulfonated Poly(Ether Ether Ketone) STM Scanning Tunneling Microscope TBAP Tetrabutylammonium Perchlorate TEM Transmission Electron Microscopy THF Tetrahydrofuran UPD Underpotential Deposition XPS X-Ray Photoelectron Spectroscopy XRD X-Ray Diffraction xxvii ACKNOWLEDGEMENTS I offer my enduring gratitude to the faculty, staff, and fellow colleagues who have made this thesis a reality. I would like to express my deepest thanks to my supervisor, Dr. Elöd Gyenge, for his innumerable assistance and treasured advices throughout the duration of my studies. ElOd has been a great role model and mentor, whom I have learned invaluable knowledge and skills. I thank my fellow colleagues in the Applied Electrochemistry and Fuel Cell Laboratory, including Dr. Alex Bauer, Derek Lycke, Vincent Lam, and Dr. Anna Ignaszak for their contributions to the laboratory, technical discussions, and the joy they brought along that made the lab a wonderful environment to work in. I owe special thanks to Horace Lam, Qi Chen, Lori Tanaka, Helsa Leong, and the staff of the Chemical and Biological Engineering Stores and Workshop for all their assistance. I also thank the Natural Sciences and Engineering Research Council of Canada for four years of financial support. Lastly, I express my warmest gratitude to my family for supporting and encouraging me throughout the hard times. This wouldn’t have been possible without you. xxviii CO-AUTHORSHIP STATEMENT The experimental work and data analysis presented in this thesis were executed by Tsz Hang Tommy Cheng. The experimental plan was co-developed by Dr. Elöd Gyenge and Tsz Hang Tommy Cheng. Dr. Elöd Gyenge also worked closely together with Tsz Hang Tommy Cheng in preparing the manuscripts for journal publications. xxix 1 INTRODUCTION 1.1 General Considerations The demand for energy is continuously increasing in an age with rapid growth in population, social, and economic development. Since the dawn of the industrial revolution, mankind has depended mainly on fossil fuels to provide energy. Coal, crude oil, and natural gas amount approximately 64% of the total electricity generated worldwide while the remaining amount is made up by renewable energy sources (such as hydro, wind, and solar energy) [1]. With the current oil production and consumption level, the global oil reserves are sufficient for around 40 years [2]. However, due to the exploding population and economic growth in many parts of the world (e.g. China and India), the current oil reserves can be expected to diminish at an even faster rate. Coal reserves, on the other hand, are expected to be sufficient for over 150 years [2]. Contrary to coal and oil, significant new methane reserves have been recently discovered in the form of hydrates, estimated to be 20 times higher than the combined reserves of coal, oil, and gas [3]. It was also predicted that the technology required to extract those methane hydrates from the ocean will be ready in 10 years. It can be seen that fossil fuels cannot be treated as a long-term solution for energy production. The ultimate goal is to employ renewable resources, such as solar, wind, hydro, biomass, and nuclear energy. Another potential alternative is the use of electrochemical power sources: batteries and fuel cells, to generate electricity. Batteries generate electricity from the stored chemical energy in the two half-cells through electrochemical reactions. Common batteries include the lead-acid, alkaline, zinc-air, molten salt, and the lithium-ion rechargeable batteries. In contrast to batteries, fuel cells generate electricity from the oxidation of a fuel and the reduction of an oxidant fed to the anode and cathode, respectively. The instantaneous refueling time of the fuel cell and the superior specific energy density of fuel cell fuels (e.g. 6000 Wh kg’ for methanol) compared to competing batteries (600 Wh kg’ for lithium-ion battery) have made the fuel cell one of the most popular and important research topics [4-5]. I In general, fuel cells are categorized into different types depending on the fuel, oxidant, and electrolyte used. The most common fuel cells are the proton exchange membrane fuel cell (PEMFC), alkaline fuel cell (AFC), phosphoric acid fuel cell (PAFC), molten carbonate fuel cell (MCFC), and the solid oxide fuel cell (SOFC). See Table 1-1 for their typical operating conditions and applications. Table 1-1: Comparison of modern fuel cell types [61. Fuel Cell Type Common Electrolyte Operating Applications Temperature (K) PEMFC Solid organic polymer 333 — 373 Backup power, poly-perfluorosulphonic portable power, acid small distributed generation, transportation AFC Concentrated solution of 363 — 373 Military, space potassium hydroxide PAFC Liquid phosphoric acid 423 — 473 Distributed generation MCFC Molten lithium, sodium, 873 — 973 Electric utility, and/or potassium distributed carbonates generation SOFC Yttria stabilized zirconia 873 — 173 Auxiliary power, electric utility, distributed generation Arguably the most vastly researched fuel cell is the hydrogen-oxygen PEMFC, which was first demonstrated in 1839 by Sir William Grove [7]. The fuel cell, which Grove called a gas voltaic battery, is comprised of two electrodes (anode and cathode) 2 and a phosphoric acid electrolyte. Hydrogen is oxidized on the anode and oxygen is reduced on the cathode. Electron transport occurs from the anode to the cathode through an external circuit to provide electrical power. Protons generated from the hydrogen oxidation reaction flow through the electrolyte towards the cathode. In 1955, Thomas Grubb, from the General Electric Company, invented the suiphonated polystyrene ion- exchange membrane, which fairly soon thereafter was used as the solid polymer electrolyte. This was the first PEMFC design [8]. In 1958, another researcher from General Electric, Leonard Niedrach, deposited platinum catalyst onto the ion-exchange membrane to catalyze the fuel cell reactions. In 1959, Francis Thomas Bacon successfully demonstrated the first 6-kW commercial alkaline fuel cell stack [9]. The fuel cell patent was later used by the US National Aeronautics and Space Administration (NASA) in the Project Apollo space program. Despite over 50 years of research and development, fuel cell commercialization is still facing numerous obstacles in terms of cost, durability, fuelling infrastructure, and fuel storage. In order to be competitive with conventional energy sources, the fuel cell system cost should not exceed $75 kW’ for automobiles, $300-450 kW’ for buses, $600- 900 kW for portable applications, and $600-900 kW’ for cogeneration systems [10]. The current fuel cell system costs ranged from $4500-7500 kW’, up to 2 orders of magnitude higher for personal vehicles. The materials used in fuel cells account for a major fraction of the overall system cost. The most commonly used proton exchange membrane is Nafion®, developed by Walther Grot of DuPont in 1960. Nafion® is a sulfonated tetrafluorethylene copolymer (Figure 1-1) containing sulfonic acid groups (- SO3H) for proton transport via proton hopping, Grotthuss mechanism, and traditional diffusion (Figure 1-2) [11-13]. The cost of Nafion® membranes accounts between $60- 300 kW’ in terms of fuel cell output [10]. 3 Figure 1-1: Structure of Nation® a) PTFE Hydronium Ion Bulk Water PTFE CFj CF3 F2 F2 S /OH PTFE 0 0 Sulfur • Oxygen o Hydrogen CF2 * Surface Water b)/ Figure 1-2: Mechanisms of proton transport in PEM. (a) Proton bopping; (b) Grottbu55 mechanism 4 For a typical hydrogen PEMFC with power density of 10 kW m2, the platinum catalyst cost amounts for approximately $20 kW’, with a total catalyst load of 5 g Pt m2 (anode: 1 g m2 and cathode: 4 g m2). However, in the cases of using alternative fuels, such as methanol in the direct methanol fuel cell (DMFC), a higher platinum catalyst load is required due to much slower anodic kinetics (to be discussed later). The catalyst cost alone becomes $600-1200 kW, already higher than the overall system cost for the hydrogen PEMFC. Aside from system cost, another major obstacle the fuel cell technology is facing is durability. It is generally accepted that the required lifetime for a personal automobile is 5000 hours. With the current technology, the achieved hydrogen PEMFC durability is about 2000+ hours [141. The main causes of fuel cell degradation are water management issues, fuel contamination with species such as CO and SO2 that can poison the catalyst, formation of peroxide intermediates that can chemically attack the polytetrafluoroethylene (PTFE) membrane backbone leading to membrane thinning and defects, catalyst sintering resulting in effective catalyst area (ECA) loss, carbon corrosion due to voltage fluctuation from startup/shutdown processes, and ruthenium dissolution/re deposition leading to higher cathode kinetic loss [14-171. Adapting the use of hydrogen as a transportation fuel requires new infrastructure for refueling, which further complicates the situation along with the fuel storage problem. Even when stored at a very high pressure or as a liquid, hydrogen has a lower volumetric energy density compared to gasoline. See Table 1-2 below for a comparison between the energy content of various fuels. Significant research has been done on hydrogen storage and some examples include hydrogen storage by reversible metal hydrides, carbon nanofibers, and metal-organic frameworks [7]. However, these alternatives at present are either too costly or cannot achieve desirable hydrogen content and adsorptionldesorption reversibility under near ambient conditions. 5 Table 1-2: Volumetric energy density of various fuels (18]. Fuel Volumetric Energy Density (kWh L’) Hydrogen (310’ MPa) 0.8 Liquid Hydrogen (20 K) 2.4 Formic Acid 2.1 Methanol 4.4 Gasoline 8.7 An alternative to using hydrogen is to use organic liquid fuels with higher volumetric energy density, such as methanol and formic acid, directly without reforming. Compared to using these fuels as hydrogen carrier by reforming, direct oxidation of these organic liquids simplifies greatly the system design and cost due to the lack of reformer. See Appendix A for general background information on methanol and formic acid. The DMFC was first pioneered by Shell Research in England and Exxon-Alsthom in France during the 1 960s and 1 970s [19] while the direct formic acid fuel cell (DFAFC) received less attention due to its lower volumetric energy density. More recently in 2003, Tekion, Inc. gained exclusive right and started to develop miniature DFAFC’s for portable applications. A general schematic of the direct liquid fuel cell (DLFC) is shown below in Figure 1-3. A conventional DLFC utilizes a multi-layer design, identical to that of the hydrogen PEMFC. The fuel cell consists of an electrolyte (proton exchange membrane), two catalyst layers (anode and cathode), two gas diffusion layers aiding the transport of reactants and products to and from the catalyst layers, respectively, and two flow fields uniformly delivering reactants to and removing products from the fuel cell. Both the DMFC and DFAFC are more advantageous than the hydrogen PEMFC in terms of fuel handling and volumetric energy density. However, they have their own drawbacks including the slower anodic kinetics leading to higher surface overpotential, significantly higher fuel crossover causing a mixed cathode potential, and CO2 accumulation in the anode compartment leading to higher mass transport overpotential. 6 Figure 1-3: Schematic of DLFC. The anode, cathode, and overall reactions for the DMFC and DFAFC are shown below in Equation 1-1 to 1-6: DMFC Anode: Cathode: CH3OH + H, 0-> CO, + 6H + 6e +6H +6e 3H,O 2 E°9gK = 0.04 V E°ggK =l.23V [Eq. 1-11 [Eq. 1-2] H20(I), CO2(g) O2(g), H2O(g), N2(g) —0 CxHyOz(aq), H20(I) Methanol: x = 1; y = 4; z = 1. FormicAcid: x=1;y=2;z=2. O2(g) N2(g) Gas Diffusion Layer Bipolar PlateIFlow Field Proton Exchange Membrane Catalyst Layer 7 Overall: CH3O4,—>2H20+C0 E2098K =1.19V [Eq. 1-3] DFAFC Anode: HCOOH — CO2 + 2H + 2e E,°98K = —0.22 V [Eq. 1-4] Cathode: 107 +2H +2e H20 E98K L23V [Eq. 1-5] Overall: HCOOH+10,-*H20+C07 E98K = 1.45 V [Eq. 1-6] For both the DMFC and DFAFC, CO is formed as a reaction intermediate [20, 21]. Unfortunately, CO can adsorb very strongly on platinum, the most commonly used catalyst in fuel cell applications. The presence of CO can lower the electrochemically active surface area, resulting in slower anodic kinetics. Details of the reaction mechanism of the electro-oxidation of methanol and formic acid are presented in Sections 1.3.1 and 1.3.2. The second problem challenging the DMFC and DFAFC is fuel crossover, a phenomenon where the fuel (methanol or formic acid) is transported from the anode unreacted to the cathode by diffusion and electro-osmotic drag. Electro-osmotic drag arises from the polar nature of methanol and formic acid; hence, as the protons are transported across from the anode to the cathode, methanol and formic acid are dragged along. The result of fuel crossover is lower fuel utilization as well as the induced presence of a mixed potential on the cathode, generating lower overall cell voltage and power density. According to Equation 1-3 and 1-6, the equilibrium cell voltages of the DMFC and DFAFC are 1.19 and 1.45 V at standard conditions. However, due to fuel crossover, the open-circuit voltage (OCV), the cell voltage at zero current, measured in experiments 8 range between 0.5-0.8 V [22-24], depending on the concentration of the fuel, type of membrane and temperature. In terms of fuel crossover, formic acid poses a smaller problem compared to methanol. For example at 1 M, the formic acid crossover flux through Nafion 11 7® at 293 K is 2x 1 mol m2 s1, which is almost two orders of magnitude lower than methanol crossover flux (3 to 6 x102 mol m2 s’) [25-271. The current research attention for mitigating fuel crossover include the use of fuel additives [28], membrane modification [29], and the development of new membrane materials that are less prone to methanol or formic acid permeation [30-35]. Regarding membrane modification, Tang et al. have anchored palladium onto the sulfonic sites in Nafion® membranes and successfully reduced the crossover flux of methanol by up to eight orders of magnitude [29]. Some alternate membrane materials that have been developed and synthesized include the sulfonated poly-ether sulfones and the sulfonated poly-ether ether ketones (SPEEK), polyphosphazene, sulfonated polyimide, and polybezimidazole [30- 34]. It was previously reported that replacing Nafion® with SPEEK yielded improved performance in a DMFC (from 500 to 750 W m2) and that a multilayer SPEEK design with a lower sulfonation level layer (trade off with ionic conductivity) could further reduce methanol permeation and crossover [35]. The third challenge for the DMFC and DFAFC is carbon dioxide accumulation that arises from the reaction product on the anode side as presented in Equation 1-3 and 1-6. The presence of CO2 can hinder the penetration of the liquid fuel to the catalyst layer. The effect of CO2 accumulation is particularly important for large cell stacks as well as for cell scale up. Argyropoulos et al. have captured visual images of the carbon dioxide gas evolution and flow behavior with the use of an acrylic cell [36, 37]. It was found that conventional Toray carbon paper, one of the most commonly used gas diffusion layer materials, due to its surface roughness making it hard for the gas bubbles to dislodge, is inefficient in removing the produced CO2. Scott et al. have proposed and demonstrated the use of stainless steel mesh as flow bed material, while Allen et al. have suggested the use of titanium mesh as a new catalyst support material to obtain the so called three-dimensional (3-D) anode design to better disengage the CO2 gas bubbles [38, 39]. 9 1.2 Theoretical Background This section provides the fundamental and theoretical background knowledge required to better understand the methodologies employed and scope of the present study. Electrochemical thermodynamics, kinetics, voltage balance, conventional DLFC anode design, and half-cell electrochemical methods are discussed and presented. 1.2.1 Electrochemical Thermodynamics In any chemical system operating at constant temperature and pressure, the change in Gibbs free energy (AG) is defined to be the difference between the change in enthalpy (All) and the product of the change in entropy (AS) and temperature (Equation 1-7). AGAH-TAS [Eq. 1-7] The enthalpy change is given by the sum of changes in internal energy of the system (AU) and the mechanical work due to volume change (Equation 1-8): All = AU + PAV [Eq. 1-8] Based on the 1st law of thermodynamics, AU is the difference between the heat transferred into the system and the total work done by the system (w < 0): AU=q+w [Eq.1-9] Since for a reversible system, using the 2nd law of thermodynamics, the heat transferred is given by TAS and for an electrochemical system the total work done by the system is the sum of the mechanical and electrical contributions (we < 0 when ‘done’ by the system): AU=TASPAV+We [Eq. 1-10] Substituting Equation 1-8 and 1-10 into Equation 1-7 yields Equation 1-1 1. AG=We [Eq. 1-11] Equation 1-11 shows that the maximum electrical work that can be extracted from a reversible closed system under constant temperature and pressure is equal to the change in Gibbs free energy. Since the electrical work is the product of the potential and total charge, the Gibbs free energy can be related to the electric potential as shown by 10 Equation 1-12, taking into account the sign convention that AG < 0 and E > 0 for a thermodynamically spontaneous reversible electrochemical system. AG=-nFE, or [Eq. 1-12] AG° = —nFE° (at298Kandactivity= 1) The effect of temperature on the equilibrium cell potential can be derived from the Maxwell Equation. AG [Eq. 1-13] Substituting Equation 1-12 into Equation 1-13 yields Equation 1-14, a relationship between the changes in cell potential due to temperature effects: -l = [Eq. 1-14] iÔTJ nF Integrating Equation 1-14 and assuming AS = constant over the integrated temperature range, the effect of temperature on cell potential is: Ee2 Eei =--(T, —T1) [Eq. 1-15] nF The Gibbs free energy at activity 1 is related to the standard Gibbs free energy according to Equation 1-16: AG=AGr0+RTln(fla1) [Eq. 1-16] Substituting Equation 1-12 into Equation 1-16 yields Equation 1-17, generally referred as the Nernst Equation, where at” is the activity of species i with a stoichiometric coefficient s: ET° _ln(Ha;”) [Eq. 1-17] nF C a f—,wherec0= 1 M Co or [Eq. 1-18] P a=f—,whereP0=1 atm 0 In practice, the activity is approximated with the molar concentration (liquid solution) and partial pressure (gas) of the reactants and products. For example, the half-cell 11 equilibrium potentials of methanol and formic acid oxidation can be calculated as shown in Equation 1-19 and 1-20. RT c6P ‘ Ee(H3O = ET CH3O — — ln H CO2 [Eq. 1-19] 6F CCHQH ,) EHCoOH = EHCOOH — ln CH* 2 CO2 [Eq. 1-20] 2F CHCH J The cell potential of an electrochemical system is the difference between the cathode and anode half-cell potentials (Equation 1-21). Therefore, the full cell potentials of the DMFC and DFAFC are given in Equation 1-22 and 1-23. E = — Ea [Eq. 1-2 1] RT P EeDMF(• = E.°DM —--—-ln c2 [Eq. 1-22]6F CCHOHPO) RT P EeDFAPC ET°DFAFC ————ln CO2 NO.5 [Eq. 1-23]2F CCHOH O2) See Table 1-3 for the equilibrium potentials of the half and full cell potentials of methanol oxidation, formic acid oxidation, DMFC, and DFAFC at different conditions. It can be seen that the higher the fuel concentration or temperature, the lower the anode equilibrium potential and thus, the higher the full cell equilibrium potential. 12 Table 1-3: Equilibrium potentials of methanol oxidation, formic acid oxidation, DMFC, and DFAFC at various conditions. Anode Equilibrium Potential for Methanol Oxidation: Ee,CH3OH (V VS. SHE) pH = 0, P2 = 1 atm CCH3OH (M) T(K) 1 3 10 298 0.040 0.035 0.030 333 0.039 0.034 0.028 363 0.038 0.032 0.026 Anode Equilibrium Potential for Formic Acid Oxidation: Ee,HCOOH (V VS. SHE) pH = 0, Pü2 = 1 atm CHCOOH (M) T(K) 1 3 10 298 -0.220 -0.225 -0.230 333 -0.235 -0.241 -0.246 363 -0.249 -0.254 -0.261 Equilibrium Cell Potential of DMFC: Ee,DMFC pH = 0, Pco = 1 atm, ccH3oH (M) T=333K P02 (atm) 1 3 10 1 1.185 1.190 1.196 2 1.190 1.195 1.201 5 1.197 1.202 1.208 Equilibrium Cell Potential of DFAFC: EeDFAFC pH = 0, PCO2 1 atm, CHCOOH (M) T=333 K P02 (atm) 1 3 10 1 1.459 1.475 1.493 2 1.464 1.480 1.497 5 1.471 1.487 1.504 13 1.2.2 Half-Cell Potential and Reference Electrode Measurement of the electrode potential is relative to a reference electrode. By the International Union of Pure and Applied Chemistry (IUPAC) standard, the reference is the potential for the standard hydrogen electrode (SHE), which is defined to be 0 V. The standard hydrogen electrode is based on the hydrogen redox reaction at standard conditions (Equation 1-24). 2H + 2e H2 [Eq. 1-24] The operating principle behind a reference electrode is that the electrode reactions are very fast in both directions. Therefore, the reference electrode has a negligible overpotential when a very small current is flowing through the measuring circuit. Hence, all the measured overpotential comes from the working electrode (in half-cell electrochemistry experiments, the electrode of interest is called the working electrode and the auxiliary electrode required to complete the circuit is called the counter electrode). See Figure 1-14 for a schematic of a typical 3-electrode setup. It should be noted that the use of SHE reference is generally uncommon in practical experimental work, especially in half cell tests with liquid electrolyte, due to the inconvenience of having a constant hydrogen gas flow. Some common reference electrodes are the calomel (Hg/Hg2C1), silver-silver chloride (Ag/AgCl), mercury-mercury oxide (Hg/HgO), and the mercury mercurous sulfate (Hg/Hg2SO4)(MSE) reference electrodes. Table 1-4 below shows the potential of these reference electrodes in conjunction with their typical electrolyte partner. 14 Figure 1-4: Schematic of a three-electrode setup. Table 1-4: Reference electrode potentials [40]. Reference Electrode Electrochemical Reaction Ee,Ref at 298 K (E vs. SHE) Hg/Hg2C1 (saturated KC1) Hg2C1 + 2e *—* 2Hg + 2C1 0.24 1 Ag/AgC1 (saturated KC1) AgC1 + e Ag + C1 0.197 Hg/HgO (20% KOH) HgO + H20 + 2e -* Hg + 20W 0.098 Hg/Hg2SO4(saturatedK2S04) HgSO4+ 2& —* Hg + so42- 0.640 1.2.3 Kinetics of Electrode Processes Starting with the equilibrium condition, at any electrode (either cathode or anode), there exists both oxidation (loss of electron) and reduction (gain of electron) proceeding at equal rate. Under non-equilibrium conditions, oxidation is the dominant reaction at the 15 anode whereas reduction dominates at the cathode. Any given half-cell reaction occurring at an electrode can be represented by the following reaction (Equation 1-25), where Ox is the species being reduced and Red is the reduced species. Ox + ne -> Red [Eq. 1-25] In the case where the forward reaction rate is greater than the reverse reaction rate, the electrode becomes a cathode in the overall electrochemical system. Conversely, when the backward reaction rate is greater, the electrode becomes an anode. At equilibrium, the forward and backward reactions have the same reaction rate. The reaction rate of electrochemical reactions is heavily influenced by the electrode potential, electrocatalytic properties of the electrode material, concentration, and temperature. Figure 1-5 below illustrates the energy level of the oxidized and reduced species in an electrochemical reaction at different potential, where E> E°. Energy Red Ox (e.g. CH3O , HCOOH) (e.g. GO2) nFE, Activation Energy Reaction Coordinate Figure 1-5: Energy of an electrochemical reaction at increasing potential, E > E°. 16 As the electrode potential is increased (becoming more positive or anodic), the activation energy of the reaction is reduced, favoring the oxidation reaction or the formation of Ox. When the electrode potential is increased from E° to E, the initial energy of the oxidized species is increased by nF(E-E°) as given by Equation 1-12. However, not all of this energy is utilized in lowering the activation barrier as shown in Figure 1-15. The fraction of energy that is used to reduce the activation barrier is often denoted as cLaF(E-E°), where aa is the anodic transfer coefficient. Thus, the activation barrier for the anodic process at E is given by: AG: =AG’° _aaF(E_E0) [Eq. 1-26] The cathodic and anodic transfer coefficients are defined in Equation 1-27 and 1-28, where 13 is called the symmetry factor and has a value between 0 and 1. aa = (i — fi)n [Eq. 1-27] = fin [Eq. 1-28] Similarly, the activation barrier for the cathodic process at E is given as follows in Equation 1-29: AG’° +acF(E_E0) [Eq. 1-29] For any chemical reaction, the reaction rate constant, k, is given by the Arrhenius expression, where AG is the free energy of activation: k k exp [Eq. 1-30] From Faraday’s law, the reaction rate, r, is related to the current density as shown in Equation 1-31: [Eq. 1-3 1] nF For a first-order reaction, the reaction rate is given by Equation 1-32: r = kcexpl’A =__ [Eq. 1-32] RT ) nF Equating Equation 1-31 and 1-32 and substituting Equation 1-26 or Equation 1-29 yields an expression relating the current density with the electrode potential as shown below, where and CRed,S denote the surface concentration of Ox and Red, respectively: 17 = nFkc0.,ex[’° —aF(E—E°) [Eq. 1-331 (z\G’° +aaF(E_E0) la = flFka CRed exPt\ RT [Eq. 1-34] The activation barrier at the equilibrium potential can be absorbed into the rate constants k’ and ka’, yielding k4° and ka°, referred to as the standard heterogeneous rate constants. Thus, Equation 1-33 and 1-34 become as follows: nFlç c0 exp RT [Eq. 1-35] (a0F(E_E°)’ la = flFka exp RT [Eq. 1-361 The net rate of the half-cell electrode reaction is given by ra — r. [Eq. 1-37] nFnFnF Rearranging Equation 1-37 yields Equation 1-38: = flFk°C exp[FTE exp[FE [Eq. 1-38] At equilibrium (E = Ee) under non-standard conditions, the net current, i, is equal to 0. However, the oxidation and reduction reactions are ongoing, but occurring at the same rate. This ongoing current at equilibrium is called the exchange current density, i0, an important parameter for the assessment of the intrinsic electrocatalytic activity. Furthermore, with a net rate of consumptionlformation of 0, the surface concentration of Ox and Red are equal to their respective bulk concentration. = flFk0Cb exp[a_ E° ) = nFkC°coXb ex[ aF(E, — E° ) [Eq. 1-39] Substituting the Nernst Equation into Equation 1-39 yields the following expression for the exchange current density, indicating its dependence on temperature (from the rate constants) and the concentrations of Ox and Red: a —a a = flFkcOfl)kaO[fl)COx bjCRed,b [Eq. 1-40] Dividing Equation 1-38 by Equation 1-39 yields the following expression: 18 flFk0CexP[ n°co ex[ aF(E— E0) = flFk0Cexp aFe_E°) — flFkC°coXb exp acF(E _E0)] [Eq. 1-41] Assuming pure kinetics control by neglecting concentration effects (i.e. c = Cb for i 0) and by rearranging yields an expression relating the net current density to the electrode potential [41, 42]. i = i0 [exp aa F(E— Ee — exp — aF(E — Ee [Eq. 1-42] The difference between the electrode and equilibrium potential (E - Ee) is the overpotential, ri. In the case of kinetic control, (E - Ee) = 1s, which is the surface or charge transfer activation overpotential. i = io[exp [Eq. 1-43] The Butler-Volmer-Erdey-Grüz (BEV) equation (Equation 1-43) can be simplified into two different forms. At high overpotential (typically > 0.050 V at 298 K), the simplification yields the Tafel Equation, where b is the Tafel slope, a measure of how fast the potential changes with current density [43]. See Table 1-5 for examples of b and i0 values from literature and their strong dependence on the catalyst and experimental conditions. = ln = b logji + a [Eq. 1-44] aF 0 where b = 2.303RT [Eq. 1-45] aF a = logi0 [Eq. 1-46] 19 =b ,‘ Intercept = log i Log 1i0 Figure 1-6: Graphical representation of Tafel plot. Under small overpotential conditions (<0.010 V at 298 K), the BEV equation leads to the linear approximation form as shown below by using a two-term Taylor Series expansion: = (a0± ) Fii = i [Eq. 1-47] The low field assumption is not commonly used as the surface overpotential for reactions of interest are usually significantly higher than the limitation for linear approximation. ——J 20 Table 1-5: Published data of exchange current density and Tafel slope of methanol and formic acid electro-oxidation. Reaction Electrode Electrolyte Temperature i0 b (K) (A m2) (V decad&’) CH3O Pt/Polypyrrole 1 M CH3O in 298 -- 0.164 Oxidation [44] 0.1 M HC1O4 Pt [45] 0.5 M CH3O 298 -- 0.104 in 0.1 MH2SO4 Pt [46] 0.005 to 0.1 M 294 3 x 10 0.077-0.440 CH3OHin1M to HC1O4 3 x i0 Pt-Ru [46] 0.005 to 0.1 M 294 1 x i0 0.075-0.105 CH3OHin1M to HC1O4 4 x i0 Pt-Ru [47] 0.5 M CH3O 293 -- 0.120-0.200 in 0.5 M H2S04 HCOOH Pt/Polypyrrole 1 M HCOOH 298 -- 0.237 Oxidation [44] in 0.1 M HC1O4 Pt [48] 0.5 M HCOOH 333 -- 0.132 ino.5MH2S04 Pt [49] 0.5 M HCOOH 295 -- 0.120 in0.5MH2SO4 Pt [50] 0.1 MHCOOH 298 -- 0.140-0.176 in 0.1 MH2SO4 Pt-Pd [50] 0.1 M HCOOH 298 -- 0.136 in 0.1 MH2SO4 Pt-Bi [51] 0.05 M 293 -- 0.120 HCOOH in 0.1 M HC1O4 21 In any surface electrochemical reaction, the adsorption and desorption of reactants and products take place at the electrode. At a given temperature and electrode potential, the adsorption and desorption follow that of an adsorption isotherm. The simplest and most common adsorption isotherm for electrochemical reactions is the Langmuir isotherm, which assumes a homogeneous surface, monolayer coverage at saturation, and no adsorbate interactions. The Langmuir equilibrium coverage of an adsorbing species, 0, is given in Equation 1-48, where AGacis. kads, and kdes are the Gibbs free energy of adsorption, kinetic constant of adsorption and kinetic constant of desorption, respectively. It is important to note that AGads is dependent on the electrode potential and hence, the equilibrium coverage of the adsorbing species is also a function of the electrode potential. =1--c exp’_AG [Eq. 1-48] 1—0 kdL a RT ) The Langmuir adsorption isotherm in conjunction with the Langmuir-Hinshelwood reaction mechanism is often used to model electrochemical reactions, such as the methanol electro-oxidation reaction on Pt-Ru [46]. Another commonly used adsorption isotherm for electrochemical reaction modelling is the Temkin-Frumkin adsorption isotherm [52, 53]. The Temkin-Frumkin adsorption isotherm takes into account adsorbate interaction as a function of coverage via an interaction factor, g where g> 0 for adsorbate repulsion and g < 0 for adsorbate attraction (see Equation 1-49 and 1-50). Reactions that are modeled with the Temkin-Frumkin adsorption isotherm include the methanol electro oxidation reaction and the hydrogen electro-oxidation in the presence of carbon monoxide on Pt [21, 46, 53]. AGadc AGadsO + gO [Eq. 1-49] 0 = exp1’O0o)’ [Eq. 1-50] 1—0 kdes. a RT 1.2.4 Mass Transfer, Ionic Conductivity, and Voltage Balance Equation 1-51 is the general voltage balance equation for an electrochemical system, comprising different terms associated with surface overpotentials of the anode 22 and cathode, concentration overpotentials of the anode and cathode, as well as ohmic loss. Eceii = — 77s anode + — + 7ld cathode — Acohm [Eq. 1—51] As introduced in Section 1.2.3, the anodic and cathodic surface overpotentials are the changes in half-cell electrode potential required to achieve a certain current density for the anode and cathode, respectively. The concentration overpotentials arise from the concentration gradient between the electrode surface and the bulk cell condition. For instance, excess water on the cathode side, which hinders the diffusion of 02 to the active sites, and CO2 accumulation on the anode side can lead to undesirable concentration overpotentials in a DLFC. From Fick’s Law, the molar flux of species j transporting by diffusion is shown in Equation 1-52, where is the diffusion layer thickness, Kmj is the mass transport coefficient of species j, and and Cbj are the concentrations of species j at the electrode surface and bulk condition, respectively. dc (c Ch) N =_D_j=_D ‘ =K(c, cbJ) [Eq. 1-52] At a surface concentration of 0, the current density becomes limited completely by mass transfer, and it is called the mass transfer limiting current density, L. i =K(c cbf) [Eq. 1-53] ZL = — nF K,flJcbJ [Eq. 1-54] Si The concentration overpotential is defined to be the difference in equilibrium potential at the surface and bulk condition (i.e. from Nernst Equation). RT —s1 lnI —u- [Eq. 1-55] nF CbJJ Rearranging Equation 1-54 (solving for cb) and substituting into Equation 1-53, the following relationship is obtained: —----=1—--- [Eq. 1-56] Cbi 1L 23 Substituting Equation 1-56 into Equation 1-55 yields a relationship between concentration overpotential and mass transport limiting current density: RT( i 1d = —ml 1— [Eq. 1-57] aF lU From Equation 1-41 and by assuming high kinetic overpotential (i.e. either anodic or cathodic dominance), Equation 1-58 and 1-59 are obtained, where k is the kinetic current density. CRedc . (aaF(E—Ee) CROdS = l exp = 1k [Eq. 1-58] CRedb . RT ) c0 (—acF(E—Ee) c0,, = —zn exp I = [Eq. 1-59] CQXb RT ) Cox, Substituting Equation 1-56 into Equation 1-58 and 1-59 and by rearranging leads to the reciprocal relationship (Equation 1-60). [Eq. 1-60] 1k 1L The last term in the general voltage balance equation is the ohmic loss associated with both electrical and ionic conductance. In general, both components are given by Ohm’s Law, where K and x are the conductivity and length of the conducting medium, respectively. ohm = [Eq. 1-6 1] The conductivity depends on the porosity, tortuosity and void fraction of the conducting medium. In the case of a liquid electrolyte with gas bubbles (such as in the case of a DMFC anode), the conductivity dependence on gas volume fraction, 6g can be determined with the Maxwell Equation (Equation 1-62) or by the Meredith-Tobias Equation (Equation 1-63) [54]. Maxwell Equation is most accurate at 8g < 0.1, but is generally used at volume fraction up to 0.6. On the other hand, the Meredith-Tobias Equation is most applicable at high gas volume fraction, 8g> 0.6. K. 1-6 [Eq. 1-62] Ic 6 1+— 2 24 0.4 0.3 0.2 0.1 Kg — (1_SgX2_6g) K (4+6gX4_6g) [Eq. 1-63] As can be seen from Figure 1-7, the effect of gas bubbles has a significant impact on the ionic conductivity of the electrolyte. In the case of DMFC and DFAFC, in order to reduce the effect of ohmic loss, it is essential to be able to efficiently remove product C02, which could potentially hold up at the anode, leading to lower ionic conductivity. 0.9 0.8 0.7 0.6 0.5 0 0 0.2 0.4 0.6 0.8 E9 Figure 1-7: Effect of gas volume fraction on the effective ionic conductivity. 25 1.2.5 Half-Cell Electrochemical Experimental Methods For fundamental electrochemical research, typically a three-electrode setup described in Section 1.2.2, is employed. Common electrochemical methods used in relation to electrocatalysis include cyclic voltammetry (CV) on static and rotating electrodes, chronopotentiometry (CP), and chronoamperometry (CA). Additionally, methods based on CV are also used to obtain specific characteristics of the working electrode of interest, such as determining the effective catalyst surface area by hydrogen adsorption, carbon monoxide stripping, and underpotential deposition of copper (Cu UPD). 1.2.5.1 Cyclic Voltammetry CV is arguably the most widely used method in electrochemistry. It involves measuring the current of the working electrode when its potential is cycled linearly at a desired scan rate, v, from an initial potential (E1) to a final potential (Ef). See Figure 1-8. The resulting CV plot is called a polarization curve. By IUPAC convention, the anodic current is treated as positive while cathodic current is considered to be negative. In the CV plot (i vs. E), shown by Figure 1-8, the forward scan is toward the anodic direction, and the reverse scan is toward the cathodic direction. 26 Diffusion Note: Exponential begins to Rate E — E E — Erise with d t controlled I CC f I — — potential omina e by IL or E0 or E dt VS Esw 4, I Ef (Non-Faradaic Current) E=Ef E Esw Signal Response Signal Input Figure 1-8: Concept of cyclic voltammetry. In Figure 1-8, in a pure kinetic control scenario, the current will rise exponentially with potential (see Section 1.2.3). Under stagnant electrolyte conditions, at the peak potential E, diffusion limitation sets in establishing either mixed or pure diffusion control. However, somewhat similar peak behavior is also observed when the reaction is under catalyst poisoning condition, i.e. the current reaches a peak value at a certain potential. For a reversible reaction, the peak current is a linear function of the square root of the scan rate, but the peak potential, E, is independent of scan rate. Under diffusion control, the relationship between the peak current density and scan rate is given by the Randles-Sevcik Equation as shown below [40]. The equilibrium potential is also related to the peak potentials of the oxidation and reduction peak potentials if the peaks are close together (Equation 1-65). EpRed 27 3/2 3/2 ___________ 1/2 1/2 = 0.446 RL2T D CbVS [Eq. 1-64] E +E 0059E = p,Red j,Ox Ep Red = at295K [Eq. 1-65]2 n For an irreversible reaction, the peak potential is a function of scan rate as shown in Equation 1-66 and the relationship between the peak current and scan rate is given by Equation 1-67 [40]. E =E _[1.04_logJ_2logk0 +logvc] [Eq. 1-66] (2.3RT’” 1/2 1/2 = 3 •10 bF j D CbVS [Eq. 1-67] Valuable information, such as the number of electrons reacted, kinetic parameters, and the diffusion coefficient can be obtained. For instance, by plotting E vs. log v based on Equation 1-66, the Tafel slope b can be obtained from the slope of the plot and standard heterogeneous rate constant k° can then be calculated from the intercept. Similarly, by plotting i vs. log vs’12 the number of electrons reacted or the diffusion coefficient can be calculated from the slope if one or the other is known. 1.2.5.2 Chronopotentiometry The chrono-technique CP involves stepping the current density from an initial value to a desired final current density and measuring the resultant potential as a function of time. A CP experiment has essentially the same setup as CV. CP is a very meaningful method as it gives very precious information about how the working electrode of interest behaves over time. It serves the purpose of being a screening method for potential fuel cell catalysts because it operates under the same constant current mode as an operating fuel cell. The stability of the potential catalyst can then be evaluated based on the CP data without incorporating other effects, whether detrimental or beneficial, during fuel cell operation. The concept of CP is shown in Figure 1-9. 28 ITime Period Used for Stabilization at E0 Signal Response Figure 1-9: Concept of chronopotentiometry. Signal Input RT 2k° RT 1/2E=E0___lnI i+__i( _thI2) / \1/2aF jrD0) ,) aF 2 E I / — — — I 2 I r2 TI t t The potential as a function of time is given in Equation 1-68 and 1-69 for a reversible and an irreversible reaction, respectively. ‘r is the time required for the concentration of the reactant species to reach zero [40]. At time t, a sharp change in potential is usually observed to drive a secondary reaction due to the current drawn. RT (‘D RT (‘ 1/2 1’2 “E = — —mi Ox + —lni ? [Eq. 1-6812nF DRed,) nF t2 ,) [Eq. 1-69] Another vital equation that is relevant to CP is the Sand’s Equation as shown in Equation 1-70 [40]. 1/2 nFAc4/iT = 2 For redox processes that are not complicated by poisoning or chemical reactions, the parameter i’r2 is constant; therefore, the Sand’s Equation is often used to determine the [Eq. 1-70] 29 diffusion coefficient of the reactant species for such redox processes. Consequently, the CP method is a very important technique in fundamental electrochemistry study. 1.2.5.3 Chronoamperometry The other chrono-technique is CA, which involves stepping the electrode potential from an initial value to a desired final potential and measuring the resultant current density as a function of time. Figure 1-10 shows the signal input and response of a typical CA experiment. Diffusion Control Mixed Control Kinetic Control t t Signal Response Signal Input Figure 1-10: Concept of chronoamperometry. Identical to the CV and CP experimental setup, CA can be used to confirm whether the electrochemical reaction of interest is diffusion controlled. From Fick’ nd Law, the change in concentration with respect to diffusion as a function of time is given by the following (see Figure 1-11 for concentration profiles): (ac” CbCS I — I = [Eq. 1-71] axJ, 30 = = nFD Cb LôxJ, t= 0 t3 Cb t3 > t2 > ti > 0 At diffusion control condition, c = 0; hence, the mass transport limiting current density can be written in the following form, which is known as the Cottrell Equation [40, 55]: [Eq. 1-72] Cs Figure 1-11: Concentration profiles of a reactant species at various times into a CA experiment. Therefore, if a reaction is diffusion control, plotting it1”2 vs. t yields a constant line with the value nFD”2cbC,which can be used to calculate the number of electrons reacted and the diffusion coefficient (see Figure 1-12). Deviations from the straight line indicate that the process is either coupled with other chemical processes or not under diffusion control. x 31 nFD2cbT(112 it1/2 t Figure 1-12: Typical Cottrell plot with diffusion control. For a reaction that is under mixed control condition, the current response is given by Equation 1-73 and the kinetic parameter ka° or k0° are often calculated by fitting the experimental transient to the equation [55], where in mathematics, the error function, erfc, is defined in Equation 1-74 [56]. i = flFk:C Red for reduction or [Eq. 1-73] i = nFk: cbOX ex[ terfc[ for oxidation erfc(y)= _= rexp(_z2Iz [Eq. 1-74] In addition, for a reaction that is not under diffusion control, CA data can be used to obtain kinetic parameters, such as the Tafel slope and the exchange current density by plotting the pseudo steady-state current density as a function of potential. 