Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Electrocatalysis on PtZn surface - fundamental studies and the applications in PEM fuel cells Sode, Aya 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_sode_aya.pdf [ 17.96MB ]
Metadata
JSON: 24-1.0061704.json
JSON-LD: 24-1.0061704-ld.json
RDF/XML (Pretty): 24-1.0061704-rdf.xml
RDF/JSON: 24-1.0061704-rdf.json
Turtle: 24-1.0061704-turtle.txt
N-Triples: 24-1.0061704-rdf-ntriples.txt
Original Record: 24-1.0061704-source.json
Full Text
24-1.0061704-fulltext.txt
Citation
24-1.0061704.ris

Full Text

ELECTROCATALYSIS ON PtZn SURFACE – FUNDATMENTAL STUDIES AND THE APPLICATIONS IN PEM FUEL CELLS by AYA SODE B.SC., DALHOUSIE UNIVERISTY, 2002 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF  THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Chemistry) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) February 2009 © AYA SODE 2009 Abstract The  PtZn  alloy  was  characterized  and  studied  as  a  possible  candidate  for  proton exchange membrane fuel cells  (PEMFCs).  Zn was electrochemically deposited on a bulk Pt substrate and the alloy was spontaneously formed at ambient conditions, as confirmed by Auger electron  spectroscopy  (AES).  PtZn  decreased  the  oxygen  reduction  reaction  (ORR) overpotential by 30 mV with respect to that measured for Pt. The alloy was also prepared in a similar manner on a membrane electrode assembly (MEA) and studied in a single stack fuel cell. The  PtZn  MEA also  showed  an  improved  performance  compared  to  the  Pt  MEA.  Carbon monoxide oxidation catalysis was also investigated. Zn modification of a Pt substrate makes the resulting  surface  rough,  and  the  roughness  was  compensated  through  the  measurement  of capacitance.  Although,  in  this  case,  Zn  modification  did  not  enhance  the  catalytic  activity compared to Pt, cycling the potential in a CO saturated acidic solution made the PtZn surface smoother and more Pt-rich. The ORR catalytic activity was also compared to Pt with equivalent roughness. PtZn showed enhancement in the kinetics of the reaction, indicating Zn modification had impact on improving the ORR catalytic activity beyond the roughness effect.  The PtZn surface becomes Zn-rich after cycling the potential in an oxygen saturated acidic solution. The stability of the alloy under fuel cell operating conditions was also studied. PtZn nanoparticles were formed on glassy carbon substrates and the surface and bulk alloy compositions were determined before and after  the alloy was polarized in  an oxygen saturated acidic  solution. Approximately 18% of Zn originally present in the alloy was lost after 40 hours of polarization at 0.8 V  vs. RHE. It was also confirmed PtZn particles adhere well to carbon and were free from sintering  effects.  The  cumulative  results  presented  support  that  the  PtZn  alloy  is  thus  a promising candidate for the cathode of PEMFCs. ii Table of Contents Abstract..........................................................................................................................................ii Table of Contents.........................................................................................................................iii List of Tables..................................................................................................................................x List of Figures...............................................................................................................................xi List of Symbols And Abbreviations..........................................................................................xxi Acknowledgments..................................................................................................................xxviii 1.   Introduction.............................................................................................................................1 1.1.  Electrochemical reactions at the electrode......................................................................4 1.2.  Outline of the thesis.........................................................................................................6 2.  Literature Review....................................................................................................................8 2.1.  Introduction.....................................................................................................................8 2.2.  Proton exchange membrane fuel cell..............................................................................8 2.2.1.  Membrane electrode assembly...............................................................................8 2.3.  Electrochemistry............................................................................................................10 2.3.1.  Metal/liquid interface and electrical double layer...............................................10 iii 2.3.2.  Metal side of interface: surface dipole.................................................................14 2.4.  Thermodynamics and kinetics of electrode reactions...................................................16 2.4.1.  Fuel cell power output.........................................................................................16 2.4.2.  Current-potential relationship in kinetics............................................................17 2.4.2.1.  Overpotential.......................................................................................................17 2.4.2.2.  Current-overpotential relationship.......................................................................19 2.4.3.  Heterogeneous catalysis.......................................................................................22 2.5.  Platinum.........................................................................................................................24 2.5.1.  Low Miller index Pt single crystals.....................................................................24 2.5.2.  Oxygen reduction reaction...................................................................................26 2.5.2.1.  Mechanisms of oxygen reduction reaction on Pt.................................................26 2.5.2.2.  ORR on rough Pt surface.....................................................................................30 2.5.3.  CO oxidation reaction..........................................................................................31 2.5.3.1.  Mechanisms of CO oxidation on Pt.....................................................................31 2.5.3.2.  CO oxidation on low Miller index Pt surfaces.....................................................32 2.5.3.3.  CO oxidation on a rough Pt surface.....................................................................35 2.5.3.4.  Adsorbate-induced surface restructuring in the Pt surface...................................35 2.6.  Pt-based alloys...............................................................................................................39 2.6.1.  Fuel cell reactions on Pt bimetallic alloy catalysts..............................................39 2.6.1.1.  ORR on Pt bimetallic alloy catalyst.....................................................................40 2.6.1.2.  Pt skin formation.................................................................................................41 2.6.1.3.  