1.2.5.4 Catalyst Surface Area Estimation by Electrochemical Techniques a) Hydrogen Adsorption The effective electrocatalytic surface area of catalysts is often measured by hydrogen adsorption, carbon monoxide stripping, and Cu UPD. Hydrogen adsorption is 32 one of the most common and traditional methods used to determine the active surface area of platinum electrocatalysts [57-59]. Hydrogen ions in acid solutions (usually sulfuric acid or perchioric acid,), will adsorb onto platinum surface and form a full monolayer at potentials near but slightly anodic to the equilibrium potential of hydrogen evolution (i.e. 0 V vs. SHE at standard conditions). This is referred to as hydrogen underpotential deposition. In order to prevent bulk hydrogen evolution that could potentially influence the surface area estimation, the hydrogen adsorption experiment is typically performed at 0.05 V at 298 K. Pt+H +e —* Pt—H [Eq. 1-75] It is known that each platinum atom has the capacity to adsorb close to one hydrogen atom, hence the number of platinum atoms in a given surface area can be estimated, from the charge associated with stripping (anodically removing) the adsorbed H after subtracting the background current due to double layer charging. It is generally accepted that the charge associated to hydrogen adsorption!desorption for polycrystalline platinum is 2.1 C m2 [57-59]. Figure 1-13 depicts the charge associated with hydrogen adsorption in a typical platinum CV scan in acidic media. Generally, two hydrogen stripping and adsorption peaks are observed for platinum CV and they are associated with different crystallographic planes. The hydrogen desorption peak located at higher potential (more strongly bound) is linked to Pt(100) while the peak located at lower potential is associated with Pt(1 10) and Pt(1 11). See also Figure 1-14, which shows a typical polycrystalline platinum CV in acidic media (both hydrogen adsorption and CO stripping) [60]. 33 1 Hydrogen Stripping Charge Figure 1-13: Hydrogen adsorption charge. Hydrogen adsorption is widely accepted as the baseline method in determining the active surface area of platinum electrocatalysts; however, it is not suitable for ruthenium- containing catalysts because the hydrogen stripping can occur in the potential region where ruthenium oxidation occurs and furthermore hydrogen can absorb into ruthenium [61]. Ru + nH + ne —f Ru — H [Eq. 1-76] b) CO Stripping A second widely used CV-based surface area determination method is carbon monoxide stripping. Due to the affinity of CO to platinum, a monolayer of CO can be adsorbed onto platinum very easily. Pt+CO—Pt-CO [Eq. 1-77] Pt+H2O—>Pt—OH+H+e [Eq. 1-78] Double Layer Region Double Layer Charging Current E Hydrogen Underpotential Adsorption at Different Crystal Planes Bulk Hydrogen Evolution 34 Pt-OH+Pt-CO--*2Pt+CO2+ +e [Eq. 1-79] 1 o,i Figure 1-14: Typical platinum CV in 0.5 M H2S04 recorded at 0.010 V s1 showing the hydrogen desorption and CO oxidation peaks [601. Reproduced by permission of The Electrochemical Society. In general, the oxidation of the adsorbed carbon monoxide is assumed to proceed along the boundaries of the carbon monoxide islands, where nucleation sites allow for electrochemical adsorption of oxygen-containing species, such as water, to oxidize the carbon monoxide to carbon dioxide as can be seen from the scanning tunneling microscope (STM) images published by Erti et al. (Figure 1-15) [62, 63]. The overall oxidation reaction is a 2-electron transfer process and the charge associated with CO stripping is thus two times that of the hydrogen adsorption, or 4.2 C m2. See Figure 1-14 for a typical platinum CO stripping CV in acid media. Similar to hydrogen adsorption, CO stripping is also not suitable for ruthenium-containing catalysts. It is known that CO can linearly, bridge-bond, and bond via three-fold hollow sites on platinum and it is ‘‘1 Hydrogen AdsorptionlStripping Region 0.1 0.3 0.5 0.7 0.9 E vs. SHE 35 generally accepted that linear bonding is the predominant mode at high CO coverage [63], resulting in a 1:1 Pt:CO ratio. However, CO tends to linearly and bridge-bond to ruthenium, invalidating the method [611. Figure 1-15: STM images recorded during the reaction of adsorbed oxygen species with co-adsorbed CO [63]. From J. Wintterlin, S. Volkening, T.V.W. Janssens, T. Zambelli, G. Erti, Science 278 (1997) 1931. Reprinted with permission from AAAS. c) Cu UPD A promising method for determining the effective surface area of Pt-Ru catalyst is Cu UPD because it is capable of identifying the charge contribution from each of the components. The Cu UPD method is similar to hydrogen adsorption and CO stripping, but employs copper as the adsorbing species [64-68]. Green and Kucernak have shown that only a monolayer of copper is deposited on both platinum and ruthenium [64]. In a separate study, Nagel et al. obtained comparable surface area estimation result with Cu UPD and CO stripping for Pt, Pt-Ru and Pt-Ru-Se [65]. Though, some researchers have indicated that there might also be issues associated with Cu UPD, such as incomplete surface coverage, the formation of more than a monolayer, and partial reduction of copper [66-67]. sos 140s 290s 36 For a monolayer of copper underpotentially deposited, the charge associated with the stripping of the monolayer is 4.2 C m -, since it is a 2-electron process and the Cu:Pt and Cu:Ru ratio is assumed to be 1:1 due to their similar atomic radii. CUUPD —> Cu + 2e [Eq. 1-801 Cu UPD is typically performed by first obtaining a blank CV in an acidic media followed by the formation of the copper monolayer by polarizing the electrode in the presence of copper ions to a potential more anodic than the Cu2/C equilibrium potential of bulk copper deposition and subsequent stripping of the copper monolayer. Typical experiments are performed with 1 0 M CuSO4 and a UPD potential near the equilibrium potential at the experimental condition (0.26 V vs. SHE) [64-681. On the anodic scan, there are two main stripping peaks for Cu UPD for Pt-Ru electrocatalysts (see Figure 1- 16). The first peak occurs at approximately 0.45V whereas the second peak, which is typically broader, occurs at approximately 0.70 V vs. SHE. The first peak corresponds mainly to the stripping of copper deposited on ruthenium and the second peak corresponds to the stripping of copper deposited on platinum. However, it must be noted that there are charge contribution from both platinum and ruthenium for each peak due to peak overlap. A schematic depicting blank and Cu UPD scans of Pt-Ru catalysts is shown in Figure 1-17. To determine the respective area of Pt-Ru, one can assume similar charge contribution from platinum for the ruthenium peak and from ruthenium for the platinum peak and simply separate the two peaks at the peak transition. Another method, suggested by Green and Kucemak, is to first determine the total Pt-Ru surface area followed by the determination of platinum area by either stripping the ruthenium off the surface or oxidizing ruthenium to ruthenium oxide by cycling to high potentials and performing a second Cu UPD test. This is based on the hypothesis that copper does not deposit onto ruthenium oxide at UPD condition [64]. However, it has been shown by Zhang et al. that copper can deposit close to a full monolayer on ruthenium oxide [68]. Furthermore, cycling to high potentials can lead to catalyst agglomeration and surface area loss [14]. Therefore, the surface area determined by the second Cu UPD test might not be representative for the original platinum surface area. 37 (a)Pt 0 0.2 0,4 0.8 0.8 1 12 1.4 E vs. SHE (b)Ru •(i) (I) Anodic Cu UPD Stripping (u) (ii) Blank Scan - .- I - o 02 04 06 08 1 12 14 Evs.STIE (c) PtRu 1 II 0 0.2 0.4 0.6 0.8 1 1.2 14 E vs. SHE Figure 1-16: Cu UPD CV in 0.5 M H2S04 recorded at 0.010 V s1. (a) Pt; (b) Ru; (c) Pt-Ru. Reprinted with permission from [641. Copyright 2002 American Chemical Society. 38 1Blank Scan I’ I I I I Figure 1-17: Cu UPD stripping charge. Cu UPD Scan It is evident that each of the three electrochemical methods discussed has limitations and that there is no well-accepted method for determining the surface area of carbon supported Pt-Ru catalysts. Though, Cu UPD appears to be the most suitable and informative technique that also allows a rough estimate of the respective charge contributions from platinum and ruthenium. 1.2.6 Conventional Fuel Cell Electrode Design One of the main goals of electrode design in any fuel cell is to ensure that reactants are readily accessible to the catalyst layer while the products are effectively removed. The most common conventional design is the gas-diffusion electrode (GDE), where a catalyst, typically supported on high surface area carbon such as Vulcan XC-72 (20-40 wt% catalyst), is sprayed or painted onto the gas diffusion layer or membrane (see Figure 1-18). The catalyst is typically prepared in a catalyst ink consisting of 10-15 wt% Nafion® or other ionomer as a binder to provide protonic conductivity and good adhesion to the gas diffusion layer or membrane [5, 36, 69]. However, the inclusion of Nafion® in the catalyst ink was reported to lead to a 40% decrease in formic acid oxidation current density and 13% decrease in active surface area [69]. A microporous 39 sublayer, typically prepared by mixing a certain amount of PTFE, typically 30%, with carbon powder, is sometimes coated onto the gas diffusion layer. Though, it must be noted that the addition of microporous sublayer is not common for DLFC’s due to the undesirable increase in hydrophobicity. The GDE is subsequently hot pressed onto the proton exchange membrane, resulting in the membrane-electrode assembly (MEA). See Figure 1-19 for a cross-section SEM image of a full MEA with catalyzed carbon cloths hotpressed onto Nafion® 1135 membrane [70]. Catalyst-Coated Membrane (CCM) 100-300 10-50 pm pm 150-200 pm 4 I • I’ b. 4I I4I t:. s . ...q.. . •:4... •) t t t Diffusion Proton-Exchange Layer Membrane Supported Catalyst Figure 1-18: Conventional GDE MEA design comparison. Catalyst-Coated Diffusion Layer (CCDL) . . . . 0 0 0 , , t t t Diffusion Proton-Exchange Layer Membrane Supported Catalyst 40 The state-of-the-art commercial MEA design is the so-called catalyst-coated membrane (CCM). CCM is often prepared by spraying or painting the catalyst ink directly onto the membrane or by the decal transfer (DT) method. The DT method involves the painting of a catalyst ink on a blank decal PTFE, followed by the transfer of the catalyst layer formed on the PTFE to the PEM by hot pressing at 433-473 K [71]. The DT method yielded better contact and lower resistance between the membrane and the catalyst layer. As a result, the cell performance was also enhanced (from a DMFC peak power density of 600 w m to 1400 W m2 at 363 K) compared to the painting technique. Carbon Cloth Anode Catalysts Membrane Cathode Catalysts Carbon Cloth Figure 1-19: Cross-section SEM image of a full MEA with catalyzed carbon cloths (CCDL) hotpressed onto Nafion® 1135 membrane (70]. Reprinted with permission from Elsevier. Copyright Elsevier (2004). 41 1.3 Literature Review [CH2OHJ [CHOH] Short—Lived Adsorbed Intermediates Pathway Not Welt-Documented Figure 1-20: Schematic of methanol electro-oxidation pathways. 1.3.1 Methanol Electro-Oxidation The electro-oxidation of methanol has been a topic of interest since the 1 960s due to the attractive properties of methanol as a fuel. The reaction mechanism has been explained by a rate determining step model composed of a series of dehydrogenation steps followed by reactions between surface hydroxyl radicals and the adsorbed intermediates [20, 21, 72-74]. Figure 1-20 summarizes the possible reaction pathways. The electro-oxidation mechanism has more than one pathway and some do not lead to the formation of CO2. In fact, research has shown that CO2 is not the major reaction product of oxidation [74]. Other electro-oxidation products include formic acid (HCOOH) and formaldehyde (CH2O) (see Table 1-6). H2C(OK) - t•l2O ,/ [CH3O] CH2O ,“ HCOOH 7 N CH3O HCOOH 7 [CHOJ 7 + CO2 [CO H] ÷O[l CO 7 42 Table 1-6: Published data of oxidation products of methanol electro-oxidation. Methanol Oxidation Product Conditions CO2 HCOOH CH2O Authors Pt(100) at 0.2 V vs. SHE, 5% * 26% 0.1 M CH3O in 0.1 M HC1O4 Pt(100) at 0.3 V vs. SHE, 50% * 29% Jarvi et al. 0.1 M CH3O in 0.1 M HC1O4 [75] Pt(100) at 0.4 V vs. SHE, 80% * 8% 0.1 M CH3O in 0.1 M HC1O4 Polycrystalline Pt at 0.65 V vs. SHE, 34% 10% 56% 0.001 M CH3O in 0.5 M H2S04 Polycrystalline Pt at 0.65 V vs. SHE, 16% 34% 50% Wang et al. 0.01 M CH3O in 0.5 MH2S04 [76] Polycrystalline Pt at 0.75 V vs. SHE, 72% 3% 25% 0.01 M CH3O in 0.5 M H2SO4 HCOOH was not analyzed by Jarvi et al. [75]. As shown in Figure 1-20, it can be seen that methanol can react via COad or non-COad pathways, indicating that the catalyst can potentially be poisoned by the CO intermediate. The adsorbed CO can react with adsorbed OH species, which forms from dissociative chemisorption of water, to form CO2. as shown in Equation 1-81 and 1-82 below. Pt+HO—*Pt—OH(ad)+H+e [Eq. 1-81] Pt — CO(ad) + Pt — OH(ad) .—> 2Pt + CO2 + H + e [Eq. 1-82] Alternatively, the intermediate CH2O can desorb from the platinum surface to form formaldehyde in solution, undergo oxidation to generate formic acid, or react with water giving the adsorbed complexH2C(OH) [72, 77]. Adsorbed formic acid and the complex H2C(OH) could also undergo dehydrogenation to form CO2 without the formation of adsorbed CO. See Equation 1-83 and 1-84. HCOOH(Od) —* CO, +2H +2e [Eq. 1-83] 43 H2C(OH)(d) —> CO, + 4H + 4e [Eq. 1-84] The dissociation of water on Pt is the rate determining step at potentials below 0.7 V vs. SHE [20]. It is generally well-accepted that Pt-Ru is the best catalyst methanol oxidation due mainly to the bi-functional mechanism of ruthenium [20, 21, 60, 72, 73, 78-89]. The dissociative chemisorption of water on Ru on the other hand, occurs at 0.2 V vs. SHE. This is the result of having the highest water binding energy among all transition metals, 2.6 eV for Ru and 1.8 eV for Pt [90]. See Equation 1-85 and 1-86. Ru+H20 —> RUOH(ad)+H +e [Eq. 1-85] PtCO(ad)+RUOH(ad) -* Pt+Ru+C02+H +e [Eq. 1-86] With the addition of Ru, the potential at which adsorbed CO is removed is significantly reduced. This is shown in Figure 1-16 by the lower oxidation onset potential for methanol electro-oxidation with Pt-Ru catalysts compared to Pt catalysts [78, 91]. 550 500 450 400 350 300 E 250 2 200 150 100 50 0 -50 1110 200 300 400 509 600 700 E/Vvs. SHE Key:. U P + Ru (5O:5O; O P + Ru (7th3O; P + Ru (aO:2O; ( ) P + Ru c9O:UY; () Pt. Figure 1-21: Linear sweep voltammograms of methanol oxidation on Pt-Ru catalysts with different compositions in 1 M CH3O and 0.5 M H2S04 at 298 K. Scan rate: 0.001 V s_i [91]. With kind permission from Springer Science + Business Media: J. Appl. Electrochem., “Electrooxidation of Methanol at Platinum-Ruthenium Catalysts Prepared from Colloidal Precursors: Atomic Composition and Temperature Effects”, Vol. 33, 2003, 419-429, L. Dubau, C. Coutanceau, E. Gamier, J.M. Keger, C. Lamy, Fig. 7. 9 44 In addition to the lower potential required for the adsorption of water, the addition and presence of Ru increases the electron density around Pt sites, leading to a weaker chemisorption of CO [20, 92]. This effect is called the Ligand Effect and further reduces the potential required to remove adsorbed CO. Even though the addition of Ru is beneficial to the adsorption of water and the removal of CO, the adsorption of methanol on Ru is less favorable compared to Pt [20, 72, 79]. In addition to Pt-Ru, other binary combinations such as Pt-Ni, Pt-Sn, Pt-W, Pt-Co, and Pt-Metal Oxides have also been investigated [71, 80, 89]. These catalysts showed enhanced activity compared to pure Pt and are believed to promote methanol electro-oxidation mainly through electronic effects, but generally their performance is weaker than Pt-Ru. The electro-oxidation of methanol is strongly dependent on the reaction temperature, mainly due to the shift in the rate-limiting step [79, 91, 93]. It is known that Ru is not favorable for the adsorption of methanol (relative to Pt) at low temperatures (< 313 K) and Pt is not favorable for CO removal at low potentials. Hence, the optimal Pt:Ru ratio for the electro-oxidation of methanol depends on the synergistic effect of temperature and potential. According to Dubau et al. [91], at potentials smaller than 0.5 V vs. RHE, an increase in temperature requires an increase of Ru content (optimal Pt:Ru is 1:1) to enhance the methanol electro-oxidation, while at potential greater than 0.5 V vs. RHE, an increase in temperature requires a decrease in Ru content (optimal Pt:Ru atomic ratio 4:1). This was also confirmed by the experimental findings of Gasteiger et al. and Dickinson et al. [79, 93]. Gasteiger et a!. found an optimal surface ruthenium content of approximately 10 atom % at 298 K and approximately 30 atom % at 333 K. It was proposed that ruthenium is less active towards methanol dehydrogenation compared to platinum and increasing the reaction temperature increases the activity of ruthenium towards methanol dehydrogenation; therefore, more Ru is beneficial at high temperature [79]. Thus, as the temperature is increased, the rate-limiting step switches from methanol adsorption and dehydrogenation to the surface reaction between adsorbed oxygen species and adsorbed methanol oxidation intermediates. Similar findings were obtained by Dickinson and co-workers, showing that platinum-rich (3:2 atomic ratio) catalyst performed better than 1:1 catalyst at 298 K while the situation was reversed at 338 K [93]. 45 The interaction between the support and the catalyst also affects the electro oxidation of methanol [80]. Umeda et al. used different catalyst substrates, such as gold, platinum, carbon, and silicon, for their Pt-Ru-W, Pt-W, and Pt-Ru catalysts. For the supported Pt-Ru-W catalyst, the order was as follows: Au> Pt> C> Si. For the Pt-W catalyst, on the other hand, the activity using different supports was found to be essentially the same. For the Pt-Ru catalyst, the different supports generated the following order in terms of activity as measured by CV: Pt > Si > Au. Though, information about their respective catalyst particle size and surface area, which could influence their relative activity, were not provided. See Figure 1-22 below. c>J E 0 E 0 D 0 E/Vvs. SHE Figure 1-22: Effect of catalyst substrate withPt80Ru2.Voltammograms measured in 0.5 M H2S04 and 1 M CH3O . Temperature: 296 K. Scan rate: 0.010 V s_i [801. Reprinted with permission from Elsevier. Copyright Elsevier (2004). As shown by the experimental results, the effect of catalyst support depends on the catalyst composition. It was believed that the catalyst support can electronically affect 0.2 0.45 0.75 46 the catalyst surface, similar to that of the ligand effect of ruthenium on platinum, which can potentially affect the CO bond strength with the catalyst surface [80]. A number of studies showed that conditions for catalyst preparation depend heavily on the support of interest, due to both chemical and physical interaction effects between catalyst and support [94-98]. Therefore, it can be concluded that the catalyst support has the ability to modify the surface of the resulting catalysts through electronic effects as well as physical effects, such as specific surface area and the number of nucleation sites available for catalyst deposition and adsorption. Adding a third or a fourth metal to the Pt-Ru catalyst can further improve the activity of methanol electro-oxidation [80, 89, 99-101]. Typical alloying metallic elements include W, Ni, Mo, Sn, Os, and Ir. Umeda et a!. have shown that Pt-Ru-W catalyst on Au support shows enhanced catalytic activity towards methanol electro oxidation compared to the typical binary Pt-Ru catalyst (see Figure 1-23) [80]. <9 0 0.2 0.45 0.75 E/Vvs. SHE Figure 1-23: Linear sweep voltammograms of catalysts on Au substrate with different compositions measured in 0.5 M H2S04 and 1 M CH3O . Temperature: 296 K. Scan rate: 0.010 V 1801. Reprinted with permission from Elsevier. Copyright Elsevier (2004). 40 j 20 0 47 Gotz and Wendt have shown similar results with carbon substrate [99]. In the findings of Gotz and Wendt, Pt-Ru-W catalyst, in the form of WO,, performed the best compared to other catalysts, including Pt-Ru-Mo, Pt-Ru-Sn, and Pt-Ru. The catalytic promotion by W was believed to be due to the rapid changes in oxidation state of W involving W(VI)/W(IV), which renders sites active for dissociative adsorption of water [89]. The promotion by Mo was rationalized to be similar to W. However, a different mechanism was suggested by Shukla et a!. [102]. Shukia et a!. states that the reversible potential of the W(VI)!W(IV) couple is approximately -0.03V vs. SHE and at the oxidation potential of methanol, W(IV) concentration was found to be negligible by x-ray photoelectron spectroscopy (XPS). They speculated that the WO, of the catalyst is present in the form of an oxyhydroxide, which undergoes a proton shift to promote the formation of oxygen-containing species on the Pt surface [102]. The cocatalytic effect of Sn is more controversial and no definitive conclusion can be made regarding Sn promotion. Gotz and Wendt reported that Pt-Ru-Sn catalyst had lower activity than binary Pt-Ru catalyst [99]. The catalyst is active for H2/CO, but not for methanol electro-oxidation. It was postulated that the CO adsorbates generated by the dehydrogenation of methanol is of a different nature from CO adsorbed directly from solution in the H2/CO system [99, 103]. However, some researchers believe that Sn is catalytic towards methanol electro-oxidation due to the presence of oxidized Sn ions in solution [90, 100, 101]. It was suggested that Sn(II) in solution catalyzes the removal of surface poison generated during methanol electro-oxidation, with 1 jiM being the optimal concentration. Gurau et al. have shown that Pt-Ru-Os-Jr quaternary catalyst is superior to Pt-Ru, as well as Pt-Ru-Os, Pt-Ru-Ir, and Pt-Os-Ir (see Figure 1-24) [1001. The cocatalytic effect of Os is due to its superior oxyphilic behavior, it is known to adsorb water at potentials slightly more negative compared to Ru [104]. However, it can also contribute to methanol dehydrogenation [104]. The promotion by Ir is different from that of Os and Ru. Ir has a high carbon bond strength, which was believed to catalyze the methanol adsorption step [100]. The quaternary catalyst prepared contained 10 atom % Os and 5 atom % Ir, which were very close to their solubility limit in the face-center cubic 48 structure of Pt. Further increase in Os and Jr content led to additional phases, with reduced catalytic activity. Figure 1-24: Mass-normalized activity as a function of methanol concentration and anode potential forPt50-Ru,Pt65-Ru20s10,andPt47-Ru290s01r.Cathode: 40 g m2 Pt black with dry air fed at 1 atm and 400 mL min’; Anode: 40 g m2Pt50Ru, Pt65Ru20s10orPt47Ru290s01rwith methanol-water solution fed at 12.5 mL mm1. Reprinted with permission from [100]. Copyright 1998 American Chemical Society. As shown by the wide range of research done on the electro-oxidation of methanol, it is evident that the general consensus is to develop a catalyst formulation that can efficiently remove the adsorbed CO reaction intermediate. As presented in the next section, it is clear that formic acid electro-oxidation catalysis shares a similar goal and focus. 49 U -4 E z —-——--— ..,.. 411 21) 1 0.2 -p (14 (*H()H (:(IncenLrt4tion .1 r (132 — (V vs. I A) SHE) 1.3.2 Formic Acid Electro-Oxidation It is generally accepted that the electro-oxidation of formic acid on Pt proceeds through a dual-path mechanism involving a non-COad direct path and a COad indirect pathway [105-108). The mechanism, which was originally proposed by Capon and Parsons in 1973, is summarized in Figure 1-25. The direct path proceeds via dehydrogenation and CO2 is directly formed through the formation of active intermediates. The indirect path involves the formation of adsorbed carbon monoxide, both an intermediate and a catalyst poison through dehydration. The adsorbed carbon monoxide is then subsequently oxidized to CO2. -ti [HC:..0] -H [(OOHj HCC)H CO, Co [ j Active IntermedIate Figure 1-25: Dual-path mechanism of the electro-oxidation of formic acid. Arenz et al. [109, 110] studied the electro-oxidation of formic acid on Pt(1 11) single crystal and observed that COad is produced at potentials less than 0.35 V (see CO peak in Figure 1-26), but no CO2 band was observed by Fourier transform infrared spectroscopy (FTIR). This indicates that at <0.35 V, the electro-oxidation of formic acid does not proceed via the direct pathway, thus, the only pathway that can lead to CO2 is via COad which is stable on the platinum surface at this potential range. 50 HCOOH -> COCd + H20 [Eq. 1-87] COad + F1,O —> CO2 + 2H + 2e [Eq. 1-881 ________/ / ________ ________ CO2 Band CO Band j CO2 Band CO Band - O65V U I osv .. [____ pt... 1osv ,2O4cm4 I E 10.4s V SHE fl I O3S’ __ _ ois 2,PSflcm1 I -- . 1. 1024v 2)38 cm / 0,13V .. O,I.3V i I .. JI 1 i /,A i i Pt(l11) Monolayer of Pd on Pt(111) Figure 1-26: Pt and Pd - FTIR spectra recorded in 0.1 M HC1O4 containing 50 mM HCOOH [1091. Reproduced by permission of the PCCP Owner Societies. In the potential range of 0.35 to 0.45 V, where COad is still stable on platinum surface, the presence of a substantial CO2 band suggests that formic acid electro-oxidation can proceed through dehydrogenation (Equation 1-89 and 1-90). 51 HCOOH -> HCOO + H + e or [Eq. 1-89] HCOOH-COOH+H +e HCOO-CO2+H +e or [Eq. 1-90] COOH -* CO +H +e At potentials higher than 0.45 V, the intensity of CO spectra is reduced and has completely disappeared at 0.60 V. This suggests that adsorbed CO is being oxidized and that CO2 observed can be produced from both dehydration and dehydrogenation. This finding is also in agreement with Osawa and co-workers [111]. A typical cyclic voltammogram of formic acid electro-oxidation is shown in Figure 1-27 and the five different waves are related to different surface processes according to Okamoto et al. [112]. Wave and peak I are associated with formic acid dehydrogenation and the accumulation of COad formed by the non-faradaic heterogeneous formic acid dehydration, respectively, whereas wave and peak II are attributed to the oxidation of COad and surface oxide/hydroxide formation. Peak III is believed to be connected to the formic acid dehydrogenation on the oxide surface. On the reverse scan, wave and peak IV arise from the formic acid oxidation on the cleaned catalyst surface, while wave V could be due to surface re-arrangement processes [112]. i (mA) to E/Vvs. SHE Figure 1-27: Typical cyclic voltammogram of formic acid electro-oxidation on Pt in 0.5 M H2S04with 0.1 M HCOOH at 315 K. Scan rate: 0.100 V s’. Reprinted with permission from [112]. Copyright 2005 American Chemical Society. 0 0.5 1.0 1.5 1.8 52 A current-oscillation behavior for formic acid oxidation at high potentials has also been previously observed (on Pt as well as other electrodes) and studied by various researchers (Figure 1-28 and 1-29). The current oscillation phenomenon due to deactivation and reactivation is believed to be related to the periodic adsorption and removal Of COad [113118]. 30 20 i(mA) 10 L._J a - • I I 0 5 10 1:5 20 Time (s) Figure 1-28: Current oscillation observed in formic acid electro-oxidation on Pt at 1.1 V vs. SHE in 0.5 M H2504 and 1 M HCOOH 11131. Copyright Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission. 100 - 80 I: o 20 0 ___________________________________ -0.04 0.24 0.44 0.64 0.84 1.04 E/Vvs. SHE Figure 1-29: Cyclic voltammogram of formic acid electro-oxidation in 0.5 M H2S04 and 1 M HCOOH at 0.010 V s [117]. (a) Pt modified by 106 M Bi; (b) Pt. Reprinted with the permission from J. Lee, J. Christoph, P. Strasser, M. Eiswirth, G. Ertl, J. Chem. Phys., Vol #. 115, 3, 1485-1492, 2001. Copyright 2001, American Institute of Physics. II 53 Another common mono-metal catalyst for formic acid electro-oxidation is palladium. The electro-oxidation of formic acid on palladium, in contrast to platinum, is believed to proceed primarily through the flOflCOad pathway [11 1, 119, 1201. It was shown by Larsen et a!. by CO stripping voltammetry that the CO buildup on Pd black was significantly lower than on Pt black [120]. Similarly, Arenz et al. did not detect any adsorbed CO by FTIR spectra (Figure 1-26) even though a high production rate of CO2 was observed [109]. As a result, Pd is seemingly more catalytic towards formic acid electro-oxidation than Pt, especially at low potential (<0.4 V) where the adsorbed CO intermediate on Pt cannot be further oxidized. However, it is important to note that Pd deactivates significantly with time as shown by many different studies [109, 120-122]. Blair et al. have studied the long term stability of Pt, Pd, and various Pd-Pt catalysts by CP and found that Pd albeit having superior initial catalytic activity, its activity falls below that of Pd-Pt and Pt within just hours (see Figure 1-30). Similarly, Li and Hsing compared the stability of Pt, Pd, and Pt-Pd by CA and found that the oxidation current on Pd dropped by ten-fold in under half an hour (Figure 1-3 1). They proposed that the deactivation of Pd was likely a result of Pd dissolution and the adsorption of intermediate species or residual amounts of acetic acid in formic acid [122]. Additionally, it was proposed by Arenz et al. [109] that the deactivation of Pd was a result of the blocking of active sites by the adsorption of spectator species (such as Had, OHad and anions of the supporting electrolyte) or Pd oxide formation. It was shown that increasing the temperature from 276 to 303 K led to catalyst deactivation as opposed to an activation observed on Pt. 54 08 0.7 —— — W 06 — r 0.5 1 a) > Pd ———— Pd:Pt94:6 W 0.4 Pd:Pt85:15 Pd:Pt 65:35 Pd:Pt 50:50 Pd:Pt25:75 0.3 _________ Pt 0.2 — 0 2 4 8 8 10 12 14 tmeihours Figure 1-30: Chronopontentiometry of formic acid electro-oxidation on Pt, various Pd-Pt, and Pd (60 g m2) in 0.1 M H2S04+ 10 M HCOOH at 293 K and 1000 A m2 11211. Reproduced by permission of The Electrochemical Society. 050 0.45 • Pd/C PtPdJC 0.20 0,25 0.20 0 200 400 600 800 1000 1200 1400 1600 1800 Tine I s Figure 1-31: Chronoamperometric curves of Pt/C, Pt05d.5IC and Pd/C in 0.25 M HC1O4 + 0.25 M HCOOH at 0.3 V [1221. Reprinted with permission from Elsevier. Copyright Elsevier (2006). 55 The electro-oxidation of formic acid has also been studied on bimetallic catalysts including Pt-Ru, Pt-Pd, Pt-Au, Pt-Bi, and Pt-Pb [109, 12 1-1301. In the case of Pt-Ru, it was shown by Markovic et al. that Ru is active towards formic acid electro-oxidation via the dehydration pathway and that the addition of Ru for formic acid electro-oxidation is shown to have the same bi-functional ligand effect on Pt for removing COad as observed in methanol electro-oxidation [1071. Hence, it was found that Pt-Ru yielded a five-fold increase in oxidation current compared to pure Pt. Similarly, Rice et al. have shown that Pt-Ru has the highest catalytic activity among Pt and Pt-Pd at high current density and anode overpotential (Figure 1-32) [123]. 0 100 200 300 Current Density, Figure 1-32: Polarization curves of DFAFC at 303 K with Pt, Pt-Pd, and Pt-Ru anode catalysts (40 g rn2). Cathode: 70 g m2 Pt black with pure humidified oxygen fed at 2 atm and 100 mL mm1; Fuel: 5 M HCOOH fed at 0.5 mL mm’ [123]. Reprinted with permission from Elsevier. Copyright Elsevier (2003). > O6 0 400 500 rnAJcm2 56 On the other hand, Pt-Pd have been demonstrated to have superior activity at low overpotentials and higher OCV in DFAFC tests, indicating a lower onset potential, due to the difference in reaction mechanisms. Larsen and Masel studied the reaction with a starting CO-covered catalyst surface for Pt, Pt-Ru, and Pt-Pd [1251 and have evaluated their respective ability to strip off adsorbed CO. The steady-state CO coverage of Pt-Pd was observed to be higher than both Pt and Pt-Ru at 0.3 and 0.5 V. Among the three different catalysts, Pt-Pd was found to have the least capability for stripping off CO; however, the catalytic activity towards formic acid electro-oxidation is higher, suggesting that formic acid electro-oxidation proceeds via a non-CO pathway on Pt-Pd similar to Pd. However, it is also evident that Pt-Pd suffers from the same deactivation problem as Pd as shown in Figure 1-31 while Pt-Ru has been shown to have higher stability in long-term tests [121, 1221. Choi et al. have compared the performance in DFAFC tests of unsupported Pt-Au with Pt-Ru, prepared by chemical reduction of H2PtC16 and HAuCL1 with sodium borohydride [124]. It was found that Pt-Au yielded a lower onset potential for formic acid electro-oxidation as well as higher oxidation current in the whole potential range tested (Figure 1-33). The promotion by Au is explained by the third-body effect, where Au is preventing the formation of adsorbed CO due to geometric hindrance, favoring the dehydrogenation pathway. Unlike Pd and Pt-Pd, Pt-Au was demonstrated to have reasonable long-term stability where the performance decay was not more than 10% for up to 500 hours. 57 0.26 020 , 0i5. E o,10 0.05 0.00 Figure 1-33: Polarization curves of formic acid electro-oxidation at 303 K with Pt- Ru and Pt-Au anode catalysts (30 g m2)with 6 M HCOOH fed at 5 mL mm1. Scan rate: 0.010 V s_i [124]. Reprinted with permission from Elsevier. Copyright Elsevier (2006). Bismuth promotes the formic acid oxidation reaction by favoring the dehydrogenation reaction via the third-body effect and electronic effects [126-129]. Tripkovic et al. studied the electro-oxidation of formic acid on Pt-Bi and have shown that the addition of Bi resulted in 0.25 V more negative onset potential and more than two orders of magnitude in oxidation current at 0.05 V vs. SHE [126]. Based on XPS data, the promotion effect by Bi was explained by hydroxylated Bi species providing OHad for CO removal in addition to the third-body effect. In agreement with Tripkovic et al., Macia et al. studied the effect of adsorbing Bi onto Pt( 111) single crystal and found that the oxidation current increased with Bi coverage up to approximately 0.25 [127]. Similarly, Kang et al. have studied the effect of Bi on polycrystalline Pt and found that Bi-modified Pt yielded superior performance by CV (Figure 1-34) and DFAFC tests while PtAu (1:1) i —C—PtRu(1:i) I • I • I • I • I • 0.0 0.1 0.2 0.3 0,4 0.5 0.6 O..i E/Vvs. SHE 58 cJ E >‘ ‘I) a) 0 C 1) C) Roychowdhury et al. presented the superior catalytic activity of Pt-Bi over Pt and Pt-Ru [128, 129]. However, no long-term stability study of Pt-Bi is presented. 250 200 150• 100 50 0 -50 Figure 1-34: Cyclic voltammograms of formic acid electro-oxidation in 1 M HCOOH and 0.5 M 112S04.Scan rate: 0.050 V s1. (a) Pt and (b) Pt modified with Bi. Reprinted with permission from [1281. Copyright 2006 American Chemical Society. Zhang and coworkers have investigated the electro-oxidation of formic acid on Pt-Pb (1:1 atomic ratio) prepared by vacuum melting [1301. Compared to Pt, the onset potential of formic acid oxidation was shifted negatively by 0.3 V and the oxidation current was increased by over 2-folds (Figure 1-35). The catalytic enhancement was explained by the increase in Pt-Pt bond length from 2.77 A to 4.24 A with the addition of Pb, which made it difficult for CO to bind in bridge site and three-fold hollow sites. -0.06 0.24 0.54 0.84 1.14 1.44 1.74 2.04 E/Vvs. SHE 59 if E/Vvs. SHE Figure 1-35: Cyclic voltammograms of formic acid electro-oxidation on Pt and Pt- Pb in 0.25 M HCOOH and 0.5 M H2S04 [130]. Scan rate: 0.050 V s1. Reprinted with permission from Elsevier. Copyright Elsevier (2006). 1.3.3 Preparation of Electrocatalysts and Nanostructured Materials Nano-sized electrocatalysts show enhanced catalytic activity due to the high specific surface area, changes in electronic structure, as well as the enhancement in surface diffusion. Methods that are generally employed in producing nano-sized metal catalyst particles for fuel cell applications include the sol-gel (SG) method, polyol method, colloidal method, and electrodeposition. 1.3.3.1 Sol-Gel Method The SG method is a process that results in a hydrated solid precursor or a hydrogel [13 1-133]. It involves the formation of inorganic networks from the gelled colloidal suspension. The precursors for synthesis consist of a metal surrounded by reactive ligands, such as acetylacetonate (-C5H702)and ethoxide (-0C2H5). In some 40 P1Pb ‘““-“Pt 30 20 10 0 -10 -0.16 0.04 0.24 0.44 0.64 0.84 1.04 1.24 1.44 60 cases, the metal-ligand precursor can be synthesized by mixing common metal precursors (such as hexachioroplatinic acid H2PtC16)with metal alkoxides (e.g. sodium ethoxide NaOC2H5and tetraethoxysilane Si(0C2H5)4.There are four main steps involved in the SG method: formation of the hydrogel, aging of the hydrogel, removal of solvent, and heat treatment. The formation of the hydrogel is given in Figure 1-36 [133]. — M OR + H20 Hydrolyis — M OH + ROH I I Water I I — M— OH + — M— OH Condensation M— O M + H2O — M— OH + M— OR ation M— 0 M + ROH M = Metal R = Carbon Chain Figure 1-36: Formation of hydrogel. The aging process can take up to 100 hours and is followed by drying, where the solvent is removed and solid particles or networks are obtained [1311. The catalyst is then activated by heat-treatment. The SG method is versatile because the morphology of the resulting solid structure may be porous, dense, and uniform, depending on the conditions of the various steps involved, such as p1-I, temperature, time of reaction, concentration, catalyst nature, and aging temperature and time. Metal films and particles of different sizes for fuel cell applications have been produced via the SG methods in literature. Kim et al. synthesized Pt and Pt-Ru particles of up to 20 urn in diameter (Figure 1-3 7) with the SG method based on homogenous Pt and Ru-sols derived from platinum acetylacetonate (Pt(C5H702)and ruthenium acetylacetonate (Ru(C5H702)3[134]. Shilova et al. synthesized nano-layers of Pt-doped 61 polysiloxane matrix (Figure 1-38), which had uniformly distributed Pt particles 0.5-2.5 nm in diameter [135]. McLeod and Birss prepared Pt-WO, catalyst for methanol electro oxidation from a Pt-sol mixed with a WO,-sol with both ethanol and water condensation [136]. The Pt-WO, films synthesized with ethanol condensation ranged from 0.3 to 0.9 urn in thickness while the thickness of the film prepared with water condensation varied from 0.2 to 5 trn (Figure 1-39). The film obtained with water condensation had superior catalytic activity towards methanol electro-oxdation due to co-catalytic effect from WO. Their results illustrate very well the importance of catalyst preparation method and conditions. Figure 1-37: SEM image of Pt powder synthesized by the sol-gel method by Kim et al. [134]. Reproduced by permission of The Electrochemical Society. 62 Figure 1-38: TEM image of Pt nanoparticles synthesized by the sol-gel method with tetraethoxysilane [135j. With kind permission from Springer Science + Business Media: Glass Phys. Chem., “Formation of Catalytic Layers from Tetraethoxysilane Based Sols for Use in Polymer Fuel Cells”, Vol. 30, 2004, 98-100, O.A. Shilova, V.V. Shilov, N.D. Koshel, E.V. Kozlova, Fig. 2. 63 Figure 1-39: SEM images of Pt-WO films synthesized by the sol-gel method. (a) Ethanol condensation; (b) Water condensation 11361. Reprinted with permission from Elsevier. Copyright Elsevier (2005). 64 1.3.3.2 Polyol Method Another approach, generally known as the polyol method, is often employed in synthesizing nanoparticles. The poiyoi method typically involves the use of an ethylene glycol solution containing the catalyst precursor salts, though other polyols such as triethylene glycol, glycerol, and tetraethylene glycol have also been employed previously [129, 137-142]. The liquid polyol acts as the solvent, reducing agent, and also as a protecting agent to limit particle sintering. A strong base, typically KOH, is added as a catalyst to promote the formation of the polymeric poiyol to limit particle growth. The solution is heated, generally with the assistance of a microwave to provide more uniform and rapid heating, so that in-situ reducing species can be generated from ethylene glycol to reduce the metal precursor salts. One advantage of this method is the relatively low temperature required for particle synthesis and the particle size and morphology can be controlled via the different experimental parameters, including temperature, reaction time, polyol type, metal precursor type, concentration, and pH. For instance, Goia et al. have studied the effects of employing different liquid polyols and metal precursor to prepare bismuth nanoparticles at the same reaction time of 4 hours [1411. It can be seen from Figure 1-40 and 1-41 that the size and shape of the particles are highly dependent on the mat&rials used. The polyol method has also been used to produce catalysts for fuel cell application. Yu et al. used this method to prepare platinum nanoparticles with uniform size range between 2-4 nm (Figure 1-42) [139]. Similarly, Liu et al. have utilized the microwave-assisted polyol process to synthesize Pt/C and Pd/C catalysts of 4-5 nm [140]. Bi-metallic nanoparticles have also been synthesized as fuel cell catalysts. Roychowdhury et al. have prepared Pt-Bi as DFAFC anode catalysts with the microwave-assisted polyol method with the precursor salts H2PtC16 and bismuth nitrate (Bi(NO3)). The particles were approximately 1 9-nm in diameter with essentially a 1:1 Pt-Bi atomic ratio (Figure 1-43). 65 Figure 1-40: SEM images of Bi nanoparticles prepared by boiling bismuth hydroxide for 4 hours in different liquid polyols (141]. (a) 1,3 Propane diol; (b) Diethylene glycol; (c) Glycerol; (d) Tetraethylene glycol. Reprinted with permission from Materials Research Society and D. Goia. 66 Figure 1-41: SEM images of Bi nanoparticles prepared by boiling for 4 hours in propylene glycol with different bismuth precursors [1411. (a) Bismuth acetate; (b) Bismuth nitrate; (c) Bismuth subcarbonate. Reprinted with permission from Materials Research Society and D. Goia. 67 Figure 1-42: TEM image of Pt nanoparticles synthesized by the microwave-assisted polyol method. Reprinted with permission from [139]. Copyright 1999 American Chemical Society. Figure 1-43: SEM image of Pt-Bi nanoparticles prepared by the microwave-assisted polyol method. Reprinted with permission from [129]. Copyright 2005 American Chemical Society. I * 68 1.3.3.3 Colloidal Organosol Method The colloidal organosol method developed by Bönneman and coworkers is a promising preparation technique capable of producing metal particles in the range of 1-10 nm [143-148]. The colloidal method involves the stabilization of the metal particles by a protective shell. A common surfactant employed is tetraoctylammonium triethyihydroborate [N(C8H17)4BEt3H]. When added to a metal salt, such as PtCI2 in tetrahydrofuran (THF), tetraoctylammonium triethyihydroborate acts as both a reductant and a protective shell around the metal particles, which prevents particle agglomeration. Hydrogen gas is evolved as a by-product. See Equation 1-91 and Figure 1-44. The metal colloid can also be deposited onto catalyst supports by adsorption, simply adding the catalyst support into the colloidal solution. PtC12 + N(C8H17)4BEt3H —+ Pt + 2N(C8H17) Cl + 2BEt3 + H, [Eq. 1-91] Figure 1-44: N(C8H17)C1stabilized metal core. H3C H3C H3C CH3 CH3 69 BOnneman and coworkers synthesized a large variety of metals (e.g. Pt, Ru, Au, Sn, Co, Os, Ir) and alloys [143-1481. Alloys prepared with this colloidal method that are relevant to DLFC include Pt-Ru (0.8 to 3.0 nm in diameter, Figure 1-45) and Pt-Sn nanoparticles (1.3 nm in diameter) supported on silica [144, 148]. Similarly, Gotz and Wendt prepared carbon-supported Pt-Ru, Pt-Ru-W, Pt-Ru-Mo, and Pt-Ru-Sn catalysts with particle size of approximately 1.7 nm as DMFC anode catalysts [99]. Coutanceau et al. used a similar method and dispersed Pt-Cr with particle sizes ranging from 4 to 5 nm onto carbon powder (Figure 1-46) [149]. In addition, the Bönneman method has also been modified by various researchers. Reetz and Quaiser replaced the solvent with an acetonitrile/THF mixture and employed an electrochemical setup for electrochemical reduction of the metal precursors [150, 151]. Pt, Rh, Ru, Os, Pd, and Mo were successfully prepared with particle sizes ranging from 2 to 5 nm. Bi-metallic nanoparticles (e.g. Pt-Rh, Pt-Sn, Cu-Pd, and Pd-Pt) were also synthesized by either using two metal precursor salts in solution or with the use of sacrificial anode. Likewise, Lycke and Gyenge modified the Bönneman method by introducing an electrophoretic driving force to deposit Pt and Pt-Sn particles on uncompressed GF for use in the direct ethanol fuel cell (DEFC) [97]. Pt-Ru - Nanoparticles Figure 1-45: High resolution TEM image of Pt-Ru nanoparticles synthesized by the colloidal organosol method. Reprinted with permission from [148]. Copyright 1997 American Chemical Society. 70 Figure 1-46: TEM images of Pt-Cr/C nanoparticles synthesized by the colloidal organosol method [149]. (a)Pt80-Cr2/ ; (b)Pt50-Cr/ . Reprinted with permission from Elsevier. Copyright Elsevier (2005). L-J t [b> 71 1.3.3.4 Electrodeposition Electrodeposition is a versatile catalyst preparation technique that can lead to different deposit surface morphology depending on the experimental conditions [152- 156]. Jayashree et al. electrodeposited Pt and Pt-Pd nanoparticles on Au in an aqueous media [152]. The surface morphology of the electrodeposited Pt differed significantly depending on the deposition potential. At -0.80 V vs. SHE, a smooth deposit film was observed in SEM while mesoporous particles were obtained at -1.8 V vs. SHE (Figure 1- 47). Saber et al. reported pulsed current electrodeposition results and found a similar trend for Zn deposits [153]. The metal deposits were found to have smaller particle size when the pulse peak current density was increased (Figure 1-48). These results are in agreement with the favorable conditions for nanoparticle production reported by Choo and coworkers [154] (i.e. high overpotential, high adion population, and low adion surface mobility). During electrodeposition, the incorporation of metal ions to kink sites or surface defects, which leads to crystal growth, is most favorable as these sites have the lowest potential energy [1551. Since other deposition sites have higher potential energy, the formation of new crystals requires additional nucleation energy. This can be overcome by electrodepositing with a high overpotential (or current density), which provides a stronger driving force for nucleation, promoting the formation of nanoparticles. Electrodeposition can proceed via two alternate paths: direct deposition of the metal ion to the kink site or the transfer of the metal ion to the crystal surface as an adion followed by surface diffusion to the kink site and subsequent deposition [155-1571. Bockris and Reddy have explained that low adion surface mobility could slow down the adions from searching for more favorable crystallization sites; thus, promoting the formation of new crystals [1561. This can be achieved by adding additives to produce a chemical or physical barrier at the cathode. The electrodeposition method can be employed in conjunction with other particle size controlling methods to synthesize nanoparticles. Though, it is important to emphasize that electrodeposition was mostly used in conjunction with flat substrates in literature; therefore, the method is generally unused for preparing fuel cell electrodes. 72 Figure 1-47: SEM images of electrodeposited Pt catalyst [1521. (a) Deposition potential of -0.8 V vs. SHE; (b-d) Deposition potential of — 1.8 V vs. SHE. Reprinted with permission from Elsevier. Copyright Elsevier (2005). jh ) Figure 1-48: SEM images of Zn electrodeposited at various pulse current densities [153j. (a) 4000 A m2; (b) 8000 A m2; (c) 12000 A m2; (d) 16000 A m2. Reprinted with permission from Elsevier. Copyright Elsevier (2003). 10 kV.x3 ID kV x 5000, 2.5 pm I 73 Elliott et al. electrodeposited Pt nanoparticles on a smooth Au wire with a hexagonal lypotropic liquid crystalline phase [158, 159]. The hexagonal nanostructure was reflected in the morphology of the metal deposits (see Figure 1-49 and 1-50). They reported that the change in deposition potential modified the surface morphology of the metal deposits. At a deposition potential of -0.76 V vs. SHE, the hexagonal surface structure was obtained. At a deposition potential of -1.76 V vs. SHE, however, the structure was disordered. At any deposition potential more negative than -1.76 V vs. SHE, completely disordered deposits were obtained. The effect of deposition temperature was also reported, the surface area of the metal film increased with temperature. Similarly, Ganesh and Lakshminarayanan electrodepsited nanostructured Ni on polished Ni using a Triton X-100/poly-aerylic acid/water hexagonal liquid crystalline template (Figure 1-51) [160]. They reported that the surface roughness was improved by close to ten-folds with the addition of poly-acrylic acid compared to the Triton X-100/water system, indicating that the morphology is strongly dependent on the template bath composition. I:?.c F1b4 Figure 1-49: Schematic of a 3-D structure of a hexagonal liquid crystalline phase (a) and the expected nanostructure of a material produced in its presence (b). Reprinted with permission from [158]. Copyright 1999 American Chemical Society. L . F •ie 4.. I 74 Figure 1-50: SEM image of nanostructured Pt deposited at -0.76 V vs. SHE at 338 K in a hexagonal liquid crystalline phase [1581. Reprinted with permission from [1581. Copyright 1999 American Chemical Society. Figure 1-51: SEM image of nanostructured Ni deposited on polished Ni at 50 A m2 at 298 K for 1 hour in a Triton X-100/poly-acrylic acid/water hexagonal liquid crystalline phase [160]. Reprinted with permission from Elsevier Copyright Elsevier (2004). 20 KV X 5,000 1 Lfl1 75 In parallel with the timeframe of this study, Bauer et al. have studied the galvanostatic electrodeposition of Pt, Pt-Ru, and Pt-Ru-Mo catalysts on compressed GF (350 jim thickness), as extended reaction zone anodes for the DMFC, using the Triton X 100 liquid crystalline and micellar phases [95, 96]. Particles in the range of 10-100 nm and a specific mass load of 43 (Pt-Ru) and 52 g m2 (Pt-Ru-Mo) were deposited on the compressed GF from the Triton X-l00 micellar phase at 333 K (Figure 1-52). The electrodeposition rate of Pt, Ru, and Mo in the presence of Triton X-100 (0 to 40 wt.%) was studied by CV. It was found that the electrodeposition of Pt was suppressed significantly with increasing Triton X- 100 content, as opposed to a lesser effect on Ru and Mo. Therefore, with the addition of Triton X-100, the Pt:Ru atomic control can be controlled by lowering the Pt content. It is important to note that Bauer et al. did not investigate different colloidal structures and focused only on the GF substrate. Figure 1-52: High resolution SEM micrographs of Pt-Ru (a) and Pt-Ru-Mo (b) deposits on the pressed graphite fiber surface [96j. Reprinted with permission from Elsevier. Copyright Elsevier (2007). 76 1.3.3.5 Chemical Reduction in Reverse Microemulsion Chemical reduction of metal salts in reverse microemulsion has received considerable attention in recent decades as a nanoparticle synthesis method. The method involves the formation of a water-in-oil microemulsion from an organic phase, an aqueous phase, a surfactant, and a co-surfactant. The metal salt is dissolved in the nano sized water pools and each acts as reactor for the chemical reduction process. The microemulsion containing the metal salt is mixed with an identical microemulsion but with a chemical reductant, typically hydrazine or sodium borohydride, in place of the metal salt [161, 162]. Continuous Oil Phase Figure 1-53: Schematic of metal nanoparticle formation in reverse microemulsion. The particle size obtained from the microemulsion method is dependent on the solution phase structure of the microemulsion employed, which in turn is a function of the salt concentration, composition, chemical nature of the different phases, and Metal Salt in Reductant in Water Water Pools in Pools in WIO WiO Microemulsion Microemulsion Metal Nanoparticle in Water Pools in WIO Microemulsion 77 temperature [163-165]. In general, the increase in salt concentration will lead to larger particles, the decrease in water content will lead to smaller particles, and the increase of temperature will lead to smaller particles [163]. Zhang and Chan have successfully synthesized Pt-Co and Pt-Ru nanoparticles (3- 5 nm) with a cyclohexane (35 vol%) / Triton X-100 (10 vol%) / isopropanol (40 vol%) / water (15 vol%) reverse microemulsion [164, 165]. The particles obtained were monodisperse, as shown in Figure 1-54 below. The nanoparticles were also supported on carbon by simple addition of carbon powder into the microemulsion. Similarly, Santos and coworkers have synthesized Pt-Ni/C (4 to 5 nm) with different Pt:Ni atomic composition from a Brij ®3 0 (polyethylene glycol dodecylether)/heptane/water reverse microemulsion [166]. However, no details were given regarding the microemulsion composition. 3 to 5 nm Pt-Co Nanoparticles Figure 1-54: TEM image of Pt-Co nanoparticles prepared by microemulsion [1641. Reproduced by permission of The Royal Society of Chemistry. 10 nm 78 1.3.4 Novel Electrode Design for Direct Liquid Fuel Cells Recent novel electrode designs found in literature include novel porous carbon catalyst support [167], the multi-layer anode design [1661, and use of titanium mesh support [39, 169-172]. As can be seen from the results discussed below, the design of the anode in DMFC or DFAFC has a significant impact on its performance. Chai et al. custom-made uniform and ordered porous carbon networks by using a colloidal silica templating technique. The carbon network was made by polymerization and carbonization of phenol and formaldehyde in a silica template followed by etching in hydrofluoric acid. The pore sizes can be controlled and ranged from 10 to 1000 nm (Figure 1-55) [167]. Figure 1-55: SEM images of silica-templated carbon support with different pore size. (a) 25 urn. (b) 68 nm; (c) 245 nm; (d) 512 nrn. Reprinted with permission from [1671. Copyright 2004 American Chemical Society. 79 The mesoporous carbon substrates have very high surface area (e.g. 450-1000 m2 g’ compared to 232 m2 g’ of Vulcan XC-72), large pore volumes, and allow for high degree of catalyst dispersion and efficient gas diffusion. The performance of the prepared anode with a high Pt-Ru load of 30 g m2 was superior to conventional catalysts (improved from 330 to 580 W m2, see Figure 1-56). In general, the performance followed the trend of decreasing pore size and hence increasing surface area. The superior performance obtained with the custom-made carbon support was suggested to be due to both their higher surface area and more efficient diffusion of fuel and products compared to the Vulcan XC-72 support. However, it is important to note that this novel support is probably not easily synthesized on a large scale. Figure 1-56: Single DMFC cell performance comparing the commercial E-Tek catalyst with catalysts prepared on custom-made carbon supports. The numbers in the legend designate the pore diameter of the porous carbon replicas in nm. Temperature: 303 K. Anode: 30 g Pt-Ru m2 with 2 M CH3O fed at 1 mL mm1. Cathode: 50 g Pt m2 with dry air fed at 500 mL mm1.Reprinted with permission from [1671. Copyright 2004 American Chemical Society. • . PIRU(E-TEK) PtRu-C26 A PtRu-C48 PtRuC245 * PtRu-G-S12 700 600 500 . 400 •3G0 200 100 60 50 40 E 30 20 0 100 0 — - — 0 50 100 150 200 250 300 Current densIty! mA 80 Wilkinson et al. recognized the importance of the anode design in reducing the methanol crossover, and they patented the multi-layer anode concept [168]. Three sheets of catalyzed carbon fiber paper (thickness 100 j.tm) with a total load of 18 g m2 Pt-Ru were stacked to give a multi-layer DMFC anode (Figure 1-57). As compared to a single- layer design, it was found that the multi-layer anode design enhanced the methanol utilization efficiency from 60% to 80%, when employed with a mixed solution of 2 M CH3O + 0.5 M H2S04, to provide protonic conductivity. The methanol utilization efficiency is defined to be the ratio of charge output due to methanol oxidation to the theoretical output if all methanol is consumed at the anode to provide electrical current. Co2 4 CH3O H20 I Diffusion Layer t Proton-Exchange Membrane Supported Catalyst Figure 1-57: Schematic representation of the multi-layer anode design for DMFC as suggested by Wilkinson et al. [1681. 81 •1 Allen et al. investigated the feasibility of using titanium mesh as 3-D anode as an alternative to conventional catalysts for reasons discussed before. They electrodeposited Pt-Ru catalysts onto titanium mesh and preliminary studies have shown very similar performance compared to conventional catalysts [39]. Recently, Shao et al. published DMFC results with their novel Pt-RuJTi mesh anode prepared by electrodeposition at 0.1 V vs. SHE from an acidic deposition bath containing 10 mM H2S04,2 mMH2PtC16,and 2 mM RuC13 [169-171]. It is evident that the use of catalyzed titanium mesh can enhance the performance of DMFC, especially at high current densities and 0.5 M methanol concentration (see Figure 1-58). > 500 600 Figure 1-58: Comparison of the performance of the DMFC at 363 K with different anodes under different methanol concentrations L1691. Anode: 40 g m2 Pt-Ru black or Pt-Ru/Ti withCH3OH-water fed at 10 mL mm’. Cathode: 35 g m2 Pt/C with dry air fed at 200 mL mm1.Reproduced by permission of the PCCP Owner Societies. The enhancement in performance at high current density is explained by enhanced catalyst utilization as well as superior mass transport of methanol and the disengagement of product CO2(g). However, due to the open structure of titanium mesh, the use of 2 M methanol concentration yielded poorer performance brought about by severe crossover. 0.6 025 M, Corwentkna anode —o— 0.25MTimesI anode —. (15 M. Corwentional anode 0.5 M. Ti mesh anode —— 2 M. Conventional anode —— 2 M. Ti mesh anode 0.4 0 00 200 300 400 Current Density .1 mA cnr2 82 Similarly, Chetty and Scott have deposited Pt-Sn and Pd catalysts on titanium mesh for use as anodes in DFAFC by thermal decomposition (363 K) and galvanostatic electrodeposition at 50 A m2 with 2.5 mM K2PdC16 [1721. The Pd and Pt-Sn/Ti mesh catalyst prepared by both thermal decomposition and electrodeposition were superior to the respective conventional PdJC and Pt-Sn/C ODE design in a DFAFC operated with 1 M HCOOH (Figure 1-59 and Figure 1-60). However, no comparison was made at higher fuel concentration to determine the relative performance of the two anode designs. Figure 1-59: Comparison of the performance of the DFAFC at 333 K [1591. Anode: 20 g m2 Pd/C ( • ) or Pd/Ti mesh (thermal decomposition, • ) or Pd/Ti mesh (electrodeposition, A) with 1 M HCOOH fed at 1 mL mm1. Cathode: 20 g m2 Pt/C with atmospheric air fed at 400 mL mm1. Reproduced with permission from [172]. Copyright 2007 Journal of New Materials for Electrochemical Systems. 20 > I S ‘ C.) 2; Current density (mA.cm2) 83 E4) -D L) & Figure 1-60: Comparison of the performance of the DFAFC at 333 K [1591. Anode: (i) 20 g m2 Pt-Sn/C or (il)Pt-Sn/Ti mesh (thermal decomposition) with 1 M HCOOH fed at 1 mL mm1.Cathode: 20 g m2 Pt/C with atmospheric air fed at 400 mL mm1. Reproduced with permission from [1721. Copyright 2007 Journal of New Materials for Electrochemical Systems. 1.4 Objectives 1 — Develop colloidal templated electrodeposition methods to produce nanostructured catalysts on 3-D substrates, including reticulated vitreous carbon (RVC), graphite felt (GF), and titanium mesh. 2 — Investigate the electrocatalytic activity of the novel catalyzed 3 -D substrates prepared with the developed methods towards methanol and formic acid electro-oxidation and develop an understanding of the interacting effects among substrate characteristics, catalyst surface morphology, and catalytic activity. Current density (mAcrn2) 84 3 — Compare the conventional (CCM, CCDL) and novel anode designs and catalysts in DMFC and DFAFC experiments to evaluate the benefits and disadvantages of the new anode design. There has been very little research done on depositing nanoparticles uniformly throughout these 3 -D substrates. Most literature studies, as presented in the Literature Review section, were concerned with deposition on smooth substrates. As a result, it becomes important first to develop methods to create the desired extended reaction zone anodes so that the anode design comparison can be made. The catalyst preparation method investigated was electrodeposition from different colloidal solutions, such as emulsion, microemulsion, and micellar solutions, involving different surfactant and co surfactant combinations, including Triton X-100, Triton X-102, and isopropanol. The combination of electrodeposition technique with colloidal phases that limit particle growth and provide penetration of the catalyst nanostructure into the 3-D substrate, is a novel approach. When used as an extended reaction zone anode by replacing both the catalyst layer and gas-diffusion layer in a DMFC or DFAFC, the catalyzed 3-D anode potentially enhances the catalyst utilization, simplifies the fuel cell design, and reduces the fuel cross-over problem, since less fuel becomes available at the membrane-electrode interface as the fuel is reacted across the 3-D anode. As a proof of concept, a sulfuric acid supporting electrolyte can be employed to provide the required ionic conductivity in fuel cell tests. Figure 1-61 illustrates the differences between the conventional and novel anode design. Consequently, the present study can lead to new methods for synthesizing nanoparticles on 3-D substrates, demonstrate the benefits from the point of view of electrocatalysis, and provide insights for improving the DMFC and DFAFC anode design. The better understanding of the synergy between electrocatalysis and electrode engineering will lead to improved fuel cell electrode design, enhanced performance and reduced cost. In addition to the application for direct fuel cell anodes, nanoparticles deposited on a 3-D substrate can be utilized in both thermochemical and other electrochemical systems, such as trickle-bed thermochemical reactors, supercapacitors and electrosynthesis reactors. 85 C024 n*H20 1) Flow Field (End Plate) 2) Diffusion Layer 3) Catalyst Layer 4) Proton Exchange Membrane CH3O + H20 HCOOH or H20 C024 n*H20 1) Flow Field (End Plate) 2) Diffusion Layer 3) Catalyst Layer 4) Proton Exchange Membrane 5) Extended Reaction Zone AnodeCH3O H20 HCOOH or Extended Reaction Zone Anode Design Figure 1-61: Conventional vs. extended reaction zone DLFC anode design. 2 3 4 3 2 Conventional Anode Design 02 5 4 3 2 02 86 1.5 References 1. World Nuclear Association, Nuclear Power in the World Today, August 2007. (http: www.worLd-nuclear.org/info/info1 .htmi) 2. BP Statistical Review of World Economy, June 2008. (http://www.bp.com!liveassets/bp_internet/globalbp/globalbp_uk_englishlreports _and_publications/statistical_energy_review_2008/STAGING/local_assets/downl oads/pdf/statistical_review_of_world_energy_full_review_2008 .pdf) 3. US Department of Energy, Report of the Methane Hydrate Advisory Committee on Methane Hydrate Issues and Opportunities, December 2002. 4. R. Dillon, S. Srinivasan, A.S. Arico, V. Antonucci, 3. Power Sources 127 (2004) 112. 5. V. Baglio, A. Di Blasi, E. Modica, P. Creti, V. Antonucci, A.S. Arico, mt. J. Electrochem. Soc. 1(2006) 71. 6. U.S. Department of Energy, August 2008. (http://wwwl .eere.energy. gov/hydrogenandfuelcells/fuelcells/pdfs/fc_comparison chart.pdf) 7. J. Larminie & A. Dicks, “Fuel Cell Systems Explained Second Edition”, John Wiley & Sons Ltd., London (2003). 8. Smithsonian Institute, National Museum of American History, September 2007. (http://americanhistory.si.edu/fuelcells/) 9. G. Wand, “Fuel Cell History — Part 2”, January 2007. (http://www.fuelcelltoday.comlmedialpdf/archive/Article_1 1 52_Fuel%2OCell%2 OHistory%20part%202%20with%20illustrations.pdf) 10. P. Zegers, J. Power Sources 154 (2006) 497. 11. P. Choi, N.H. Jalani, R. Datta, J. Electrochem. Soc. 152 (2005) E123. 87 12. M. Eikerling, A.A. Kornyshev, A.M. Kuzenetsov, J. Ulstrup, S. Waibran, J. Phys. Chem. B 105 (2001) 3646. 13. A.A. Kornyshev, A.M. Kuzenetsov, E. Spohr, J. Ulstrup, J. Phys. Chem. B 107 (2003) 3351. 14. R. Borup et. a!., Chem. Rev. 107 (2007) 3904. 15. J. St-Pierre, D.P. Wilkinson, S. Knights, M.Bos, J. New Mater. Electrochem. Syst. 3 (2000) 99. 16. M.V. Lauritzen, P. He, A.P. Young, S. Knights, V. Colbow, P. Beattie, J. New Mater. Electrochem. Syst. 10 (2007) 143. 17. L. Gancs, B.N. Hult, N. Hakim, S. Mukerjee, Electrochem. Solid-State Left. 10 (2007) B 150. 18. W. Zittel, R. Wurster, HyWeb: Knowledge - Hydrogen in the Energy Sector (1996). 19. E. Chen, “Fuel Cell Technology Handbook - Chapter 2 History”, 2003. (http://www.unibo.edu.ar/biblioteca/digital/archivos/montanari/Corso%2oArgenti na%2005/3%20- %20Produzione%2Odi%20energia%2Oda%20celle%20a%2ocombustibile/Approf ondimenti/Fuel%2OCell%20Technology%20Handbook!0877-Ch02 .pdf) 20. B.D. McNicol, D.A.J. Rand, K.R. Williams, J. Power Sources 83 (1999) 15. 21. A.S. Arico, S. Srinivasan, V. Antonucci, Fuel Cells 1(2001)133. 22. Z. Qi, A. Kaufman, J. Power Sources 110 (2002) 177. 23. Y.W. Rhee, S.Y. Ha, R.I. Masel, J. Power Sources 117 (2003) 35. 24. K.J. Jeong, C.M. Miesse, J.H. Choi, J. Lee, J. Han, S.P. Yoon, S.W. Nam, T.H. Lim, T.G. Lee, J. Power Sources 168 (2007) 119. 25. L.J. Zhang, Z.Y. Wang, D.G. Xia, J. Alloys Compd. 426 (2006) 268. 88 26. X. Li, I.M. Hsing, Electrochim. Acta 51(2006) 3477. 27. J.H. Choi, K.J. Jeong, Y. Dong, J. Han, T.H. Lim, J.S. Lee, Y.E. Sung, J. Power Sources 163 (2006) 71. 28. A. Lam, D.P. Wilkinson, 44th Annual Conference of Metallurgists (C0M2005), Proceedings of the First International Symposium on Fuel Cells and Hydrogen Technologies, Editor D. Ghosh, pg. 19 - 33 , Calgary, Alberta, August 2 1-24, 2005. 29. H.L. Tang, M. Pan, S.P. Jiang, R.Z. Yuan, Mat. Lett. 59 (2005) 3766. 30. A. Kuver, K. Potje-Kamloth, Electrochim. Acta 43 (1998) 2527. 31. J. Kerres, w. Cui, R. Disson, W. Neubrand, J. Membr. Sci. 139 (1998) 211. 32. J. Kerres, A. Ulrich, F. Meier, Solid State lonics 125 (1999) 243. 33. H. Tang, P.N. Pintauro, Q. Gup, S. O’Connor, J. Appl. Polym. Sci. 71(1999) 387. 34. D.J. Jones, J. Rozierre, J. Membr. Sci. 185 (2001) 41. 35. R. Jiang, H.R. Junz, J.M. Fenton, J. Electrochem. Soc. 153 (2006) A1554. 36. P. Argyropoulos, K. Scott, W.M. Taama, J. Appl. Electrochem. 29 (1999) 661. 37. P. Argyropoulos, K. Scott, W.M. Taama, Electrochim. Acta 44 (1999) 3575. 38. K. Scott, P. Argyropoulos, P. Yiannopoulos, W.M. Taama, J. Appi. Electrochem. 31(2001) 823. 39. R.G. Allen, C. Lim, L.X. Yang, K. Scott, S. Roy, J. Power Sources 143 (2005) 142. 40. A.J. Bard & L.R. Faulkner, “Electrochemical Methods: Fundamentals and Applications”, John Wiley & Sons, New York (2000). 41. J.A.V. Butler, Trans. Faraday. Soc. 19(1924)729. 89 42. T. Erdey-Gruz, M. Volmer, Physik. Chem. 15A (1930) 203. 43. E.L. Gyenge, “CHBE 477: Fuel Cells and Electrochemical Engineering Notes”, 2004. 44. I. Becerik, F.Kadirgan, Turk. J. Chem. 25 (2001) 373. 45. N.A. Tapan, J. Prakash, Turkish J. Eng. Env. Sci. 29 (2005)95. 46. M. Metikos-Hukovic, R. Babic, Y. Piljac, J. New Mater. Electrochem. Syst. 7 (2004) 179. 47. C. Bock, M.A. Blakely, B. MacDougall, Electrochim. Acta 50 (2005) 2401. 48. J. Jiang, A. Kucernak, J. Electroanal. Chem. 520 (2002) 64. 49. A.V. Tripkovic, K.D. Popovic, J. Lovic, J. Serb. Chem. Soc. 68 (2001) 849. 50. G.Q. Lu, A. Crown, A. Wieckowski, J. Phys. Chem. B 103 (1999) 9700. 51. M.D. Macia, E. Herrero, J.M. Feliu, J. Electroanal. Chem. 554-555 (2003) 25. 52. A. Frumkin, Z. Phys. Chem. 116 (1925) 466. 53. A.S. Arico, E. Modica, P. Creti, P.L. Antonucci, V. Antonucci, J. New Mater. Electrochem. Syst. 3 (2000) 207. 54. G. Prentice, “Electrochemical Engineering Principles”, Prentice Hall, New Jersey (1991). 55. D. Pletcher, “A First Course in Electrode Processes”, The Electrochemical Consultancy, Hants (1991). 56. R.A. Adams, “Calculus of Several Variables Fourth Edition”, Addison Wesley Longman Ltd., Don Mills (2000). 57. C.L. Scortichini, C.N. Rilley, J. Catal. 79 (1983) 138. 58. R. Woods, J. Electroanal. Chem. Interface. Electrochem. 49 (1974) 217. 90 59. K. Yamamoto, J. Electroanal. Chem. Interface. Electrochem. 96 (1979) 233. 60. C. Bock, B. MacDougall, Y. LePage, J. Electrochem. Soc. 151 (2004) A1269. 61. K. Kinoshita, P.N. Ross, J. Electroanal. Chem. 78 (1977) 313. 62. H.A. Gasteiger, N. Markovic, P.N. Ross, E.J. Cairns, J. Phys. Chem. 98 (1994) 617. 63. J. Wintterlin, S. Volkening, T.V.W. Janssens, T. Zambelli, G. Erti, Science 278 (1997) 1931. 64. C. Green, A. Kucernak, J. Phys. Chem. B 106 (2002) 1036. 65. T. Nagel, N. Bogolowski, H. Baltruschat, J. App!. E!ectrochem. 36 (2006) 1297. 66. C. Nishihara, H. Nozoye, J. Electroanal. Chem. 386 (1995) 75. 67. M. Wunsche, H. Meyer, R. Schumacher, E!ectrochim. Acta 40 (1995) 629. 68. Y. Zhang, L. Huang, T.N. Arunagir!, 0. Ojeda, S. F!ores, 0. Chyan, R.M. Wa!lace, Electrochem. So!id-State Lett. 7 (2004) C 107. 69. M.S. McGovern, E.C. Garnett, C. Rice, R.I. Masel, A. Wieckowski, J. Power Sources 115 (2003) 35. 70. M. Schuize, A. Schneider, E. Gu!zow, J. Power Sources 127 (2004) 213. 71. S.Q. Song, Z.X. Liang, W.J. Zhou, G.Q. Sun, Q. Xin, V. Stergiopoulos, P. Tsiakaras, J. Power Sources 145 (2005) 495. 72. W. Vie!stich, J. Braz. Chem. Soc. 14 (2003) 503. 73. D. Cao, G.Q. Lu, A. Wiechowski, J. Phys. Chem. B. 109 (2005) 11622. 74. T. Iwasita, Electrochim. Acta 47 (2002) 3663. 75. T.D. Jarvi, S. Sriramu!u, E.M. Stuve, J. Phys. Chem. B. 101 (1997) 3649. 76. H. Wang, T. Loffler, H. Baltruschat, J. Appi. Electrochem. 31(2001) 759. 91 77. C. Korzeniewski, C.L. Childers, J. Phys. Chem. B. 102 (1998) 489. 78. P. Liu, J.K. Norskov, Fuel Cells 1 (2001) 192. 79. H.A. Gasteiger, N. Markovic, P.N. Ross, E.J. Cairns, J. Electrochem. Soc. 141 (1994) 1795. 80. M. Umeda, H. Ojima, M. Mohamedi, I. Uchida, J. Power Sources 136 (2004) 10. 81. L. Jiang, G. Sun, X. Zhao, Z. Zhou, S. Yan, S. Tang, G. Wang, B. Zhou, Q. Xin, Electrochim. Acta 50 (2005) 2371. 82. J. Jiang, A. Kucernak, J. Electroanal. Chem. 533 (2002) 153. 83. J. Jiang, A. Kucernak, J. Electroanal. Chem. 543 (2003) 187. 84. T. Hyeon, S. Han, Y.E. Sung, K.W. Park, Y.W. Koon, Angew. Chem. mt. Ed. 42 (2003) 4352. 85. S. Han, Y. Yun, K.W. Park, Y.E. Sung, T. Hyeon, Adv. Mater. 15 (2003) 1922. 86. X. Zhang, K.Y. Chan, J. Mater. Chem. 12 (2002) 1203. 87. X. Zhang, K.Y. Chan, Chem. Mater. 15 (2003) 451. 88. K.W. Park, J.H. Choi, B.K. Kwon, S.A. Lee, Y.E. Sung, H.Y. Ha, S.A. Hong, H. Kim, A. Wieckowski, J. Phys. Chem. B 106 (2002) 1869. 89. J. Zeng, J.Y. Lee, J. Power Sources 140 (2005) 268. 90. A.B. Anderson, E. Grantscharova, S. Seong, J. Electrochem. Soc. 143 (1996) 2075. 91. L. Dubau, C. Coutanceau, E. Gamier, J.M. Keger, C. Lamy, J. Appl. Electrochem. 33 (2003) 419. 92. D.B. Kang, C.K. Lee, Bull. Korean Chem. Soc. 21(2000) 87. 92 93. A.J. Dickinson, L.P.L. Carrette, J.A. Collins, K.A. Friedrich, U. Stimming, J. Appi. Electrochem. 34 (2004) 975. 94. T.T. Cheng, E.L. Gyenge, Electrochim. Acta 51(2006) 3904. 95. A. Bauer, E.L. Gyenge and C.W. Oloman, Electrochim Acta 51(2006) 5356. 96. A. Bauer, E.L. Gyenge and C.W. Oloman, J. Power Sources 167 (2007) 281. 97. D.R. Lycke, E.L. Gyenge, Electrochim. Acta 52 (2007) 4287. 98. T.T. Cheng, E.L. Gyenge, J. Appi. Electrochem. 38 (2008) 51. 99. M. Gotz, H. Wendt, Electrochim. Acta 43 (1998) 3637. 100. B. Gurau, R. Viswanathan, R. Liu, T.J. Lafrenz, E. Reddington, A. Sapienza, B.C. Chan, T.E. Mallouk, S. Sarangapani, J. Phys. Chem. B. 102 (1998) 9997. 101. A.S. Arico, P. Creti, N. Gordano, V. Antonucci, P.L. Antonucci, A. Chuvilin, J. Appl. Electrochem. 26 (1996) 959. 102. A.K. Shukia, M.K. Ravikumar, A.S. Arico, G. Candiano, V. Antonucci, N. Gordano, A. Hamnett, J. Appi. Electrochem. 25 (1999) 528. 103. K. Wang, H.A. Gasteiger, N.M. Markovic, P.N. Ross, Electrochim. Acta 41(1996) 2587. 104. J. Kua, W.A. Goddard, J. Am. Chem. Soc. 121 (1999) 10928. 105. A. Capon and R. Parsons, J. Electroanal. Chem. 45 (1973) 205. 106. Y.X. Chen, M. Heinen, Z. Jusys, R.J. Bebm, Angew. Chem. mt. Ed. 45 (2006) 981. 107. N.M. Markovic, H.A. Gasteiger, P.N. Ross, X. Jiang, I. Villegas, M.J. Weaver, Electrochim. Acta 40 (1995) 91. 93 108. M. Weber, J.T. Wang, S. Wasmus, R.F. Savinell, J. Electrochem. Soc. 143 (1996) L158. 109. M. Arenz, V. Stamenkovic, T.J. Schmidt, K. Wandelt, P.N. Ross, N.M. Markovic, Phys. Chem. Chem. Phys. 5 (2003) 4242. 110. M. Arenz, V. Stamenkovic, P.N. Ross, N.M. Markovic, Surf. Sci. 573 (2004) 57. 111. A. Miki, S. Ye, M. Osawa, Chem. Commun. 14 (2002) 1500. 112. H. Okamoto, W. Kon, Y. Mukouyama, J. Phys. Chem. B 109 (2005) 15659. 113. G. Samjeske, M.Osawa, Angew. Chem. mt. Ed. 44 (2005) 5694. 114. G. Samjeske, A, Miki, S. Ye, A. Yamakata, Y. Mukouyama, H. Okamoto, M. Osawa, J. Phys. Chem. B 109 (2005) 23509. 115. G. Samjeske, A, Miki, S. Ye, M.Osawa, J. Phys. Chem. B 110 (2006) 16559. 116. G.G. Lang, M. Seo, K.E. Heusler, J. Solid State Electrochem. 9 (2005) .3 117. J. Lee, J. Christoph, P. Strasser, M. Eiswirth, G. Erti, J. Chem. Phys. 115 (2001) 1485. 118. N. Markovic, P.N. Ross Jr., J. Phys. Chem 97 (1993) 9771. 119. D. Capon and R. Parsons, Electroanal. Chem. Interfacial Electrochem. 44 (1973) 239. 120. R. Larsen, S. Ha, J. Zakzeski, R.I. Masel, J. Power Sources 157 (2006) 78. 121. S. Blair, D. Lycke, C. lordache, Electrochem. Soc. Trans. 3 (2006) 1325. 122. X. Li, I.M. Hsing, Electrochim. Acta, 51(2006) 3477. 94 123. C. Rice, S. Ha, R.I. Masel, A. Wieckowski, J. Power Sources 115 (2003) 229. 124. J.H. Choi, K.J. Jeong, Y. Dong, J. Han, T.H. Lim, J.S. Lee, Y.E. Sung, J. Power Sources 163 (2006) 71. 125. R. Larsen, R.I. Masel, Electrochem. Solid-State LeU. 7 (2004) A148. 126. A.V. Tripkovic, K. Dj. Popovic, R.M. Stevanovic, R. Socha, A. Kowal, Electrochem. Comm. 8 (2006) 1492. 127. M.D. Macia, E. Herrero, J.M. Feliu, J. Electroanal. Chem. 554-555 (2003) 25. 128. 5. Kang, J. Lee, J.K. Lee, S.Y. Chung, Y. Tak, J. Phys. Chem. B. 110 (2006) 7270. 129. C. Roychowdhury, F. Matsumoto, P.F. Mutolo, H.D. Abruna, F.J. DiSalvo, Chem. Mater. 17 (2005) 5871. 130. L.J. Zhang, Z.Y. Wang, D.G. Xia, J. Alloys Compd. 426 (2006) 268. 131. M. Campanati, G. Fornasari, A. Vaccari, Catal. Today 77 (2003) 299. 132. K.J. Smith, CHBE 563 Course Notes (2004). 133. K. Mauritz, Sol-Gel Chemistry (2004) (http://www.psrc.usm.edulmauritz/solgel.html) 134. J.Y. Kim, Z.G. Yang, C.C. Chang, T.I. Valdez, S.R. Narayanan, P.N. Kumta, J. Electrochem. Soc. 150 (2003) A1421. 135. O.A. Shilova, V.V. Shilov, N.D. Koshel, E.V. Koziova, Glass Phys. Chem. 30 (2004) 98. 136. E.J. McLeod, V.1. Birss, Electrochim. Acta 51(2005) 684. 95 137. H. Grisaru, 0. Paichik, A. Gedanken, V. Palchik, M.A. Slificin, A.M. Weiss, Inorg. Chem. 42 (2003) 7148. 138. A.K. Sra, T.D. Ewers, R.E. Schaak, Chem. Mater. 17 (2005) 758. 139. W. Yu, W. Tu, H. Liu, Langmuir 15 (1999) 6. 140. Z. Liu, L. Hong, M.P. Tham, T.H. Lim, H. Jiang, J. Power Sources 161 (2006) 831. 141. C. Goia, E. Matijevic, D. Goia, J. Mater. Res. 20 (2005) 1507. 142. Z.C. Ore!, E. Matijevic, D.V. Goia, J. Mater. Res. 18 (2003) 1017. 143. H. Bonnemann, W. Brijoux, R. Brinkmann, E. Dinjus, T. Jouben, B. Korall, Angew. Chem. Tnt. Ed. Engi. 30(1991)1312. 144. H. Bonnemann, P. Britz, W. Vogel, Langmuir 14 (1998) 6654. 145. H. Bonnemann, K.S. Nagabhushana, J. New Mater. Electrochem. Syst. 7 (2004) 93. 146. H. Bönneman, R.M. Richards, Eur. J. Inorg. Chem. (2001) 2455. 147. H. Bönneman, R.M. Richards, Synth. Met. Org. Inorg. Chem. 10 (2002) 209. 148. T.J. Schmidt, M. Noeske, H.A. Gasteiger, R.J. Behm, P. Britz, W. Bjoux, H. Bönneman, Langmuir 13 (1997) 2591. 149. R.C. Koffi, C. Coutanceau, E. Gamier, J.M. Leger, C. Lamy, Electrochim. Acta 50 (2005) 4117. 150. M.T. Reetz, S.A. Quaiser, Angew. Chem. Tnt. Ed. Engi. 34 (1995) 2240. 151. M.T. Reetz, W. Helbig, J. Am. Chem. Soc. 116 (1994) 7401. 152. R.S. Jayashree, J.S. Spendelow, J. Yeom, C. Rastogi, M.A. Shannon, P.J.A. Kenis, Electrochim. Acta 50 (2005) 4674. 96 153. Kh. Saber, C.C. Koch, P.S. Fedkiw, Mater. Sci. Eng., A 34 (2003) 174. 154. R.T.C. Choo, J.M. Toguri, A.M. El-Sherik, U. Erb, J. App!. Electrochem. 25 (1995) 384. 155. J.V. Petrocelli, ACS Fuels 11(1967)24. 156. J.O’M. Bockris, A.K.N. Reddy, “Modem Electrochemistry — 2”, Plenum Press, New York (1970). 157. B.E. Conway, J.O’M. Bockris, Electrochim. Acta 3 (1960) 340. 158. J.M. Elliott, G.S. Attard, P.N. Bartlett, N.R.B. Coleman, D.A.S. Merckel, J.R. Owen, Chem. Mater 11(1999) 3602. 159. J.M. Elliott, P.R. Birkin, P.N. Bartlett, 0.5. Attard, Langmuir 15 (1999) 7411. 160. V. Ganesh, V. Lakshminarayanan, Electrochim. Acta 49 (2004) 3561. 161. M. Boutonnet, J. Kizling, P. Stenius, Colloids Surf. 5 (1982) 209. 162. C.H. Lu, H.C. Wang, H. J. Electrochem. Soc. 152 (2005) C341. 163. J. Kizling, M. Boutonnet-Kizling, P. Stenius, R. Touroude, G. Maire, Electrochem. Colloids Dispersions (1992) 333. 164. X. Zhang, K.Y. Chan, J. Mater. Chem. 12 (2002) 1203. 165. X. Zhang, K.Y. Chan, Chem. Mater. 15 (2003) 451. 166. L.G.R.A. Santos, C.H.F. Oliveira, I.R. Moraes, E.A. Ticianelli, J. Electroanal. Chem. 596 (2006) 141. 167. G.S. Chai, S.B. Yoon, J.S. Yu, J.H. Choi, Y.E. Sung, J. Phys. Chem. B 108 (2004) 7074. 168. D. P. Wilkinson, M.C. Johnson, K.M. Colbow and S.A. Campbell, US Patent 5,874,182, February 13 (1999). 97 169. Z.G. Shao, F. Zhu, W.F. Lin, P.A. Christensen, H. Zhang, Phys. Chem. Chem. Phys. 8 (2006) 2720. 170. Z.G. Shao, F. Zhu, W.F. Lin, P.A. Christensen, H. Zhang, Inter. J. Hydrogen Energy 31(2006)1914. 171. Z.G. Shao, F. Zhu, W.F. Lin, P.A. Christensen, H. Zhang, B. Yi, J. Electrochem. Soc. 153 (2006) A1575. 172. R. Chetty, K. Scott, J. New Mater. Electrochem. Syst. 10 (2007) 135. 98 2 RESULTS AND DISCUSSION: ELECTRODEPOSITION OF Pt-Ru ON RVC FROM REVERSE EMULSIONS AND MICROEMULSIONS* 2.1 Introduction The synthesis of nano-sized metal particles is an extremely active area of research targeting a variety of applications including electrocatalysts for diverse electrochemical reactions relevant for fuel cells (e.g. 02 electroreduction, electro-oxidation of CH3O , C2H5OH). At present, colloidal particles (1 to 50 nm) well-dispersed on carbon supports are arguably the most common form of electrocatalyst structure used in the gas diffusion electrodes of fuel cells. Synthetic methods of metal colloids typically involve the formation of protective shells to avoid particle agglomeration (by electrostatic and/or steric effects), followed by adsorption on fine carbon particle supports and finally removal of the shell by a combination of solvent washing and heat treatment methods. One successful variant applied to a variety of metals (e.g. Pt, Ru, Au, Sn, Co, Os, Ir) and some of their alloys has been developed by BOnneman and co-workers, and it is based on the formation of organosols using cationic surfactants such as trioctylammonium hydroborate [1-4]. After removal of the particle stabilizer shell, the carbon-supported catalyst slurry containing controlled amounts of a hydrophobic agent (e.g. Teflon) and the liquid form of the polymer electrolyte (e.g. Nafion) is applied conventionally by mechanical techniques such as painting, or spraying onto either the membrane or gas diffusion layer (various carbon papers or cloth) to form thin (about 5 to 50 p.m) fuel cell catalyst layers bonded by hot pressing to the polymer electrolyte membrane and gas diffusion layer. Liquid crystal templated formation of nanostructured (e.g. mesoporous) fuel cell catalyst layers (e.g. Pt, Pt-Ru) by electrochemical or chemical reduction has also been investigated with respect to electrocatalytic activity toward both CH3O oxidation and 02 reduction [5-9]. It must be noted that liquid crystal templating has been mostly applied * A version of this chapter has been published: T.T. Cheng, E.L. Gyenge, “Electrodeposition of Mesoseopic Pt-Ru on Reticulated Vitreous Carbon from Reverse Emulsions and Microemulsions: Application to Methanol Electro-Oxidation”, Electrochim. Acta 51(2006)3904-3913. 