CO-oxidation on Pt-alloy surfaces.......................................................................42 2.6.1.4.  Stability of Pt-based alloys under a fuel cell operation conditions.......................43 iv 2.6.1.5.  Adsorbate induced surface segregation................................................................44 2.7.  PtZn...............................................................................................................................45 2.7.1.  Pt, Zn, and bulk PtZn “alloy”..............................................................................45 2.7.2.  PtZn alloy formation............................................................................................47 2.7.2.1.  Zn underpotential deposition on Pt......................................................................47 2.7.2.2.  Transformation of Zn UPD into Pt-Zn alloy........................................................48 2.7.2.3.  Stability of PtZn – Theoretical point of view.......................................................50 2.7.2.4.  Pt alloying effects................................................................................................50 2.7.2.5.  Comparison of PtZn to other alloys.....................................................................52 2.7.2.5.1.  Pt-transition metal alloys...........................................................................52 2.7.2.5.2.  RuZn..........................................................................................................52 2.8.  Summary.......................................................................................................................53 3.  Materials And Methods.........................................................................................................55 3.1.  Introduction...................................................................................................................55 3.2.  Electrochemistry............................................................................................................55 3.2.1.  Rotating disc electrodes.......................................................................................56 3.2.2.  Kinetic and mass transport-limited current relationship for RDE.......................58 3.2.3.  Cyclic voltammetry..............................................................................................59 3.2.4.  Anodic stripping voltammetry.............................................................................59 3.2.5.  Differential pulse voltammetry............................................................................60 3.2.6.  Electrochemistry – experimental consensus........................................................62 v 3.2.7.  PtZn alloy formation............................................................................................66 3.2.7.1.  Zn deposition and PtZn alloy formation..............................................................67 3.3.  Analytical instrumentation............................................................................................68 3.3.1.  Auger electron spectroscopy................................................................................68 3.3.1.1.  Operation principles.............................................................................................68 3.3.1.2.  Nomenclature of Auger transition........................................................................72 3.3.2.  Secondary (Scanning) electron microscopy .......................................................73 3.3.2.1.  Scanning electron microscopy (SEM).................................................................73 3.3.2.2.  Backscattered electron (BSE) imaging................................................................73 3.3.3.  Energy dispersive X-ray spectroscopy ................................................................74 3.3.4.  Inductively coupled plasma – Mass spectrometry...............................................75 3.4.  Single fuel cell...............................................................................................................76 3.4.1.  MEA fabrication..................................................................................................76 3.5.  Summary.......................................................................................................................81 4.  Oxygen Reduction Reactions on Pt and PtZn.....................................................................82 4.1.  Introduction...................................................................................................................82 4.2.  Experimental section.....................................................................................................84 4.2.1.  General electrochemistry and the PtZn alloy preparation...................................84 4.2.2.  Oxygen reduction reaction and the data processing............................................84 4.2.3.  Auger electron spectroscopy on Zn modified electrode .....................................85 vi 4.2.4.  Membrane electrode assembly preparation and testing.......................................85 4.3.  Results and discussion...................................................................................................86 4.3.1.  Formation of surface PtZn alloy characterized by AES......................................87 4.3.2.  Preparation of the Zn-modified Pt RDE. ............................................................96 4.3.2.1.  Ar saturated CVs of Pt and PtZn..........................................................................98 4.3.2.2.  ORR kinetics for Pt and Zn-modified Pt electrodes.............................................99 4.3.3.  Zn modification of a gas diffusion electrode ....................................................103 4.3.4.  Effect on the electronic nature of the surface due to alloying Zn with Pt.........107 4.4.  Summary and conclusions...........................................................................................110 5.  CO Oxidation Catalysis, Surface Roughness And Restructuring on PtZn.....................111 5.1.  Introduction..................................................................................................................111 5.2.  Experimental ...............................................................................................................112 5.2.1.  Electrochemistry and PtZn alloy formation.......................................................112 5.2.2.  CO stripping voltammetry.................................................................................113 5.2.3.  Oxygen reduction reaction ................................................................................114 5.2.4.  Pt electrodeposition............................................................................................114 5.3.  Results and discussion.................................................................................................115 5.3.1.  The nature of the PtZn surface after electrodeposition......................................115 5.3.2.  Ar saturated cyclic voltammograms of Pt and PtZn..........................................117 vii 5.3.3.  