99 for deposition on flat substrate surfaces (e.g. Au and glassy carbon); consequently, the resulting catalysts were not tested in fuel cells. Thus, the gas diffusion electrode with supported nano-sized catalyst is the dominant technology for electrochemical reactions of gases. While it has been adopted virtually in the same design for use as anodes in direct liquid fuel cells (e.g. methanol and ethanol), the applicability of the gas diffusion design for the latter application has been recently called into question and the possibility of using anodes which could be called, based on the accepted nomenclature, as three dimensional electrodes has been proposed [10, 11]. Scott and collaborators explored Ti mesh as the anode for direct methanol fuel cells with Pt-Ru made by electrodeposition and thermal decomposition from aqueous phase [101, whilst Gyenge and co-workers investigated the liquid crystal templated electrodeposition of Pt-Ru nanoparticles on graphite fibers (thickness between 300 and 200 .im) and the applicability as direct methanol fuel cell anodes [11]. The three- dimensional anode concept presents the opportunity to create an extended reaction zone for the oxidation of the fuel thereby minimizing the fuel crossover rate to the cathode and improving the CO2 disengagement. However, it poses the challenge of creating both a uniform catalyst deposit and ionic conductor network across the thickness of the electrode. Campbell and co-workers used templating by polyethylene oxide nonionic surfactants to electrodeposit mesoporous Ni films (pore to pore distance of 7.5 nm) on foamed (i.e. reticulated) Ni [12]. Whilst the transfer of the bulk templating pattern (e.g. hexagonal surfactant phase) to the surface deposit was not conclusively established by experimental means, the templated deposit showed a three-times higher surface area compared to the undeposited Ni foam. Alternatively to surfactant-based particle stabilization and templating, reverse (water-in-oil) microemulsions stabilized by non-ionic surfactants could be of great interest for nanoparticle synthesis due to the possibility of carrying out the material synthesis in the sub-micron size aqueous droplets [13,14]. Employing reverse microemulsions of cyclohexane and water stabilized by Triton X-100 and isopropanol, Pt-Ru and Pt-Co nanoparticles of 2.5 to 4.5 nm diameter were prepared by chemical 100 reduction using a second microemulsion of the same composition but containing hydrazine in the aqueous phase [15, 16]. The goal of the present investigation was to study the electrodeposition of Pt-Ru on RVC using both reverse emulsion and microemulsion systems with the aim of developing novel three-dimensional anodes for direct methanol fuel cells. Furthermore, RVC with nanostructured catalysts could be useful for other applications besides fuel cells such as electrosynthesis and heterogeneous thermochemical processes using trickle beds. 2.2 Experimental Procedure 2.2.1 RVC Pretreatment The RVC used in the present work had 39 pores per centimeter (ppc) and a specific surface area of 6600 m2 m3 supplied by Electrolytica Inc. The surface pretreatment of various carbon specimens can have a major role on the electrodeposition. Table 1 summarizes the explored RVC pretreatment methods. In the first pretreatment method, the RVC was washed in dc-ionized water for 5 minutes, methanol for 5 minutes, and immersed in 35 vol% nitric acid at 333 K for 30 minutes. In the second method, the RVC was immersed in a PdC12 + SnC12 solution for 1 minute after pretreatment method 1. The PdC12 + SnC12 solution used was a Shipley type solution, and was prepared by mixing 0.1 g of PdC12 (Sigma-Aldrich), 5 g of SnC12HO(Sigma-Aldrich), 60 mL of de-ionized water and 30 mE of concentrated hydrochloric acid (Fisher Scientific) [17]. See Table 2-1. The last pretreatment method was electrochemical, involving potential cycling of the RVC between 1.44 to 2.09 V vs. SHE in concentrated (98%) H2S04 solution at a scan rate of 0.001 V s1 repeated fifty times [18]. The pretreated RVC samples (5 cm2 geometric area, thickness 2 mm) were washed thoroughly with de-ionized water followed by drying in air. 101 Table 2-1: RVC pretreatment methods. Pretreatment Immersion in 35 Immersion in 6 mM Electrochemical Cycling Method vol% HNO3 at 333 PdC12,0.3 M SnC12, in 98% H2S04 K for 30 minutes and 4 M HC1 solution for 1 minute 1 + - - 2 + + - 3 - - + 2.2.2 Emulsion and Microemulsion Preparation for Electrodeposition The emulsion and microemulsion systems were prepared using as organic phase cyclohexane (Fisher Scientific) with 0.001 M tetrabutylammonium perchiorate (TBAP, [(CH3C2)4N 1O])to impart ionic conductivity. The surfactant was Triton X 100 [14H21O(CI-LO)9.5] (Sigma-Aldrich) while iso-propanol (Sigma-Aldrich) was also added as co-surfactant for microemulsification. The aqueous phase was composed of equimolar 0.01 M H2PtC16 (Sigma-Aldrich) and RuC13 (Sigma-Aldrich) or (NH4)2RuC16 solution. All chemicals used were reagent grade and were used as delivered without further purification processes. The emulsions were prepared by mixing the aqueous and organic phases with the surfactant in a water-jacketed glass vessel connected to a circulating water bath at 323, 333, 341, or 343 K for 30 minutes. The electrode assembly composed of the RVC working electrode (5 cm2) sandwiched at a distance of 2 mm between two Pt/Ti counter electrodes (5 cm2 geometric area each) was inserted into the glass vessel after no visible change was observed in the emulsion for a period of 15 minutes. The electrode assembly was then connected to a dc power supply to carry out the electrodeposition (see Figure 2-1). 102 Deposnc Bath Contr E1cctrodc SubsLrrse (,zedTitauu) Figure 2-1: Schematic of electrodeposition vessel and electrode assembly. After the deposition experiment, the RVC was sonicated in THF (Reagent Grade, Sigma Aldrich) for 5 minutes to break the emulsions retained in the porous matrix [19]. The deposited RVC was then washed thoroughly with deionized water and dried. Some of the RVC supports were then heat-treated in a N2 stream for 1 hour at 573 K to remove traces of adsorbed organic compounds. A partial phase diagram for the ternary system containing the mixed surfactant phase (1:4 volume ratio of Triton X- 100 and isopropanol), cyclohexane, and the aqueous Wer Flow 103 phase was established to determine the microemulsion region. The emulsion and microemulsion boundary was determined by titrating the aqueous solution into a mixture of a given cyclohexane-to-surfactant ratio as described by Lu and Wang [201. The composition at which a visual transition between opaque and transparent behavior occurred was recorded and it was attributed to microemulsification. 2.2.3 Electrochemical Measurements Voltammetry and CP were carried out at 298 K in a water-jacketed electrochemical cell connected to a circulating water bath. The experiments were carried out using the computer-controlled VoltaLab PGZ4O2 potentiostat with the VoltaMaster 4 software (Radiometer Analytical). The three-electrode setup was used with a 1 cm2 geometric area plated RVC working electrode, Hg/Hg2SO4K2S04,std.(MSE) reference electrode and a platinum wire counter electrode. All potentials in the present work are reported against the SHE reference. Before the voltammetry and CP experiments, the catalyst surface was chemically and electrochemically pre-treated to remove surface impurities. The working electrodes were first immersed in a 1:1 v/v. concentrated H2S04 and 30 vol% H20 (Fisher Scientific) solution five times for a few seconds each, and rinsed thoroughly with deionized water. Afterwards, using 0.5 M H2S04 potential steps for 10 s each were applied to clean the surface at 1.28 V, 1.20 V and 0.05 V, respectively. The electrochemical pretreatment steps were repeated three times. The methanol electro oxidation voltammetry and CP experiments were carried out in a 50 mL solution of 1.0 M CH3O and 0.5 M H2S04. The active surface area of the electrocatalysts was determined by the Cu UPD, which was shown by Kucernak and Green to be a promising method for Pt-Ru electrocatalysts [211. The working electrodes were prepared with the same method as for the electrochemical tests. First reference voltammograms were obtained in 0.5 M H2S04 with cycles between —0.04 to 0.91 V at a scan rate of 0.050 V s. The Cu UPD experiments were carried out in a 0.5 M H2S04 and 0.002 M CuSO4 solution. The working electrodes were polarized at 0.26 V for 300 seconds to form a monolayer of 104 copper on the catalyst surface. A linear voltammetric scan with a scan rate of 0.050 V s1 was then performed from 0.26 to 0.91 V to remove the adsorbed copper monolayer [21, 22]. The charges obtained for the copper stripping were corrected for the charges associated with background processes and oxide growth by subtracting the charge obtained from the reference scan in the same potential range. All Cu UPD measurements were performed at room temperature (298 K). Whether Cu could deposit onto RVC was evaluated to determine the feasibility of utilizing this surface area determination method. Figure 2-2 below compares the anodic scans obtained on blank (i.e. pure) RVC after it was subjected to Cu deposition for 300 s at 0.06 V vs. SHE (bulk deposition) and 0.26 V vs. SHE (Cu UPD), respectively. In the case of the underpotential deposition, there was no anodic stripping current observed indicating the absence of Cu underpotential deposition on vitreous carbon. Therefore, the anodic stripping currents observed for Cu UPD experiments were due exclusively to the oxidation of Cu underpotentially deposited on the Pt-Ru sites. As noted in Section 1.2.5.4, there is currently no perfect method for determining the surface area of Pt-Ru catalysts. Therefore, there exists no validation tool for the Cu UPD method. However, from the experimental observations, it was found to be a feasible method to systematically estimate the surface area of the prepared catalysts. 100 ____________________________________ Blank Run: Blank RvC so. ——— CuUPD:6lankRVC — — — — Cu Bulk Deposition: Blank RVC 60 1 Blank scan of RVC in j 40 0.6MHSO 1 20 dicscanfromblankRVC Anodic scan from blank RVC subjected to Cu UPD -40 subjected to Cu bulk deposition -0.2 0.0 0.2 0.4 0.6 0.6 1.0 E/Vvs. SHE Figure 2-2: Anodic scans obtained on blank RVC subjected to Cu bulk deposition and UPD. Temperature: 298 K; Scan rate: 0.050 V s. 105 2.2.4 Electrical Conductivity Measurements The electrical conductivity of the emulsions was measured using a Thermo Orion 1 O5Aplus conductivity meter. The conductivity probe was immersed in the solutions for 30 seconds before recording to conductivity measured. The conductivity probe was calibrated using a 0.01 M KC1 standard solution supplied by Thermo Orion. 2.2.5 Surface and Analytical Characterization of Catalysts Images of the catalyst surfaces were obtained by SEM using a Hitachi S4700 SEM. RVC fragments from the samples were flush mounted with carbon stickies on SEM stubs. The visual images were generated using a beam accelerating voltage of 2000 V and an emission current of 1 .25x 1 0 A at a working distance of 0.0025-0.0035 m. The mass loading of the catalysts was obtained by inductively coupled plasma atomic emission spectroscopy (ICP-AES) using a Perkin Elmer Optima, model 3300DV instrument. A 1 cm2 piece of the catalyst substrate was weighed and digested in aqua regia (HC1-HNO3volume ratio of 3:1) for four hours to solubilize the Pt and Ru metal deposits. The Pt and Ru content in the solution was determined by ICP-AES, which yielded the mass loading and bulk Pt:Ru ratio of the deposits. 2.3 Results and Discussion 2.3.1 Electrodeposition from Reverse emulsion 2.3.1.1 The Effect of Temperature The temperature at which the electrodeposition experiments are carried out can have a significant impact on the deposition process not only by affecting the kinetics of the various deposition steps (e.g. nucleation, growth) but also by influencing the phase behavior and stability of the emulsion. The goal here was to perform the electrodeposition in a water-in-oil (W/O) emulsion (also referred to as reverse emulsion) to obtain discrete and dispersed metal deposits. The emulsion was composed of 90 vol% 106 cyclohexane with 0.001 M TBAP, 7 vol% Triton X-100, and 3 vol% aqueous phase with H2PtC16and RuC13,each in 0.01 M concentration. To gain preliminary insights into the phase behavior and to determine the phase inversion temperature, the electrical conductivity of the emulsion was measured between 319 and 345 K. Figure 2-3 shows that the electrical conductivity of the emulsion within the temperature range of 319 to 327 K was approximately constant, between 3.2 to 2.1x104 S m’. Once a temperature of 329 K was reached, an abrupt increase in conductivity was observed, from 2.1x104 S m’ to 7.8x103 S m1. The conductivity increased until a maximum of 1.0x102 S m1 was reached at 333 K followed by an abrupt decrease in conductivity between 335 to 337 K, when the conductivity dropped from 9.5x103to 1.9x103 S m1. It is proposed that 333 K corresponds to the phase transition from mostly oil continuous (W/O) to water continuous (0/W) emulsion, which explains the high electrical conductivity [23]. Interestingly, at 343 K a local conductivity maximum was obtained followed by a sharp decrease between 343 to 344 K, from 2.1 x103 S m1 to 3.0x105 S m1. Possibly at 343 K, which exceeded the cloud point temperature of Triton X- 100 (339 K) [24], the dehydration of Triton X- 100 has occurred [25]. The cloud point temperature of a surfactant is defined to be the temperature at which the surfactant and aqueous phase begin to separate. Due to the dehydration of Triton X-100, the hydrophilic-lipophilic balance was first re-established with a bicontinuous phase with a weaker or less interactive aqueous network, resulting in a lower conductivity compared to that at 333 K, before re-establishing as a W/O emulsion at temperatures at and above 343 K. The differences in electrical conductivity and the abrupt transitions clearly suggest that the emulsion systems had very different phase structures as a function of temperature ranges [23, 26]. 107 Cl) > 0 C-) U Li uJ 0.012 0.010 0.008 0.006 0.004 0.002 w/o 0.000 Figure 2-3: Electrical conductivity of 7 vol% Triton X-l00I 90 vol% cyclohexane I 3 vol% water emulsion at a temperature range of 319 to 345 K. Pt-Ru electrodeposition experiments were carried out at 26 A m2 on nitric acid pretreated RVC samples (Table 1) at 323, 333 and 343 K. Figure 2-4 shows the voltammograms for methanol oxidation obtained from the resulting Pt-Ru catalysts. Interestingly, the only sample which showed activity was the one deposited at 343 K. It is likely that the phase structure at 343 K was favorable not only from the point of view of deposition kinetics but also regarding the wetting of the RVC surface to induce the adsorption and nucleation of the Pt(IV) and Ru(III) ions. 320 325 31 335 340 345 Temperature / K 108 250 200 A Deposition Temperature: 23 K — Deposition Temperature: 333 1< __ — — Deposition Temperature: 343 K 150. / E c/”100 V 05 0.6 0 0.9 E/ V vs. SHE Figure 2-4: Voltammograms of methanol oxidation at 298 K on Pt-Ru/RVC in 1 M CH3O and 0.5 M H2S04.Effect of temperature during catalyst electrodeposition via emulsion with a superficial deposition current density and time of 26 A m2 and 60 minutes. Current density is given in A m2 geometric area basis. Scan rate: 0.005 Vs1. 2.3.1.2 The Effect ofDeposition Current Density The effect of deposition superficial current density on the mass loading and activity of the catalysts was studied. Three electrodeposition experiments were performed in the emulsion media at 343 K, but at different deposition current and time, such that the electric charge passed during deposition was constant 1 .4x1 C m2. Thus, the deposition superficial current densities were 10, 26, and 40 A m2 with a deposition time of 234, 90, and 59 minutes, respectively. 109 Figure 2-5 shows the SEM images of the three samples under consideration. Both the particle size and deposit structures of the samples were very different. Deposition at 10 A m2 yielded a coating composed of 10-50 nm particles together with a few larger aggregates (Figure 2-5 a). The large clusters were found to be rare and they appeared on random locations throughout the RVC. In Figure 2-5b, the sample prepared at 26 A m2 is shown. The morphology can be characterized as an interconnected mesoporous network with pore diameters of approximately 10 nm. Deposition at 40 A m2 yielded a sparsely covered surface with large aggregates approximately 500 to 2000 nm in size (Figure 2- Sc). The activity of the catalysts characterized in Figure 2-5 was studied with respect to methanol electro-oxidation by voltammetry (Figure 2-6). The highest methanol electro-oxidation superficial current densities were obtained with the catalyst electrodeposited at 26 A m2 corresponding to the mesoporous coating, whilst deposition at 40 A m2 yielded virtually no activity. Thus, the voltammetry results are corroborating the morphology as revealed by high resolution SEM. 110 Figure 2-5: Effect of current density on the Pt-Ru morphology electrodeposited on RVC pretreated in HNO3 at 343 K using W/O emulsion with constant deposition charge. (a) Deposition current and time of 10 A m2 and 234 minutes; (b) Deposition current and time of 26 A m2 and 90 minutes; (c) Deposition current and time of 40 A m2 and 59 minutes. 111 250 200 150 E 100 50 0 0.3 0.9 E/ V vs. SHE Figure 2-6: Voltammograms of methanol oxidation on Pt-RuJRVC in 1 M CH3O and 0.5 M H2S04. Effect of electrodeposition current density and time at 343 K. Current density is given in A m2 geometric area basis. Temperature: 298 K. Scan rate: 0.005 V s1. 2.3.1.3 Effect ofRu Precursor and Pt:Ru Ratio Initially, in addition to H2PtC16,RuC13 was used in the aqueous phase of the emulsion for Ru deposition. However, in spite of the equimolar RuC13 : H2PtC16 ratio in the aqueous phase, the ICP-AES analysis of the samples showed that smaller amounts of Ru was deposited compared to Pt. To investigate if this finding was brought about by the emulsion electrolyte or by the Ru salt component of the aqueous phase, an experiment was performed where (NH4)2RuC16in equimolar ratio with H2PtC16 (i.e. 0.01 M) was used in the aqueous phase. The rest of conditions were identical to the best established 0.4 0.5 0.6 0.7 o.e 112 Econditions in the previous section, i.e. superficial current density 26 A m2, for 90 mm at 343 K. The mass loading of the sample prepared with (NH4)2RuC16was found to be 2.5 g m2, compared to 1.5 g m2 for the sample prepared with RuCI3 and the bulk Pt:Ru atomic ratio was 0.8:1. Thus, using (NH4)2RuC16decreased the Pt:Ru ratio by improving the Ru deposition. It was observed that (NHRuCl dissolved completely in the aqueous phase while RuC13 formed a colloidal dispersion. The improved Ru content of the deposit is also supported by the CV of methanol electro-oxidation comparing the two deposits, prepared with RuC13 and (NH4)2RuC16, respectively (Figure 2-7). The onset oxidation potential of the sample prepared with (NH4)2RuC16was found to be about 0.2 V more negative than the sample prepared with RuC13,showing clearly the ruthenium effect on methanol electro-oxidation [27-29]. 1.5 q m Pt-Ru (7.3:1 Aomic Ratio) Deposion Current Densfty: 2. A rn Deposion Time: 90 mm Ruthenium Precursor: RuCI3 — — — 2.59 m Pt-Ru (0.0:1 .4omic Ratio) Deposion Current Densy: 26 A mC Deposion Time: 90 mm Ruthenium Precursor: (NHjRuCI 400 300 200 100 0 Onset Potentials I I I I I 0.4 0.5 0.6 0.7 0.00.3 0.9 E/Vvs. SHE Figure 2-7: Voltammograms of methanol oxidation on Pt-RuIRVC in 1 M CH3O and 0.5 M H2S04 at 298K. Effect of ruthenium precursor (RuCl3 vs. (NH4)2RuCl6. Electrodeposition was carried out at 343 K with a current density and time of 26 A m2 and 90 minutes. Scan rate: 0.005 V s1. Current density is given in A m2 geometric area basis. 113 2.3.2 Effect of Isopropanol on Solution Phase Behavior: Microemulsification As shown in the previous section, the electrodeposition of Pt-Ru is not very favorable in the emulsion system as can be seen from the very low mass load (only up to 2.5 g m2). Therefore, it is of great interest to further investigate improvements to the deposition media and conditions. The addition of isopropanol in conjunction with Triton X-100 (i.e. so-called co-surfactant addition to the surfactant in a 4:1 volume ratio) can lead to microemulsification for certain composition ranges, as shown by the phase diagram determined by titration at 341 K and presented in Figure 2-8. In the resulting reverse microemulsion system (isotropic in appearance), isopropanol acted as a co-surfactant and helped stabilize the aqueous phase in the continuous oil phase [15,16]. The aqueous phase contained 0.01 M H2PtC16 and (NH4)2RuC16,each. In Figure 2-8 the curve marks the microemulsion-emulsion boundary and the system is optically transparent under the curve. The microemulsion used in the electrodeposition experiments had a composition of 72 vol% cyclohexane (with 0.001 M TBAP), 25 vol% mixed surfactant, and 3 vol% aqueous phase. A low aqueous volume fraction was used since the water droplet size would be smaller [14,191, and could result in smaller deposit particles. The reverse nature of the microemulsion (i.e. W/O) is also shown by its very low ionic conductivity 20 1iS m1. 114 Aqueous o.cA.I.0 0.4J y ‘y \ x \0.2 o.’ S’°/ ‘•x Y ‘ ‘ \ Mixed io V V V V V Triton X.i0 0.0 0.1 0.2 0.3 OA 0.5 0.6 0.7 0.8 0.9 1.0 cydohexane Is opropano I (1:4 Volume Ratio) Figure 2-8: Partial phase diagram of the cyclohexane, Triton X-100/isopropanol, and aqueous phase with 0.01 MH2PtC16and (NH4)2RuCL5ternary system at 341 K. 2.3.3 Electrodeposition from Reverse Microemulsion Due to the extremely low ionic conductivity of the reverse microemulsion, the highest superficial current density that could be utilized for electrodeposition was 10 A m2 (i.e. a total current of 5 mA with a corresponding applied voltage of 152 V). The deposition time was therefore, 234 minutes to have the same total charge density as in the emulsion deposition experiments for easy comparison. Furthermore, the electrodeposition was carried out at 341 K, since at temperatures starting at 343 K, the vapor pressure of the reverse microemulsion was too high to carry out deposition experiments for several hours. 115 Using a nitric acid pretreated RVC, the electrodeposition from microemulsion yielded a mass loading of only 0.6 g m2 with a Pt:Ru atomic ratio of 0.9:1. However, the mass-specific surface area of the catalyst determined by Cu UPD was 40 m2 g’, i.e. over three times higher than in the case of reverse emulsion deposition (12 m2 g’). It must be noted that in most cases in the present study, only one broad Cu stripping peak was observed, representative of both Pt and Ru sites. Resolving the anodic peak into the two substrate constituents (Pt and Ru) was not possible by electro-oxidative methods as suggested by Green and Kucernak [211. Figure 2-9 shows the SEM images of the Pt-Ru catalyst produced in the microemulsion system. Comparing Figure 2-9 with Figure 2-5a (emulsion deposition at the same deposition current density of 10 A m2) reveals that the proportion of about 10 nm or smaller particles was much higher for the microemulsion system, while aggregates were fewer and smaller than in the case of deposition from emulsion. Figure 2-9: The effect of microemulsion on the Pt-Ru morphology electrodeposited on RVC pretreated in 35 vol% HNO3 at 341 K with a superficial deposition current density and time of 10 A m2 and 234 minutes. 116 Figure 2-10 compares the voltammograms of the electro-oxidation of methanol on a mass activity basis for the two samples obtained by deposition from microemulsion and emulsion, respectively. The mass activity of the sample prepared by microemulsion deposition at 0.6 V was about three times higher, while the intrinsic catalytic activity based on real surface area was virtually identical, indicating that the superior catalytic activity was mainly attributed to an increase in active surface area. 350 _________________________________________ 12 2.5 gm-2 Pt-Ru (Atomic Ratio 0.8:1) Media: Reverse Emulsion 300 10 Deposition Current Density: 26 A rn-2 Deposition Time: 90mm Deposition Temperature: 343 K 260 0.6 gm-2Pt-fu (Atomic Ratio 0.9:1) E Media: Reverse M icroem ulsion Deposition Current Density: 10 A rn-2 200 4 Deposition Time: 234 mm Deposition Temperature: 341 K 150 030 0.3 0.40 0A 053 0. 100 E/ViS. SHE ‘liii E/Vvs. SHE Figure 2-10: Voltammograms of electrodeposited RVC in 1 M CH3O and 0.5 M H2S04. Effect of colloidal media composition, microemulsion vs. emulsion on the catalytic activity of Pt-RuIRVC towards methanol oxidation. Temperature: 298 K. Scan rate: 0.005 V s. Inset: Intrinsic catalytic activity based on real surface area. Compared to the emulsion media, the use of reverse microemulsion was more beneficial in the preparation of high specific surface area catalysts; therefore, the effects of RVC pretreatment and postdeposition treatment were studied using the microemulsion media. 117 2.3.4 Effect of RVC Pretreatment and Postdeposition Heat Treatment In order to increase the Pt-Ru deposit mass load using the microemulsion system, various pretreatments methods were applied to the reticulated vitreous carbon substrate (Table 2-1). Interestingly, both the mass loading and the Pt:Ru ratio differed as a function of RVC pretreatment technique (Table 2-2). Table 2-2: Mass loading and bulk Pt:Ru ratio of samples prepared by the microemulsion method. Microemulsion — 10 A m2, 234 mm, 341 K Pretreatment Method Mass Loading (g m2) Pt:Ru Atomic Ratio Nitric Acid 0.6 0.9:1 Shipley 0.1 4.7:1 Electrochemical Cycling 2.3 1.3:1 The electrochemical cycling pretreatment method yielded the highest mass loading for the microemulsion deposition method, almost four times higher compared to the deposition carried out on nitric acid treated RVC i.e. 2.3 g m2. The Pt:Ru ratio was 1.3:1 The increased deposit load was a result of an increased number of deposition sites due to the rougher RVC surface after electrochemical cycling. Figure 2-11 compares the SEM picture of unpretreated RVC and the one pretreated by the electrochemical method described in Table 2-1 (see Section 2.2.1). The electrochemical cycling pretreatment significantly roughened the RVC surface and the surface oxygen content increased more than six times as shown by XPS (i.e. from 4.5 to 27.4 at.%). The formation of OH—C0, —C=O, and —C—OH functional groups was observed (see Figure 2-12). It has also been previously observed by surface FTIR that electrochemical cycling could increase the oxygen content of the RVC due to the formation of—C=O and —COOH surface functional groups, resulting in a higher capacity for metal ion nucleation and deposition [181. 118 Figure 2-11: SEM images of the RVC surface. (a) No pretreatment; (b) Pretreated with electrochemical cycling from 1.44 to 2.09 V vs. SHE at 0.001 V s’ repeated 50 times in 98% H2S04. 119 R’. bok!I3 c2 Name Pos. FWHM LSh. %Area Grapitoic Carbon 28438 0.76 A(0.38,0.7,I0)GL(10) 100.0 (6 24 294 292 290 289 288 284 281 2 94990E969g () SIR FACE SCIFSCE (VEST I/RN(a) [OW 286/15 Name Pos. FWHM LSh. %Area OH-C=O 289.26 1.18 GL(30) 7.0 C=O 287.85 1.18 GL(30) 4.8 (:(j•1 286.63 1.18 GL(30) 16.1 CII 285.03 1.18 GL30) 72.2 70 00 58 40 50 20 292 290 288 286 284 282 29.) 278 Bi.d/gE,wrg (e9 (b) Figure 2-12: XPS spectra. (a) No pretreatment; (b) Pretreated with electrochemical cycling from 1.44 to 2.09 V vs. SHE at 0.001 V s1 repeated 50 times in 98% H2S04. 120 It has been proposed originally by Frumkin and co-workers that the adsorption properties of high specific surface area carbons are dependent on the surface potential [30, 311. Therefore, the surface changes induced by the electrochemical oxidation pretreatment were also reflected in the open circuit potential values of the RVC before and after electrochemical pretreatment in 0.5 M H2S04, i.e. —0.26 and +0.65 V, respectively. The same measurement has been used by Goldin et al. to characterize the surface state of various activated carbon particles [30, 311. The higher OCV observed indicates an increased amount of oxidized species on the carbon surface, in agreement with the XPS results shown in Figure 2-12. Compared to the electrochemical cycling and nitric acid pretreatment, the presence of Pd and Sn sites on the RVC surface due to the Shipley pretreatment inhibited especially the electrodeposition of Ru as shown by the 4.7:1 Pt:Ru atomic ratio (Table 2- 2). Figure 2-13a shows the SEM images of the deposit obtained with electrochemical oxidation pretreatment of the RVC. In comparison with the nitric acid treatment shown by Figure 2-9, a denser deposit has been obtained composed of uniform less-than-i 0-nm- diameter interconnected particles forming a mesoporous network. Figure 2-14 shows the effect of RVC pretreatment method in conjunction with post deposition heat treatment at 573 K for 1 hour in a N2 stream, on the activity toward methanol electro-oxidation. In accordance with the data in Table 2-2, electrochemical cycling in conjunction with heat treatment increased the oxidation superficial current density, whilst the pretreatment with the Shipley method resulted in very low activity. Performing heat treatment on the RVC sample in nitrogen at 573 K for 1 hour after electrodeposition increased the activity for methanol electro-oxidation, most likely due to the removal of adsorbed residues of organic compounds (Figure 2-14 inset). The heat treatment caused some degree of particle sintering, as revealed by the SEM image shown in Figure 2-13b, however, the mesoporous network has been maintained and the electrochemically active surface area was virtually the same 40 m2 g’ (Figure 2-15). Thus, the beneficial effect of a cleaner surface prevailed over some degree of particle sintering. 121 Figure 2-13: SEM images of Pt-Ru electrodeposited on RVC pretreated with electrochemical cycling. Electrodeposition was carried out at 341 K using microemulsion with a superficial deposition current density and time of 10 A m2 and 234 minutes. (a) No heat treatment; (b) After heat treatment in N2 at 573 K for 1 hour. 122 400 - Electrochemical Cycling Pretreatmenttwith No Heat Treatment 350 - 800 Electrochemical Cycling Pretreatmentwith 1 HourPostdepition HeatTreatment in N2 at 573 K 300 - 400 / CX C —200 ... V 250 - 0 200- -200 7 -0.1 00 01 02 0.3 0.4 0 00 0.7 OS 0.9 150- •1 EIVs. SHE 0 .0’ 100- —— A Nric Acid Pretreatment +Heat Postdeposftion Treatment — Shiey Pretreatment + Heat Postdeposition Treatment — — — — Flectrochemical Cycling Pretreatment + Heat Postdeposition Treatment I I I I I 1 0.3 0.4 0.5 0.6 0.7 0.6 0.9 E / V vs. SHE Figure 2-14: Effect of different RVC pretreatment methods for Pt-Ru electrodeposition via mieroemulsion. Voltammograms of methanol oxidation for Pt RuIRVC prepared at 341 K with microemulsion. Superficial deposition current density and time of 10 A m2 and 234 minutes. Electrolyte: 1 M CH3O and 0.5 M H2504. Current density is given in A m2 geometric area basis. Temperature: 298 K. Scan rate: 0.005 V sO’. Inset: Effect of postdeposition heat treatment. 123 100 80 - 60 - E 40 - 20 - 0- 0.0 0.4 0.6 0.8 1.0 El V vs. SHE Figure 2-15: Blank scan and Cu UPD stripping curves of Pt-Ru electrodeposited on RVC prepared at 341 K with microemulsion with a superficial deposition current density and time of 10 A m2 and 234 minutes. Electrolyte: 0.5 MH2S04(blank); 0.5 MH2S04+ 0.002 M CuSO4 (Cu UPD) Temperature: 298 K; Scan rate: 0.050 V s. The activity of the catalyst prepared with the microemulsion method with electrochemical cycling pretreatment and postdeposition heat treatment was compared with the activity of commercial E-TEK Pt-Ru (1:1) catalyst reported by Hyeon et a!. on both mass and active surface area basis [32, 33]. The experimental conditions for the electrochemical measurements were identical except for the use of 2 M CR3OH in the literature data. At 0.6 V, the catalyst prepared by microemulsion exhibited a mass- specific oxidation current of 24 A g1 compared to 35 A g1 for the commercial E-TEK supported catalyst [32]. On an active surface area basis, the specific oxidation current for the catalyst prepared by microemulsion was 0.61 A m2 compared to 0.30 A m2 for the Iank Scan — — — Cu UPD Sffipping / / / \ N P I/ 0.2 124 commercial catalyst [32, 33]. This implies that the active sites of the catalyst prepared by microemulsion had superior catalytic effect on methanol electro-oxidation compared to the commercial catalyst, but had a lower specific active surface area. Lastly, a comparison was made by CP, mimicking the constant current operation of a DMFC anode, between the electrocatalytic activities toward CH3O oxidation of Pt- Ru produced by deposition from a pure aqueous solution composed of 0.01 M H2PtCI6 and (NH4)2PtC16and reverse microemulsion at 341 K, using in both cases 10 A m2 for 234 minutes. Both samples were identically pretreated by electrochemical oxidation cycling and subjected to postdeposition heat treatment. SEM images of the catalyst prepared with aqueous solution are shown in Appendix B. It can be seen that the sparse coating of electrodeposits are significantly larger in diameter (up to 500 nm) compared to catalysts prepared with microemulsion. As shown by Figure 2-16, both the open circuit potential and the anode potential corresponding to 50 A m2 current density step were more negative for the sample prepared by microemulsion deposition, by about 0.2 V and 0.05 V, respectively. These results were supported by the characteristic data of the two samples given in Table 2-3. Both the mass loading and the specific surface area of the deposit were lower for the sample produced by aqueous phase electrodeposition. Furthermore, the Pt:Ru atomic ratio in the case of aqueous deposition was very high, i.e. 15:1 compared to 1.3:1 for microemulsion, clearly demonstrating the benefits of employing a colloidal deposition media. 125 0.6 Figure 2-16: Chronopotentiometry data of methanol oxidation on Pt-RuLRVC in 1 M CH3O and 0.5 M H2S04. Temperature: 298 K. Current density: 50 A m2. Samples prepared in microemulsion and aqueous solution. Both samples were pretreated with electrochemical cycling, prepared at 341 K with a superficial deposition current density and time of 10 A m2 and 234 minutes, and heat treated after deposition. Table 2-3: Mass loading, Pt:Ru atomic ratio, and active surface area prepared by electrodeposition in microemulsion and aqueous solution. of samples Deposition Current Density: 10 A m2 Mass Pt:Ru Cu UPD Deposition Time: 234 mm Loading Atomic Area Deposition Temperature: 341 K (g m2) Ratio (m2 g’) Pretreatment: Electrochemical Cycling Postdeposition Treatment: Heat Treatment Microemulsion 2.3 1.3:1 40 Aqueous 1.0 15:1 24 w I If) > > w 0.5 0.4 0.3 0.2 0.1 hrZZZZZZZZZZZZZ / 2.3 g m Pt-Ru (1.3:1 Atomic Ratio) Prepared from Mtcroernulsion — — — 1.0 g m Pt-Ru (15:1 Atomic Ratio) Prepared from Aqueous Bath 0 50 100 150 200 250 Time 1 S 126 2.4 Conclusion The electrodeposition of Pt-Ru on RVC with 39 ppc was studied using reverse emulsion and microemulsion systems as the plating media. The electrocatalytic activity toward methanol electro-oxidation has been assessed by voltammetry and CP. It was found that the pretreatment of the RVC had a major influence on the morphology and catalytic activity of the Pt-Ru deposits. For the microemulsion system, an electrochemical oxidative cycling pretreatment method of the RVC yielded a mesoporous Pt-Ru coating characterized by the highest catalyst specific surface area (40 m2 g’) and catalytic activity toward methanol oxidation. The performance of the catalyst produced from microemulsion exceeded those obtained by deposition from either reverse emulsion or pure aqueous phase plating solutions. 127 2.5 References 1. H. Bönneman, K.S. Nagabhushna, J. New Mater. Electrochem. Syst. 7 (2004) 93. 2. H. Bönneman, R.M. Richards, Eur. J. Inorg. Chem. (2001) 2455. 3. H. Bönneman, R.M. Richards, Synth. Met. Org. Inorg. Chem. 10 (2002) 209. 4. L. Dubau, C. Coutanceau, E. Gamier, J-M. Lèger, C. Lamy, J. App!. Electrochem. 33 (2003) 419. 5. G.S. Attard, C.G. Göltner, J.M. Corker, S. Henke, R.H. Templer, Angew. Chem. Tnt. Ed. 36 (1315) 1997. 6. G.S. Attard, S.A.A. Lec!erc, S. Maniguet, A.E. Russell, I. Nandhakumar, P.N. Bartlett, Chem. Mater. 13 (1444) 2001. 7. J. Elliott, G.S. Attard, P.N. Bartlett, N.R.B. Coleman, D.A.S. Merckel, J.R. Owen, Chem. Mater. 11(1999) 3602. 8. J.M. Eliott, P.R. Birkin, P.N. Bartlett, G.S. Attard, Langmuir 15(1999)7411. 9. A. Kucemak, J. Jiang, Chem. Eng. J. 93 (2003) 81. 10. R.G. Allen, C. Lim, L.X. Yang, K. Scott, S. Roy, J. Power Sources 143 (2005) 142. 11. A. Bauer, E.L. Gyenge, C.W. Oloman, Electrochim Acta 51(2006) 5356. 12. AR. Campbell, M.G. Bakker, C. Treiner, J. Chevalet, J. Porous Mater. 11 (2004) 63. 13. A.J. Zarur, J.Y. Ying, Nature 403 (2000) 65. 14. A.J. Zarur, N.Z. Mehenti, A.T. Heibel, J.Y. Ying, Langmuir 16 (2000) 9168. 15. X. Zhang, K.Y. Chan, J. Mater. Chem. 12 (2002) 1203. 128 16. X. Zhang, K.Y. Chan, Chem. Mater. 15 (2003) 451. 17. T.D. Tran, S.H. Langer, Electrochim. Acta 38 (1993) 1551. 18. E. Gyenge, J. Jung, B. Mahato, J. Power Sources 13 (2003) 388. 19. J. Kizling, M. Boutonnet-Kizling, P. Stenius, R. Touroude, G. Maire, Electrochem. Colloids Dispersions (1992) 333. 20. C.H. Lu, H.C. Wang, J. Electrochem. Soc. 152 (2005) C341. 21. C.L. Green, A. Kuccernak, J. Phys. Chem. B 106 (2002) 1036. 22. C. Scortichini, C.N. Reilley, J. Electroanal. Chem. 139 (1982) 233. 23. Y.X. Pang, X. Bao, J. Mater. Chem. 12 (2002) 3699. 24. T. Gu, P.A. Galera-Gomez, Colloids Surf. A 147 (1999) 365. 25. H.C. Teh, G.H. Ong, S.C. Ng, J. Dispersion Sci. Technol. 6 (1985) 255. 26. K. Shinoda, H. Sagitani, J. Colloid Interface Sci. 64 (1978) 68. 27. H.A. Gasteiger, N. Markovic, P.N. Ross, E.J. Cairns, J. Electrochem. Soc. 141 (1994) 1795. 28. C. Bock, M.A. Blakely, B. MacDougall, Electrochim. Acta 50 (2005) 2401. 29. M. Umeda et. al., J. Power Sources 136 (2004) 10. 30. M.M. Goldin, A.G. Volkov, D.N. Namychkin, J. Electrochem. Soc. 152 (2005) E167. 31. M.M. Goldin, A.G. Volkov, D.N. Namychkin, E.A. Filatova, A.A. Revina, J. Electrochem. Soc. 152 (2005) E172. 32. T. Hyeon, S. Han, Y.E. Sung, K.W. Park, Y.W. Koon, Angew. Chem. mt. Ed. 42 (2003) 4352. 33. S. Han, Y. Yun, K.W. Park, Y.E. Sung, T. Hyeon, Adv. Mater. 15 (2003) 1922. 129 •1 3 DIRECT METHANOL FUEL CELLS WITH RETICULATED VITREOUS CARBON, UNCOMPRESSED GRAPHITE FELT, AND TITANIUM MESH ANODES* 3.1 Introduction The DMFC has attracted much research attention over the past two decades due to high theoretical energy density (4.4 kWh L’) and potential suitability as power source for both automotive transportation and portable electronic devices as a consequence of simpler, liquid-based, fuelling infrastructure [1]. However, the sluggish CR3OH electro oxidation kinetics coupled with catalytic activity loss over time due mainly to the accumulation of the CO intermediate on the catalyst surface (e.g. the reaction rate decreases by a factor of four to five during the first 100 ms of reaction [2]) remains a major challenge even after four decades of methanol electro-oxidation fundamental research, hampering the commercialization of the DMFC. Pt-Ru catalysts acting most likely by a combination of bifunctional mechanism and electronic coordination effect, with optimal Pt:Ru atomic ratios between 1:1 and 4:1 (depending on temperature, electrode potential and CH3O concentration), show lower susceptibility for CO poisoning and higher CH3O oxidation rates at low anode potentials (e.g. below 0.6 V vs. SHE). Moreover, Pt-Ru catalysts enhance the complete 6 e electro-oxidation rate compared to pure Pt. The area of CH3O electro-oxidation kinetics and catalysis has been extensively discussed in the literature (see Refs. [3-6] and Section 1.3.1). It must be noted that the slow anode kinetics causing a low CH3O conversion per pass in the thin catalyst layer (thickness about 20 .tm) of the gas diffusion electrode contributes indirectly to the enhanced CH3O diffusion across the solid polymer electrolyte, which in turn compromises the cathode performance by establishment of a mixed potential on the cathode surface [7, 8]. * A version of this chapter has been published: T.T. Cheng, E.L. Gyenge, “Direct Methanol Fuel Cells with Reticulated Vitreous Carbon, Uncompressed Graphite Felt and Titanium Mesh Anodes”, J. Appi. Electrochem. 38 (2008) 5 1-62. 130 In addition to fundamental electrocatalytic aspects and development of improved electrocatalyst formulations (e.g. ternary and quaternary compositions such as Pt-Ru-Rh Ni [9], Pt-Ru-Os-Ir [101) the synergy between the catalyst layer and the overall anode structure (e.g. presence and/or type of support, catalyst-support interaction, ionic conductor load, hydrophobic-hydrophilic pore balance) has a significant impact on the fuel cell performance-catalyst load (i.e. cost) relationship. The utilization of the catalyst load in a typical membrane-gas diffusion electrode assembly, defined as the ratio between the effective electrochemically available surface area and the total catalyst surface area, is between 10-50% [11] and it could further diminish during extended fuel cell operation. Furthermore, the counter-current two-phase (L: CH3O and H20; G: C02) flow in the porous electrode affects the diffusion overpotential and the effective ionic conductivity [12]. Whilst numerous studies have been devoted to understanding and improving the electrocatalysis of CH3O oxidation, it could be argued that the engineering of both the catalyst layer and the overall anode structure received much less attention. The vast majority of DMFC experiments have been carried out with gas diffusion type anodes comprised of carbon cloth or paper diffusion-backing layer and carbon-black supported catalyst layer containing Nafion ionic conductor. Mastragostino et al. showed the importance of optimizing the anode catalyst layer composition for DMFC’s with respect to Nafion load and type of carbon support together with their combined effect on the electrochemically available surface area [13]. Utilizing carbon nanocoil supported Pt-Ru (1:1) catalyst promising DMFC results were obtained by Hyeon et al. with a load of 20 g m2 (e.g. peak power density of 2300 W m2 was achieved at 333 K) [14]. Wilkinson et al. recognized the importance of the anode design in reducing the methanol crossover, and they patented the multi-layer anode concept [15]. Three sheets of carbon fiber paper (thickness 100 m) with Pt-Ru supported on carbon black applied onto them (total load of 18 g m2) were stacked to give a multi-layer DMFC anode. Employing a constant volume of 250 mL 2 M CH3OH-0.5 M H2S04they found that the methanol utilization increased from 60% to 80% with the multi-layer anode compared to a single-layer electrode. Recently, Gyenge and collaborators proposed three-dimensional monolithic carbon-based electrodes of about 200 to 2,000 urn thickness, as direct alcohol 131 fuel cell anodes [16-181. The three-dimensional matrix supporting well-dispersed electrocatalysts with various morphologies such as discrete nanoparticles or thin mesoporous coatings (pore diameter between 2 and 50 nm according to IUPAC nomenclature [19]) could assure an extended reaction zone for CH3O oxidation compared to the gas diffusion design providing an ionic (i.e. H3O) conductor network is established to link the catalytically active sites and the proton exchange membrane. Both the multi-layer and the three-dimensional anode concepts are similar since they provide an extended reaction zone (volume) for the electrochemical reaction. However, the monolithic three-dimensional electrode eliminates the possible contact resistance between the individual layers. Structural differences and surface physico-chemical properties of the three- dimensional electrode matrices such as various graphite felts and reticulated vitreous carbons, could improve the two-phase (L/G) flow dynamics. Three-dimensional electrodes can accommodate a wide range of L/G flow regimes [20] as opposed to the gas diffusion electrode, which operates best in the gas continuous regime and therefore, is susceptible to flooding. These issues would gain significance with scale-up of DMFC (e.g. to geometric electrode areas of 25 to 100 cm2) and large stack development. The three-dimensional electrode concept poses two major challenges. The first one relates to synthesizing and uniformly depositing nano-sized electrocatalyst throughout the thickness of the three-dimensional matrix. Second, is the formation of the proton conductor network across the three-dimensional electrode connecting the electrocatalytic sites and the PEM membrane, as discussed also by Wilkinson et al. In order to address the first challenge, the goal of the present investigation was to study the electrodeposition of Pt-Ru nanostructures on different three-dimensional electrodes, such as RVC, uncompressed GF and Ti mesh, using colloidal deposition media to control the crystallite size [17]. Furthermore, the larger aim of this study was to determine the applicability of the various three-dimensional substrates as DMFC anodes by a combination of fundamental electrochemical and surface analytical techniques in conjunction with fuel cell experiments. The issue of protonic conductivity in the fuel cell anode was addressed by employing a liquid electrolyte, 0.5 M H2S04 — 1 M CH3O 132 solution. The liquid electrolyte employed here serves the purpose of demonstrating the concept of three-dimensional fuel cell electrodes. 3.2 Experimental Section 3.2.1 Pretreatment of Three-Dimensional Catalyst Supports The three-dimensional electrodes employed in the present work were: RVC (Electrolytica Inc., 39 pores per centimeter, thickness 2 mm), uncompressed GF (Test Solutions, thickness 3 mm) and Ti mesh (VWR Canlab, 200 gm). Figure 3-1 shows the SEM images of the structures and surfaces of the different substrates. Figure 3-1: SEM images of 3-D substrates. (a) RVC structure; (b) RVC surface; (c) GF structure; (d) GF surface; (e) Ti mesh structure (21]; (f) Ti mesh surface. Reproduced by permission of The Electrochemical Society. 133 It was shown in Chapter 2 the importance of RVC pretreatment for nanostructure electrodeposition [17]. Table 3-1 summarizes the pretreatment methods applied for the different supports. The electrochemical pretreatment method involved potential cycling of the RVC between 1.44 to 2.09 V vs. SHE in concentratedH2S04solution (see Table 1) at a scan rate of 0.001 V s repeated fifty times [17]. The electrochemical pretreatment could not be carried out on GF due to severe weakening of the physical integrity of the material. Therefore, the only pretreatment applied to the felt electrode was sonication in methanol for 30 minutes. The pretreatment method for Ti mesh involved etching in boiling HC1 for 30 s to remove the well-adhered surface oxide layer [22]. Upon pretreatment, all samples (5 cm2 geometric area) were washed thoroughly with distilled water followed by drying in air. Table 3-1: Substrate pretreatment methods. Substrate Electrochemical Sonication in Etching in Boiling Cycling in Methanol Hydrochloric Acid Concentrated Sulfuric Acid RVC + - - GF - + - TiMesh - - + 3.2.2 Electrodeposition Procedure Two colloidal electrodeposition media were investigated and compared, a novel micellar solution and a microemulsion system discussed extensively previously in Chapter 2 [17]. The micellar deposition solution was composed of Triton X-100 non- ionic surfactant[C8H17640(C)95], iso-propanol, and an aqueous phase with various concentrations ofH2PtC16 and (NH4)2RuC16.A typical deposition bath had the following composition: 5 vol% Triton X-100, 20 vol% iso-propanol, and 75 vol% aqueous phase with H2PtC16 and (NH4)2RuC16.The concentration of the Pt and Ru compounds was each 0.25 mM in the total homogenous solution. In Section 3.3.1, further 134 details are given regarding the role of isopropanol and a comparison is made with the microemulsion and pure aqueous deposition media. The chemicals used were reagent grade obtained from Sigma-Aldrich and were used as delivered without further purification processes. The micellar solutions were prepared by mixing the aqueous and surfactant phases in a water-jacketed glass vessel connected to a circulating water bath at 341 K for 30 minutes. The electrode assembly composed of the working electrode (geometric area of 5 cm2) placed at a distance of 1 0_2 m between two Pt/Ti counter electrodes (5 cm2 geometric area each) was inserted into the glass vessel and connected to a dc power supply (Xantrex XHR15O-7 DC Power Supply, 0-150 V, 0-7 A) to carry out the electrodeposition. After the deposition experiment was completed, the working electrode was sonicated in THF (Reagent Grade, Sigma Aldrich) for 5 minutes to wash out the organic compounds retained in the porous matrix. The deposited substrate was then washed thoroughly with distilled water and dried, followed by heat treatment in a N2 stream for 1 hour at 573 K to remove traces of adsorbed organic compounds [17]. 3.2.3 Electrochemical Measurements Voltammetry was carried out at 298 K, with a three-electrode setup, in a water- jacketed electrochemical cell connected to a circulating water bath. The test solution was 1 M CH3O and 0.5 M H2S04 with a volume of 50 mL. The Hg/Hg2SO4,K2S04 std.(MSE) electrode and a platinum wire were used as reference and counter electrodes, respectively. A computer-controlled VoltaLab PGZ4O2 potentiostat with the VoltaMaster 4 software by Radiometer Analytical was used in the experiments. The catalyst surface was chemically and electrochemically pre-treated to remove surface impurities. The chemical pretreatment involved the immersion of the working electrodes in a 1:1 v/v. concentrated H2S04and 30 vol% H20(Fisher Scientific) solution five times for a few seconds each, and rinsed thoroughly with distilled water. Following the chemical pretreatment, the working electrodes were repeatedly cleaned three times 135 electrochemically in 0.5 M H2S04 by applying potential steps for 10 s each at 1.28 V, 1.20 V and 0.05 V, respectively. The effective electrochemically active Pt-Ru surface area was estimated by the Cu underpotential deposition and anodic stripping technique [17, 231. In our previous investigations we found that Cu does not underpotential deposit on vitreous carbon [17]. Therefore, assuming complete Cu monolayer coverage on the Pt and Ru sites, the effective surface area can be calculated from the anodic stripping charge (i.e. 4.2 C m2). Reference voltammograms between the potential range of —0.04 V to 0.91 V were first obtained in 0.5 M H2S04 at a scan rate of 0.050 V s1. The Cu UPD experiments were carried out in 0.5 M H2S04 and 0.002 M CuSO4 at 298 K. The underpotential deposited Cu monolayer was formed on the catalyst surface by polarizing the working electrodes at 0.26 V for 300 s. Afterwards, a linear voltammetric scan in the anodic direction was applied between 0.26 V and 0.91 V with a scan rate of 0.050 V s1 to remove the adsorbed copper monolayer. The charge differences in the same potential range between the reference scan and Cu stripping were used to calculate the active surface area. 3.2.4 Surface and Analytical Characterization of the Catalysts Hitachi S4700 high resolution SEM was used to capture visual images of the prepared catalysts. Fragments of the deposited substrates were flush mounted onto SEM stubs with carbon adhesive. An accelerating voltage and emission current of 2000 V and 1.25x107 A, respectively, were employed at a working distance of 0.0025-0.0035 m to obtain the images. ICP-AES using a Perkin Elmer Optima, model 3300DV instrument, was used to determine the mass loading of the electrodeposits. 3.2.5 Membrane Electrode Assembly and Fuel Cell Experiments The MEA was prepared with a half MEA (cathode, Fideris Inc., Nafion® 117 membrane) with 40 g m2 of pre-painted Pt black and the 3-D anode (5 cm2). The DMFC was assembled with two gold-plated end plates, an Elat® carbon cloth (E-Tek Inc.) as the cathode backing layer, the custom MEA with the 3-D anode, a silicone-coated gasket, 136 and a carbon cloth as the anode backing layer. The DMFC was held together by insulated bolts. The fuel cell tests were performed using a Fideris Inc. MTK fuel cell test station, equipped with corrosion-resistant fittings and operated using the FC Power® software. The fuel cell tests were carried out at 333 K with an oxygen flow rate of 500 mL min1 at 2.5 bar absolute pressure. Dry medical grade oxygen was used as supplied by Praxair Inc. The anode was fed with a liquid solution of 0.5 M H2S04and 1 M CH3O at 2 mL min1 without preheating. With the flow system and temperature control turned on at open circuit, membrane conditioning was performed for 2 hrs. This step assured a stable and reproducible operation of the cell. Current was then progressively drawn and the cell voltage was recorded after 2 minutes of continuous operation at constant current. 3.3 Results and Discussion 3.3.1 Electrodeposition from Micellar Solution vs. Microemulsion: RVC support In Chapter 2, the microemulsion media was investigated for the galvanostatic electrodeposition of Pt-Ru on RVC. The microemulsion was composed of 72 vol% cyclohexane (with 0.001 M tetrabutylammonium perchlorate), 25 vol% mixture of Triton X- 100 and isopropanol in 1:4 volume ratio, and 3 vol% aqueous phase with 0.01 M H2PtC16and (NH4)2RuC16[17]. The galvanostatic deposition mode was chosen instead of the controlled potential variant, for superior industrial applicability. Moreover, the constant deposition current density could also assure better reproducibility of the catalyst mass load throughout the three-dimensional electrode structure. Due to the very low ionic conductivity of the microemulsion (i.e. 2x105 S m’) the deposition current density was limited to 10 A m2, generating a Pt-Ru load on RVC of 2.3 g m2 in 234 minutes at 341 K. The Pt:Ru atomic ratio was 1.3:1. Comparative Pt-Ru electrodeposition in an aqueous solution carried out under identical conditions of surface pretreatment, deposition current density, temperature and time, yielded a Pt rich catalyst (Pt:Ru atomic ratio of 15:1) with much lower catalyst load (i.e. 1 g m2) and mass specific surface area (24 vs. 40 m2 g’) [171. Thus, the presence of a colloidal system and a surfactant in particular are essential 137 to control the catalyst load and composition (Pt:Ru ratio) on the three-dimensional substrate as it was also conclusively demonstrated in other work from this group [16]. The goal here was to simplify the colloidal electrodeposition bath composition while further improving the electrocatalytic activity by affecting the catalyst morphology and Pt:Ru ratio. Therefore, a new deposition bath was developed without the presence of cyclohexane, the majority component in the previously employed microemulsion system. The non-ionic surfactant Triton X-100 and isopropanol were retained in order to create a micellar deposition media containing aqueous H2PtC16 and (NH4)2RuC16.Teh et al. reported that isopropanol eliminates the possible formation of liquid crystal gel in the Triton X-100/aqueous system, forming a micellar solution [24, 25]. The hydrodynamic radius of the micelles decreased by the addition of isopropanol and further decrease was observed with increased temperature at isopropanol concentration of 14 to 26 wt%. For example, the hydrodynamic radius at 293 K decreased from 4.13 nm (0 wt% isopropanol) to 1.95 mu (26 wt%). At 298 K and 26 wt% isopropanol, the hydrodynamic radius was 1.74 nm [24, 25]. In our current study, the micellar system was composed of 75 vol% aqueous phase and 25 vol% Triton X-100 / isopropanol in a 1:4 volume ratio. The weight percentages were: 6 wt% Triton X-l00 and 17 wt% isopropanol. Dynamic light scattering experiments (using a Brookhaven Instruments Corporation FOQELS Particle Size Analyzer) were performed to determine the hydrodynamic radius of the micelles of the investigated deposition media; however, the size of the micelles could not be determined, suggesting possibly a micellar radius smaller than the detection limit of the instrument (>2 mn). Comparative galvanostatic electrodeposition experiments were performed in the two media, i.e. microemulsion and micellar, with 0.25 mM H2PtC16 and (NH4)2RuC16 each. The deposition superficial current density was 10 A m2, applied continuously for 240 minutes at a constant temperature of 341 K. Table 3-2 shows the Pt-Ru characteristics obtained, while Figure 3-2 compares the electrocatalytic activity toward methanol oxidation by voltammetry. 138 Table 3-2: Electrodeposition of Pt-Ru on RVC from colloidal media: comparison between micellar and microemulsion methods. Electrodeposition conditions: 10 A m2, 240 mill, 341 K, flat plate counter electrodes. Deposition Media Mass Loading Bulk Pt:Ru atomic Specific surface area (g m2) ratio (m2g1) Microemulsion 2.3 1.3:1 40 Micellar 1.5 7.3:1 46 c1 E 800 600 400 200 0 -200 0.9 Figure 3-2: Voltammograms of methanol electro-oxidation using Pt-Ru electrodeposited on RVC: Effect of colloidal electrodeposition media. Electrolyte: 1 M CH3O and 0.5 MH2S04.Temperature: 298 K. Current density is given in A m2 geometric area basis. Scan rate: 0.005 V s. -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 E/Vvs. SHE 139 The methanol electrooxidation superficial current density on Pt-RuIRVC at 298 K was higher for the catalyst prepared by electrodeposition from micellar media compared to microemulsion (Figure 3-2). This result is interesting considering the lower catalyst load obtained using the micellar media (1.5 g m2) and the approximately identical specific surface area of the two catalysts (Table 3-2). High resolution SEM showed (Figure 3-3) that the micellar solution produced a more porous, interconnected open cell deposit morphology compared to the denser particulate-like structure due to microemulsion. However, these apparent morphological differences were only slightly reflected in the measured specific surface area by the Cu UPD technique, the mass- specific surface area increased from 40 to 46 m2 g’. Figure 3-4 shows the blank scan and Cu UPD stripping curve of the sample prepared from the micellar media. In contrast to Figure 2-16 (blank scan and Cu UPD stripping curve of catalyzed RVC prepared from microemulsion, where a broad peak at —0.58 V was observed), an inflection point was found at 0.44 V showing two distinct peaks at 0.4 and 0.7 V. The peak potentials correspond to those of Pt (0.7 V) and Ru (0.4 V) as suggested by Green and Kucernak [221. The contribution from Ru was estimated by integrating the difference between the stripping curve and blank scan up to the potential at which the inflection point was observed, namely 0.44 V. Similarly, the contribution from Pt was calculated by integration starting from 0.44 V. The total Cu stripping charge obtained was 284 C m2. The Cu stripping charge up to 0.44 V was found to be 37 C m2. This resulted in a Pt-Ru surface ratio of approximately 6.8:1, which is in close agreement with the bulk Pt-Ru atomic ratio obtained by ICP-AES (7.3:1). It must be noted that most Cu UPD experiments done in this study did not result in distinct peaks, which are required for peak de-convolution to estimate the Pt-Ru surface atomic ratio. However, from the current result, it might be reasonable to believe that the surface atomic ratio is close to that of the bulk atomic ratio obtained by ICP-AES. 140 Figure 3-3: SEM images of electrodeposited Pt-Ru on RVC prepared with microemulsion and micellar solution. (a) Microemulsion; (b) Micellar solution. 141 70 60 50 i 40 E 3D 20 10 0 1 .0 Figure 3-4: Blank scan and Cu UPD stripping curves of electrodeposited RVC prepared at 341 K from micellar media with a superficial deposition current density and time of 10 A cm2 and 240 minutes. Test solution: 0.5 M H2S04 (blank); 0.5 M H2S04+ 0.002 M CuSO4 (Cu UPD) Temperature: 298 K; Scan rate: 0.050 V s1. Changing the phase structure of the electrodeposition media affected not only the deposit load and morphology but also the Pt:Ru atomic ratio (Table 3-2). The micellar system produced a catalyst with high Pt:Ru atomic ratio 7.3:1, whilst the microemulsion favored more the Ru deposition generating a Pt:Ru ratio of 1.3:1. This could explain the differences in electrocatalytic activity observed by voltanimetry at 298 K (Figwe 3-2). The optimal Pt:Ru ratio is a long-standing issue in the methanol electrooxidation literature. Gasteiger et al. found using ultra-high vacuum experiments on sputter-cleaned alloys, that surfaces with 7-10 at.% Ru are the most active at room temperature, with a shift to 50 at.% Ru at high temperatures (e.g. 333 K) due to a change in the rate- 0.0 0.2 0.4 0.6 0.8 ElVvs. SHE 142 determining step from adsorptionldehydrogenation of CH3O to oxidation of CO [26- 28]. Therefore, the capability to easily control the Pt:Ru ratio by adjusting certain variables of the catalyst preparation procedure (such as solution composition, deposition current density, temperature and electrode geometry) is of paramount significance. Further experiments were carried out using the micellar deposition media aimed at controlling both the catalyst load and the Pt:Ru atomic ratio. 3.3.2 Pt-Ru Electrodeposition on RVC from Micellar Media: Effect of Perforated Counter Electrodes and Deposition Current Density During electrodeposition significant 02 gas evolution occurs on the counter electrodes. The inefficient release of gas bubbles from the cell decreases the effective conductivity and therefore, alters the potential distribution which impacts the deposit characteristics such as load, morphology and penetration throughout the three- dimensional matrix. In order to improve the gas disengagement on the counter electrodes, perforated Pt/Ti plates were employed with 10 holes per cm2 and hole dimension of about 0.03 cm2 (Figure 3-5). ....... ....... ....... ....... ....... ....... 2.2 cm 2.2 cm Non-Perforated Counter Electrode Perforated Counter Electrode Figure 3-5: Schematic of counter electrodes used in colloidal electrodeposition. 143 For a deposition superficial current density of 10 A m2 applied for 240 minutes at 341 K, the perforated counter electrode design yielded on the RVC substrate a three fold increase of Pt:Ru mass loading from 1.5 to 4.6 g m2 (Table 3-3). The Pt:Ru ratio in the deposit was found to be 7.2:1, virtually identical to the catalyst sample produced without the use of perforated counter electrodes. Moreover, the catalyst morphology remained the same as shown by SEM imaging (compare Figures 3-6a and 3-3b). However, the specific surface area decreased to 29 m2 g’ indicating the build up of a thicker film. The blank scan and Cu UPD stripping curve of the sample prepared with perforated counter electrodes is given in Figure 3-7. An inflection point was observed at approximately 0.52 V. The total Cu UPD stripping charge obtained was 568 C m2. Following the same procedure as described earlier, the Ru charge contribution by integrating up to 0.52 V was 70 C m2, resulting in a Pt-Ru surface atomic ratio of 7.1:1. This is almost identical to the bulk Pt-Ru ratio obtained by ICP-AES (7.2:1), further supporting the notion that the Pt-Ru surface ratio was similar to the bulk ratio for the catalysts prepared by colloidal electrodeposition. Table 3-3: Effect of perforated counter electrodes on the electrodeposition of Pt-Ru from micellar solution on various substrates. Temperature: 341 K. Substrate Deposition Time Mass Loading Bulk Pt:Ru Specific current - atomic surface area(mm) (g m2) 2density ratio (m g) (A m2) RVC 10 240 4.6 7.2:1 29 RVC 20 120 12.0 3.6:1 12 RVC 40 60 8.7 4.4:1 16 RVC 60 40 4.9 3.5:1 25 GF 20 120 9.8 4.0:1 36 Ti Mesh 20 120 2.8 4.5:1 32 144 -:.,-w4 -— a: _I s• i..-- Figure 3-6: SEM images of electrodeposited Pt-Ru on RVC prepared from micellar solution with perforated counter electrodes at 341 K. Effect of deposition current density (a) 10 A m2; (b) 20 A m2; (c) 40 A m2; (d) 60 A m2. 145 250 200 150 E 100 SD 0 1.0 E/Vvs. SHE Figure 3-7: Blank scan and Cu UPD stripping curves of electrodeposited RVC prepared at 341 K from micellar media with perforated counter electrodes with a superficial deposition current density and time of 10 A cm2 and 240 minutes. Test solution: 0.5 M H2S04 (blank); 0.5 M H2S04 + 0.002 M CuSO4 (Cu UPD) Temperature: 298 K; Scan rate: 0.050 V s. The catalyst produced with perforated counter electrodes showed enhanced activity toward CH3O electrooxidation compared to the one obtained using un perforated, flat, counter electrodes (Figure 3-8 curves A and B). This is likely due to the higher catalyst surface area enhancement factor a3 (defined as the total catalyst surface area per geometric area of the substrate, Equation 3-1), i.e. 133 m2 m2 vs. 69 m2 nf2 for the catalysts produced with perforated vs. flat counter electrodes, respectively. a = a,,, m, [Eq. 3-1] 0.0 0.2 0.4 0.6 0.8 146 where a is the area enhancement factor (m2totai m2 geom), am is the mass specific catalyst surface area (m2totai g’), and m is the catalyst load (gm2geom). Mcellar - 10 Am2 t42n- P-frated Counter ettrcdes MeIIr- 10 Am2 PerI’oraed Counter Bectrodes McIIer- ) Am Perfored Counter Berodes McIIar- Am2 Ferfored Counter Bectrodes McIIer- Am2 Perfored Counter Bectrodes ¶ - —scici — I I I I I I I -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 E/Vvs. SHE Figure 3-8: Voltammograms of methanol electro-oxidation using Pt-Ru electrodeposited on RVC. Effect of electrodeposition conditions: counter electrode type and superficial current density. 1 M CH3O and 0.5 M H2S04.Temperature: 298 K. Scan rate: 0.005 V s1. The application of higher deposition current density (20 A m2, 40 A m2, and 60 A m2 vs. 10 A m2), which was permitted by the high ionic conductivity of the micellar solution and the more effective gas release due to the perforated counter electrode, was found to have significant effects on the catalyst morphology, total load and Pt:Ru ratio (Table 3-3). Interestingly, whilst the total charge applied during deposition remained the same, the deposit mass load on RVC changed with superficial current density, up to 12 g A 8 _ç;__ D E 2000 1500 - 1000 - 500 - 0 E C 0 147 m2 at 20 A m2and the Pt:Ru atomic ratio decreased to 3.5:1 at 60 A m2. Therefore, the higher current density favored the electrodeposition of Ru from the micellar solution. However, the Pt:Ru atomic ratio leveled off to —4:1 and a further increase in current density above 20 A m2 did not have a significant impact on the Ru deposition. The surface area enhancement factor (Equation 3-1) on RVC was the highest for the catalyst prepared at 20 A m2 compared to the 10 A m2 case, i.e. 144 m2 m2vs. 133 m2 m2, respectively. The deposit morphology at 20 A m2 was characterized by a mesoporous structure (pore diameter equal or less than 50 nm), formed by interconnected nanoparticles of approximately 10-20 nm diameter (Figure 3-6 b). At 40 A m2, the mesoporous structure began to disappear and completely vanished at 60 A m2. The resulting morphology was a rough catalyst coating made up of particles of 20-5 0 nm diameter (Figures 3-6 c and d). Figure 3-8 shows the catalyst prepared at 20 A m2 using perforated counter- electrodes gave the highest superficial anodic current densities at 298 K for 1 M CH3O oxidation in 0.5 M H2S04.This result is in accordance with the surface area enhancement factor, however, the contribution of the different Pt:Ru atomic ratio cannot be disregarded as well. 3.3.3 Effect of Catalyst Support Micellar electrodeposition of Pt-Ru was carried out in addition to RVC, on uncompressed GF and Ti Mesh supports. The experiments were carried out at 341 K with a deposition superficial current density and time of 20 A m2 and 120 minutes, respectively. The mass loading, Pt:Ru ratio, and specific surface area (determined by Cu UPD and stripping) of the catalysts prepared on the different substrates are presented in Table 3-3. The highest mass loading was obtained with the RVC support, 12 g nf2, which was attributed to the large number of deposition sites created by the electrochemical cycling pretreatment method as shown previously [17]. On GF the load was lower, 9.8 g m2, since surface pretreatment could not be carried out due to a loss of mechanical integrity. The load on Ti mesh was only 2.8 g m2 due probably to incomplete removal of 148 the non-conductive surface oxide layer and/or incomplete wetting of the surface by the micellar media. The morphology of the deposits on all three supports could be characterized as mesoporous coating composed of nanoparticle agglomerates (Figure 3-9). The highest catalyst surface area enhancement factor was obtained in the case of uncompressed GF (i.e. 353 m2 m2 compared to 144 m2 m2 for RVC and 90 m2 m2 for Ti mesh). The Pt:Ru ratio varied only slightly with deposition substrate type, in the range of 3.5:1 to 4.5:1. Figure 3-9: SEM images of Pt-Ru electrodeposits prepared from micellar solution with different substrates at 20 A m2, 341 K and perforated counter electrodes. (a) RVC; (b) uncompressed GF; (c) Ti Mesh. 149 Interestingly, voltammetric experiments carried out at 298 K revealed the highest CH3O electrooxidation superficial current density for to the RVC supported catalyst, followed by uncompressed GF, and lastly Ti mesh (Figure 3-10). The relative performance was further supported by CP data obtained at a constant current of 50 A m2 at 298 K (Figure 3-1 1). Thus, the order of electrocatalytic activity between RVC and uncompressed GF did not follow the catalyst surface area enhancement factor a. This indicates that there must be more subtle differences between the Pt-Ru catalysts supported on RVC and uncompressed GF, respectively. Figure 3-10: Effect of three-dimensional support on the electrocatalytic activity of Pt-Ru toward methanol electro-oxidation. Solution: 1 M CH3O and 0.5 M H2S04. Temperature: 298 K. Scan rate: 0.005 V s1. Samples prepared at a temperature of 341 K and a deposition current density of 20 A m2 for 120 mm with perforated counter electrodes in micellar solution. c1 E 2000 1500 1000 500 0 -500 -0.1 0.0 0.1 0.2 0.3 E/V 0.4 0.5 0.6 0.7 0.6 vs. SHE 0.9 150 0.5 0.4 0.3 w I(n 0.2 > w 0.1 0.0 -0.1 Figure 3-11: Chronopotentiometric data for methanol electro-oxidation of catalyzed RVC, GF, and Ti mesh in 1 M CH3O and 0.5 M H2S04at 298 K. Current density: 50 A m2. Samples prepared at a temperature of 341 K and a deposition current density of 20 A m2 for 120 mm with perforated counter electrodes in micellar solution. In order to unveil certain differences between the Pt-Ru/RVC and Pt-RuJGF systems, let us express the superficial current density for methanol oxidation I (Am2geom) in the three-dimensional electrode in terms of catalyst physico-chemical properties, catalyst utilization efficiency and local current density: i = a Jic’x = a, - IdX [Eq. 3-2] o To 0 50 100 160 200 250 300 tis 151 where a is the volume specific catalyst surface area (i.e. the total catalyst area as determined by ex-situ physico-chemical/electrochemical methods per geometric volume of the three-dimensional electrode (m2totainf3geom)), am is the catalyst surface area per unit catalyst weight (referred to as the mass specific surface area (m2totai g’)), i is the local current density (A m2eff), m is the catalyst load (g m2geom), Yc is the catalyst utilization efficiency reflecting the effective area participating in the electrochemical reaction (m2eff m2totaj), and ‘r is the electroactive reaction zone (or catalyst layer) thickness (mgeom). Note that r may or may not be equal to the physical thickness of the three-dimensional electrode, depending on the catalyst deposition penetration across the thickness and the availability of the ionic conductor network linking the catalytic sites. The local current density i is related to the local anode overpotential i according to various electrode polarization conditions, such as intrinsic electrode kinetic control, or mixed control involving intrinsic electrode kinetics together with reactant mass transfer and/or ionic conductivity effects leading to multiple apparent Tafel slopes [29]. Considering in Equation 3-2 the simplest case, namely intrinsic kinetic control described by Tafel polarization, and introducing the catalyst area enhancement factor a3 defined by Equation 3-1), the superficial current density for methanol oxidation in the three- dimensional electrode becomes: = yc expI dx [Eq. 3-31b3 ) where b and are the local Tafel slope (V dec’) and exchange current density (A m2eff) for methanol electrooxidation, respectively, at a point x in the electroactive zone. Thus, the spatial inhomogeneity of the electrode reaction rate in the three- dimensional electrode is accounted for. Equation 3-3 shows that even if a3 is higher for Pt-RuJGF compared to Pt RuJRVC, there are a number of other variables that contribute also to the measured superficial current density i. An important factor is the catalyst penetration depth in the three-dimensional matrix determining the electroactive zone thickness r and moreover, the homogeneity of the Pt-Ru catalyst composition across ‘r. Lycke and Gyenge studied 152 this issue in the case of Pt-Sn nanoparticle deposition on uncompressed OF using an electrochemical organosol method [18]. It was found that catalyst particles situated on the outer face of the substrate have different atomic ratios of the constitutive elements compared to particles situated inside, in the middle of the felt electrode. Hence, it is hypothesized that both r and the local Pt- Ru ratio could be different for Pt-Ru/OF vs. Pt-Ru/RVC in spite of an average 4:1 atomic ratio for both. The catalyst composition gradient across the electroactive zone thickness will effect the kinetic parameters b and i0, which in turn leads to different methanol oxidation rates. Furthermore, differences in catalyst/support interaction between Pt Ru/OF, Pt-Ru/RVC and Pt-RU/Ti could also affect the kinetic parameters of methanol oxidation, such as electronic effects and/or different crystallographic features of the Pt- Ru electrodeposit induced by the support. The crystallographic features, in terms of Equation 3-3, will also impact the area based catalyst utilization efficiency c since formation of crystallites richer in faces that are more active toward methanol oxidation will increase (see Chapter 4). These considerations point toward future experimental studies that are required in order to better understand the synergies between three- dimensional support and electrocatalytic activity. 3.3.4 Direct Methanol Fuel Cell Experiments The extended reaction zone three-dimensional anodes with low Pt-Ru catalyst load prepared using perforated counter electrodes and a deposition current density of 20 A m2 were also investigated in single-cell DMFC’s operated at 333 K and fed on the anode side with 1 M CH3O and 0.5 M H2S04 solution. Figure 3-12 shows the cell voltage vs. superficial current density, whilst Figure 3-13 expresses the specific power output on area and catalyst mass basis, respectively. 153 07 - 0.7 0.6 • RviD (12 g m) 06 v GF8gmj I • Ti Mesh (2.8 g m) 0.5 0.5 U > 0.4 0.4 > - iO.3- • • V • I V 0.2 0.2 I 0.1 . 0.1 0.0 I I I I I I 0.0 0 200 400 600 800 1000 1200 1400 1600 1800 i/A rn2 Figure 3-12: Effect of three-dimensional anode with electrodeposited Pt-Ru on DMFC performance. Catalysts electrodeposited at 341 K with a superficial current density and time of 20 A m2 and 120 mm, respectively. Fuel: 1 M CH3O in 0.5 M H2S04at 2 mL mm1.Cathode: 40 g Pt m2, dry 02 fed at 2.5 bar and 500 mL mm1. Temperature: 333 K. 154 (a) 600 600 • R’t 500 • Ti Mesh 400 400 N V V NV 300 V 30U 200 200 V • • . 100 v 100 B 0 I I I 0 0 500 1000 1500 2000 2500 3000 / A m2(b) 60 • PVC v SF 50 • Ti Mesh I 40 •• 40V V. 7:. V 0) 30 • 30 a. 20 20 V a 10 v 10 47 0 I I I I I I 0 0 50 100 150 200 250 300 350 400 i/A g1 Figure 3-13: DMFC power density. Three-dimensional anode support comparison. Conditions idem Figure 3-12. (a) Area specific power density; (b) Mass specific power output. 155 The highest fuel cell peak power density at 333 K was obtained with the RVC substrate (486 W m2 at 2250 A m2), followed by uncompressed GF and lastly Ti mesh (Figure 3-1 3a). While these results are in accordance with the voltammetry study, the fuel cell results are reflective of other phenomena as well, in addition to electrode kinetics. It is expected that the CH3O crossover to the cathode, compromising the cell voltage output due to the establishment of a mixed cathode potential, was the most severe in the case of the thin Ti mesh anode which also had the lowest Pt-Ru load (only 2.8 g m2). On the other hand, the ohmic voltage drop loss was the lowest for Ti mesh support. The high methanol crossover for the Pt-Ru/Ti system was reflected by the low open circuit cell voltage (i.e. 0.54 V) compared to Pt-Ru/RVC and Pt-Ru/GF. Qi and Kaufman discussed the relationship between open circuit cell voltage and methanol crossover [30]. The power output on a catalyst mass basis was virtually identical for the three- anodes for currents up to about 200 A g’ (Figure 3-13 b). At higher currents the performance of Pt-Ru/RVC and Pt-RuJGF leveled off, while the Ti mesh supported catalyst yielded a maximum catalyst mass specific power output of 50.4 W g’ (Figure 3- 13 b). Thus, the mass specific activities of the investigated anodes were comparable even though there was more than a four times difference in the Pt-Ru load between Ti mesh (2.8 g m2) and RVC (12 g m2) in conjunction with obvious differences in substrate physico-chemical properties. In other words, increasing the anode catalyst load with the developed electrodeposition procedure on the various three-dimensional substrates did not lead to a decrease of catalyst mass specific activity. Comparative fuel cell experiments were performed under identical operating conditions using a conventional CCM with 10 and 40 g m2 Pt-Ru load. Table 3-4 summarizes the performance of the reference gas diffusion anodes obtained in-house together with representative literature results in order to better asses the effectiveness of the novel, extended reaction zone, three-dimensional anodes proposed in the present work. The literature results were selected such that to be comparable as much as possible with the conditions employed here, e.g. Nafion® 117 membrane, operating temperature of 333 K and Pt-Ru catalysts with loads between 10 to 40 g nf2. In all the selected reference DMFC experiments on the cathode side a gas diffusion electrode was employed with some variation regarding the cathode catalyst loading and 02 pressure. 156 However, a major difference between our experimental conditions and pertinent literature relates to the use of 0.5 M H2S04 for ionic conductivity. Most of the gas diffusion electrode studies in the literature were carried out without liquid ionic conductor (i.e. the proton exchange polymer supplies the ionic conductivity in the catalyst layer). The 0.5 M H2S04 electrolyte while improves the ionic conductivity, which was essential especially for the RVC and uncompressed OF anode substrates, it could also increase the CH3O crossover flux to the cathode due to increased electro osmotic drag. Table 3-4 shows the performance of the conventional anode design with CCM utilized in the present work compares favorably with literature results, hence, validating the DMFC testing protocol. The catalyst mass specific peak power output of the reference CCM with a 1:1 Pt:Ru atomic ratio was 17.6 W g’ and 44.1 W g’ at loads of 40 g m2 and 10 g m2, respectively. Thus, in the case of the conventional anode, there was a significant loss in mass specific activity with catalyst load increase. The three- dimensional anodes gave maximum power outputs of 38.6 to 50.4 W g1. However, the Pt-Ru atomic ratios obtained by the micellar media assisted deposition on the three- dimensional substrates were around 4:1, while a lower ratio (e.g. 1:1) is more favorable at 333 K [26-28, 34]. Therefore, it could be argued that the performance of the DMFC could be further improved with the extended anodic reaction zone provided by the three-dimensional substrate, if a 1:1 Pt:Ru atomic ratio can be achieved by the electrodeposition method. This was also proven experimentally by Bauer et al. in the case of a different type of OF (i.e. compressed) subjected to an electrodeposition procedure carried out at high non- ionic surfactant concentration (i.e. 40 %) [16]. There has been very little literature information on the use of three-dimensional extended reaction zone anodes in direct fuel cells. Two recent studies employed Ti mesh substrates [21, 34]. Experiments by Allen et al. showed virtually no difference between the fuel cell polarization of Pt-Ru/Ti mesh (10 g m2) and conventional GDE (10 g m2 Pt) in DMFC operated at 363 K [21]. The power output on both area and anode catalyst mass basis was nearly identical for the Ti mesh and ODE. Shao et al. on the other hand, working with 0.25 and 0.5 M methanol concentration at 363 K as well, observed that the 157 Ti mesh supported Pt-Ru performed better than the GDE in the high current density range, where mass transfer related effects gain significance [34]. The mass specific activities of their catalysts were lower than those obtained in the present study. Therefore, it could be concluded that the electrode preparation method using micellar media gives a Pt-Ru catalyst morphology that provides high catalyst utilization on various three- dimensional substrates. Table 3-4: DMFC performance comparison between published data and results obtained in the present work. Temperature: 333 K. Anode feed: 1 M CH3O - 0.5 M H2S04 (present work), 1 M CH3O in water (literature). Membrane: Nafion® 117. Anode Pt-Ru Pt-Ru Peak Catalyst Mass Cathode Reference Catalyst Load Atomic Power Specific Peak pressure Layer (g m2) Ratio Output Power Output (barabs) (W m2) (W g’) Anode Type: Conventional Anode Pt-Ru Black 10 1.0:1 441 44.1 2.5 present work Pt-RuBlack 40 1.0:1 703 17.6 2.5 present work Pt-RulVulcan 10 1.0:1 300 30.0 1.0 [31] XC-72 Pt-Ru/Vulcan 35 1.0:1 510 14.6 1.0 [31] XC-72 Pt-Ru Black 40 1.0:1 500 12.5 1.0 [32] Pt-Ru Black 40 1.0:1 740 18.5 1.0 [33] Anode Type: Three-Dimensional Substrate (Extended Reaction Zone) RVC 12 3.6:1 486 40.5 2.5 present work GF 9.8 4.0:1 379 38.6 2.5 present work Ti Mesh 2.8 4.5:1 141 50.4 2.5 present work Ti Mesh 40 2.3:1 450 11.3 1.0 [34] 158 3.4 Conclusion The electrodeposition of Pt-Ru on three different three-dimensional substrates (Ti mesh, RVC and OF) was studied using colloidal media. The resulting electrodes were tested for electrocatalytic activity toward methanol oxidation. The combination of Triton X-100/isopropanol aqueous micellar electrodeposition media and perforated counter electrodes yielded a mesoporous Pt-Ru deposit morphology with high anode catalyst mass specific activity toward methanol oxidation as shown by both voltammetry and fuel cell experiments. The anode catalyst mass specific peak power output at 333 K was in the range of 38.6 to 50.4 W g’ corresponding to current loads of 170 to 220 A g’. The highest power densities were obtained with Pt-RuJRVC. It must be noted however, that RVC is a brittle material hence, the anode gasket thickness and cell compression have to be designed such that to avoid the crushing of the RVC. The electrode design proposed in the present work opens up the possibility of lowering the precious metal catalyst load in direct fuel cell anodes below 10 g m2. 159 3.5 References 1. A.S. Arico, S. Srinivasan, V. Antonucci, Fuel Cells 1(2001)133. 2. E. Herrero, K. Franaszczuk and A. Wieckowski, J.Phys.Chem. 98 (1994) 5074. 3. T. Iwasita, Electrochim. Acta 47 (2002) 3663. 4. A.B. Anderson, E. Grantscharova and S. Seong, J. Electrochem. Soc. 143 (1996) 2075. 5. H. Wang, C. Wingender, H. Baltruschat, M. Lopez and M.T. Reetz, J. Electroanal. Chem. 509 (2001) 163. 6. D. Cao, G.-Q. Lu, A. Wieckowski, S.A. Wasileski and M. Nuerock, J. Phys. Chem. B, 109 (2005) 11622. 7. B. Gurau and E. Smotkin, J. Power Sources 112 (2002) 339. 8. J. Ling and 0. Savadogo, J. Electrochem. Soc. 151 (2004) A1604. 9. K-W. Park, J-H. Choi, S-A. Lee, C. Pak, H. Chang and Y-E. Sung, J. Catal. 224 (2004) 236. 10. B. Gurau, R. Viswanathan, R. Liu, T.J. Lafrenz, K.L. Ley and E.S. Smotkin, J. Phys. Chem. B 102 (1998) 9997. 11. T.R. Ralph, G.A. Hards, J.E. Keating, S.A. Campbell, D.P. Wilkinson, M. Davis, J. St-Pierre and M.C. Johnson, J. Electrochem. Soc. 144 (1997) 3845. 12. P. Argyropoulos, K. Scott and W.A. Tanma, J. Appl. Electrochem. 29 (1999) 661. 13. M. Mastragostino, A. Missiroli and F. Soavi, J. Electrochem. Soc. 151(2004) A1919. 14. T. Hyeon, S. Han, Y.E. Sung, K.W. Park, Y.W. Kim, Angew. Chem. Tnt. Ed. 42 (2003) 4352. 160 15. D.P. Wilkinson, M.C. Johnson, K.M. Colbow and S.A. Campbell, US Patent 5,874,182, February 13 (1999). 16. a) A. Bauer, E.L. Gyenge and C.W. Oloman, Electrochim Acta 51(2006) 5356; b) A. Bauer, E.L. Gyenge and C.W. Oloman, J. Power Sources 167 (2007) 281. 17. T.T. Cheng, E.L. Gyenge, Electrochim. Acta 51(2006) 3904. 18. D.R. Lycke and E.L. Gyenge, Electrochim. Acta 52 (2007) 4287. 19. L.B. McCusker, F. Liebau and G. Engelhardt, Pure Appi. Chem. 73 (2001) 381. 20. I. Hodgson and C. Oloman, Chem. Eng. Sci. 54 (1999) 5777. 21. Z.G. Shao et a!., J. Electrochem. Soc. 153 (2006) A1575. 22. R.G. Allen et a!., J. Power Sources 143 (2005) 142. 23. C.L. Green, A. Kucernak, J. Phys. Chem. B 106 (2002) 1036. 