CO oxidation and Pt enrichment of the alloy surface........................................120 5.3.4.  ORR and Zn enrichment of the alloy surface ...................................................126 5.3.5.  Surface reorganization and the CO oxidation reaction mechanisms.................132 5.4.  Summary and conclusions...........................................................................................133 6.  Stability of PtZn Alloy in Acidic Aqueous Environment..................................................134 6.1.  Introduction.................................................................................................................134 6.2.  Experimental method:.................................................................................................135 6.2.1.  Preliminary test 1 – Gravimetric determination of Zn loss...............................135 6.2.1.1.  Preparation of PtZn foil.....................................................................................136 6.2.1.2.  Stability test.......................................................................................................136 6.2.2.  Preliminary test 2 – Determination of Zn loss by anodic stripping method......137 6.2.2.1.  Anodic Stripping- differential pulse voltammetry..............................................137 6.2.3.  PtZn nanoparticle stability test:.........................................................................138 6.2.3.1.  Pt particle deposition.........................................................................................138 6.2.3.2.  PtZn particle preparation...................................................................................138 6.2.3.3.  PtZn stability test ..............................................................................................140 6.2.3.3.1.  Polarization of PtZn particles...................................................................140 6.2.3.3.2.  Auger electron spectroscopy ...................................................................140 6.2.3.3.3.   Inductively coupled plasma – mass spectroscopy measurement.............140 6.3.  Results and discussion:................................................................................................141 6.3.1.  Preliminary test 1 – Determination of Zn loss by gravimetry...........................141 6.3.2.  Preliminary test 2 – Determination of Zn loss by anodic stripping voltammetry viii ......................................................................................................................................142 6.3.3.  Pt particle deposition.........................................................................................143 6.3.3.1.  Zn deposition on the Pt particles........................................................................146 6.3.3.2.  Zn-modified Pt particles....................................................................................148 6.3.3.3.  Polarization of PtZn particles.............................................................................149 6.4.  Summary and conclusions...........................................................................................159 7.  Conclusions And Future Work...........................................................................................160 7.1.  Summary of the thesis.................................................................................................160 7.2.  Future work.................................................................................................................163 7.2.1.  Hydrogen peroxide production on the PtZn surface and the reaction mechanisms ......................................................................................................................................163 7.2.2.  Density functional theory calculations..............................................................167 7.2.3.  Alternative PtZn alloy formation – A bulk alloy or alloy particles....................168 7.2.4.  Application of PtZn in the PEMFC....................................................................170 7.2.5.  In-situ Neutron reflectometry ...........................................................................170 8.  References.............................................................................................................................172 ix List of Tables Table 1-1  Types of fuel cells and their characteristics...................................................................3 Table 2-1 Auger notations. Different electronic states are expressed in terms of their quantum numbers, n, l, and j [126]...........................................................................................72 Table  5-1  Electrochemical  surface  areas  and  the  charge  densities  of  CO  stripping voltammograms on a smooth Pt electrode and a CO-treated PtZn electrode (over 4 days).  As PtZn electrode was treated with CO, the electrochemical surface areas decreased and the charge densities increased. Reprinted from [8]. Copyright 2009 with permission from Elsevier Ltd...........................................................................125 Table 6-1 Stability of the PtZn alloy at various fixed potential in a 0.1 M H2SO4 solution for 24 hours determined by gravimetry...............................................................................142 Table 6-2 Relative Zn loss from an initial concentration on the nanoparticles calculated from ICP-MS measurements (Zn:Pt was 0.653 :1 before polarization)...........................155 Table 6-3 Pt-Pt and M-M bond lengths  (M denotes the transition metal) in the pure forms or in the Pt alloys; percent change in bond lengths before and after alloying. The values for Pt, PtCr, PtCo, and PtNi were taken from [70], whereas the ones for PtZn were taken from [101]. PtCr, PtCo, and PtNi show approximately 3 % decrease in Pt-Pt bond distance after alloying, although the Pt-Pt bond distance in PtZn increased by 1.4 %. On contrary, M-M bond distances in PtCr, PtNi, and PtZn increased. This may be because only transition metal electrons contribute to the bimetallic bonding. Pt-M bond distance in PtZn is considerably shorter in PtZn than Pt-Pt or Zn-Zn. Pt appears to contribute to the bimetallic bonding in PtZn, which may be important in designing a stable catalyst in fuel cells....................................................................156 x List of Figures Figure 1-1 The diagram of a PEM fuel cell with hydrogen as the  fuel. Hydrogen molecules are oxidized at the anode and oxygen molecules are reduced to water at the cathode. Electrons go through the external circuit to generate electricity. The anode, the solid electrolyte, and the cathode layers are the parts of a membrane electrode assembly (MEA)..........................................................................................................................4 Figure 2-1 Structure of Nafion™. Fluorinated carbon chains with sulfonic acid terminals. When Nafion is hydrated, sulfonic acid becomes ionized, and the protons will be involved in the ion exchange process [2]....................................................................................9 Figure 2-2 The double layer structure based on Gouy-Chapman-Stern model. The double layer capacitance, Cd, is considered as two capacitors in series, a capacitor at the inner layer and the diffuse layer capacitor; CH and CD, respectively. The figure shows the potential drop across the double layer, calculated for a 0.01 M 1:1 electrolyte at 298 K, assuming a potential of 100 mV at the Outer Helmholtz Plane. The compact layer is an alternative name for the inner layer.  