24. H.C. Teh, G.H. Ong, S.C. Ng, L.M. Gan, Phys. Lett. A 78 (1980) 487. 25. H.C. Teh et a!., J. Dispersion Sci. Technol. 6 (1985) 255. 26. H.A. Gasteiger et. al., 3. Electrochem. Soc. 141 (1994) 1795. 27. H.A. Gasteiger et. al., J. Phys. Chem. 97 (1993) 12020. 28. P.N. Ross, in ‘Electrocatalysis’, p.66-’72, Wiley-VCH, New York (1998). 29. E.L. Gyenge, J. Power Sources 152 (2005) 105. 30. Z. Qi and A. Kaufman, J. Power Sources 110 (2002) 177. 31. V. Bag!io et. a!., mt. J. Electrochem. Sci. 1(2006) 71. 32. R. Jiang, H.R. Kunz, J.M. Fenton, J. Electrochem. Soc. 153 (2006) A1554. 33. B. Gurau and E.S. Smotkin, J. Power Sources 112 (2002) 339. 34. Z.G. Shao et a!., Phys. Chem. Chem. Phys. 8 (2006) 2720. 161 4 EFFICIENT ANODES WITH LOW Pt-Ru LOAD FOR DIRECT METHANOL AND FORMIC ACID FUEL CELLS* 4.1 Introduction The DMFC and DFAFC are attractive alternatives to the hydrogen PEMFC due to the high thermodynamic energy densities of methanol (4.4 kWh L’) and formic acid (2.1 kWh L’) coupled with having a simpler fuel storage, transportation and refueling infrastructure [1-5]. However, both systems, at present stage of development, are most suitable for portable electronic applications, due to much lower power outputs compared to the hydrogen cell. There are a number of significant differences between the two fuel cells making them potential competitors. The crossover flux of HCOOH across Nafion® membranes is lower compared to CH3O . The steady-state flux of 1 M CH3O and HCOOH through Nafion® 117 at room temperature are about 3-6 xlW2 mol m2 s’and 2x104 mol m2 s, respectively [5-7]. Thereby, the cathode potential and the fuel utilization are both theoretically higher for DFAFC vs. DMFC. Regarding the anode kinetics, while both methanol and formic acid oxidation reactions can proceed according to the dual pathway mechanism either by a COad or non COad intermediate route, the network of surface reactions and intermediates is more complex for methanol due to the total 6e exchange. The electrocatalytic aspects of both methanol and formic acid have been intensely investigated [2, 8-15]. See also Section 1.3.1 and 1.3.2. Due to the formation of CO intermediate/poison and the resulting sluggish reaction kinetics of methanol electro-oxidation, DMFC’s with the conventional GDE anode design coupled with a common carbon support for the catalyst, such as Vulcan XC-72, typically require Pt-Ru mass load of 30 g m2 or higher to achieve power densities in the range of 500 to 750 W m2 at 333 K whilst at 10 g m2 the ower densities fall below 500W m2 [1, 16-18]. * A version of this chapter has been published: T.T. Cheng, E.L. Gyenge, “Efficient Anodes for Direct Methanol and Formic Acid Fuel Cells: The Synergy between Catalyst and Three-Dimensional Support”, J. Electrochem. Soc. 155 (2008) B819-828. 162 DFAFC’s on the other hand can achieve power density exceeding 700 W m2 at 303 K with Pt-Ru mass load of 30 g m2, while higher power outputs are achievable with other catalyst formulations that more suitable for HCOOH oxidation kinetics (Pt-Sn, Pt Au, Pt-Bi, Pt-Pd, Pd, and Pd-Au) [4, 7, 19-21]. Recently, in addition to electrocatalysis, the anode design has received increased attention. Scott et al. investigated the feasibility of using Ti mesh as anode substrate for DMFC and DFAFC, aiming to enhance the CO2 gas disengagement by having a wider range of two-phase flow regimes than in the typical GDE [22, 23]. Similarly, Shao et a!. electrodeposited Pt-Ru on Ti mesh in 10 mM H2S04 in a 3-electrode electrochemical setup and obtained enhanced performance in a DMFC with a low concentration of methanol (0.5 M or lower) [24-26]. At a cell voltage of 0.3 V and an operation temperature of 363 K, the current density of the conventional anode was approximately 2200 A m2 whereas the Ti mesh anode reached 2500 A m2. Gyenge and collaborators proposed the use of carbon-based three-dimensional substrates such as RVC and GF, and demonstrated their potential as DMFC anodes [18, 27, 28]. The concept of 3-D extended reaction zone anode poses two major challenges. Firstly, a protonic conductive network is required across the three-dimensional electrode (thickness between 200 to 2,000 pm), which cannot be easily achieved with the typically employed Nafion solid polymer electrolyte due to high ohmic loss. Secondly, the 3-D substrates require a catalyst preparation method that is capable of producing nanoparticle deposits with uniform Pt-Ru ratio and mass load throughout the extended anode thickness. In other words, support specific catalyst preparation techniques must be developed. The electrodeposition of high surface area metallic (such as Pt, Pd, and Co) nanoparticles onto flat gold working electrode from liquid crystal solutions, based on non-ionic surfactants Briji® 56 and octaethyleneglycol monohexadecyl ether, have been previously studied and published by Attard and coworkers [29-33]. Likewise, the preparation of Ni electrodeposits onto polished Ni foil substrate using Triton X-100 liquid crystalline templating technique was demonstrated by Ganash and Lakshminarayanan [34]. In contrast to electrodepositing onto flat substrate, the authors of the present study have previously employed the use of micellar solution-based (Triton X 1 00/isopropanol) electrodeposition media and synthesized mesoporous Pt-Ru catalyst on 163 three-dimensional RVC, OF, and Ti mesh supports [18, 281. The goal of the present investigation was to enhance both the area-specific and anode catalyst-mass specific performance of the three-dimensional GF anode by modifying the catalyst characteristics, (such as Pt:Ru ratio), morphology, and crystallography. This strategy, allows significant reduction of the precious metal catalyst load with improved power output compared to the conventional GDE designs (either CCM or CCDL). 4.2 Experimental Section 4.2.1 Anode Matrix: Graphite Felt Catalyst Support The OF, with uncompressed thickness of 3 mm, was provided by Test Solutions through Electrolytica Inc. The GF was produced using polyacrylonitrile fibers and had a carbon content of 99.7%. The surface area measured by BET nitrogen adsorption was 0.7 m2 g’. The OF substrates were pretreated by sonication in methanol for 30 minutes and were washed thoroughly in distilled water followed by drying in air. 4.2.2 Electrodeposition Procedure The Pt-Ru electrocatalyst supported on OF was prepared by galvanostatic electrodeposition using a micellar solution composed of a non-ionic surfactant (either Triton X-100: C14E09.5or Triton X-102: C14E02)and an aqueous phase with individual H2PtC16 and (NH4)2RuC16 concentrations from 0.25 to 0.75 mM (depending on the desired catalyst load and composition). A typical deposition media contained 12.5 vol% Triton X-102 and 87.5 vol% aqueous phase with H2PtC16 and (NH4)2RuC16.The chemicals used were reagent grade from Sigma-Aldrich and were used as delivered without further purification processes. The micellar solutions were prepared by mixing the Triton X- 102 and metal precursor solutions in a water-jacketed glass vessel at 341 K for 30 minutes. The electrode assembly for deposition consisted of the OF working electrode and two perforated platinized titanium counter electrodes (anodes), with a geometric surface 164 area each of 5 cm2. The counter electrodes had --10 holes per cm2 and hole dimensions of --‘0.03 cm2. The electrodes were placed at a distance of 1 cm and were inserted into the glass vessel after no visible change was observed. The electrodepositon was carried out with a Xantrex XHR15O-7 DC power supply capable of operating at 0-150 V and 0-7 A. The post-deposition treatment was discussed previously elsewhere [18, 35]. The GE was sonicated in THE (Reagent Grade, Sigma Aldrich) for 5 minutes to wash out the surfactant retained in the porous matrix. The GE was then washed thoroughly with distilled water and dried in air, followed by heat treatment in a N2 stream for 1 hour at 573 K [35]. 4.2.3 Electrochemical Measurements Voltammetry, CP, and CA were carried out at 298 and 333 K in a water-jacketed electrochemical cell connected to a circulating water bath with a three-electrode setup. Reference electrode Hg/Hg2SO4,K2S04,std.(MSE), working electrode 1 cm2 geometric area GF with electrodeposited Pt-Ru and a platinum wire (—-0.5 cm2) acting as counter electrode. The electrolyte was 50 mL of 1.0 M CH3O (or HCOOH) and 0.5 M H2S04. The VoltaMaster 4 software was used in conjunction with the computer-controlled VoltaLab PGZ4O2 potentiostat in the experiments. All potentials in the present work are reported vs. SHE reference: E (V vs. SHE) = E (V vs. MSE) + 0.640. The catalyst was electrochemically pre-treated for surface conditioning and removal of impurities. Three potential steps were successively applied for 10 s each in 0.5 M H2S04 at 1.28 V, 1.20 V and 0.05 V, respectively. The electrochemical conditioning/cleaning cycle was repeated three times. The effective electrochemically active Pt-Ru surface area was estimated by the Cu UPD and stripping technique presented by Kucernak and Green [36]. It was previously shown to be a promising method to determine the active surface area for Pt-Ru catalysts deposited on three-dimensional carbon substrates [18, 27, 28, 35]. Assuming complete Cu monolayer coverage on both Pt and Ru sites and an anodic Cu stripping charge of 4.2 C m2, the charge differences between the reference scan and Cu stripping was used to calculate the active surface area. 165 4.2.4 Conductivity Measurements The ionic conductivity of sulfuric and formic acid mixtures was measured at 333 K using a YSI 3200 conductivity meter. The conductivity probe was immersed in the solutions for 30 seconds before recording to conductivity measured. A 0.01 M KC1 standard solution (Thermo Orion) was used to calibrate the conductivity probe. 4.2.5 Surface and Analytical Characterization of the Catalysts Visual images of the GF samples were captured by the Hitachi S4700 high resolution SEM. Fragments of the deposited GF were flush mounted onto SEM stubs with carbon adhesive. The visual images were generated using a beam accelerating voltage of 2000 V and an emission current of 1.25x107A along with a working distance of 0.0025-0.0035 m. The catalyst mass load was determined by ICP-AES using a Perkin Elmer Optima, model 3300DV instrument with Pt-Ru solutions obtained from aqua regia digestion (HC1:HNO3 volume ratio of 3:1) of the prepared GF samples. The crystallographic features of the prepared catalysts were determined by X-ray diffraction (XRD) using an Advanced Bruker powder X-ray diffractometer with Cu K radiation wavelength of 1.5418 A. The XRD experiments were performed with 20 values from 10 to 85° with a stepping of 0.04°. 4.2.6 Membrane Electrode Assembly and Fuel Cell Experiments Fuel cell experiments were carried out with a Fideris Inc. MTK test station using the FC Power® software. The MEA was prepared with a commercial half MEA (Lynntech Inc.) pre-painted on the cathode side with 40 g m2 unsupported Pt (i.e., Pt black). The Pt-Ru electrodeposited GF anode was not physically bonded to the membrane and no backing layer was used in conjunction with the GF. These aspects contribute to cell assembly simplification that was one of the goals of this study. The electrodes had a geometric surface area of 5 cm2. In the cases of DLFC, the membrane (Nafion® 117) was conditioned in 0.5 M H2S04both ex-situ (for 24 hours at 298 K half-MEAs were exposed to the acid solution) and in-situ (i.e., MEAs installed in the fuel cell with the GF anode for 2 hours at 333 K). 166 The fuel cell was assembled with two gold-plated end plates with serpentine flow fields, the custom MEA containing the extended reaction zone anode, and an Elat® carbon cloth (E-Tek Inc.) functioning as the cathode backing layer. For anode design comparison, commercial CCM and CCDL, both from Lynntech Inc., with pre-painted 10 or 40 g m2 unsupported Pt-Ru (1:1 atomic ratio) on the anode side and identical cathodes were also tested. The fuel cell experiments were carried out at a temperature of 333 K with oxygen flow rate of 500 mL min1 at 2.5 bar absolute pressure on the cathode side. The oxygen was of dry medical grade, supplied by Praxair Inc. For DMFC, the anolyte consisted of 1 M CH3O and 0.5 M H2S04fed in the cell at 2 mL min1 and atmospheric pressure. For the DFAFC, HCOOH (1, 3, and 10 M) with or without 0.5 M H2S04was fed at a rate of 6 mL min* 4.3 Results and Discussion 4.3.1 Effect of Non-Ionic Surfactant Type and Micellar Media on the Pt-Ru Electrodeposition A Triton X-100/isopropanol-based micellar media was previously investigated (Chapter 3) for the electrodeposition of Pt-Ru on various 3-D substrates, including RVC, GF, and Ti mesh [18]. The micellar solution was composed of 75 vol% aqueous phase and 25 vol% Triton X-100 / isopropanol mixture in a 1:4 volume ratio (i.e., 6 wt% Triton X-100 and 17 wt% isopropanol). The precursor metal salt concentration was 0.25 mM for both H2PtC16 and (NH4)2RuC16 (i.e., 1:1 bulk atomic ratio). Since the deposition temperature of 341 K exceeded the cloud point temperature of Triton X- 100 (339 K) [18, 37, 38], the initial intention of isopropanol addition was to avoid the surfactant-aqueous phase separation. However, due to the reductive nature of isopropanol the effective metal ion concentration available for electrodeposition was lower then the initial amount present in the system. Furthermore, it has been observed that the colloidal Pt-Ru formed by homogeneous reduction with isopropanol was not adsorbing efficiently on any of the explored three-dimensional substrates. 167 The goal here therefore, was to develop an isopropanol-free micellar solution for templated electrodeposition of Pt-Ru. In order to prevent phase separation in the new media, Triton X-102[C8H17640(C)2]with 12 hydrophilic ethylene oxide units per chain (vs. 9 — 10 of Triton X- 100) and a higher cloud point temperature of 361 K was chosen. Comparative galvanostatic electrodeposition experiments on GF were performed at a deposition superficial current density of 20 A m2, temperature of 341 K and duration of 120 minutes (these conditions are identical to those employed previously using Triton X-100 / isopropanol [18]). Table 4-1 compares the mass load, bulk Pt-Ru atomic ratio, and specific surface area of the Pt-Ru samples obtained from the two media at different concentrations, while Figure 4-1 shows their respective voltammograms of methanol electrooxidation based on the real catalyst surface area. Comparative specific surface area data by Cu UPD in Table 4-1 showed that the catalyst samples prepared using Triton X 102 yielded lower specific surface areas, which was attributed to the lack of mesoporous structure (observed in the case of Triton X-100 I isopropanol [18]) and the build up of a thicker catalyst layer. It must be noted that the surface Pt-Ru ratio obtained by Cu UPD was essentially identical to the bulk ICP-AES results (see Figure 4-2). For instance, the catalyst prepared with 5 vol% Triton X-102 had a surface Pt-Ru ratio of 9.3:1 by Cu UPD (see below). 168 Table 4-1: Electrodeposition of Pt-Ru on GF from micellar solution: comparison between Triton X-100/isopropanol and Triton X-102. N E Electrodeposition Conditions — 20 A m2, 120 mm, 341 K; Precursor Concentrations — Pt: 0.25 mM, Ru: 0.25 mM Surfactant Mass Loading Bulk Pt:Ru Specific Surface (g m2) Atomic Ratio Area (m2 g’) A 5vol%TritonX-100/ 9.8 4.0:1 36.3 20 vol% Isopropanol B 5vol%TritonX-102 18 9.2:1 7.6 C 12.5 vol% TritonX-102 14 7.9:1 9.2 D 25 vol% Triton X-102 4.3 7.3:1 16.2 16 A B C D 6 vol% Triton X-100/ 20 vol% Isopropanol S vol% Triton X-102 12.5 vol% Triton X-102 25 vol% Triton X-1 02 3.0 2.5 2iJ N E 1.5 i.o 0.5 0.0 0.0 __________ C 14 12 - 10 /// 6- 1 0 -2 0.0 0.2 0.4 0.6 0.6 I— — ElY vs. SHE Figure 4-1: Voltammograms of methanol electro-oxidation using Pt-Ru electrodeposited on GF: Effect of micellar electrodeposition media. Solution: 1 M CH3O and 0.5 M H2S04. Temperature: 298 K. Scan rate: 0.005 V s. Inset: enlarged view from 0 to 0.5 V. 169 250 200 150 E 100 50 0 1.0 E/ V vs. SHE Figure 4-2: Blank scan and Cu UPD stripping curves of electrodeposited GF prepared at 341 K from 5 vol% Triton X-102 micellar media with a superficial deposition current density and time of 20 A cm2 and 120 minutes. Test solution: 0.5 M H2S04 (blank); 0.5 M H2S04+ 0.002 M CuSO4 (Cu UPD) Temperature: 298 K; Scan rate: 0.050 V s1. In spite of smaller specific surface area, the intrinsic catalytic activity of methanol electrooxidation on real surface area basis, was higher for the catalysts electrodeposited in the presence Triton X-102 compared to the Triton X-lOO/isopropanol media. Firstly, the higher Pt:Ru ratio obtained ( 7 to 9:1) for the catalysts prepared with Triton X-102 was more favorable as the optimal surface ruthenium content for methanol electro oxidation at 298 K was previously reported by Gasteiger et al. to be 10 at.% [39, 40]. Secondly, the crystallographic features of the catalysts prepared with the two micellar media also differed. The XRD spectra (Figure 4-3) shows that electrodeposition 0.0 0.2 0.4 0.6 0.0 170 from the two micellar solutions yielded polycrystalline Pt with Miller indices of (111), (100), (110), and (311). As shown by Table 4-2, electrodeposition using Triton X- 102 led to a higher fraction of Pt(1 11), Pt(1 10), and Pt(3 11). Thus, the superior activity of the catalysts prepared using Triton X-102 relative to those obtained with Triton X-100 could be partially attributed to the differences in crystallographic features. It was reported by various research groups that the methanol electro-oxidation activity decreased and the oxidation onset potential increases in the order of Pt( 111) > Pt(110) > Pt(100) [41-44]. Furthermore, higher-index facets were found to possess excellent catalytic activity for the electro-oxidation of small organic molecules [45]. Therefore, the changes in the crystallographic features (Table 4-2) agreed well with the differences in intrinsic catalytic activity (Figure 4-1). Using the XRD data along with Scherrer’ s equation, the Pt( 111) crystallite size was estimated to be 6.0 nm for the Triton X-100/isopropanol case and 14.2 nm for 12.5 vol% Triton X-102, explaining the difference in specific surface area measurements by Cu UPD (Table 4-1). See Appendix C for crystallite size calculations. Table 4-2: Effect of the micellar deposition media on the Pt crystallographic for the GF supported catalysts. Electrodeposition Conditions — 20 A m2, 120 mm, 341 K Miller indices 5 vol% Triton X-100 / 12.5 vol% Triton 12.5 vol% Triton 20 vol% Isopropanol X- 102 X- 102 (Pt: 0.25 mM (Pt: 0.25 mM (Pt: 0.75 mlvi Ru: 0.25 mM) Ru: 0.25 mlvi) Ru: 0.75 mM) (111) 34.6% 39.7% 38.2% (100) 43.2% 27.6% 24.6% (110) 14.4% 17.1% 17.2% (311) 7.8% 15.5% 20.0% 171 >Cl) C -I C > Cl) C a) C 35 40 45 50 55 60 2-Theta Figure 4-3: XRD spectra of Pt-Ru prepared from 5voI%/2Ovol% Triton X 100/isopropanol and 12.5 vol% Triton X-102. Precursor concentration — Pt: 0.25 mM, Ru: 0.25 mM. 65 70 75 80 85 172 From the shift in Pt reflection and Vegard’s law, the degree of alloying of the catalyst samples was determined. The lattice parameter, a, for pure Pt is 3.92 A and its dependence on the Ru content has previously been published [16, 46-48]. In the case of carbon-supported Pt-Ru alloy, which is of relevance of the present study, the change in lattice parameter was found to be more significant at low Ru content compared to unsupported Pt-Ru alloy [47]. The lattice parameter of the samples prepared from Triton X-100/isopropanol and Triton X-102 were estimated to be 3.91 and 3.90 A, respectively. The small changes in lattice parameter could be translated to < 10% of Ru in the Pt face- centered cubic structure. Considering the presence of the Ru hexagonal phase and the Pt- Ru atomic ratio of the catalysts, it is evident that the catalysts prepared with the present method consisted of Pt-Ru alloy and in higher proportion dispersions of metallic Pt and Ru. The trends shown by Table 4-1 whereby both the catalyst mass load and the Pt-Ru atomic ratio decreased with increasing Triton X-102 concentration, was further studied by carrying out electrodeposition on GF using linear voltammetry. Figure 4-4 a exemplifies the effect of Triton X-102 on the Pt electrodeposition voltammogram. The voltammetric wave corresponding to Pt electrodeposition, along with that of hydrogen evolution, was greatly suppressed with the addition of Triton X-102. In contrast, the addition of Triton X-1 02 suppressed to a lesser extent the electrodeposition of Ru (see Figure 4-4 b). This voltammetry study agreed very well with the characteristics of the current Pt-Ru samples and with findings obtained in a previous study on Triton X- 100 by Gyenge and collaborators [28]. 173 A 0voI%TrcnX-102 — — 6.25vo1%TritonX-102 —-a—— 12.SvoI%TritonX-11J2 ——u-——-- 18.75vo1%TritonX-102 -0.2 0.0 02 0.4 0.6 E/Vvs. SHE A 0voI%TritonX-102 6.25 vol%Triton X-102 12.5 vol%Triton X-102 ——a——— 10.75 vol%Tritan X-102 (a) ct E (b) N E 10 0 -10 -20 -30 — -GA 10 — 0 -10 -20 30. -40 . — -0.4 -0.2 0.0 0.2 0.4 0.6 E/Vs. SHE Figure 4-4: Effect of Triton X-102 content. (a) Cathodic voltammograms of Pt electrodeposition on GF with 0.75 mMH2PtC16;(b) Cathodic voltammograms of Ru electrodeposition on GF with 0.75 mM (NH4)RuC1 Temperature: 341 K. Scan rate: 0.005 V s. 174 4.3.2 Effect of Metal Precursor Concentration on the Electrodeposition from Triton X-102 Micellar Media and the Associated Electrocatalytic Activity for Methanol Oxidation The metal precursor concentrations along with the Pt-Ru ratio in solution had a major impact on the catalyst dispersion and the resulting mass load. The electrodeposited catalyst characteristics are presented in Figure 4-5. Mass Loading (g m2): 30 Mass Loading (g m2): 10 Pt-Ru Atomic Ratio: 8.8:1 Pt-Ru Atomic Ratio: 4.0:1 0.75 — Specific Surface Area (m2 gj: 3.0 Specific Surface Area (m2g1): 16.3 Pt(mM) Mass Loading (g m2): 14 Mass Loading (g m2): 3.8 Pt-Ru Atomic Ratio: 7.9:1 Pt-Ru Atomic Ratio: 3.1:1 Specific Surface Area (m2 g’): 9.2 Specific Surface Area (m2 gj: 17.5 0.25 I I 0.25 0.75 Ru (mM) Figure 4-5: Effect of metal precursor concentration in the electrodeposition media on the Pt-Ru catalyst characteristics. Superficial current density 20 A m2,341 K. The Pt-Ru amount on GF increased more than two folds (from 14 to 30 g m2) when the concentration of H2PtC16 in solution increased from 0.25 mM to 0.75 mM, while the concentration of (NH4)2RuCl6was held constant at 0.25 mM. The increase in mass load was associated with a catalyst morphology that could be characterized as intergrown nanoparticles forming a fairly dense film as shown by high-resolution SEM 175 images (Figure 4-6). This is further supported by the low specific surface area of 3.0 m2 g’. The Pt-Ru atomic ratio increased only slightly to 8.8:1 from the original 7.9:1 in spite of a three-fold increase in platinum precursor concentration in the solution. This clearly shows the increase ofH2PtC16 concentration enhanced not only the deposition of Pt but that of Ru as well. Such an effect could possibly be due to an increase in sensitization of the graphite felt surface creating extra nucleation sites. 0.75 - Pt(mM) 0.25 0.25 Ru (mM) 0.75 Figure 4-6: SEM images of Pt-Ru electrodeposited on GF prepared with 12.5 vol% Triton X-102 and different precursor concentrations. Superficial current density 20 Am2,341K. I I 176 When the concentration of (NH4)2RuC16in solution was increased from 0.25 mM to 0.75 mM, while the concentration ofH2RuC16was held constant at 0.25 mM, the mass load dropped to 3.8 g m2 and the Pt-Ru atomic ratio decreased to 3.1:1, signifying the inhibitory effect of the excess (NH4)2RuC16concentration on the electrodeposition of Pt. The low catalyst load on the GF surface is also reflected by the comparatively sparse deposit observed using high-resolution SEM (Figure 4-6). Interestingly, in the case of both metal precursors at the highest concentration of 0.75 mM, the total mass load remained fairly low (average of 10 g m2, see Table D-1 in Appendix D for details) and the Pt-Ru atomic ratio was 4.0:1 (Figure 4-5). The decrease in mass load further supported the inhibition effect by (NH)2RuC16However, the inhibition effect led to a more uniform catalyst coating as well as higher specific surface area (16.3 m2 grn’). The area increase was attributed to the morphology and enhanced catalyst dispersion shown in Figure 4-6. Instead of intergrown particles forming a network, well-dispersed individual particles and agglomerates made up of small particles (<10 nm) with diameters of 10 to 20 nm were obtained. Antonlini et al. and Jiang et al. have also previously reported a similar effect, i.e. Pt agglomeration was found to be inhibited by Ru, which resulted in finer and more evenly dispersed particles [47, 48j. The catalytic activity toward methanol oxidation for the Pt-Ru samples presented by Figures 4-5 and 4-6, was evaluated by comparative voltammetry at 298 K (Figure 4-7) and CP at both 298 K and 333 K (Figure 4-8). The intrinsic catalytic activity at 298 K (as measured by both voltammetry and CP), was the highest for the sample prepared with 0.75 mM H2PtC16 and 0.25 mM (NH4)2RuC16(curve B). This was attributed to its Pt-Ru atomic ratio of 8.8:1. On the other hand, the sample prepared with 0.75 mM H2PtCI6and 0.75 mM (NH4)2RuCl6(Pt-Ru atomic ratio of 4.0:1), curve D, had a higher catalytic activity at 333 K. This agreed well with the well-known synergy between temperature effect and surface Ru content caused by the change in rate-determining step from methanol adsorption to surface reaction between COad and OHad [16, 39, 40]. 177 40 10 0 1W -100 50 30 20 A Deposition Current: 20A m 0.25 mM H2PtC 0.25 mM (N HR uC -2 — — — Deposition Current: 20A m 0.75mM l-t2PtC 0.25 mM (N HRuC — —— — DepositionCurrent:20Am 025mM D 0.75 rnM(NHRuC Deposition Current: 20A m2 0.75mMH2PtC 0.75 mM (N HRuC (a) N E (b) -10 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 E/Vvs. SHE 5W 4W A B C Depition Current Am2 0.25 mM H2PtC 0.25 mM(NH4)RuCI Depition Current A m2 0.75 mMH2PtC 0.25 mM(NH4)RuCI -2Depition Current Am 0.25 rrmMHPtC 0.75 mM(NHRuCI -2Depition Current Am 0.75 mMH2PtC 0.75 rnM(NHRuCl 2W D 0 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 E/Yvs. SHE Figure 4-7: Voltammograms of methanol electro-oxidation using Pt-Ru electrodeposited on GF: Effect of precursor concentration and deposition current. (a) Real surface area basis; (b) mass basis. Solution: 1 M CH3O and 0.5 M H2S04. Temperature: 298 K. Scan rate: 0.005 V s. 178 0.4 03 1;’ I ft It 0.2 I I ‘1 ii I,(I -f — - c - -A — t .-r.rrz=z.rzrz.tr: A .. -2Deposnion Current 20 Am Precursor Concentration: 025 mM H-RC and 0.25mM (NH4)2RuC 5 -2— — — Depoeion Current 20 Am Precursor Concentration: 0.75mM HPtC and 025 mM(NH4)RuCI C_ -2— — — Depoetion Current 20 Am Precursor Concentration: 0.25 mM H2PtC and 015 mM (NH4)2RuCI5 D -2 Depoetion Current 20 A m Precursor Concentration: 075 mM H-PtC end 075 mM (Nl-t4)2RuCI5 (a) LU 0) ci, > > LU 0.1 0.0 -0.1 (b) 0.35 0.30 0.25 LU o 0.20 > > 0.15 0.10 0.05 0.00 0 50 100 150 t/s 200 250 300 350 A Deposition Currant 20 Aol2 — 0.25mM HPtCl6end 0.25mM (NHj2RuC — — Deposition Current 20 Ag2 D 0.75 mM HPtCI6 and 0.25mM (NH4)2RuCC — — — Deposition Current 20 Ag2 9 0.25mM H2PtCI5 and 0.75mM (NH4)2PuCb — — — — Deposition Current 20 Am2 015 mM H2PtC and 0.75mM NH4)2PuC I I I I I I 0 50 100 150 200 250 300 350 tls Figure 4-8: Chronopotentiometry data of methanol electro-oxidation using Pt-Ru electrodeposited on GF: Effect of precursor concentration and deposition current. Solution: 1 M CH3O and 0.5 M H2S04. Current Density: 50 A m2. (a) Temperature: 298 K; (b) Temperature: 333 K. 179 Furthermore, as shown by Figure 4-9 and Table 4-2, there was a subtle difference in the crystallographic features of the catalysts prepared with low and high Pt-Ru precursor concentrations (i.e., 0.25 mM each vs. 0.75 mM each). It was found that the Pt(31 1) content was higher for the sample prepared using 0.75 mM Pt and Ru salt precursor concentrations. Moreover, the lattice parameter of the sample prepared with 0.75 mM Pt and Ru precursor was 3.92 A, virtually identical to that of pure Pt, indicating that it was composed of dispersed Pt and Ru and not Pt-Ru alloy. It was previously reported by Dubau et al. that dispersed Pt and Ru (or bimetallic Pt-Ru) on carbon support had higher activity than supported Pt-Ru alloy [16]. The higher fraction of bimetallic Pt- Ru as well as Pt(3 11) content translated into higher activity on both real surface area and catalyst mass basis (compare curves D and A in Figure 4-7). The mass specific current density reached 150 A g’ at 0.6 V vs. SHE (see Table 4-3 for kinetic parameters and Appendix D for Tafel plot). As shown in Table 4-3, the apparent exchange current density was in the order of i0 as compared to i0 _108 found in some published studies (see Table 1-5). Even though the test solution was not identical, the significant increase in exchange current might suggest enhanced kinetics with the novel anode. >, U) C a) C 35 40 45 50 55 60 65 2-Theta Figure 4-9: XRD spectra of Pt-Ru prepared from 12.5 vol% Triton X-102. Precursor concentration — Pt: 0.75 mM, Ru: 0.75 mM. 70 75 80 85 180 Table 4-3: Apparent Tafel slope and exchange current density of methanol electro oxidation at 298 K on Pt-Ru/GF prepared from 12.5 vol% Triton X-102 with 0.75 mMH2PtC16and (NH4)2RuC16.Solution: 1 M CH3O and 0.5 MH2504. Catalyst Tafel Slope, b Exchange Current Density, i0 (V) (A m2reai) Pt-Ru!GF 0.165 l.7x103 4.3.3 Catalyst Distribution and Penetration in the Three-Dimensional Graphite Felt Support The Pt-Ru distribution and penetration for the GF supported catalyst showing the highest activity at 333 K (prepared with 0.75 mM H2PtC16 and 0.75 mlvi (NH4)2RuC16 was explored using Hi-Res SEM and ICP-AES. Catalyzed fibers from the front face, middle, and back face were randomly pulled out at the top left, bottom left, middle right, and the center of the GF for SEM imaging (see Figure 4-10 for schematic of locations labeled as 1 to 12). See also Appendix B (Figures B-i c,d) for the SEM images Pt-Ru/GF prepared without the presence of Triton X-102. In the case with 12.5 vol% Triton X-102, as shown from the corresponding SEM images (Figure 4-11), the catalyst morphology was found to be very uniform throughout the GF regardless of location and depth. On the other hand, the Pt-Ru atomic ratio was found to vary somewhat with depth, with the middle section being slightly more platinum-rich (i.e., the Pt:Ru atomic ratio was approximately 0.3 points higher in the middle than on the exterior faces adjacent to the counter-electrodes). This is due to the non-uniform potential distribution in the three-dimensional matrix during electrodeposition. In the case without Triton X- 102, large and rough catalyst agglomerates (up to 1000 nm) was found on the exterior face of the GF, while sparse coatings of 50-nm particles were observed in the interior, leading to significantly lower overall and specific surface area of 58.4 m2 m2, and 14.6 m2 g’, respectively (the catalyst load and Pt-Ru ratio was 4.0 g m2 and 5.2:1); hence, resulted in inferior catalytic activity (see Appendix D). 181 Front Face 22mm 4 3mm Back Face Catalyzed Graphite Felt Location of Fiber Pullout Figure 4-10: Locations of fiber pullout for SEM imaging of the Pt-Ru catalyst dispersion and penetration in the three-dimensional GF support (see Figure 4-11). 182 Figure 4-11: Pt-Ru dispersion and penetration into the three-dimensional support: SEM images of catalyzed GF at locations 1 to 12 (see Figure 4-10). Sample prepared at 20 A m2 with 12.5 vol% Triton X-102, 0.75 mM H2PtC16, and 0.75 mM (NH4)RuCI6,341 K. 183 4.3.4 Direct Methanol Fuel Cell Experiments Catalyst-Coated Membrane (CCM) Catalyst-Coated Diffusion Layer (CCDL) Extended Reaction Zone (e.g. GF) ___ Diffusion Layer (Carbon Cloth) Catalyst (Pt-Ru) Membrane ‘ (Nafion®1 17) Figure 4-12: Schematic of Different Anode Designs — CCM, CCDL, and extended reaction zone anodes. The CCM anodes with 10 and 40 g m2 Pt-Ru black mass load and with a Pt-Ru atomic ratio of 1:1 (which is the most common and arguably optimal at intermediate and high temperature [2, 16, 49]) had peak power densities on geometric area basis of 442 and 703 W m2, respectively (or 44.2 and 17.6 W g’ vs. the anode catalyst load) as The best performing GF-supported catalyst at 333 K (see Figure 4-10) was compared in single-cell DMFC experiments with conventional CCM and CCDL. Figure 4-12 shows schematically the differences between the three anode designs, outlining the extended reaction zone created with the three-dimensional support. 184 indicated by Figure 4-13 and Table 4-4. The other typical electrode design, CCDL, performed worse than the CCM, gave a peak power density of 278 W m2 (Figure 4-13). 0.70 0.65 0.60 0.55 0.50 0.45 > 0.40 0.35 1) w 0.30 0.25 0.20 0.15 0.10 0.05 0.00 600 750 700 650 600 550 500 450 400 350 300 250 200 150 100 50 0. 0.70 0.65 0.60 0.55 0.50 0.45 0.40 > 0.35 0.30 125 0.20 0.15 0.10 0.05 600 1750 700 650 600 550 500 450 400 350 300 250 200 150 100 50 —0 3000 Figure 4-13: DMFC performance at 333 K — effect of extended reaction zone vs. conventional anode design. Fuel: 1 M CH3O in 0.5 M H2S04 at 2 mL mh11. Cathode: 40 g Pt m2, dry O fed at 2.5 bar and 500 mL mm1.(a) Polarization curve; (b) Power density. (a) • CCM-10 gnf2Pt-Ru(1.0:1) v CCM-40 gm2 Pt-Ru (1.0:1) n ccDL-4o9mCPtRu(1.0:1) 0 GF -10 q m2 Pt-Ru (4.0:1) ly V ________________ 9 V 5 8 C 0 .007 . C V C . (b) 0 250 500 750 1000 1250 1500 1750 2000 2250 2500 2750 3(130 I / A m2 n fin 0 V Q ye ye 0. • 0• . V • 8 55 C • CCM-10gmPt-Ru(1.01) v CcM-4oqrn2Pt-Ru(1.o:1) 5 000L-4OgmPt-Ru(1.0:1) 0 SF-lU g m Pt-Ru (4.0:1) cI 6 0 250 500 750 1000 1250 1500 1750 2000 2250 2500 2750 / A m2 185 Table 4-4: DMFC performance comparison between published data and results obtained in the present work. Temperature: 333 K; Anode feed: 1 M CH3O - 0.5 M H2S04 (present work), 1 M CH3O in water (literature). Membrane: Nafion® 117. Anode Catalyst Pt-Ru Pt-Ru Peak Catalyst Mass Cathode Ref. Layer Load Atomic Power Specific Peak pressure (g m2) Ratio Output Power Output (barabs) (W m2) (W g’) Anode type: Conventional CCM/CCDL Pt-Ru Black 10 1.0:1 441 44.1 2.5 present (CCM) work Pt-Ru Black 40 1.0:1 703 17.6 2.5 present (CCM) work Pt-Ru Black 40 1.0:1 278 7.0 2.5 present (CCDL - No Hot work Press) Pt-RulVulcan 10 1.0:1 300 30.0 1.0 [1] XC-72 (CCDL) Pt-Ru/Vulcan 35 1.0:1 510 14.6 1.0 [1] XC-72 (CCDL) Pt-Ru Black 40 1.0:1 500 12.5 1.0 [50] (CCDL) Pt-Ru Black 40 1.0:1 740 18.5 1.0 [51] (CCDL) Anode Type: Three-Dimensional Substrate (Extended Reaction Zone) Pt-Ru/GF 10 4.0:1 741 74.1 2.5 present work 186 The results for the conventional electrode designs are corroborated by literature information as detailed in Table 4-4. The literature results had conditions identical to the present work, e.g. Nafion® 117 membrane, operating temperature of 333 K and Pt-Ru catalysts with loads between 10 to 40 g m2, except for some variation regarding the cathode catalyst loading and 02 pressure. The novel OF extended reaction zone anode with 10 g m2 Pt-Ru (4.0:1 atomic ratio) had an OCV for the DMFC of 0.62 V and yielded a peak power density of 741 W m2, which was 68% higher than the commercial CCM with the same catalyst load. On the anode catalyst load basis the power output of the OF was 74.1 W g’. The experimental variation for the fuel cell tests was estimated to be around 0.005 to 0.010 V (see Appendix D for replicate data). Interestingly, even in the case of a CCM with four times higher Pt-Ru load the three-dimensional anode gave a 5% higher power density. Compared to the CCDL design, the advantage of the novel OF anode was even more dramatic. The more significant performance difference between the CCDL and OF as compared to the CCM and OF designs was in part due to methanol crossover effects, which subsequently impacted the cathode. Furthermore, the presence of Nation on the carbon fibers in the case of CCDL might have impeded the mass transfer and possibly CO2 disengagement as opposed to the OF and CCM. Albeit at four times the catalyst load, the OCV for the CCDL design was 0.60 V, i.e. 0.02 V lower than the OF anode. In both anode designs, there was no catalyst on the membrane surface that could prevent methanol crossover. The effect of having catalysts on the membrane surface was evident as the OCV observed for the CCM design was highest (0.655 V). The difference in OCV between the CCDL and OF anode designs suggested a crossover inhibition effect from the extended reaction zone. Thus, the advantage of the novel three-dimensional OF anode concept has been clearly demonstrated, showing the possibility of reducing significantly the precious metal load while improving the fuel cell performance. Figure 4-13 shows at current densities above 600 A m2, the cell voltage of the DMFC utilizing the conventional design (10 g m2) started to drop at a higher rate compared to the novel OF. This indicates that the overall anode structure and the synergy among various factors such as electrocatalyst preparation, catalyst-support interaction 187 combined with improved reactant mass transport and CO2 disengagement, are key aspects for improving the catalyst utilization efficiency and the power output. 4.3.5 Direct Formic Acid Fuel Cell Experiments The GF anode with 10 g m2 Pt-Ru (4.0:1) was tested and compared with a commercial CCM having the same anode mass load in a DFAFC (Figure 4-14). The effect of supporting electrolyte was investigated. It must be noted although Pt-Ru has been investigated for electrocatalysis of formic acid oxidation [7, 14, 52], it is not the most active catalyst for this reaction. Generally, Pd and Pd-based catalysts such as Pt-Pd are kinetically superior to Pt-Ru with respect to formic acid oxidation, albeit long-term stability related issues might arise with these catalyst formulations [52]. The goal here was to investigate the extended reaction zone anode concept to another type of direct fuel cell. Future work will address the electrocatalytic aspects of formic acid oxidation in conjunction with the novel anode structure. Using 1 M formic acid the conventional CCM design showed better performance in the low-overpotential region (corresponding to superficial current densities below 1000 A m2). This could be explained by the formic acid oxidation kinetic differences between the 4:1 Pt:Ru atomic ratio (on the GF) and the 1:1 Pt:Ru atomic ratio (on the CCM). The maximum power output without 0.5 M H2S04 acting as supporting electrolyte was virtually identical for the thick (- 2 mm) GF and the thin ( 20 jim) CCM designs. Addition of 0.5 M H2S04 improved the effective ionic conductivity of the GF anode (Table 4-4). Hence, the advantages related to catalyst utilization and mass transfer effects, as described previously for the DMFC, came to the fore. The peak power output of the DFAFC with GF and 0.5 M H2S04was about 860 W m2, while the 526 W m2 was obtained with CCM. Thus, a 63% enhancement of the maximum power output was observed with the novel anode. Comparing DMFC and DFAFC (Figures 4-13 and 14), generally the formic acid fuel cell was superior. Considering that Pt-Ru is not the ideal catalyst formulation for DFAFC, further improvements could be expected by combining a better electrocatalyst of formic acid oxidation with the GF anode. 188 (a) 0.70 0.70 0.65 • CCM-lOgm2Pt-Ru(lfl:1),1 MHCOCH 0.65 0.60 • SF- lOg rff2 Pt-Ru 4fl:1), 1 M HCOOH 0.60 0.55 . V SF- lOg rvf2 Pt-Ru (4.0:1), 1 M HCOOH -‘-0.5 M H2S04 0.55 0.50 0.509. 0.45 0.45 . >0.40 1 0.40> jo.35 ••:, O.35j uJ0.30 : 0.30w “-‘C. Vy LLLJ 0.20 0.20 0.15 $ 0.15 0.10 0.10 0.05 0.05 0.00 I I I I I I 0.00 0 500 1000 1500 210 2500 3000 3500 4000 / A m2 (b) 900 — 900 VT, 800 v -800 V V 700 V -700 600 ,‘ •600 3: • 3: 400 . •400 0- 1 0- 300 1 -300 r 200 • CCM-lOgm2Pt-Ru(1.0:1),1 MH000H -200 100 , • SF- 109 m2 Pt-Ru (4fl:1), 1 M HCOOH . 100 • V SF- lOg rrf2 Pt-Ru (4fl:1),l MHCOOH+0.5MH29 0 I I I I I -0 0 500 1000 1500 2000 2500 3000 3500 410 I I A m2 Figure 4-14: DFAFC performance — effect of extended reaction zone vs. conventional anode design. Fuel: 1 M HCOOH with or without 0.5 M H2S04 at 6 mE mm1. Cathode: 40 g Pt m2, dry 02 fed at 2.5 bar and 500 mE miii. Temperature: 333 K. (a) Polarization curve; (b) Power density. 189 1000 oo 0 800 700 600 500 400 300 200 100 0 HCOQH Concentration I M Figure 4-15: DFAFC performance at 333 K — effect of formic acid concentration on peak power density. Extended reaction zone vs. conventional anode design. Fuel: 1, 3, and 10 M HCOOH with or without 0.5 M H2S04at 6 mL mm’. Cathode: 40 g Pt m2, dry 02 fed at 2.5 bar and 500 mL min1. Figure 4-15 shows the impact of increasing the formic acid concentration from 1 M to 10 M. For the DFAFC using the conventional CCM, the OCV recorded with 3 and 10 M formic acid were 0.62 V and 0.56 V respectively, indicating a larger fuel crossover flux with increasing fuel concentration (see Appendix E for polarization and power density curves). Even though fuel crossover was more severe with higher concentration of formic acid, the peak power density for the CCM was highest with 10 M formic acid, due to reduced mass transport limitations. The same trend was observed previously by I 3 10 190 Choi et al. and 10 M was reported as the optimum concentration of formic acid for the best performance [7]. The novel GF based DFAFC using 3 and 10 M formic acid without supporting electrolyte had an OCV of 0.62 and 0.52 V, respectively. The peak power density increased slightly from 530 to 546 W m2 when the concentration of formic acid was increased from 1 to 3 M and 0.5 M H2S04was not present. Since the novel DFAFC was not as restricted by fuel mass transfer like the conventional DFAFC, the performance increase could be attributed to the slight increase in ionic conductivity leading to somewhat lower ohmic voltage loss (Table 4-5). Table 4-5: Ionic conductivity of formic acid and sulfuric acid solutions at 333 K. Ionic Conductivity at 333 K (S m’) 0.5 M H2S04 14.9 1MHCOOH 0.8 3MHCOOH 1.2 1OMHCOOH 1.4 1M HCOOH + 0.5 M H2S04 13.3 3M HCOOH + 0.5 M H2S04 11.6 1OM HCOOH + 0.5 M H2S04 7.5 However, when 10 M formic acid was employed, the peak power density of the fuel cell with the GF anode dropped to 408 W m2 (without supporting electrolyte) and 468 W m2 (with 0.5 MH2S04)(Figure 4-14). The noticeable drop in peak power density as well as OCV with formic acid concentration above 3 M was due to the open porous structure of the GF substrate prone to fuel crossover in the cases with large excess of unconsumed fuel. Chetty and Scott studied the effect of formic acid concentration (from 2 to 10 M) using Ti mesh anode substrate [20]. These authors observed also that 10 M formic acid yielded the lowest peak power density, which is in agreement with the current findings. 191 4.4 Conclusion The galvanostatic electrodeposition of Pt-Ru on GF was investigated using a Triton X- 102 micellar solution. The aim was to produce three-dimensional, extended reaction zone anodes for direct methanol and formic acid fuel cells. It was found that the type of micellar media employed during electrodeposition had an impact on the crystallographic features of the Pt-Ru deposit and conversely on the electrocatalytic activity toward methanol oxidation. The highest activity was observed for the Pt-Ru deposits with the largest fraction of Pt (111) and (311) Miller index facets. The micellar media using Triton X-102 as opposed to Triton X-100/isopropanol favored the formation of the above-mentioned high-activity crystal faces during galvanostatic deposition at 20 A m2, 120 mm, 341 K. The exchange current density was found to be two or more orders of magnitude higher than those found in literature, suggesting improved methanol oxidation kinetics. DLFC experiments revealed clearly the advantages of the novel anode. In the case of the DMFC, with the Pt-Ru!GF anode a maximum power output on geometric area basis of 741 W m2 was obtained at 333 K with 10 g m2 Pt-Ru (4:1 at. ratio) mass load. Under similar conditions a commercial Pt-Ru catalyst in the catalyst coated membrane design and with four times higher load (40 g m2) gave only 703 W m2. Thus, with the novel anode the precious metal load on the anode side could be lowered four times, while still improving the DMFC performance. In DFAFC experiments, depending on fuel composition, the GF anode yielded up to 63% increase in peak power density (up to 860 W ma), compared to 526 W m2 when conventional catalysts with the same mass load were employed. The performance enhancement was most notable at high current density and could be a result of enhanced oxidation kinetics as a result of more active crystalline structure, better CO2 disengagement, and reduced fuel crossover. 192 4.5 References 1. V. Baglio, A.D. Blasi, E. Modica, P. Creti, V. Antonucci, A.S. Arico, Tnt. J. Electrochem. Sci. 1 (2006) 71. 2. A.S. Arico, S. Srinivasan, V. Antonucci, Fuel Cells 1 (2001) 133. 3. P. Zegers, J. Power Sources 154 (2006) 497. 4. 5. Kang, J. Lee, J.K. Lee, S.Y. Chung, Y. Tak, J. Phys. Chem. B 110 (2006) 7270. 5. L.J. Zhang, Z.Y. Wang, D.G. Xia, J. Alloys Compd. 426 (2006) 268. 6. X. Li, I.M. Hsing, Electrochim. Acta 51(2006) 3477. 7. J.H. Choi, K.J. Jeong, Y. Dong, J. Han, T.H. Lim, J.S. Lee, Y.E. Sung, J. Power Sources 163 (2006) 71. 8. B.D. McNicol, D.A.J. Rand, K.R. Williams, J. Power Sources 83 (1999) 15. 9. W. Vielstich, J. Braz. Chem. Soc. 14 (2003) 503. 10. D. Cao, G.Q. Lu, A. Wiechowski, J. Phys. Chem. B. 109 (2005) 11622. 11. T. Iwasita, Electrochim. Acta 47 (2002) 3663. 12. A. Capon and R. Parsons, J. Electroanal. Chem. 45 (1973) 205. 13. Y.X. Chen, M. Heinen, Z. Jusys, R.J. Behm, Angew. Chem. mt. Ed. 45 (2006) 981. 14. N.M. Markovic, H.A. Gasteiger, P.N. Ross, X. Jiang, I. Villegas, M.J. Weaver, Electrochim. Acta4O (1995) 91. 15. M. Weber, J.T. Wang, S. Wasmus, R.F. Savinell, J. Electrochem. Soc. 143 (1996) L158. 16. L. Dubau, C. Coutanceau, E. Gamier, J-M. Leger, C. Lamy, J. App!. Electrochem. 33 (2003) 419. 193 17. J. Ge, H. Liu, J. Power Sources 142 (2005) 56. 18. T.T. Cheng, E.L. Gyenge, J. App!. Electrochem. 38 (2008) 62. 19. R. Larsen, S. Ha, J. Zakzeski, R.I. Masel, J. Power Sources 157 (2006) 78. 20. Z. Liu, L. Hong, M.P. Tham, T.H. Lim, H. Jiang, J. Power Sources 161 (2006) 831. 21. C. Roychowdhury, F. Matsumoto, P.F. Mutolo, H.D. Abruna, F.J. DiSalvo, Chem. Mater. 17 (2005) 5871. 22. R.G. Allen, C. Lim, L.X. Yang, K. Scott, S. Roy, J. Power Sources 143 (2005) 142. 23. R. Chetty, K. Scott, J. New. Mater. Electrochem. Sys. 10 (2007) 135. 24. Z.G. Shao, W.F. Lin, P.A. Christensen, H. Zhang, Tnt. J. Hydrogen Energy 31 (2006) 1914. 25. Z.G. Shao, F. Zhu, W.F. Lin, P.A. Christensen, H. Zhang, B. Yi, J. Electrochem. Soc. 153 (2006) A1575. 26. Z.G. Shao, F. Zhu, W.F. Lin, P.A. Christensen, H. Zhang, Phys. Chem. Chem. Phys. 8 (2006) 2720. 27. A. Bauer, E.L. Gyenge and C.W. Oloman, Electrochim Acta 51(2006) 5356. 28. A. Bauer, E.L. Gyenge and C.W. Oloman, J. Power Sources 167 (2007) 281. 29. G.S. Attard, P.N. Bartlett, N.R.B. Coleman, J.M. Elliott, J.R. Owen, Langmuir 14 (1998) 7340. 30. J.M. Elliott, G.S. Attard, P.N. Bartlett, N.R.B. Coleman, D.A.S. Merckel, J.R. Owen, Chem. Mater. 11(1999) 3602. 31. J.M. Elliott, P.R. Berkin, P.N. Bartlett, G.S. Attard, Langmuir 15(1999)7411. 194 32. P.N. Bartlett, B. Gollas, S. Guerin, J. Marwan, Phys. Chem. Chem. Phys. 4 (2002) 3835. 33. P.N. Bartlett, P.N. Birkin, M.A. Ghanem, P. Groot, M. Sawicki, J. Electrochem. Soc. 148 (2001) C119. 34. V. Ganesh, V. Lakshminarayanan, Electrochim. Acta 49 (2004) 3561. 35. T.T. Cheng, E.L. Gyenge, Electrochim. Acta 51(2006) 3904. 36. C.L. Green, A. Kucernak, J. Phys. Chem. B 106 (2002) 1036. 37. H.C. Teh, G.H. Ong, S.C. Ng, L.M. Gan, Phys. Lett. A 78 (1980) 487. 38. H.C. Teh, G.H. Ong, S.C. Ng, L.M. Gan, J. Dispersion Sci. Technol. 6 (1985) 255. 39. H.A. Gasteiger, N. Markovic, P.N. Ross and E.J. Cairns, J. Electrochem. Soc. 141 (1994) 1795. 40. H.A. Gasteiger, N. Markovic, P.N. Ross and E.J. Cairns, J. Phys. Chem. 97 (1993) 12020. 41. A.V. Tripkovic, S. U. Gojkovic, K. DJ. Popovic, J.D. Lovic, J. Serb. Chem. Soc. 71(2006)1333. 42. T. Iwasita, J. Braz. Chem. Soc. 13 (2002) 401. 43. T.D. Jarvi, S. Sriramulu, E.M. Stuve, Colloids Surf., A 134 (1998) 145. 44. X.H. Xia, I. Iwasita, F. Ge, W. Vielstich, Electrochim. Acta4l (1996) 711. 45. N. Tian, Z.Y. Zhou, S.G. Sun, Y. Ding, Z.L. Wang, Sci 316 (2007) 732. 46. C. Bock, M.A. Blakely, B. MacDougall, Electrochim. Acta 50 (2005) 2401. 47. E. Antolini, L. Giorgi, F. Cardellini, E. Passalacqua, J. Solid-State Electrochem. 5 (2005) 131. 195 48. L. Jiang, G. Sun, X. Zhao, Z. Zhou, S. Yan, S. Tang, G. Wang, B. Zhou, Q. Xin, Electrochim. Acta 50 (2005) 2371. 49. A.J. Dickinson, LP.L. Carrette, J.A. Collins, K.A. Friedrich, U. Stimming, J. Appi. Electrochem. 34 (2004) 975. 50. R. Jiang, H.R. Kunz, J.M. Fenton, J. Electrochem. Soc. 153 (2006) A1554. 51. B. Gurau and E.S. Smotkin, J. Power Sources 112 (2002) 339. 52. R. Larsen, R.I. Masel, Electrochem. Solid-State Lett. 7 (2004) A148. 196 5 COMPARISON OF Pd AND Pt-Ru SUPPORTED ON GRAPHITE FELT FOR THE DIRECT FORMIC ACID FUEL CELL* 5.1 Introduction The direct formic acid fuel cell (DFAFC) has been shown to be an excellent candidate for powering portable electronic devices. The DFAFC not only has a higher theoretical thermodynamic potential (1.45 V) than the direct methanol fuel cell (DMFC, 1.19 V), but also possesses a smaller crossover flux due to anodic repulsion between HCOO and the sulfonic groups in Nafion® membrane, resulting in higher fuel utilization [1-31. It is generally accepted that the electro-oxidation of formic acid on Pt proceeds through a dual-path mechanism originally proposed by Capon and Parsons in 1973, involving a non-CO direct path and a CO indirect pathway (see Section 1.3.2) [4-7]. The direct path proceeds via dehydrogenation and CO2 is formed through active non-COad intermediates. The indirect path involves the formation of adsorbed carbon monoxide, both an intermediate and a catalyst poison, through dehydration. The adsorbed carbon monoxide is then subsequently oxidized to CO2. The electro-oxidation of formic acid on Pd is believed to proceed primarily through a non-CO pathway [8]. Evidence for it was obtained by CO stripping voltammogram, revealing that the CO buildup on Pd black was significantly lower than on Pt black [9]. Furthermore, it was shown by Arenz et al. that no adsorbed CO could be detected by FTIR spectra even though a high production rate of CO2 was observed [10]. As a result, Pd has better performance than Pt, especially at low anode potential (<0.4 V) where the adsorbed CO intermediate on Pt cannot be further oxidized. However, it is interesting and important to note that Pd deactivates significantly with time [2, 9-11]. Blair et al. have published CP data on the long term stability of Pt, Pd, and various Pd-Pt catalysts. They found that Pd albeit having superior initial catalytic activity, its activity falls below that of Pd-Pt and Pt within just hours (see Section 1.3.2) [11]. Li and Hsing * A version of this chapter has been submitted for publication: T.T. Cheng, E.L. Gyenge, submitted to 3. Appi. Electrochem. in October 2008. 197 presented a CA study comparing the stability of Pt, Pd, and Pt-Pd and found that the oxidation current on Pd dropped ten folds in 30 mm [2]. It has been proposed that the deactivation of Pd was a result of the blocking of active sites by the adsorption of spectator species such as Had and OHad, in conjunction with anions from the supporting electrolyte and/or Pd oxide formation [10]. More studies are required in this area. The DFAFC anode structure design has also received research attention. It has been shown previously that the utilization of the catalyst load in a typical ODE anode is only between 10-50% [12]. Therefore, research was carried out to improve the catalyst utilization and anode performance demonstrated by Wilkinson et al. [13] and Gyenge et al. [14-18] (see also Chapter 3 and 4 of the present thesis) the benefits of employing a multi-layer and three-dimensional monolithic carbon-based electrode design, respectively. Using catalyzed (10 g m2 Pt-Ru) OF (3 mm thickness) as the anode, the peak power density of a DMFC operated at 333 K reached 741 W m2 compared to 442 W m obtained with a commercial GDE Pt-Ru anode with the same catalyst load and operating conditions [18] (Chapter 4). The same catalyzed OF anode was also employed in a DFAFC, with the peak power density reaching up to 860 W m vs. 528 W m2 obtained with commercial CCM Pt-Ru, both at 333 K. In addition, Chetty and Scott have also demonstrated the use of Ti mesh as DFAFC anode support for thermally deposited Pd and Pt-Sn catalysts [19]. The goal of the present investigation was to employ the previously studied catalyst preparation method, electrodeposition from Triton X- 102 micellar solution [18], to prepare Pd nanoparticles on GF for comparison with the previously prepared Pt-Ru/OF for the DFAFC. Even though formic acid has inherent protonic conductivity, it was shown earlier in Chapter 4 that in order to fully utilize the whole catalyst layer, a supporting electrolyte was needed. For the present study as well, a liquid electrolyte, 0.5 M H2S04— 1 M HCOOH was used to address the protonic conductivity challenge. 198 5.2 Experimental Section 5.2.1 Graphite Felt Pretreatment The GF (thickness 3 mm) was provided by Test Solutions through Electrolytica Inc. Produced with polyacrylonitrile fibers, the GF substrate had a carbon content of 99.7% and a surface area of 0.7 m2 g’ (measured by nitrogen adsorption). GF technical data presented were provided by Electrolytica Inc. In the pretreatment, the GF substrates were first sonicated in methanol for 30 minutes and were washed thoroughly in distilled water. An additional pretreatment step involving the immersion of the GF in a PdC12 (6 mM) + SnCJ2 (0.3 M) solution for up to 48 hours at 303 K was performed for selected samples. The PdCl2 + SnCl2 solution (i.e. Shipley type solution [16, 201), was prepared by mixing 0.1 g of PdCl2(Sigma-Aldrich), 5 g of SnCl2HO(Sigma-Aldrich), 60 mL of de-ionized water and 30 mL of concentrated hydrochloric acid (Fisher Scientific). Upon pretreatment, all samples were washed thoroughly with distilled water followed by drying in air. The Shipley solution pretreatment is based on the surface redox of Pd, Sn, and carbon. As shown in Equation 5-1 to 5-4, Pd2 can spontaneously reduce to Pd with the 2+ 4+coupled oxidation of carbon (to C02, OH—C=’O, or —C0) and/or Sn (to Sn ). The resulting potential can also cause the reduction of Sn2 to Sn. Pd2+2e-Pd Ee=0.85V [Eq.5-1] Sn+2e’c-*Sn Ee=0•15V [Eq.5-2] Sn4+2e’c->Sn Ee=0.15V [Eq.5-3] CO7+4e+4H -*C+2H2O Ee=0•19V [Eq.5-4] 5.2.2 Pd Electrodeposition Procedure The electrodeposition media employed in the present study was the same Triton X- 102 micellar solution studied extensively previously (Chapter 4) [18]. The micellar deposition solution was composed of Triton X-102 non-ionic surfactant 199 [C8H17640(C2)2],and an aqueous phase with various concentrations of metal precursors. The concentration of PdC12 in the colloidal solution varied from 0.75 to 4.5 mM. The Triton X- 102 content ranged from 0 to 25 vol % and the aqueous phase made up the remaining portion. The chemicals used were reagent grade obtained from Sigma- Aldrich and were used as delivered without further purification processes. The micellar solutions were prepared by mixing the aqueous and surfactant phases in a water-jacketed glass vessel connected to a circulating water bath at 341 K for 30 minutes. The electrode assembly consisted of the GF working electrode and two perforated platinized titanium counter electrodes (anodes), each with a geometric surface area of 5 cm2 (with --10 holes per cm2 and hole dimensions of -0.3 cm2). The benefits of using perforated counter electrodes were described in Chapter 3 [17]. The electrodes were placed at a distance of 1 cm and were inserted into the glass vessel after no visible change in the colloidal solution was observed. The electrodepositon was carried out with a Xantrex XHR15O-7 DC power supply capable of operating at 0-150 V and 0-7 A. The GF was sonicated in THF (Reagent Grade, Sigma Aldrich) for 5 minutes to help wash out the organic compounds retained in the porous matrix after the electrodeposition. The GF was then washed thoroughly with distilled water and dried in air. After drying, the GF was heat treated in a N2 stream for 1 hour at 573 K to remove traces of adsorbed organic compounds. The effect of post-deposition chemical and electrochemical treatments were discussed extensively in an earlier chapter (Chapter 1) [16]. 5.2.3 Electrochemical Measurements Electrochemical measurements, including CV and CA were carried out at 298 K in a water-jacketed electrochemical cell connected to a circulating water bath. The test assembly was composed of a three-electrode setup: Hg/Hg2SO4,K2S04, std.(MSE) electrode as reference electrode, platinum wire as counter electrode, and the OF of interest as working electrode. The electrolyte was 1 M HCOOH and 0.5 M H2S04with a volume of 50 mL. After post-deposition treatment and before electrochemical tests, the catalyst surface was pretreated electrochemically by applying potential steps of 1.28, 200 1.20, and 0.05 V for 10 s each, repeated three times. A Radiometer Analytical VoltaLab PGZ4O2 potentiostat with the VoltaMaster 4 software was used in all experiments. All potentials in the present work are reported against the standard hydrogen electrode (SHE) reference. The electrochemically active Pd surface area was estimated using Cu UPD, a technique that was shown in Chapters 2 to 4 to be effective for Pt-Ru deposited on RVC, GF, and Ti mesh [16-18]. Therefore, in the present study, the effective surface area was estimated from the anodic stripping charge of a monolayer of Cu (i.e. 4.2 C m2) [21] for a consistent comparison with Pt-Ru!GF. In addition, Cu UPD has been employed by Rusanova et al. to determine the surface area of Pd electrodeposits [22]. The Cu UPD experiments were carried out in 0.5 M H2S04and 0.002 M CuSO4at 298 K. Prior to Cu UPD, reference voltammograms between the potential range of -0.04 V to 0.91 V were obtained in 0.5 M H2S04at a scan rate of 0.050 V srn’. A monolayer of Cu was underpotential deposited on the catalyst surface by polarizing the GF at 0.26 V for 300 s. An anodic linear voltammetric scan at 0.050 V s was then applied from 0.26 V to 0.91 V to remove the adsorbed Cu. The charge differences between the reference and anodic linear scan were then used to calculate the active surface area. A systematic voltammetric study of Pd electrodeposition from Triton X-102 micellar solution was carried out at 341 K with the same deposition setup discussed in Section 2.2. One of the perforated counter electrodes was replaced by a Ag/AgC1, KC1, std.(SSC) reference electrode. A cathodic linear voltammetric scan at 0.005 V s’ was applied from 0.60 V to -0.40 V. The effect of Triton X-102 concentration and GF pretreatment was determined by comparing the resulting voltammograms with and without Triton X- 102 and pretreatment. 5.2.4 Surface and Analytical Characterization of the Catalysts A Hitachi S4700 high resolution SEM set with an accelerating voltage and emission current of 2000 V and 1 .25x 1 0 A was used to capture visual images. Fragments of the GF with Pd electrodeposits were flush mounted onto SEM stubs with carbon adhesive and a working distance of 0.0025-0.0035 m was employed. A Hitachi S4500 201 field-emission scanning electron microscope (FESEM) using a 5000 V electron beam voltage was also used to capture high resolution images of selected samples. Catalyst mass load was determined by ICP-AES using a Perkin Elmer Optima, model 3300DV instrument. Pd/GF samples with a geometric area of 1 cm2 was weighed and digested in aqua regia (HC1-HNO3volume ratio of 3:1) for four hours to solubilize the Pd metal deposits. The resulting solution was then diluted and the Pd content was determined by ICP-AES compared to a predetermined reference Pd solution. The crystallographic features of the prepared catalysts were determined by XRD using an Advanced Bruker powder X-ray diffractometer with Cu K radiation wavelength of 1.5418 A. The XRD experiments were performed with 20 values from 10 to 85° with a stepping of 0.04°. The GF substrates pretreated with Shipley solution were also analyzed by XPS using a Kratos Axis Ultra spectrometer with a probing depth of 7 to 10 nm, and detection limits ranging from 0.1 to 0.5 at.% depending on the element. Survey scan spectra were obtained from an analysis area of -0.300 x 0.700 nm in size and with a pass-energy of 160 eV. High resolution spectra were obtained with a pass-energy of 10 eV. 5.2.5 Membrane Electrode Assembly and Fuel Cell Experiments MEA with electrode area of 5 cm2 containing 40 g m2 of pre-painted Pt black (cathode), and 40 g m2 of pre-painted Pd black (anode) in addition to half MEA’s without pre-painted Pd black were provided by Lynntech Inc. The half-MEA’s were used in conjunction with the extended reaction zone Pd anode prepared in the present work. The MEA’s were pre-conditioned in 0.5 M H2SO4 for 24 hours before use. The cathode backing layer was a piece of Elat® carbon cloth (E-Tek Inc.). In the case of full MEA, a piece of carbon cloth was employed as the anode backing layer whereas a backing layer was not needed in the fuel cell based on the GF anode. A piece of Teflon®-coated gasket was used on both anode and cathode. The DFAFC was assembled with two gold-plated end plates and held together by insulated bolts. The fuel cell tests were performed using a Fideris Inc. MTK fuel cell test station, equipped with corrosion-resistant fittings and operated using the FC Power® software. 202 The tests were carried out at 333 K with an oxygen (dry medical grade, Praxair Inc.) flow rate of 500 mL min1 at 2.5 bar absolute pressure on the cathode side and an anolyte (0.5 M H2S04and 1 M HCOOH) flow rate of 6 mL min1.No preheating was utilized for both reactant streams. The system was allowed to stabilize for 2 hours before recording the current-potential data. Current was progressively drawn and the cell voltage was recorded after 10 s of continuous operation at constant current. 5.3 Results and Discussion 5.3.1 Pd electrodeposition on GF from Triton X-102 Micellar Solution: Effect of Metal Precursor Concentration Pd was electrodeposited on unpretreated GF from a Triton X- 102 micellar solution using the same galvanostatic electrodeposition technique described in Section 5.2.2 (see also Chapter 3). The micellar solution was composed of 12.5 vol% Triton X 102 and 87.5 vol% aqueous phase with PdC12 concentration of 0.75 mM in the mixed solution. The electrodeposition was carried out a 341 K with a deposition superficial current density and time of 20 A m2 and 120 minutes, respectively. In contrast to Pt-Ru (Chapter 3), for Pd, the deposition procedure resulted in a sparse coating of Pd nanoparticles with particle sizes ranging from 5 to 50 nm (see Figure 5-1 a). The mass load determined by ICP-AES was only 0.2 g m2 and the surface area was 2.4 m2 m2. In order to achieve a denser Pd coating, the concentration of the metal precursor, PdC12,was increased six folds from 0.75 mM to 4.5 mM. The mass load increased from 0.2 g m2 to 8.0 g m2 and resulted in a denser coating (Figure 5-1 b). The mass increase could be attributed to the higher initial surface concentration of Pd2 favoring the deposition kinetics and an increase in mass-transport limiting current density for deposition. The overall surface area increased to 68.6 m2 m2. Generally, the particle size remained the same, approximately 5 to 50 nm. In hope to obtain higher Pd surface area, the electrodeposition of Pd was further investigated by studying the effects of GF pretreatment and Triton X- 102 concentration. 203 Figure 5-1: SEM images of Pd catalyzed GF (unpretreated) prepared with 12.5 vol% Triton X-102 at 341 K with a deposition current density of 20 A m2 for 120 mm. (a) Pd (0.75 mM PdC12); (b) Pd (4.5 mM PdC12). 204 5.3.2 Pd electrodeposition on GF from Triton X-102 Micellar Solution: Effect of GF Pretreatment and Triton X-102 Concentration Triton X- 102 concentration and GF pretreatment were investigated to determine their effects on the Pd morphology and mass load. As shown in Chapter 4, the Triton X 102 concentration had a direct impact on the Pt-Ru surface morphology [18]. Therefore, it was of interest to investigate if the same effect is operative for Pd. It was also found that the substrate pretreatment had a tremendous effect on the resulting catalyst morphology since the substrate surface could be significantly roughened and its surface potential could be modified (Section 2.3.4) [16]. In the case of GF, electrochemical cycling pretreatment technique (demonstrated to be effective for RVC) could not be utilized as the mechanical integrity of GF deteriorated after pretreatment. Therefore, a surface sensitization technique involving the Shipley type solution [20] was employed (Section 5.2.1). To better understand and assess the potential synergistic effect of these variables, a set of factorial experiments (2 variables at 3 levels, total of 9 experiments plus replicates) was carried out with varying surface pretreatment time of 0, 24, and 48 hours, and Triton X-102 concentration of 0, 12.5, and 25 vol%. 5.3.2.1 Effect ofShipley Solution Pretreatment on GF Substrate The Pd and Sn mass load, Pd:Sn atomic ratio and their respective deposition yield are shown in Table 5-1 and Figure 5-2. The deposition yield of the Shipley type pretreatment was very low as expected since it is only a surface modification process (sensitization — providing nucleation sites for electrodeposition). From Figure 5-2, it can be seen that the deposition yield of Sn was leveling off with pretreatment time, indicating that longer pretreatment time favored the deposition of Pd. As shown in Figure 5-3, the surface pretreatment had a huge impact on the GF surface. Compared to un-pretreated GF, the GF pretreated in the Shipley type solution for 24 hours resulted in a fairly dense coating across the felt thickness with Pd-Sn nanoparticles with diameter ranging from 10 to 30 nm (Figure 5-3 b). Pretreating the GF for 48 hr led to a denser Pd-Sn coating with essentially the same morphology and a higher Pd:Sn atomic ratio (Figure 5-3 c). It is important to note that the hydrophilicity of the GF samples increased with pretreatment 205 time. With no pretreatment, the GF sample was hydrophobic. With 24 or 48 hours of pretreatment, the GF samples became more hydrophilic and could be completely wetted by water, which would be beneficial for DFAFC anode application for enhanced fuel mass transport. From XPS spectra, it was found that the Pd and Sn surface content was 2.4 and 8.2 at.% (24-hr pretreatment), and 3.9 and 10.0 at.% (48-hr pretreatment), respectively. The oxygen content on the surface also increased from 4.8 at.% without pretreatment to 26.6 at.% with 24-hr pretreatment and 33.5 at.% with 48-hr pretreatment (Table 5-2). This explains the hydrophilicity of the pretreated carbon. Table 5-1: Pd and Sn mass load PdC12,0.3 M SnC12,4 M HC1) 2 on GF from Shipley type pretreatment (6 mM Pretreatment Time Pd Mass Load Sn Mass Load Pd:Sn Atomic Deposition (hr) (g m2) (g m2) Ratio Yield (wt%) (from XPS) 24 0.6 0.7 1:3.4 Pd: 2.0% Sn: 0.04% 48 1.0 0.8 1:2.6 Pd: 3.3% Sn: 0.05% 3.5 3 2.5 S >- 0 1.5 0 0. S 0 0.5 0 0.06 0.05 0.04 0.03 U, 0 0. 0.02 Co 0.01 0 48 Figure 5-2: Pd and Sn deposition yield as a function of pretreatment time. 0 12 24 36 Deposition Time (hr) 206 (a) (b) (c) SSW 5 kV X1eJerm Pd-Sn Nanoparticles Figure 5-3: The effect of Shipley (6 mM PdC12 and 0.3 M SnC12 in 4 M HC1) pretreatment on the GF surface (middle of the felt). (a) Unpretreated; (b) 24-hr pretreatment; (c) 48-hr pretreatment. 207 Table 5-2: Pd, Sn, and oxygen surface content of GF before and after Shipley type solution pretreatment. Pretreatment Time Pd Surface Content Sn Surface Content Oxygen Surface (hr) (at. %) (at. %) Content (at. %) 0 0 0 4.8 24 2.4 8.2 26.6 48 3.9 10.0 33.5 5.3.2.2 The Synergistic Effect ofGF Pretreatment and Triton X-102 Concentration on the Mass Load, Surface Area, and Morphology ofElectrodeposited Pd: Factorial Experimental Design The mass load and overall surface area of the electrodeposited Pd/GF samples as a function of Triton X- 102 concentration and Shipley pretreatment time are given in Figure 5-4. It is interesting to note that in the case without any surfactant and surface pretreatment, the resulting Pd coating flaked off the GF surface during post-deposition treatment and hence, no accurate mass load could be obtained. Thereby, the surface pretreatment is essential for the strong adherence of the electrodeposited Pd to the GF surface. Evidently, the amount of Pd deposited increased with surface pretreatment time because of the increased number of nucleation sites available from surface roughening. This trend was apparent regardless of Triton X- 102 concentration. In general, the amount of Pd decreased with increasing Triton X-102 content (from 12.5 to 25 vol%). The same behaviour was also observed for Pt-Ru in the previous study (Chapter 4) [18]. The addition of Triton X-102 (from 0 to 12.5 vol%) did not have a significant impact on the Pd mass load (Figure 5-4); however, the surfactant was essential for the penetration depth of the deposit (Figure 5-5 to 5-7). 208 Mass Loading Mass Loading Mass Loading 48 — (g m2): 50 (g m2): 57 (g m2): 14 Surface Area Surface Area Surface Area (m2 m2): 112 (m2 m2): 291 (m2m2): 128 Pretreatment Mass Loading Mass Loading Mass Loading Time (hr) (g m2): 43 (g m2): 31 (g m2): 15 — Surface Area Surface Area Surface Area (m2 m2): 58 (m2 m2): 201 (m2m2): 71 Mass Loading Mass Loading Mass Loading (g m2): N/A (g m2): 8.0 (g m2): 4.0 Surface Area Surface Area Surface Area (m2 m2): N/A (m2 m2): 69 (m2m2): 46 I I I o 12.5 25 Triton X-102 Concentration (vol %) Figure 5-4: Effect of Triton X-102 concentration and Shipley solution pretreatment time on the Pd mass load and surface area obtained by electrodeposition at 341 K with a deposition current density of 20 A rn2: Experimental design matrix with 2 variables at 3 levels. 209 Figure 5-5: SEM images of electrodeposited Pd/GF with 0% Triton X-102. Effect of Shipley solution pretreatment time. (a, b) no pretreatment (exterior vs. interior); (c, d) 24 hr pretreatment (exterior vs. interior); (e, 1) 48 hr pretreatment (exterior vs. interior). Exterior Surface Inner Surface 210 Figure 5-6: SEM images of electrodeposited Pd/GF with 12.5% Triton X-102. Effect of Shipley solution pretreatment time. (a, b) no pretreatment (exterior vs. interior); (c, d) 24 hr pretreatment (exterior vs. interior); (e, f) 48 hr pretreatment (exterior vs. interior). Exterior Surface Inner Surface 211 Exterior Surface Inner Surface Figure 5-7: SEM images of electrodeposited Pd/GF with 25% Triton X-102. Effect of Shipley solution pretreatment time. (a, b) no pretreatment (exterior vs. interior); (c, d) 24 hr pretreatment (exterior vs. interior); (e, 1) 48 hr pretreatment (exterior vs. interior). 212 From the SEM images (Figure 5-5 to 5-7), it is clear that both Triton X-102 and surface pretreatment time had an impact on the homogeneity of Pd throughout the substrate thickness. Without Triton X-102, thick continuous coatings of Pd were observed on the exterior faces of the GF substrate while the interior had no deposit, regardless of Shipley pretreatment time (Figure 5-5 a to 1). Interestingly, however, irrespective of the amount of Triton X- 102 added, the Shipley pretreatment was essential for enhanced deposition penetration depth (Figure 5-5 and 5-7). Comparing Figure 5-6 a and b, it is observed that without surface pretreatment, the Pd coverage in the interior was noticeably lower than on the outer surface of the felt. The results suggest a positive interaction effect between Shipley pretreatment and Triton X-102 on the deposit penetration depth, i.e. both are essential for good penetration. Besides the notable difference in deposit homogeneity throughout the substrate thickness, the catalyst morphology varied with Triton X-102 content and surface pretreatment. Without Triton X-102, a Pd coating made up of large and smooth agglomerates (1000 nm) was observed for the case with 24 hours of pretreatment (Figure 5-5 c and d). With 48 hours of pretreatment, the agglomerates were smaller (— 500 nm) and more regular in shape (Figure 5-5 e and f). In both cases, bulk deposition was evident and the resulting surface area was low (58 and 112 m2 m2, respectively). With 12.5 vol% Triton X-102 and 24 hours of surface pretreatment, the resulting Pd deposits were well dispersed on the GF surface with particle size ranging from 20 to 100 nm. Increasing the pretreatment time to 48 hours resulted in a thicker catalyst film (average of 57 g m2, see Appendix D for details) with 20 to 40-nm particles and the overall surface area was increased from 201 to 291 m2 m2, but due to the higher deposit thickness, the mass-specific surface area was reduced from 6.5 to 5.1 m2 g’. For the cases with 25 vol% Triton X-102 and 24 or 48 hours of pretreatment, the resulting Pd was well dispersed on GF with particle size of-40 to 20 nm (Figure 5-7 c to f). Note that with 25 vol% Triton X-l02 and no pretreatment, there was very little Pd deposition (Figure 5-7 a and b). 213 5.3.2.3 Effect ofTriton X-102 Concentration on PdElectrodeposition The observations presented above were in agreement with the Pd deposition study by CV (Figure 5-8). The addition of Triton X-102 suppressed both hydrogen evolution and the electrodeposition of Pd significantly (i.e. lower deposition current density at the same potential, curves A and C of Figure 5-8). Interestingly, pretreating the GF led also to a suppression of the hydrogen evolution and Pd deposition rates (curve B, Figure 5-8). As discussed in 5.3.2.2, the poor PdIGF adhesion found for the catalyst prepared without Triton X- 102 and surface pretreatment could be explained by the higher rate of hydrogen evolution, which might physically impede the Pd from depositing on GF. The effect of deposition potential on the deposition morphology has been studied for various metals, such as Pt, Pd, and Zn [23, 24]. It was shown that a higher cathodic overpotential generally favored smaller deposits, which agreed well with the present study. Due to the combined effect of Triton X-102 and Shipley surface pretreatment, based on Figure 5-8, a higher cathodic overpotential is required during galvanostatic deposition of Pd compared to the case without additive and pretreatment. The equilibrium potential of Pd2/P at the experimental conditions is 0.88 V, which leads to a cathode overpotential of 0.29, 0.33, 0.37, and 0.39 V, respectively for cases A to D in Figure 5-8, for a deposition current density of 20 A m2. 214 0• C -100 A -200 E -00 • 0 vol% Triton X-l 02; No Pretreatment .400 — — — 0 vol% Triton X-l 02; 46 hr Pretreatment — — 12.5 vol% Triton X-102; No Pretreatment — 12.5 vol% Triton X-102; 46 hr Pretreatment -SOD. I I I I -0.4 -0.2 0.0 0.2 0.4 0.6 E/Vvs. SHE Figure 5-8: Voltammograms of Pd electrodeposition on GF: Effect of Triton X-102 content and surface pretreatment. PdC12 concentration: 4.5 mM. Temperature: 341 K. Scan rate: 0.005 V s. 5.3.2.4 Crystallography ofElectrodeposited Pd Both SEM images and surface area estimation by Cu UPD agreed well with the XRD results (see Figure 5-9). For the sample prepared with 48 hours of pretreatment and 12.5 vol% Triton X- 102, the crystallite size estimated by Scherrer’ s formula for Pd( 111), Pd(100), Pd(1 10), and Pd(31 1), was found to be 27.8, 19.0, 22.6, and 18.5 nm, respectively (Appendix C). After taking into account the fraction of the different crystal planes (52.3% Pd(111), 21.7% Pd(100), 12.5% Pd(100), and 13.5% Pd(311)), the theoretical surface area based on the assumptions of i) spherical crystallite, ii) lack of agglomeration, and iii) all area is electrochemically active, was found to be 1185 m2 m2. 215 The Pd surface area estimated by Cu UPD (Figure 5-10) was 291 m2 m2, 25% of the theoretical surface area and the discrepancy was attributed mainly to agglomeration. The result was realistic as compared to Pt-Ru prepared from the same deposition media (Chapter 4), which had less agglomeration, a lower catalyst film thickness, and an associated Cu UPD to theoretical surface area of 34%. Pd/GF - 12.5 vol% Triton X-102 with Pd (111) 48-hour Shipley Solution Pretreatment > Pd (100) C a) Pd (110) Pd (311) 40 50 0 70 80 2-Theta Figure 5-9: XRD spectra of Pd/GF electrodeposited in the presence of 12.5 vol% Triton X-102 after 48 hours of Shipley solution pretreatment. 216 1000 600 600 S 400 200 0 E/ V vs. SHE Figure 5-10: Blank scan and Cu UPD stripping curves of PdIGF electrodeposited in the presence of 12.5 vol% Triton X-102 after 48 hours of Shipley solution pretreatment. Test solution: 0.5 M H2S04 (blank); 0.5 M H2S04+ 0.002 M CuSO4 (Cu UPD) Temperature: 298 K; Scan rate: 0.050 V s1. 5.3.2.5 Effect ofGF Pretreatment and Triton X-1 02 Concentration on the Intrinsic Catalytic Activity ofElectrodeposited Pd As shown in Figure 5-11, the Pd/GF prepared without Triton X-102 had a very high formic acid oxidation onset potential (0.6 V vs. SHE) as opposed to that prepared with the presence of Triton X-102. The difference can be explained by the particle size and roughness of the electrodeposited Pd. As shown from the SEM images (Figure 5-5 c), the Pd/OF prepared in the absence of Triton X-102 was made up of large and smooth agglomerates ( 1000 nm) whereas the sample prepared with 25 vol% Triton X-102 had 0.0 0.2 0.4 0.6 1 .0 217 much finer particles (< 40 nm). In the case of smooth Pd agglomerates, the CV characteristics (high onset potential at - 0.6 V and having a high-current reverse peak) were essentially the same as that of Pt as published by Liu et al. (Figure 5-12) [25]. The high onset potential in the forward scan was due to COad and the high-current reverse scan was a result of having a cleaned catalyst surface after scanning to high potentials. The result suggests that the difference in catalyst morphology caused the formic acid oxidation reaction to follow different reaction pathways (i.e. smooth Pd behaving like Pt proceeding via the COad pathway in contrast to the flOflCOad pathway for nano-sized Pd). This hypothesis is further supported by the low onset potential observed for the other nano-sized Pd/GF (Appendix F). It is interesting to note that some of the Pd/GF catalysts exhibited current-oscillation behavior at high potentials (> 1.0 V vs. SHE) (see Figure 5- 13 and Appendix F). Current-oscillation during formic acid oxidation has been observed and studied by various researchers [26-29]. The phenomenon is explained by the periodic adsorption and removal of surface poison as described in Section 1.3.2. 120 100 en 60 40 20 0 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 E/Vvs. SHE Figure 5-11: Cyclic voltammograms of formic acid electro-oxidation using Pd electrodeposited on GF. (a) no Triton X-102 and 24-hr pretreatment; (b) 25 vol% Triton X-102 and no pretreatment. Solution: 1 M HCOOH and 0.5 M H2S04. Temperature: 298 K. Scan rate: 0.005 V s. 218 I 5 to oQ E/Vvs. SHE Figure 5-12: Cyclic voltammograms of formic acid electro-oxidation with Pt/C and Pd/C in 3 M HCOOH and 1 MH2S04at 298 K L251 Scan rate: 0.010 V s1. Reprinted with permission from Elsevier. Copyright Elsevier (2006). — PVC PdC 0 0.2 0.4 0.6 0.8 1.0 1.2 219 120 100 ec 20 0 Figure 5-13: Oscillatory phenomena during formic acid electro-oxidation using Pd electrodeposited on GF prepared with 25 vol% Triton X-102 with 24-hr Shipley pretreatment. Solution: 1 M HCOOH and 0.5 M H2504.Temperature: 298 K. Scan rate: 0.005 V Figure 5-14 summarizes the intrinsic catalytic activity of formic acid electro oxidation for the different Pd/GF samples prepared with different concentration of Triton X-102 and Shipley solution pretreatment time. It is important to emphasize that the intrinsic catalytic activity increased with the presence of Triton X- 102, which was shown and discussed earlier in Chapter 4 for Pt-Ru/GF [18]. It is interesting to note the effect of Shipley solution pretreatment on the intrinsic catalytic activity. In the cases without surfactant, pretreating the GF surface with Shipley solution led to smaller and more uniform Pd deposits. Hence, the catalytic activity improved due to a reaction pathway change as discussed above. In contrast, in the cases with Shipley pretreatment as well as 0.0 0.2 0.4 0.S o.e 1.0 1.2 1.4 E/Vvs. SHE 220 the presence of Triton X- 102, it is evident that the intrinsic activity per real area basis decreased with Shipley solution pretreatment (Figure 5-14). This could be explained by the change in the GF hydrophobicity (from hydrophobic to hydrophilic) after pretreatment, causing the surfactant to adsorb in a different configuration on the surface (e.g. laterally as opposed to via the hydrophobic headgroup) and consequently leading to a change in deposit morphology. Furthermore, the presence of Sn could also contribute to the decrease of the intrinsic kinetic activity by electronic effects. UI 0 > > 0 (a 25 vol% Triton X-102 12.5 vol% Triton X-102 0 vol% Triton X-1 02 Figure 5-14: Intrinsic formic acid oxidation current density at 0.3 V vs. SHE on electrodeposited PdIGF catalysts in 1 M HCOOH and 0.5 MH2504at 298 K. 5.3.3 Catalytic Activity and Long-Term Stability of Pd vs. Pt-Ru The Pd/GF prepared with 12.5 vol% Triton X-102 after 48 hours of pretreatment had the highest overall performance per geometric area (Figure 5-15) due to its high specific surface area and excellent penetration throughout the 3-D electrode. It also did not show any peak or current oscillation behavior (Appendix F, Figure F-3). The lack of 48 hr Pretreatment 24 hr Pretreatment 0 hr Pretreatment 221 peak in CVs is a general attribute observed for catalysts with very high activity [16-18]. Therefore, it was selected as the representative for Pd/GF in the comparison with Pt Ru/GF. Figure 5-15: Superficial formic acid oxidation current density at 0.3 V vs. SHE on electrodeposited PdIGF catalysts in 1 M HCOOH and 0.5 MH2S04at 298 K. In order to evaluate the applicability of Pd/GF as DFAFC anode, its long-term stability and catalytic activity towards formic acid electro-oxidation were compared by CV and CA to the Pt-Ru/GF prepared from the same deposition media as presented in Chapter 4. The intrinsic and mass-specific catalytic activity of the highest-performing electrodeposited Pt-Ru and Pd on GF are shown in Figure 5-16 below. Lii I Cl, lb > > 0 E 0 25 vol% Triton X-102 12.5 vol% Triton X-102 0 vol% Triton X-102 Pretreatment 0 hr Pretreatment 222 (a) ______________ 40 _ _ _ _ _ ____ _ _ _ _ __ A Pt-Ru/GF 35 —-p-— Pd/OF A 30 25 It a’ E 20 15 •1• 10 —— —- 0 I I I I I I -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 (b) EIVvs. SHE 600 A Pt-Ru/OF ——- Pd/OF A 500 _ _ _ _ _ 400 ,,_/ / / 200. E /V vs. SHE Figure 5-16: Voltammograms of formic acid electro-oxidation using Pt-Ru/GF and Pd/GF (prepared with 12.5 vol% Triton X-102 with 48 hours of Shipley pretreatment). (a) Real surface area basis; (b) Mass basis. Solution: 1 M HCOOH and 0.5 MH2S04.Temperature: 298 K. Scan rate: 0.005 V s’. 