Reproduced with permission from [9]. Copyright 2001 John Wiley & Sons Ltd....................................................................12 Figure 2-3 Double layer structure of metal/solution interface. The line at x1 is inner-Helmholtz plane (IHP) and the line at x2 is the outer-Helmholtz plane (OHP). Only water and specifically adsorbed ions are found at IHP, and ions can only reach to the OHP in solvated form. Reproduced in part with permission from [9]. Copyright 2001 John Wiley & Sons, Ltd......................................................................................................13 Figure  2-4  Electron  density  at  a  metal  surface.  Electrons  “spill  over”  at  the  metal  surface showing the deficit in the electron density below the Fermi level, creating the net surface  dipole.  Distance  is  shown in  Fermi  wavelength,  which  is  direction-  and element-dependent but is  approximately 0.5 nm. Reprinted with permission from [10].  Copyright  1970  by  American  Physical  Society. http://prola.aps.org/abstract/PRB/v1/i12/p4555_1.....................................................15 Figure 2-5 A polarization curve showing the three mechanisms of overpotential observed in a fuel cell. The integral of the curve is essentially the power output from the fuel cell. The theoretical potential of the cell is calculated by the Nernst equation, and the potential is ideally constant regardless the current flow. However, the potential is lost in  overcoming  the  reaction  irreversibility  (activation  polarization),  the  internal electrical resistance (ohmic polarization), and the reactant mass-transfer limitation to the catalyst surface (concentration polarization). ......................................................18 xi Figure 2-6 The transition state of the forward and backward reactions. ∆G0‡ shows the standard Gibbs energy of the reaction at the equilibrium potential. When a potential is applied externally,  the  potential  curve  shifts,  as  shown  in  the  top  figure.  The  transfer coefficients,  α for the forward reaction or (1-α) for the reverse reaction, describes the contribution to the net current from either side of the reaction. Reproduced with permission from [9]. Copyright 2001 John Wiley & Sons Ltd..................................21 Figure 2-7 Energy diagram showing the dissociative chemisorption process in heterogeneous catalysis. The adsorption energy of A onto a site on a substrate is lowered to ∆Eact by the presence of catalyst. Adapted from [14]...............................................................22 Figure 2-8 Face centrerd cube crystal (e.g., Pt) cut in three low Miller index planes. Pt (111) is the most densely packed, followed by Pt (100) and Pt (110).....................................25 Figure 2-9 Possible reaction pathways for ORR. Final products are either water or hydrogen peroxide.  Reprinted  from  [21].  Copyright  1987  with  permission  from  Elsevier Sequoia S.A................................................................................................................27 Figure 2-10 Three possible binding arrangements of molecular oxygen adsorbed onto a catalyst surface, Me. The end-on arrangement is believed to lead to the series pathway in ORR. The interactions between oxygen molecular orbitals and the Pt d orbitals are shown.  Reprinted  from  [23].  Copyright  1986  with  permission  from  Elsevier Sequoia.......................................................................................................................28 Figure  2-11  Examples  of  different  CO  stripping  voltammograms  recorded  under  identical conditions. a) Voltammogram with multiple peaks; b) A single peak with a shoulder. Edep is the CO preconcentration potential applied to the electrode. The difference is probably due to the electrode surface characteristics which are dynamically changed. The shoulder in b) is possibly due to the oxidation of bridge bonded CO, whereas the main peak is due to the oxidation of linearly bonded CO [53]..................................34 Figure 2-12 Atomic arrangements of Pt (100)-hex and Pt (100)-(1×1) structures. Pt (100)-hex is formed over Pt (100) to minimize the surface energy of the crystal; however, when CO is adsorbed onto the hex structure, surface reconstruction occurs, transforming the overlayer into (1×1) structure...............................................................................37 Figure 2-13 Energy diagram showing the surface reconstruction of Pt(1×1) ↔  Pt- hex structure with  the  CO adsorption.  When  a  clean  Pt  (100)  surface  is  present  without  CO adsorbed,  Pt-hex  is  energetically  more  stable  than  (1×1)  by  “∆E”  and  the transformation  to  Pt-hex  requires  the  activation  energy,  E*.  However,  when CO molecules are adsorbed onto the Pt (1×1) surface, its energy is lowered significantly xii that  the  Pt  (1×1)  exhibits  a  lower energy.  Reprinted  with permission  from [66]. Copyright 1983 American Institute of Physics..........................................................38 Figure 2-14 The volcano plot (theoretically determined) showing the ORR activity with respect to  the  oxygen  binding  energy.  The  ORR  activity  corresponds  to  the  energy  of activation barrier of the rate determining step calculated. The closer the element is to the summit of the plot, the more efficient the ORR catalysis is. Reproduced in part with permission from [35]. Copyright 2004 American Chemical Society.................41 Figure 2-15 PtZn phase diagram. Reproduced in part from [102]. Copyright 1991 with kind permission of Springer Science and Business Media................................................46 Figure 2-16 The cyclic voltammograms recorded by Despic  et al.,  in 1.5 M ZnSO4 solution (pH=3.5) on a Pt electrode. Peak potential region assignments are as follows: Peak I – Pt oxide reduction; Region II – The Zn UPD formation on the Pt; Region III – Post-UPD  potential  region  reflecting  the  Pt-Zn  alloy  formation;  Peak  I'  –  Zn stripping  from the  bulk  Zn;  Peak II'  –  UPD Zn stripping;  Peak III'  –  hydrogen desorption; Peak IV' – Oxidation of PtZn alloy; Peak V' – Zn stripping from the PtZn  alloy.  The  PtZn  alloy  was  formed  when  the  cathodic  potential  limit  was extended  in  Region  III  while  the  CVs  were  recorded,  indicating  a  rapid transformation  of  the  UPD  Zn.  Reprinted  from  [106].  Copyright  1982  with permission from Elsevier (Pergamon Press Ltd).......................................................49 Figure 3-1 Schematic view of a rotating disk electrode (bottom view on the right and the cross section on the left). A disk of metal (e.g., Pt) is connected to a stainless steel shaft and shielded in a Teflon or plastic insulating material. The RDEs employed in this thesis had the Pt disc diameter of 3 mm (total diameter of the electrode is 10 mm). 57 Figure 3-2 Hydrodynamics at the surface of an RDE. Reprinted in part with permission from [122]. Copyright 1971 Oxford University Press........................................................58 Figure 3-3 Potential and current profiles with respect to time in anodic stripping voltammetry. Analyte in a solution is first preconcentrated at a deposition potential, followed by stripping (oxidation) of the deposited film by scanning the potential in the anodic direction,  resulting in a peak in a voltammogram. The peak potential,  Ep can be calculated by the Nernst equation, and the current,  ip is measured. Because of the preconcentration step, the sensitivity of detection is enhanced, making this method suitable for trace metal analysis (< 1 µM).................................................................61 Figure 3-4 Potential profile with respect to the time in differential pulse method. Potential is applied  to  the  electrode  for  a  certain  period  (“interval  time”;  typically  0.5~4 xiii seconds), a potential step (“modulation amplitude”; typically 10~100 mV) is applied to  the  electrode  and  fixed  at  the  stepped  potential  for  some  period  of  time (“modulation time”; typically 5-100 milliseconds). Current is measured just before the  step and just  before the  potential  is  stepped back to  the  base  (at  S1 and  S2, respectively), and the difference is plotted as a function of potential. ......................61 Figure  3-5  Home-made  reversible  hydrogen  electrode  (RHE).  A small  compartment  at  the bottom of the electrode contains a coiled Pt wire which is connected to the extended tungsten wire. The compartment is filled with an acid solution from the opening at the tip of the electrode. Hydrogen gas is evolved in the compartment, resulting in the Pt wire in equilibrium with hydrogen gas and protons in solution. This acts as a hydrogen electrode.....................................................................................................64 Figure 3-6 Schematic view of electrochemical cell. The counter electrode (CE) was a coiled Pt wire and the reference electrode (RE) was typically the RHE. The working electrode (WE)  and  other  electrodes  were  connected  to  potentiostat  which  was  computer controlled...................................................................................................................65 Figure 3-7 A cyclic voltammogram of polycrystalline Pt recorded in 0.1 M H2SO4 deaerated solution at the scan rate of 50 mV/s is presented. Integration of hydrogen adsorption or desorption peaks give approximately 210 µC/cm2 of charge density. This is used as a benchmark CV recorded in a clean H2SO4 solution............................................66 Figure  3-8  Distribution  of  electrons  in  various  analytical  techniques.  Auger  electrons, backscattered electrons, X-ray (fluorescence in EDX) were analyzed in this thesis. Adapted  from [124]. Incident electrons (beam diameter of ~ 10 nm) impinge on the solid surface and penetrate causing various types of excitation of the atoms in the sample. Typical path lengths for the excitation phenomena are shown in the figure. ....................................................................................................................................70 Figure 3-9 Auger electron transition. Incident beam hits the core electron in Ew in (1 and 2) followed by the relaxation of an electron in the upper orbital, Ex (3). This relaxation provides enough energy to the electron in EY to be ejected as an Auger electron. Φe is the  work  function  and  EWXY is  the  energy of  Auger  electron,  which  is  typically 50~3000 eV................................................................................................................71 Figure 3-10 The components of MEA are shown. Solid electrolyte is sandwiched between two catalyst  layers that  are  attached to  GDLs. These units  are hot pressed to  form a MEA.  It  is  further  sandwiched  by  field  flow  plates  which  are  responsible  for delivering reactant gases to the catalyst layers..........................................................78 xiv Figure 3-11 Cross section of solid electrolyte, catalyst layer, and GDL hot pressed together with Nafion™  bonding  (“proton  conducting  media”  in  the  figure).  Better  ionic conduction  and  electron  transfer  are  achieved  by  having  the  conducting  media supporting the catalyst layer. Cross section of solid electrolyte, catalyst layer, and GDL hot pressed together with Nafion™ bonding (“proton conducting media” in the figure).  Better  ionic conduction and electron transfer are achieved by having the conducting media  supporting  the  catalyst  layer.  Reprinted from [133].  Copyright 2004 with permission from Elsevier Ltd....................................................................79 Figure 3-12 Electrochemical setup to perform Zn modification onto a catalyst-attached GDL. GDL was  connected  to  potentiostat  and  immersed  into  a  ZnSO4 solution.  The counter  electrode  was  stainless  steel  mesh  and  the  reference  electrode  was Hg/Hg2SO4. ...............................................................................................................80 Figure  4-1 SEM micrograph after  Ar+ etching  of  a  thick  electrodeposited Zn layer  on a  Pt electrode. Three regions are indicated on the SEM which correspond to Pt (I), Zn (II), and Zn/Pt alloy (III). Reproduced with permission from [7]. Copyright 2006 American Chemical Society.......................................................................................88 Figure 4-2 Higher magnification SEM images of (a) region I (Pt), (b) region II (Zn), and (c) region III (Zn/Pt alloy) identified in Figure 4-1. Reproduced with permission from [7]. Copyright 2006 American Chemical Society......................................................89 Figure 4-3 Auger electron survey spectra for the regions I (Pt), II (Zn), and III (Zn/Pt alloy) (bottom,  middle,  top,  respectively)  identified  in  Figure  4-1.  Reproduced  with permission from [7]. Copyright 2006 American Chemical Society...........................90 Figure 4-4 High-resolution Auger electron spectra of (a) Pt from regions I and III and (b) Zn from regions II and III. Pt Auger transition peak heights at 1960.5 and at 2040.8 eV from region I give an area ratio of 1.86, whereas region III has a peak area ratio of 1.14. The metallic Zn Auger transition spectrum shows a shoulder by the peak at 980.6 eV but the shoulder diminished when Pt was modified by Zn. Reproduced with permission from [7]. Copyright 2006 American Chemical Society...................93 Figure 4-5 Auger electron survey spectra of regions I (Pt), II (Zn), and III (PtZn) after leaving in a vacuum (2.5 × 10-9 mbar) for (a) 2 and (b) 10 h. After 2 hours, the amount of oxygen  on  the  PtZn  surface  was  greater  than  the  amount  on  the  Pt  surface. Reproduced with permission from [7]. Copyright 2006 American Chemical Society. ....................................................................................................................................94 Figure 4-6 Auger electron spectra of a) metallic Zn and b) PtZn; before and after the surfaces xv interact with oxygen in the UHV chamber. The peaks observed at 500 eV on the spectra  in  the  top  panels  are  the  indication  of  the  presence  of  oxygen.  