223 From the CV results, it could be seen that Pd/GF had higher catalytic activity based on real surface area in the low potential region while the Pt-RuJGF had higher oxidation current at high potentials (> 0.75 V vs. SHE). The observation that Pt-Ru had better performance than Pd-based DFAFC at high overpotentials and low cell voltage has been mentioned by Rice and coworkers [30]. The noteworthy difference in the onset potential for the two different catalysts indicates the difference in reaction mechanism as discussed in Section 1.3.2. CA data reveals that the kinetics of PdJGF is superior compared to that of Pt RuJGF (Figure 5-17). See also Figure 5-18 and Table 5-3 for the Tafel plots and parameters. (a) 5000 4500 4000 3500 C’4 3000 2500 2000 1500 1000 500 0 (b) 4000 3500 3000 2500 2000 1500 1000 500 0 Figure 5-17: Chronoamperometry data of formic acid electro-oxidation using Pt-Ru and Pd electrodeposited on GF. (a) PdIGF; (b) Expanded view of PdIGF at 0.45 V vs. SHE; (c) Pt-Ru/GF; (d) Expanded view of Pt-Ru/GF at 0.45 V vs. SHE; Solution: 1 M HCOOH and 0.5 MH2S04.Temperature: 298 K. PdIGF 005 Vvs.SI-E Pd/GE 0.15 Vvs. SI-E .—-. Pd/GE 025 V vs. SI-E Pd/GE 0.35 Vvs.SI-E —a-- Pd/GE 0.45Vvs.SHE F Pd/GE 0.55Vvs. SHE E D C B A 40 50 (b) 2600 2550 2500 2450 2400 2350 (d) 600 550 500 450 E : 400 350 300 250 0 10 20 30 t /s _J- Pt-Ru’GF 0.35 V vs. SHE Pt-RuIGF 0.45Vvs. SHE ..C Pt-Ru/GF 0.55 V vs. SHE Pt-Ru/GF 0.65Vvs. SHE Pt-Ru/GE 0.75 V vs. SHE F F Pt-Ru/GF 0.SSVvs. SHE 0 10 20 30 40 50 Us C B 0 10 20 30 40 60 Us 0 10 20 30 40 t /s 50 224 The Tafel parameters were calculated from the slope and x-intercept of the Tafel plots (see Section 1.2.3), which were based on the pseudo-steady state CA current response normalized to the active surface area at the low overpotential regions relative to the respective onset potential for the two catalysts (0.35 to 0.55 V for Pt-Ru and 0.05 to 0.25 for Pd) (ii E — Ee). 0.9 — —__________________ 0.8 -0.2002 OA 06 0.8 1 Iog(i) I A m2 Figure 5-18: Tafel plots of formic acid electro-oxidation in 1 M HCOOH and 0.5 M H2S04at 298 K. (a) 10 g m2 (4:1) Pt-Ru/GF; (b) 57 g m2 Pd/GF. Table 5-3: Apparent Tafel slope and exchange current density of formic acid electro-oxidation on Pt-Ru/GF and PdIGF. Catalyst Tafel Slope, b Exchange Current Density, i0 (V) (A m2 reai) Pt-Ru/OF 0.224 2.4x103 Pd/OF 0.260 6.9x102 225 5000 4500 4000 3500 C4 3000 E 2500 2000 1500 1000 500 0 0.0 Figure 5-19: Long-term chronoamperometry data of formic acid electro-oxidation using Pt-Ru and Pd electrodeposited on GF. Solution: 3 M HCOOH and 0.5 M H2504.Temperature: 298 K. Constant potential: 0.65 V vs. SHE. Even though Pd has higher catalytic activity than Pt-Ru, it deactivates at an undesirable rate compared to Pt-Ru. As can be seen from the long-term CA tests (Figure 5-19, see also Figure 5-17 b), Pt-Ru/GF retained approximately 88% of its pseudo steady- state current density (the current density value after the initial drop) after constant operation at 0.65 V vs. SHE for 3 hours. In contrast, the catalytic activity of Pd/GF declined steadily and went below that of Pt-Ru in about 4000 s resulting in a total drop in current density of 79% in 3 hours. This rate of deactivation is in agreement with literature publications by various researchers. Li and Hsing prepared supported Pd catalyst on Vulcan XC-72 by a surfactant-stabilized method and the prepared Pd/C lost 90% of its initial oxidation current density in 1200 s [2]. Similarly, Larsen et al. reported the CA data of an Alfa Aesar commercial Pd black catalyst at 0.3 V vs. SHE and the current 0.5 1.0 1.5 2.0 2.5 3.0 t/hr 226 density decreased by over 80% and 95% after 1 and 3 hours of operation, respectively [9]. The deactivation was proposed to be the result of reaction intermediate adsorption. However, other researchers have presented long-term results of more stable Pd catalysts. For example, Larsen et al. have presented the CA data of a Sigma-Aldrich commercial Pd black catalyst and it retained about half of its original catalytic activity after 3 hours of operation. The difference in stability was proposed to be a result of particle size and morphology effect. 5.3.4 DFAFC Performance of Pd vs. Pt-Ru Comparative polarization and power density curves of DFAFC tests at 333 K are shown in Figure 5-20. The steady-state OCV of Pd CCM and Pd/GF was 0.83 and 0.84, respectively. The corresponding OCV of Pt-Ru CCM and Pt-RuJGF was 0.64 and 0.63. Thus, regardless of electrode design, the OCV of Pd was significantly higher than Pt-Ru, which is in agreement with the more negative onset potential for formic acid oxidation on Pd. In general, Pd was found to have better performance than Pt-Ru at low current densities. In the case of CCM, Pt-Ru started to have higher performance at> 300 A m2, whereas in the case of the novel GF anodes, Pt-Ru overtook Pd at above 2600 A m2. The results agreed well with those presented by Rice et al. where Pt-Ru showed higher performance at high current densities compared to Pd-based catalysts [30]. Unfortunately, concurring with the half-cell electrochemical study, the deactivation of Pd was evident. While the performance of the Pt-Ru catalysts was essentially steady during the experiment, the cell voltage of the Pd-based DFAFC was decreasing at a rate of approximately 1 mV s1 at current densities> 200 A m2. It is also important to note that the mass-specific performance of Pt-Ru was superior compared to Pd for both GF and CCM (Figure 5-21). With regards to electrode design, it is clear that the prepared three dimensional Pd/GF provided significantly higher performance than the Pd CCM on both geometric (peak power density of 852 vs. 392 W m2) and mass-specific basis (15 vs. 9.8 W g’), further justifying the effectiveness of the novel anode design as demonstrated by Pt-RuJGF presented earlier in Chapter 4. 227 (a) 0.9 0.8 0.7 0.6 a>0.4 0.3 0.2 0.1 0.0 (b) 900 800 700- 600 Cl ‘ 500 400- 300 200 100- 0- 0.9 0.8 0.7 0.6 >0.5-S. a> 0.4 0.3 0.2 0.1 900 800 700 600 500 400 A B C D CCM-40 gm2 Pd,1 MFICOOH CCM- 10 g m2 Pt-Ru (1.0:1),1 M -ICOCH GF-57gm2Pd,1 MH000H÷a5MHSO4 GF -10 g m2 Pt-Ru (4.0:1), 1 M HCDOH +0.5 M H2S04 -S. S.’ 4% .t—--,-_ \ 0.0 0 500 1000 1500 2000 2500 3000 3500 4000 i/Am2 — —— — — Cl//,, A k. A CCM-40gm2Pd,1 HCOOH 1 CCM- lOg m2 Pt-Ru (1.0:1), 1 M HCOOH/ SF-Gig m2Pd,1 MH000H+0.5MH3 SF- lOg m2 Pt-Ru (4fl:1), 1 M HCOOH +0.5 M 1-12504 I I I I I I I I 1,.? 0 500 1000 1500 2000 2500 3000 3500 4000 i/Am2 Figure 5-20: DFAFC performance — effect of extended reaction zone vs. conventional anode design and Pd vs. Pt-Ru. The Pd/GF was prepared with 12.5 vol% Triton X-102 and 48 hours of Shipley pretreatment. Fuel: 1 M HCOOH with or without 0.5 M H2504 at 6 mE mm1.Cathode: 40 g Pt m2, dry 02 fed at 2.5 bar and 500 mE mlii. Temperature: 333 K. (a) Polarization curve; (b) Power density. 228 0.9 0.8 0.7 0.6 > 1) 0.4 0.3 0.2 0.1 0.0 90 80 70 60 50 40 30 20 10 0 0.9 0.8 0.7 0.6 r > V.,., - 00.4 90 80 70 60 50 40 30 20 10 0 400 (a) A CCM-4Ogm2Pd,1 MHCOOH — CCM-lOgmPt-Ru(1fl:1),1 MHCOOH GF-S7gmPd1MHCOOH+0.5MHS0 C ——— GF - 10 g m2 Pt-Pu (4.0:1), 1 M HCOOH +0.5 M H2S04 \ \ A\l --D 0.3 I I I I I I I P0.0 0 50 100 1.0 200 250 300 350 400 i/Ag1(b) I> D — — — B — —. — — A 2j,. CCM-4Orn PdIMHCOOH . — CCM -10gmPt-Ru(1.0:1),1 M HCOOHC GF-5lgm2Pd,1 M I-fCOOH ÷0.5M1-12S04 A _D.._ GF-lOgmPt-Ru(4.0:1),1 M HCOOH+O.5MHS0 0 50 100 150 200 250 300 350 i/A g1 Figure 5-21: Mass-specific DFAFC performance — effect of extended reaction zone vs. conventional anode design and Pd vs. Pt-Ru. The Pd/GF was prepared with 12.5 vol% Triton X-102 and 48 hours of Shipley pretreatment. Fuel: 1 M HCOOH with or without 0.5 M H2S04at 6 mL mm1.Cathode: 40 g Pt m2, dry 02 fed at 2.5 bar and 500 mL mm’. Temperature: 333 K. (a) Polarization curve; (b) Power density. 229 5.4 Conclusion The galvanostatic electrodeposition of Pd on OF was investigated using a Triton X-102 micellar solution. Pretreating the OF with a Shipley solution in conjunction with the use of Triton X-102 micellar media resulted in more uniformly dispersed nanoparticles throughout the thickness of the OF. The initial catalytic activity of the Pd/OF towards formic acid oxidation at low overpotentials was superior compared to Pt-Ru/OF prepared with the same colloidal media. However, the long-term stability of the Pd/OF was vastly inferior compared to Pt- RU/OF as shown in half-cell electrochemical and fuel cell tests. In DFAFC experiments, the advantage of the novel anode was once again demonstrated. With the Pd/OF (57 g m2) anode prepared in the presence of 12.5 vol% Triton X- 102 and 48 hours of Shipley pretreatment, a maximum power output on a geometric area basis of 852 W m2 was obtained at 333 K compared to 392 W m2 obtained with a commercial CCM with Pd anode catalysts (40 g m2). Comparatively, the Pt-Ru/OF (10 g m2, 4:1 atomic ratio) showed the highest peak power density of 860 W m2. From both half-cell electrochemical and fuel cell tests, it is evident that Pt-Ru is a more promising catalyst compared to Pd for the DFAFC, due to its higher long-term stability. 230 5.5 References 1. L.J. Zhang, Z.Y. Wang, D.G. Xia, J. Alloys Compd. 426 (2006) 268. 2. X. Li, I.M. Hsing, Electrochim. Acta 51(2006) 3477. 3. J.H. Choi, K.J. Jeong, Y. Dong, J. Han, T.H. Lim, J.S. Lee, Y.E. Sung, J. Power Sources 163 (2006) 71. 4. A. Capon and R. Parsons, J. Electroanal. Chem. 45 (1973) 205. 5. Y.X. Chen, M. Heinen, Z. Jusys, R.J. Behm, Angew. Chem. Tnt. Ed. 45 (2006) 981. 6. N.M. Markovic, H.A. Gasteiger, P.N. Ross, X. Jiang, I. Villegas, M.J. Weaver, Electrochim. Acta 40 (1995) 91. 7. M. Weber, J.T. Wang, S. Wasmus, R.F. Savinell, J. Electrochem. Soc. 143 (1996) L158. 8. D. Capon and R. Parsons, Electroanal. Chem. Interfacial Electrochem. 44 (1973) 239. 9. R. Larsen, S. Ha, J. Zakzeski, R.I. Masel, J. Power Sources 157 (2006) 78. 10. M. Arenz, V. Stamenkovic, T.J. Schmidt, K. Wandelt, P.N. Ross, N.M. Markovic, Phys. Chem. Chem. Phys. 5 (2003) 4242. 11. S. Blair, D. Lycke, C. lordache, Electrochem. Soc. Trans. 3 (2006) 1325. 12. T.R. Ralph, G.A. Hards, J.E. Keating, S.A. Campbell, D.P. Wilkinson, M. Davis, J. St-Pierre and M.C. Johnson, J. Electrochem. Soc. 144 (1997) 3845. 13. D. P. Wilkinson, M.C. Johnson, K.M. Colbow and S.A. Campbell, US Patent 5,874,182, February 13 (1999). 14. A. Bauer, E.L. Gyenge and C.W. Oloman, Electrochim Acta 51(2006) 5356. 231 15. A. Bauer, E.L. Gyenge and C.W. Oloman, J. Power Sources 167 (2007) 281. 16. T.T. Cheng, E.L. Gyenge, Electrochim. Acta 51(2006) 3904. 17. T.T. Cheng, E.L. Gyenge, J. App!. E!ectrochem. 38 (2008) 51. 18. T.T. Cheng, E.L. Gyenge, J. E!ectrochem. Soc. 155 (2008) B819. 19. R. Chetty, K. Scott, J. New Mater. Electrochem. Syst. 10 (2007) 135. 20. T.D. Tran, S.H. Langer, E!ectrochim. Acta 38 (1993) 1551. 21. C.L. Green, A. Kucernak, J. Phys. Chem. B 106 (2002) 1036. 22. M.Y. Rusanova, G.A. Tsirlina, O.A. Petrii, T.Y. Safonova, S.Y. Vasil’ev, Russ. J. Electrochem. 36 (2000) 457. 23. R. Chetty, K. Scott, J. New Mater. Electrochem. Syst. 10 (2007) 135. 24. R.S. Jayashree, J.S. Spendelow, J. Yeom, C. Rastogi, M.A. Shannon, P.J.A. Kenis, E!ectrochim. Acta 50 (2005) 4674. 25. Z. Liu, L. Hong, M.P. Tham, T.H. Lim, H. Jiang, J. Power Sources 161 (2006) 831. 26. G. Samjeske, M.Osawa, Angew. Chem. mt. Ed. 44 (2005) 5694. 27. G. Samjeske, A, Miki, S. Ye, A. Yamakata, Y. Mukouyama, H. Okamoto, M. Osawa, J. Phys. Chem. B 109 (2005) 23509. 28. G. Samjeske, A, Mild, S. Ye, M.Osawa, J. Phys. Chem. B 110 (2006) 16559. 29. G.G. Lang, M. Seo, K.E. Heusler, J. Solid State Electrochem. 9 (2005) 347. 30. C. Rice, S. Ha, R.I. Masel, A. Wieckowski, J. Power Sources 115 (2003) 229. 232 6 CONCLUSIONS AND RECOMMENDATIONS 6.1 Conclusions The importance of employing colloidal media and substrate pretreatment in the electrodeposition of Pt-Ru and Pd nanoparticles was clearly demonstrated in the present study. The phase behavior of the deposition bath and surface pretreatment affected the Pt- Ru and Pd morphology, Pt-Ru atomic ratio, particle size, mass load, and crystallographic features of the resulting catalysts. Three different substrates, RVC, GF, and Ti mesh were also investigated. In the case of RVC, it was found that pretreatment had a major influence on the morphology and mass load of the resulted Pt-Ru catalysts. The electrochemical cycling method roughened the RVC surface and created more nucleation sites; hence leading to higher catalyst load and smaller particles. Performing electrodeposition in the aqueous phase resulted in a Pt-Ru atomic ratio of 15:1 [1]. When the Triton X-100 / isopropanol / cyclohexane microemulsion deposition bath was employed, the Pt-Ru atomic ratio dropped to 1.3:1. From CV studies, the reduction in Pt-Ru ratio was a result of the inhibition of Pt deposition compared to Ru due to the presence of surfactant. The particle size obtained was found to decrease with the use of microemulsion and micellar deposition media. In the case of aqueous deposition, the particle size varied from 50 to 400 nm, leading to a mass specific surface area of 24 m2 g’. With the investigated microemulsion, the catalyst particles were 10 to 20 nm in diameter, resulting in a higher specific surface area of 40 m2 g’. With both aqueous phase and microemulsion, the resulting catalyst mass load remained very low (< 2.3 g m2) for practical direct liquid fuel cell application. Therefore, a Triton X- 1 00/isopropanol micellar system was developed, leading to mass loads of up to 12 g m2 [2]. The use of Triton X 100/isopropanol micellar media yielded a uniform mesoporous coating made up of interconnected nanoparticles of approximately 10 to 20 nm throughout the RVC thickness. Pt-Ru catalysts electrodeposited on RVC (12 g m2, 3.6:1 atomic ratio), GF (9.8 g m2, 4.0:1 atomic ratio), and Ti mesh (2.8 g m2, 4.5:1 atomic ratio) with the Triton X 233 100/isopropanol micellar media were tested in a DMFC. Pt-Ru/RVC and Pt-Ru/OF had similar geometric and mass-specific peak power densities (486 vs. 379 W m2 and 40.5 vs. 38.6 W g’, respectively). Due to the low mass load of Pt-Ru/Ti mesh, the geometric performance was inferior, only reaching a peak power density of 141 W m2. Among RVC and OF, which had desirable catalyst mass loads, the OF substrate was selected for further studies due to its more favorable mechanical properties, such as mechanical strength under compression. To further improve the catalyst activity and morphology of Pt-Ru electrodeposited on OF, a Triton X-102 micellar media with a higher cloud point temperature of 361 K (vs. 339 K of Triton X-100) was investigated. In the Triton X-102 concentration study with Pt and Ru precursor concentrations of 0.25 mM, the Pt-Ru ratio of the resulting catalysts decreased with increasing Triton X-102 content, from 9.2:1 (5 vol% Triton X 102) to 7.3:1 (25 vol% Triton X-102). The specific surface area increased from 7.6 to 16 m2 g’, due to the reduction in agglomeration with increasing Triton X-102 content. In the case of 0.75 mM Pt and Ru precursor concentration, the micellar media consisting of 12.5 vol% Triton X- 102 yielded a Pt-Ru ratio of 4.0:1 with a specific surface area of 16.3 m2 g’. When Triton X-102 was absent, at the same conditions, the Pt-Ru ratio and the specific surface area were 5.2:1 and 14.6 m2 g’, respectively. In addition to the difference in Pt-Ru ratio, the catalyst morphology and uniformity with and without the presence of Triton X- 102 differed significantly. In the case with Triton X-102, individual particles ranging from 10 to 20 nm were obtained throughout the thickness of the unpressed OF. Without Triton X-102, catalyst agglomerates (up to 1000 nm) were found on the exterior of the OF, while sparse coatings of 50-nm particles were observed in the interior, thus concurring with the RVC results and signifying the importance of surfactants for preparing uniform catalyst coatings throughout the thickness of 3-D substrates. Furthermore, as compared to the Triton X-100/isopropanol micellar media, the Triton X-102 deposition bath, based on XRD spectra, led to catalysts richer in Pt(11 1) and Pt(3 11) crystalline facets, leading to higher catalytic activity. Compared to selected literature data, the exchange current density of the novel catalyst for methanol oxidation was also found to be two or more orders of magnitude higher, indicating enhanced oxidation kinetics. 234 In the factorial study of Shipley solution pretreatment time and Triton X- 102 concentration, their effects on Pd electrodeposition were clearly demonstrated. Pretreating the unpressed OF with Shipley solution (6 mM PdC12 and 0.3 M SnC12 in 4 M HC1) led to more uniform deposits, as well as enhanced penetration depth throughout the OF thickness. However, the Shipley solution pretreatment led to lower intrinsic catalytic activity based on real catalyst area due most likely to the different surfactant adsorption orientation caused by the change in the OF surface hydrophobicity, leading to different catalyst morphologies. The presence of Sn is also believed to be a contributing factor. Similar to the Shipley pretreatment, the presence of Triton X-102 yielded more uniform deposits and a lesser degree of catalyst agglomeration. With a Shipley solution pretreatment of 48 hours, the specific surface area increased from 2.2 to 5.1 m2 g’ when 12.5 vol% Triton X-102 was added. It was also found that increasing the Triton X-102 concentration generally reduced the overall catalyst load due to the suppression of Pd deposition, most notably when the Triton X-102 content was increased from 12.5 to 25 vol%. The Pd catalyst mass load deposited on OF pretreated with Shipley solution dropped from 31 to 15 g m2 (24 hours pretreatment) and 57 to 14 g m2 (48 hours pretreatment). The electrodeposition of Pt-Ru on OF using a Triton X- 102 micellar solution resulted in novel anodes with enhanced performance for both the DMFC and DFAFC. The best extended reaction zone anodes prepared with this method, which had a mass loading of 10 g m2 (4:1 Pt:Ru atomic ratio) showed superior performance in fuel cell tests over commercial Pt-Ru (1:1 atomic ratio) with a mass load of either 10 or 40 g m2. The peak power output of the Pt-Ru/OF reached up to 741 W m2 and 74.1 W g’ in a DMFC operating at 333 K, whereas the 10 and 40 g m2 commercial CCMs only reached peak power densities of 442 and 703 W m2 (corresponding to a mass-specific power density of 44.2 and 17.6 W g’), respectively. The results imply a 68% increase in power density for catalysts with the same mass load and a four-fold reduction in precious metal mass load while having superior area-specific performance. In DFAFC experiments, depending on fuel composition, the OF anode yielded up to 63% increase in peak power density (up to 860 W m2), compared to 528 W m2 when conventional catalysts with the same mass load were employed. Similarly, Pd/OF (57 g m2) provided significantly 235 higher performance than the Pd CCM (40 g m2) on both geometric (peak power density of 852 vs. 392 W m2) and mass-specific basis (15.0 vs. 9.8 W g1). Compared to commercial CCMs, the use of 3-D anodes was clearly demonstrated to improve the fuel cell performance, most notable at high current density. The improvement was most likely due to combinatory effects of enhanced oxidation kinetics/catalyst utilization as a result of modified crystalline structure and the absence of Nafion®, fuel mass transport owing to better CO2 disengagement, and reduced fuel crossover at low fuel concentration. In the comparative study of Pt-Ru/GF and Pd/GF, it was clearly demonstrated that Pd had superior catalytic activity at low overpotentials for formic acid electro-oxidation compared to Pt-Ru, which can be seen from the higher performance obtained with Pd/OF at < 2600 A m2. At a current density of 2000 A m2, the DFAFC based on the Pd/OF (57 g m2) reached peak power (852 W m2) with a cell voltage of 0.426 V. At the same current density, the cell voltage was 0.3 14 V with the Pt-RuJGF (10 g m2) anode. Unfortunately, the stability of Pd/OF was vastly inferior compared to Pt-RuJGF, notably in fuel cell and half-cell tests. The novel approach of employing the three-dimensional electrode design in conjunction with the studied colloidal deposition method was clearly shown to be a promising way to improve the catalyst morphology, modify the crystallographic features, and increase penetration throughout the 3-D matrix. These aspects led to enhanced catalyst utilization efficiencies in direct liquid fuel cells. The innovative anode catalyst layer design is an effective way to reduce precious metal load in DLFC’s, which can lower the overall fuel cell system costs, thereby facilitating fuel cell commercialization. 6.2 Recommendations With respect to the current study, the following supplementary investigations could bring further insights to the 3-D anode design: 1) TEM can be utilized to visualize or quantify the catalyst crystallites. 2) XPS can be used to determine the surface Pt-Ru ratio and the oxidation states of the metallic elements. 236 3) Post-mortem analysis including XRD, TEM, and XPS can be performed to examine the 3-D anode after fuel cell experiments to help understand the effects of fuel cell operation stressors on the novel 3-D anode. The degree of Ru crossover to the cathode catalyst layer can also be evaluated by using ICP-AES on scrapped-off cathode catalysts. 4) Reference electrodes can be added to the anode or cathode to measure the half-cell potentials as functions of operating conditions and time. The degree of fuel crossover can be monitored through the cathode potential. 5) The effect of substrate pore size can be investigated with commercially available RVC with different pore sizes. In terms of engineering design, the use of 3-D anodes, depending on the substrate thickness, could increase the overall thickness of the fuel cell stack, which might be undesirable for compact fuel cell systems used in small electronic devices, such as cell phones and laptops, where the design space is highly limited. Instead of using relatively thick substrates such as uncompressed OF or RVC, thinner substrates could be used while still maintaining the 3-D and extended reaction zone concepts. Possible commercially- available candidates are AvCarb® carbon-based fiber products with different thickness, 1O0 to 400 pm. The overall performance could be enhanced by balancing the degree of reaction zone extension (i.e. thickness) and ohmic loss. The use of sulfuric acid as supporting electrolyte could lead to more sophisticated fuel storage for the DMFC and DFAFC, which originally did not require the use of supporting electrolyte with conventional anode designs. As shown in the DFAFC experiments, the use of high formic acid concentration (10 M) was detrimental due to more severe crossover due to the open pore structure of the OF substrate as well as a reduction in ionic conductivity [3]. If an ionic conductive network could be formed on the novel catalyst layer, the use of supporting electrolyte would become unnecessary. Solid ionic conductor such as Nafion® and heteropolyacid hydrate could be used [4-6]. However, this is an extremely difficult task for OF as any ionic conductive network linking the individual GF fibers would without a doubt lower significantly the electronic conductivity of the substrate due to the reduction of electronic conductive contact points between the fibers. This problem could be mitigated with the use of continuous substrates 237 in conjunction with a preparation method involving a deposition media composed of an ionomer (such as Nafion®) and the catalyst precursor, co-depositing the ionomer and catalysts simultaneously. It must be noted that the extended reaction zone anode concept is particularly applicable to fuel cell systems where the fuel solution inherently contains or requires the use of an ionic conductive component, including the alkaline DMFC and the direct borohydride fuel cell (DBFC) [7-14]. In the alkaline DMFC and DBFC, sodium hydroxide is typically used to adjust the solution pH to the alkaline region. In the case of DBFC, it is also required to increase the stability of sodium borohydride to prevent hydrolysis [15-18]. It is also important to note that the development of new ionic conductors could also benefit the 3-D electrode concept. New or modification to existing catalyst formulations should be developed to improve the long-term stability of the DMFC and DFAFC. Even though Pt-Ru is more stable compared to Pd, it is well-known that Ru dissolution can occur during cell operation. The dissolved Ru can crossover and deposit on the cathode side, causing a negative impact on the oxygen reduction kinetics and leading to poorer overall cell performance [19-23]. Therefore, a Ru-free catalyst formulation might be more beneficial for better cell durability. One possible approach has been demonstrated by Xi et al. where Pt nanoparticles was physically mixed with transition metal oxides (such as Ti02) and it was shown that the catalytic activity of methanol electro-oxidation and CO tolerance was improved compared to Pt/C [24]. 238 6.3 References 1. T.T. Cheng, E.L. Gyenge, Electrochim. Acta 51(2006) 3904. 2. T.T. Cheng, E.L. Gyenge, J. Appi. Electrochem. 38 (2008) 51. 3. T.T. Cheng, E.L. Gyenge, J. Electrochem. Soc. 155 (2008) B819. 4. J. Chou, E.W. McFarland, H. Metiu, J. Phys. Chem. B 109 (2005) 3052. 5. J. Chou, S. Jayaraman, A.D. Ranasinghe, E.W. McFarland, S.K. Buratto, H. Metiu, J. Phys. Chem. B 110 (2006) 7119. 6. H. Hatakeyama, H. Sakaguchi, K. Ogawa, H. Inoue, C. Iwakura, T. Esaka, J. Power Sources 124 (2003) 559. 7. J. Liu, J. Ye, C. Xu, S.P. Jiang, Y. Tong, J. Power Sources 177 (2008) 67. 8. K. Scott, E. Yu, G. Viachogiannopoulos, M. Shivare, N. Duteanu, J. Power Sources 175 (2008) 452. 9. A. Verma, S. Basu J, Power Sources 174 (2007) 180. 10. G.H. Miley, N. Luo, J. Mether, R. Burton, G. Hawkins, L. Gu, E. Byrd, R. Gimlin, P.J. Shrestha, G. Benavides, J. Laystrom, D. Caroll, J. Power Sources 165 (2007) 509. 11. R.K. Raman, S.K. Prashant, A.K. Shukia, J. Power Sources 162 (2006) 1073. 12. J.H. Wee, J. Power Sources 161 (2006) 1. 13. A. Verma, S. Basu, J. Power Sources 145 (2005) 282. 14. N.A. Choudhury, R.K. Raman, S. Sampath, A.K. Shukia, J. Power Sources 143 (2005) 1. 15. J.C. Walter, A. Zurawski, D. Montgomery, M. Thornburg, S. Revankar, J. Power Sources 179 (2008) 335. 239 16. Z.P. Li, B.H. Liu, K. Arai, K. Asaba, S. Suda, J. Power Sources 126 (2004) 28. 17. S.C. Amendola, S.L. Sharp-Goldman, M.S. Janjua, N.C. Spencer, M.T. Kelly, P.J. Petillo, M. Binder, mt. J. Hydrogen Energy 25 (2000) 969. 18. Y. Kojima, K.I. Suzuki, K. Fukumoto, M. Sasaki, T. Yamamoto, Y. Kawai, H. Hayashi, mt. J. Hydrogen Energy 27 (2002) 1029. 19. P. Piela, C. Eickes, E. Brosha, F. Garzon, P. Zelenay, J. Electrochem. Soc. 151 (2004) A2053. 20. L. Gancs, B.N. Hult, N. Hakim, S. Mukerjee, Electrochem. Solid-State. Left. 10 (2007) B150. 21. M. Inaba, M. Sugishita, J. Wada, K. Matsuzawa, H. Yamada, A. Tasaka, J. Power Sources 178 (2008) 699. 22. M.K. Jeon, K.R. Lee, K.S. Oh, D.S. Hong, J.Y. Won, S. Li, S.I. Woo, J. Power Sources 158 (2006) 1344. 23. Z.B. Wang, H. Rivera, X.P. Wang, H.X. Zhang, P.X. Feng, E.A. Lewis, E.S. Smotkin, J. Power Sources, 177 (2006) 386. 24. J. Xi, J. Wang, L. Yu, X. Qiu, L. chen, Chem. Commun. (2007) 1656. 240 APPENDIX A - GENERAL BACKGROUND OF METHANOL AND FORMIC ACID The annual global production rate of methanol is above 25 billion liters per year and in Canada alone, around 1.6 billion liters per year is produced [1]. Currently, a large amount of methanol is used as a solvent in chemical cleaners, approximately 40% of the methanol produced is used to make formaldehyde, 20% is used to manufacture the fuel addictive methyl tertiary-butyl ether (MTBE), and only 2% is used as fuel [2]. The production of methanol is a well-established technology. Methanol can be produced from syngas obtained from various sources, such as natural gas (by steam reforming, Equation A-i), biomass, and coal (both by gasification). CH4+H20->CO+3H [Eq. A-i] Steam reforming typically operates at high temperatures (973 to 1273 K) and high pressure (10 to 20 bar) over nickel catalysts [1]. The ratio of CO and H2 can be adjusted to obtain the desired stoichiometry through the water-gas shift reaction, which takes place at a lower temperature (typically at 473 to 673 K) shown in Equation A-2 [2]. CO + H20 -> CO2 + H2 [Eq. A-2] The syngas mixture can then be used to produce methanol according to Equation A-3 using a copper, zinc oxide, or alumina catalyst at high temperature (> 500 K) and pressure (50-100 bar). CO + 2H -CH3O [Eq. A-3] With a well-developed and economical method for production and readily available sources, methanol remains a low-cost fuel. From ICIS Chemical Business (rebranded Chemical Market Reporter), it has an average price of $0.40 per liter [3]. 241 Formic acid is primarily used as silage addictive as a preservative and an antibacterial agent. It is also used in rubber synthesis from organic sap as well as by the textile industry for the tanning of leather. It has a considerable global production rate of around 0.27 billion liters per year. Approximately 60% of all formic acid is produced from methyl formate, with the remaining fraction produced from liquid-phase oxidation of hydrocarbons and formate salt-based processes [4]. The methyl formate precursor employed in the major production route is obtained from carbonylation of methanol (Equation A-4) with carbon monoxide at elevated temperature (353 K) and pressure (45 bar) in the presence of a strong base, such as sodium methoxide. CH3O + CO -* HCOOCH3 [Eq. A-4] Formic acid is then produced by either the direct hydrolysis of methyl formate and later separated from the liquid mixture by liquid-liquid extraction with a nitrogen base (see Equation A-5) or via an indirect pathway by first reacting methyl formate with ammonia to form formamide, which is then hydrolyzed in the presence of sulfuric acid (Equation A-6 and A-7). HCOOCH3+ H20-+ HCOOH + CH3O [Eq. A-5] HCOOCH3+ NH3 -> HCONH2+ CH3O [Eq. A-6] HCONH2+H20+HS04-> HCOOH+(NH4)2S [Eq. A-7] These complex chemical processes have made formic acid a relatively expensive chemical, especially compared to methanol. Formic acid has a price of $0.90 per liter [3], significantly higher than that of methanol, especially considering its lower energy content. 242 References 1. A. English, J. Rovner, J. Brown, S. Davies, “Kirk-Othmer Encyclopedia of Chemical Technology”, John Wiley & Sons Inc., (2000), Vol. 16: 299-3 16. 2. J. Larminie & A. Dicks, “Fuel Cell Systems Explained Second Edition”, John Wiley & Sons Ltd., London (2003). 3. ICIS Indicative Chemical Prices, 2008. (http://www.icis.com/StaticPages/k o.htm#M) 4. D.J. Drury, “Kirk-Othmer Encyclopedia of Chemical Technology”, John Wiley & Sons Inc., (2000). (http://www.mrw.interscience.wiley.comlemrw/978047123 8966/kirk/article/for mdrur.aO 1 /current/pdf) 243 APPENDIX B - SEM IMAGES OF Pt-Ru/RVC and Pt-RuIGF Figure B-i: SEM images of Pt-RuIRVC and Pt-RuIGF prepared without surfactant at 34i K. (a, b) Pt-Ru/RVC prepared with 0.25 mM H2PtC16 and (NH4)2RuC16and deposition current density iO A m2 (a: exterior vs. b: interior); (c, d) Pt-Ru/GF prepared with 0.75 mMH2PtC16and (NH4)RuC16and deposition current density 20 A m2 (C: exterior vs. d: interior). Exterior Surface Inner Surface 244 Figure B-2: SEM images of Pt-RuIRVC and Pt-Ru/GF prepared with surfactants at 341 K. (a, b) Pt-Ru/RVC prepared with 0.25 mM H2PtC16 and (NH4)2RuC16in 5 vol% Triton X-100 I 20 vol% isopropanol micellar media and a deposition current density 10 A m2 (a: exterior vs. b: interior); (c, d) Pt-RuIGF prepared with 0.75 mM H2PtC16 and (NH4)RuC16 in 12.5 vol% Triton X-102 micellar media and a deposition current density 20 A m2 (C: exterior vs. d: interior). Exterior Surface Inner Surface — 44 245 APPENDIX C - THEORETICAL SURFACE AREA CALCULATIONS FROM XRD The crystallite size, d, was estimated using Scherrer’ s formula (Equation C-i), where is the radiation wavelength of the X-ray, B is the width of the peak at half maximum in radians, and 0 is the diffraction angle: dz 0.92 [Eq.C-1] B cos For the Pt-Ru sample prepared with 0.75 mM H2PtC16 and (NH4)2RuC16in 12.5 vol% Triton X-102 micellar media, the crystallite sizes obtained were 5.8, 7.4, 5.7, 4.5, and 9.7 nm for the Pt(1 11), Pt(i00), Pt(1 10), Pt(3 ii), and Ru(Hexagonal), respectively. Along with fractions of Pt crystallographic facets data, the average crystallite size of Pt was found to be 5.9 nm. Assuming spherical particles, the surface area and volume of each crystallite were found to be 1.10x10’6m2 and 1.09x1025 m3. With a Pt density of 21450 kg m3, the specific surface area of Pt was found to be 47268 m2 kg’. With an atomic ratio of 4:1 and a total catalyst load of 10 g m2, the specific Pt mass load was 8.9 g m2. The theoretical surface area is the product of the specific surface area based on crystallite size and the specific mass load. Hence, the theoretical surface area of Pt was estimated to be 418 m2 m2. Similarly, with a Ru crystallite size of 9.7 urn, the surface area and volume of each crystallite were found to be 2.94x10’6m2 and 4.74x1025 m3. With a Ru density of 12450 kg m3, the specific surface area of Pt was found to be 49803 m2 kg* The specific Ru mass load was 1.2 g m2. Therefore, the theoretical surface area of Ru was estimated to be 57 m2 m2. The total theoretical surface area was then calculated to be 475 m2 m2. For the Pd sample prepared with 12.5 vol% Triton X-102 micellar media and a Shipley solution pretreatment time of 48 hours, the crystallite size for Pd( 111), Pd( 100), Pd(110), and Pd(311), was found to be 27.8, 19.0, 22.6, and 18.5 urn, respectively. Along with fractions of Pd crystallographic facets data, the average crystallite size of Pt was found to be 24.0 nrn. Assuming spherical particles, the surface area and volume of each crystallite were found to be 1.81x10’5m2 and 7.24x102 m3. With a Pd density of 12023 kg m3, the specific surface area of Pd was found to be 20789 m2 kg’. The specific Pd 246 mass load was 57 g m2. The theoretical surface area is the product of the specific surface area based on crystallite size and the specific mass load. Hence, the theoretical surface area of Pd was estimated to be 1185 m2 m2. 247 APPENDIX D - ADDITIONAL SUPPORTING DATA (a) ____________________ ___ 30 _ _ _ ______ _ _ _ _ _ ______ _ _ _ _ A Aqueous Deposition Media 25 Deposition Current 20 Am2 0.75 mM H2PtCI6 20 B 0.75 mM (NH4)RuCI — — — 12.5 vol% Triton X-102 Deposition Current: 20 Am2 0.75 mM H2PtCI6 0.75 mM (NH4)RuCI / E 10 / io01:1 02 03 5 0 6 Q17 08 0 9 (b) ElVvs. SHE 500 A Aqueous Deposition Media Deposition Current 20 Am2 400 0.75 mM H2PtCI6 0.75 mM (NH4)PuCI — — — 12.5 oI%Triton X-102 B 300 Deposition Current 20 Am2 / 0.75 mM I-I2PtCI6 0.75 mM (NH4)RuCI6 200- / / 100 —100 I I I I I -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 E/Vvs. SHE Figure D-1: Voltammograms of methanol electro-oxidation using Pt-Ru electrodeposited on GF: Triton X-102 vs. aqueous deposition media. (a) Real surface area basis; (b) Mass basis. Solution: 1 M CH3O and 0.5 M H2S04.Temperature: 298 K. Scan rate: 0.005 V s. 248 2200 A Pt-Ru/CF 0.35 V vs. SHE 2000 B Pt-Ru/CF 0.45 V vs. SHEC Pt-Ru/CF 0.55 V vs. SHE 1800 D Pt-Ru/CF 0.65 V vs. SHE E Pt-Ru/CF 0.75 V vs. SHE 1600 F Pt-Ru/CF 0.85 V vs. SHE 1400 ____ F 1200 E 1000 800 D 600 — 400 C 200 B A 0 I I I 0 10 20 30 40 50 t/s Figure D-2: Chronoamperometry data of methanol electro-oxidation using Pt RuIGF (10 g m2, 4:1 atomic ratio, prepared in 12.5 vol% Triton X-102 with 0.75 mMH2PtC16and (NH)RuC16in 1 M CH3O and 0.5 MH2S04.Temperature: 298 K. 249 O1 Iog(i) I A m2 Figure D-3: Tafel plots of methanol electro-oxidation in 1 M HCOOH and 0.5 M H2S04at 298 K with Pt-RuIGF prepared in 12.5 vol% Triton X-102 with 0.75 mM H2PtC16and (NH4)RuC16(10 g m2, 4:1 atomic ratio). 250 (a) 0.65 . ___________________________ 0.65 0.60 • SF-lOgm2Pt-Ru@.0:1) 0.60 055 I V Replicate 1Replicate2 0.55 0.50 0.50 0.45 9 0.45 0.40 . 0.40 0.35 0.35 - 0.30 0.30 uJ w 0.25 0.25 0.20 0.20 0.15 0.15 0.10 0.10 0.05 0.05 0.00 I I I I I I I I I 0.00 0 250 500 7 1000 1250 1J0 1750 2000 2250 2500 2750 3000 / A m2 (b) 900 BOO 750 750 700 700 650 650 600 q 600 550 550 500 500 450 450 400 400 350 • SF-lOg mPt-Ru(4.0:1) 350 300 V Replicate 1 300 Replicate 2 250 ______ __________ 250 200 200 150 150 100 100 50 V 50 0 I I I I I I I I I U 0 250 SOD 750 1000 10 1500 1750 2000 2260 2500 2750 3000 if A m2 Figure D-4: DMFC performance of 10 g m2 Pt-Ru/GF. Fuel: 1 M CH3O with 0.5 M H2S04 at 2 mL mm1. Cathode: 40 g Pt m2, dry 02 fed at 2.5 bar and 500 mL mm1.Temperature: 333 K. (a) Polarization curve; (b) Power density. 251 2000 1900 1800 1700 1600 1500 - 1400 - 1300- 1200• 1100• 1000- > 900- — 800 ______________________________________ 700- 600 500 400 __ _ __ __ __ _ __ __ __ _ __ __ _ __ __ __ _ 300 200 100 0 0.30 0.28 0.26 0.24 0.22 0.20 0.18 LU 0.160.14- 0.12- 0.10- > 0.08 —- 0.06- 0.04 0.02 0.00 -0.02 -0.04 -0.06 -0.08 -0.10 Figure D-5: Catalytic activity of Pd/GF deposited at 20 A m2 in the presence of 12.5 vol% Triton X-102 and 4.5 mM PdC12 at 341 K for 120 mm after 48 hours of Shipley solution pretreatment. Solution: 1 M HCOOH and 0.5 M H2504. Temperature: 298 K. (a) CA data, Potential: 0.25 V vs. SHE; (b) CP data, Current density: 50 A rn2. (a) Pd (12.5 val%TritonX-102After48 hr Pretreatment — — — Pd (Replicate) — — — — Pd (Replicate 2) (b) I I I I I 0 5 10 15 20 25 30 35 40 45 50 tf S Pd (12.Svol% Triton X-102 .fler4B hr Pretreatment — — — Pd (Replicate) Pd(Replicate2) 0 25 50 75 100 125 150 175 200 225 250 275 300 t/s 252 Table D-1: Mass load and Pt:Ru atomic ratio of Pt-Ru/GF deposited at 20 A m2 in the presence of 12.5 vol% Triton X-102, 0.75 mM H2PtC16, and 0.75 mM (NH4)2RuC16at 341 K for 120 mm. Pt-Ru Mass Load (g m2) Pt:Ru Atomic Ratio Sample 1 9.4 4.3:1 Sample 2 8.4 4.0:1 Sample3 11 3.8:1 Average 10 4.0:1 Table D-2: Mass load of Pd/GF deposited at 20 A m2 in the presence of 12.5 vol% Triton X-102 and 4.5 mM PdCl2 at 341 K for 120 mm after 48 hours of Shipley solution pretreatment. Pd Mass Load (g m2) Sample 1 55 Sample 2 62 Sample 3 53 Average 57 253 APPENDIX E - DFAFC PERFORMANCE WITH 3 AND 10 M HCOOH 0.65 0.65 0.60 0.60 0.55 0.55 0.50 0.50 0.45 0.45 0.40 0.40 0.35 0.35 0.30 0.30 LU 0.25 0.25 0.20 0.20 0.15 0.15 0.10 0.10 0.05 0.05 0.00 0.00 4000 750 750 700 700 650 650 600 600 550 550 500 500 450 450 E4 400 350 350 °- 300 300 °- 250 . 250 200 . 200 150 .150 100 100 50 •50 0 (a) > a.) Li LU (b) • CCM- 10 g m2 Pt-Ru (1.0:1),3 M HCOOH • GF- 10 g m2 Pt-Ru (4JJ:1),3 M HCOOH v GF- 10 g m2 Pt-Ru (4.I]:1),3 M HCOOH ÷05 M H2S04 . 9. 9. aT • VT * VT, $1. VT. 0 500 1000 1500 2W0 2500 3000 3500 /A m2 V ‘V V V V : : 9 • CCM-lOgm2Pt-Ru(1.0:1)3MHCODH • GF-10gm2Pt-Ru(4.0:1),3 MHCOOH GF -10 g rn2 Pt-Ru (4.0:1),3 M HCOOH ÷0.5 M H2S04 0 500 1000 1500 2000 2500 3000 3500 4W0 i/A m2 Figure E-1: DFAFC performance — effect of extended reaction zone vs. conventional anode design. Fuel: 3 M HCOOH with or without 0.5 M H2S04 at 6 mL mm1. Cathode: 40 g Pt m2, dry 02 fed at 2.5 bar and 500 mL mm1.Temperature: 333 K. (a) Polarization curve; (b) Power density. 254 (a) 0.60 0.60 0.55 • CCM- 10 g m2 Pt-Ru (1.0:1), 10 M HCOOH 0.55 • GF- 10 g rn2 Pt-Ru (4.0:1), 1DM HCOOH V GF- lOg m2 Pt-Ru (4.0:1), 1DM HCQOH -i-U.S M H2S04 0.45 9 _____________________________________________________ 0.45 • 0.40 • 0.40 > 0.35 • 0.35 0.30 • • 0.30 LI LI W 0.25 * • 0.25 W •* 0.20 0.20 •TVv •. 0.15 • • v 0.15 . 0.10 0.10 0.05 0.05 0.00 0.00 0 500 1000 1500 210 2500 3000 3500 4000 if A m2 (b) 650 . 600 . 600 550 550 500 500 450 • v VVy •450 ‘V V 400 • • . • • • 400 350 350 •300 300 250 250 200 9 200 150 CCM-lOgm2Pt-Ru(1.0:1)1OMHCQOH 150 100 . CF - lOg m2 Pt-Ru (4.0:1), 10 M HCOQH ‘100 SD . V GF - lOg rn2 Pt-Ru (4.0:1), 1DM HCOOH +0.5 M H2S04 60 0 ‘ I I I I I 0 0 500 1000 1500 2000 2500 3000 3500 410 i/A m2 Figure E-2: DFAFC performance — effect of extended reaction zone vs. conventional anode design. Fuel: 10 M HCOOH with or without 0.5 M H2S04 at 6 mL mm1. Cathode: 40 g Pt m2, dry 02 fed at 2.5 bar and 500 mL mm1.Temperature: 333 K. (a) Polarization curve; (b) Power density. 255 APPENDIX F - CYCLIC VOLTAMMOGRAMS OF FORMIC ACID ELECTRO. OXIDATION USING ELECTRODEPOSITED Pd/GF 120 ___________________________________________ Triton X-102: 12.5 vol%; No Pretreatment — — — Triton X-102: 2EvoI%; No Pretreatment 100 _______ _______ Si 2: D02O 0 0 011.4 E/Vvs. SHE Figure F-i: Cyclic voltammograms of formic acid electro-oxidation using Pd/GF with no pretreatment: Effect of Triton X-i02 concentration. Solution: i M HCOOH and 0.5 MH2S04.Temperature: 298 K. Scan rate: 0.005 V s1. 256 120 Triton X-102: OvoI%; 24-hr Pretreatment — — Triton X-102: 12.5voI%; 24-hr Pretreatmentirn C — — — — Triton X-102: 2 vol /; 24-hr Pretreatment 2: 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 E/Y vs. SHE Figure F-2: Cyclic voltammograms of formic acid electro-oxidation using Pd! GF with 24-hr Shipley pretreatment: Effect of Triton X-102 concentration. Solution: 1 M HCOOH and 0.5 MH2S04.Temperature: 298 K. Scan rate: 0.005 V s. 257 60 40 w L4 E 9 —,. 4 0 Figure F-3: Cyclic voltammograms of formic acid electro-oxidation using Pd/with 48-hr Shipley pretreatment: Effect of Triton X-102 concentration. Solution: 1 M HCOOH and 0.5 MH2S04.Temperature: 298 K. Scan rate: 0.005 V s. 0.0 0.2 0.4 0.6 0. 1.0 1.2 1.4 E/Vvs. SHE 258

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}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            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:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0058650/manifest

Comment

Related Items