Auger transition peaks of Zn in 900 ~ 1100 eV region are shown in the bottom. The peak positions shift to lower kinetic energy with the presence of oxygen on the metallic Zn, indicating oxygen is strongly interacting with the Zn surface. On the other hand, the  signals  from the PtZn surface  were not  affected  by the presence  of  oxygen, indicating  very weak or no interaction between Zn in PtZn and oxygen..................95 Figure 4-7 (a) Cyclic voltammetry (100 mV/s, 22 oC) of Pt in 0.1 M ZnSO4 showing deposition and stripping processes. (b) Current transients for the deposition (dashed line) of Zn onto the Pt electrode and the resulting current transient for the stripping (solid line) of the Zn remaining on the Pt electrode after allowing the formation of the Pt/Zn alloy.  Reproduced with permission from [7]. Copyright 2006 American Chemical Society........................................................................................................................97 Figure 4-8 Comparison of Ar saturated CVs recorded on the Pt (dashed line) and the PtZn (solid line) electrodes at 22 oC. Hydrogen adsorption/ desorption peaks are disappeared in the PtZn CV, and the Pt (or PtZn) oxide formation is shifted anodically. Reprinted from [8]. Copyright 2009 with permission from Elsevier Ltd...................................98 Figure 4-9 Comparison of  the ORR activity measured at  2000 rpm (22  oC) with an anodic voltage sweep rate of 10 mV/s for a smooth Pt RDE (solid line) and a Zn-modified Pt RDE (dashed line). Approximately 30 mV improvement in the overpotential with the  PtZn  alloy  compared  to  with  the  Pt.  Reproduced  with  permission  from [7]. Copyright 2006 American Chemical Society...........................................................101 Figure 4-10 Calculated kinetic currents for the ORR on smooth Pt (solid line, o) and on Zn- modified Pt surface (dashed line, ) for an anodic scan (22 oC). The data points were determined from the  Koutecky-Levich analysis.  The  solid  and dashed lines  were mass  transport  corrected  currents  measured  at  2000  rpm  shown  in  Figure  4-9. Reproduced with permission from [7]. Copyright 2006 American Chemical Society. ..................................................................................................................................102 Figure 4-11 (a) Cyclic voltammogram (100 mV/s, 22 oC) of a 1 mg cm-2 Pt GDE in 0.1 M ZnSO4 demonstrating the deposition and oxidation of the deposited Zn (solid line, cathodic limit of -1.6 V/Hg/Hg2SO4; dashed line, cathodic limit of -2.0 V/Hg/Hg2SO4). (b) The deposition and stripping current transients following a step to - 1.6 and -1.0 V vs Hg/Hg2SO4,  respectively.  Reproduced  with  permission  from [7].  Copyright  2006 American Chemical Society.....................................................................................105 Figure 4-12 MEA polarization curves (74 oC) for Zn-modified (●) and unmodified () cathodes: (a)  measured  using  the  same inlet  gas  pressures  for  the  anode and cathode;  (b) xvi measured using the higher cathode inlet gas pressure (350 kPa for the cathode and 101 kPa for the anode). In each case, the inset shows the low current kinetic region of the polarization curve.  Reproduced with permission from [7].  Copyright 2006 American Chemical Society.....................................................................................106 Figure 4-13 ORR mechanisms suggested by Fernandez, et al. The oxygen is first adsorbed onto transition metal sites on the surface, followed by a rapid transferring of the oxygen. The gray circles in lattice represent the transition metals, and the dark circles in (b) represent oxygen atoms. The reaction of ORR is believed to improve. Reprinted in part with permission from [114]. Copyright 2005 American Chemical Society......109 Figure  5-1  The  electrode  surface  area  can  be  determined  by  measuring  the  double  layer capacitance and comparing it to the one with the known area. The electrode area on the smooth Pt was determined by integrating Hads peaks.......................................113 Figure 5-2 SEM images of a) smooth Pt and b) PtZn surfaces. EDX analysis showed both Pt and Zn on the white particles in b), whereas the flat gray part contains only Pt. PtZn particle  densities  are  different  and  seem to  depend  on  the  different  Pt  surfaces. Region i) and ii) have approximately 40% and 80% coverage, respectively, whereas no  particles  were  seen  in  Region  iii).  Reprinted  from [8].  Copyright  2009  with permission from Elsevier Ltd...................................................................................116 Figure 5-3 a) SEM image of the PtZn bead electrode, and b) backscattered SEM of the same region of the electrode. Backscattered SEM shows heavier atoms in darker colors relative to lighter atoms. The white particles in the regular SEM appear in a darker color in the backscattered SEM, indicating the white particles contain Zn. The grain boundaries appear to be decorated with PtZn. Also the middle bright patch in b) shows in backscattered image showed dark “shadow” around the grain boundaries which gradually fades away from the grain boundary. This indicates Zn is migrating from either the grain boundaries or from the neighboring Zn-rich region into Pt.. .118 Figure 5-4 Ar saturated CVs of smooth Pt (dotted line) and PtZn (solid line) recorded at 50 mV/s in Ar saturated 0.1 M H2SO4 at 22 oC. PtZn electrode shows a positive shift in Pt- oxide  formation.  Also  hydrogen  adsorption/desorption  peaks  on  PtZn  are diminished. Reprinted from [8]. Copyright 2009 with permission from Elsevier Ltd. ..................................................................................................................................119 Figure 5-5 a) CO stripping voltammograms recorded at 50 mV/s in CO saturated 0.1 M H2SO4 on a smooth Pt (dotted line) and on a PtZn (solid line) electrodes recorded at 22 oC. A 200 mV negative shift was observed on the PtZn electrode relative to Pt; b) CO stripping  voltammograms  on  Pt  and  PtZn  electrodes  of  comparable  roughnesses (recorded in the same manner as [a]). “Roughness” were calculated by comparing xvii the double layer capacitances measured with Ar saturated CVs to that of smooth Pt. The negative shift in the CO stripping peak observed with PtZn in a) was purely due to roughness. Reprinted from [8]. Copyright 2009 with permission from Elsevier Ltd............................................................................................................................121 Figure 5-6 Ar saturated CVs recorded with a) smooth Pt and b) PtZn. All CVs presented in here were recorded at 50 mV/s in fresh electrolyte (Ar saturated 0.1 M H2SO4 ). On both electrodes, the charge under the CVs decreased as more CO-treatment was applied, suggesting the surface became smoother.; c) CVs recorded (in fresh electrolyte) on PtZn  after  further  CO-treatment  was  applied.  Pt  oxide  formation  was  shifted negatively, and hydrogen adsorption/desorption peaks were enhanced (indicated by arrows), suggesting the surface became Pt-rich after CO-treatment. Reprinted from [8]. Copyright 2009 with permission from Elsevier Ltd..........................................122 Figure 5-7 CO stripping voltammograms recorded at 50 mV/s in CO saturated 0.1 M H2SO4 (22 oC) on PtZn electrode over four days. CO-treatment was applied for approximately 3 hours on the electrode on each day. The dotted line at 0.92 V corresponds to the CO stripping peak on smooth Pt. As more CO-treatment was applied the stripping peaks recorded on PtZn were moving towards the peak potential measured on smooth Pt. Reprinted from [8]. Copyright 2009 with permission from Elsevier Ltd................124 Figure 5-8 ORR curves recorded at 20 mV/s with the rotation speed of 2000 rpm in oxygen saturated  0.1  M  H2SO4 (22  oC)on  smooth  Pt,  rough  Pt,  and  PtZn  electrodes. Roughnesses of PtZn and “rough Pt” are equivalent. The rough Pt showed a 15 mV positive shift, whereas PtZn showed a 35 mV positive shift,  compared to the smooth Pt electrode at current density of -1.0 mA/cm2. Reprinted from [8]. Copyright 2009 with permission from Elsevier Ltd...........................................................................127 Figure 5-9 ORR curves in kinetic-limited regions recorded at 20 mV/s with the rotation speed of 2000  rpm  in  oxygen  saturated  0.1  M  H2SO4 (22  oC)  on  smooth  Pt  and  PtZn electrodes  before  (solid  line)  and  after  (dotted  line)  CO-treatment.  ORR  curve recorded on the smooth Pt surface is  plotted on the PtZn graph for comparison. There is a slight negative shift observed after CO-treatment was applied to the PtZn surface. Reprinted from [8]. Copyright 2009 with permission from Elsevier Ltd...128 Figure 5-10 ORR polarization curves recorded at 20 mV/s with the rotation speed of 2000 rpm in oxygen saturated 0.1 M H2SO4 (22  oC) on PtZn that has been treated with CO (PtZn 1), then with oxygen (PtZn 2), and then retreated with CO (PtZn 3). Negative shift in the onset potential of ORR on PtZn after CO treatment was reproducible. ORR curve recorded on a smooth Pt surface is shown for comparison. Reprinted from [8]. Copyright 2009 with permission from Elsevier Ltd.................................131 xviii Figure 6-1 Home-made working electrode used for a PtZn nanoparticle stability test. A sheet of glassy carbon was machined in a top hat shape with a surface area of 1.23 cm2. The substrate was then accommodated into an electrode holder where the bottom face of the substrate is held tight to an aluminum plate. Teflon o-ring is placed in between the substrate and the electrode holder to ensure a tight seal....................................139 Figure 6-2 Current transient curve showing the deposition of Pt particles onto a glassy carbon substrate.  The  working  solution  was  deaerated  0.1  M  H2SO4 containing  1  mM H2PtCl6.  Nucleation  of  Pt  particle  was  performed  by  applying  the  nucleation potential of 0.0 V for 10 seconds, followed by applying the growth potential of 0.16 V for 10 seconds.......................................................................................................144 Figure 6-3 SEM image of Pt particles dispersed on a glassy carbon substrate. The particles are spherical with diameters typically 50 ~ 100 nm. ....................................................145 Figure 6-4 Zn thin film deposition over the Pt dispersed glassy carbon substrate was performed by sweeping the potential between -0.8 V and -0.63 V 20 times in a deaerated 0.1 M zinc sulfate solution at the sweep rate of 50 mV/s (shown in a dotted line). The CV in a solid line shows a complete oxidation of deposited Zn, whereas partial oxidation of deposited Zn leaves a thin film of Zn over the glassy carbon substrate. This will be useful when a bulk Zn deposition is applied in the following experimental step as hydrogen evolution from Pt particles are suppressed..............................................147 Figure 6-5 Auger electron spectra of Zn (left) and Pt (right) recorded from a single Zn-modified Pt particle after the heat treatment. The average % amount of Zn on the surface on as-prepared particles was 4.2 %...............................................................................149 Figure 6-6 SEM image of PtZn particles on the glassy carbon substrate. Unlike Pt particles that appeared  round and  showed  poor  adhesion,  PtZn  particles  appeared  in  irregular shapes  and adhered to  the substrate  well.  Generally,  Pt  particles grew in size as longer polarization was applied due to sintering, but this was not an issue with PtZn particles....................................................................................................................152 Figure 6-7 The change in surface concentration of Zn with respect to duration of polarization at the  potential  of  0.75 V (filled  circle),  1.0  V (square),  or  1.25  V (empty circle). Polarization was performed in an oxygen saturated 0.1 M sulfuric acid solution at 22 oC.  Zn  concentration  on  the  surface  increases  as  the  duration  increased  for  the samples  polarized  at  0.75 and 1.0  V.  They even reached a  plateau,  but  it  never decreased during the polarization period, indicating the surface became Zn-saturated with or without having more Zn leached out into the solution. On the other hand, the samples polarized at 1.25 V showed a rapid decrease in the surface Zn concentration indicating PtZn is unstable at this potential and Zn immediately leaches out from the xix bulk into the solution. ..............................................................................................153 Figure 6-8 Typical Pt Auger spectra from the PtZn particles polarized at a) 1.25 V for 1 hours, b) 1.0 V for 20 hours, and c) 0.75 V for 14.5 hour. The ratios of Pt MN3 to MN2 peak heights are shown on the right hand side of the spectra. As shown in Chapter 4, 1:1 ratio of those peaks implies the interaction between Pt and Zn. Spectra b) and c) show almost 1:1 peak height ratios whereas the spectrum a) (the sample polarized at 1.25 V) shows an obvious difference in the peak heights.  This indicates that  the samples  polarized  at  0.75  and  1.0  V after  14.5  and 20  hours,  respectively,  still maintain Pt-Zn bimetallic bonds, whereas the sample polarized at 1.25 V for 1 hour does not. ..................................................................................................................158 Figure 7-1 Substrate generation/ tip collection mode in SECM. The microelectrode is connected to piezo-controller to allow rastering of the electrode just above the substrate at a constant height in a controlled manner. The microelectrode is held at the appropriate oxidation potential while that of the substrate is changed. Once oxidizable products are produced at the substrate, they are detected at the microelectrode, generating the oxidative current. Reproduced in part with permission from [ref]. Copyright 2007 Elsevier Ltd. ............................................................................................................166 Figure 7-2 SEM images of the PtZn alloy deposited on a tungsten substrate from the ZnCl2- EMIC ionic solution at a) E=-0.2 V vs. Zn (Pt atomic % = 15.69) ; b)  E=-0.22 V vs. Zn (Pt atomic % = 7.71); and c) E=-0.25 V vs. Zn (Pt atomic % = 5.67). The SEM image b) shows irregular-shaped particles, similar to the PtZn particles on glassy carbon reported in Chapter 6. Reprinted from [177]. Copyright 2004 with permission from  Elsevier Ltd. ..................................................................................................169 xx List of Symbols And Abbreviations A Area Aeject Area on the surface where Auger electrons are ejected a Intercept of a Tafel plot aiσ Activity coefficient of species i in the interface aisol'n Activity coefficient of species i in a bulk solution α The transfer coefficient AES Auger electron spectroscopy AFC Alkaline Fuel Cell B Sensitivity factor b Slope of a Tafel plot bj The number density of nuclei, j CD Diffuse layer capacitance Cd Double layer capacitance CH Inner layer capacitance Co* Concentration of oxidized species, o, in a bulk solution (in mol/L) CR* Concentration of reducible species, R, in a bulk solution (in mol/L) χ The change in enthalpy of adsorption χi The change in adsorption enthalpy of species, i,  with respect to a coverage CE Counter electrode CNS The lattice coordination number of the substrate CO Carbon monoxide COads Adsorbed carbon monoxide COOHads Adsorbed formic acid CV Cyclic voltammogram xxi D0 Diffusion coefficient of species, o d Radius of the microelectrode d0, Pt The interatomic bond distance of Pt d0, Zn The interatomic bond distance of Zn DFT Density functional theory ∆E Energy difference between Pt (1×1) and the hexagonal structure of Pt in the absense of CO ∆Eact The energy of the activation barrier in the presence of a catalyst ∆Hads The enthalpy of adsorption ∆Hoformation The standard enthalpy of formation ∆Hi The enthalpy of adsorption at a certain coverage ∆HS-M The enthalpy of bond formation between the metal and a substrate ∆H0,i The enthalpy of adsorption at zero coverage ∆G0‡ The standard Gibbs free energy of the activation δE Energy difference between Pt (1×1) and the hexagonal structure of Pt in the presence of CO DMFC Direct Methanol Fuel Cell DPV Differential pulse voltammetry E Potential Eactual The potential applied to the system where the reaction produces net current Eanode Potential measured at the anode Ecathode Potential measured at the cathode Ecell Potential of the cell Edep Carbon monoxide deposition potential ENernst Nernst (equilibrium) potential Ep Peak potential EUPD The onset potential of underpotential deposition xxii EW Energy of the core orbital in an atom EWXY Energy of an Auger electron EX Energy of the electron in an upper orbital EY Energy of the electron in a valence orbital Eo Standard potential for the half-reaction E* The activation energy of transformation of Pt-hex to Pt (1×1) EDX Energy dispersive X-ray spectroscopy EMIC 1-ethyl- 3- methelimidazolium chloride e The charge on the electron; 1.602×10−19 C ε Dielectric constant of the solution medium ε0 Permitivity of free space F Faraday's constant; 96485 C/mol f Flux of the incident electron beam f The lattice mismatch Φe The work function of an atom φ2 The electrostatic potential at OHP fcc Face centred cubic GC Glassy carbon GDE Gas diffusion electrode GDL Gas diffusion layer Γi Surface excess concentration of species, i Γmax Maximum surface excess concentration γ Surface tension Hads Hydrogen adsorption Hdes Hydrogen desorption HClO4 Perchloric acid xxiii hcp Hexagonal close pack H2O2 Hydrogen peroxide H2SO4 Sulfuric acid η Overpotential I The intensity contribution to the Auger electrons spectrum i Current ib Current generated by a backward reaction if Current generated by a forward reaction ik Kinetic current il Mass transfer-limiting current inet Net current of a reaction ip Peak current iT The current measured at the SECM tip iT,∞ The current measured at the SECM tip far from the sample surface i0 The exchange current ICP-MS Inductively coupled plasma- mass spectrometry IHP Inner Helmholtz plane j Current density or spin-orbit coupling quantum number j0 Exchange current density KCO Exchange equilibrium constant for carbon monoxide Keq Exchange equilibrium constant KOH Exchange equilibrium constant for hydroxyl or an oxygen containing species kB Boltzmann constant; 1.3806 × 10−23 J/K kCO Rate constant of CO oxidation reaction k0 Standard rate constant KL Koutecky Levich l Orbital quantum number xxiv λ Inelastic mean free path LM2 The Zn Auger electron spectral peak at the kinetic energy of 990 eV m The number of atoms per unit volume of the sample M Transition metal i Electrochemical potential of species, i MCFC Molten Carbon Fuel Cell MEA Membrane Electrode Assembly MN2 The Pt Auger electron spectral peak at the kinetic energy of 2040.8 eV MN3 The Pt Auger electron spectral peak at the kinetic energy of 1960.5 eV N Total number of adsorption sites on the surface n A number of electrons transferred in a redox reaction or the principle quantum number niσ Amount of species, i, in the interface nj Coherent neutron scattering length of species j n0  The population of an ion in the interfacial region ν The kinematic viscosity NR Neutron reflectometry OHads Adsorbed hydroxyl or an oxygen containing substance OHP Outer Helmholtz plane ORR Oxygen Reduction Reaction pCO Carbon monoxide pressure PAFC Phosphoric Acid Fuel Cell PEMFC Proton Exchange Membrane Fuel Cell Pt/C Pt particles supported on carbon Pt-hex  The hexagonal structure of Pt formed on a Pt (100) surface θ Surface coverage xxv θCO Carbon monoxide coverage θT Total surface coverage θUPD The coverage of a UPD layer θ1E1 Difference in the surface energy of the hexagonal Pt structure with or without the presence of carbon monoxide θ2E2 Difference in the surface energy of Pt (1×1) with or without the presence of carbon monoxide R Gas constant; 8.3144 J/K mol r The perpendicular distance from the surface of a rotating disk electrode ρ The scattering-length density RDE Rotating disk electrode RE Reference electrode RHE Reversible hydrogen electrode rpm Rotation per minute RRDE Rotating ring disk electrode σ Cross section of the Auger process S1 The point at which the current is measured before the potential step is applied in the differential pulse mode S2 The point at which the current is measured after the potential step is applied in the differential pulse mode SECM Scanning electrochemical microscopy SEM Scanning (Secondary) electron microscopy SHE Standard hydrogen electrode S-M Substrate - metal SOFC Solid Oxide Fuel Cell T Temperature Ttrans Instrumental constant describing the transmission efficiency of the electron xxvi analyzer TDS Thermal desorption spectrometry UHV Ultra high vacuum UPD Underpotential deposition WE Working electrode ω The rotational frequency x2 The distance from the metal to OHP yiσ Activity coefficient of adsorbed species, i, which also accounts for adsorbate- adsorbate interaction in the interface z The parallel distance from the surface of a rotating disk electrode zval Valence charge on an ion xxvii Acknowledgments My graduate  school  career  has  been  supported by many people whom I  just  cannot express my gratitude well enough in words.  First of all, I would like to thank my supervisor, Dr. Dan Bizzotto, for his guidance throughout the project. What you taught me was not only science but also the positive attitude towards facing challenges, both in research and in everyday life. I also wish I had quarter of your patience and scientific talent. My project was a continuation of work by former students: Dr. Ed Guerra and Yanguo Yang. Some contents of this thesis were performed by two undergraduate students, Winton Li (supervised  by  Dr.  Elod  Gyenge,  Department  of  Chemical  Engineering)  and  Olivia  Ye (supervised by Dr.  Dan Bizzotto  and myself,  Department  of  Chemistry).  Without  their  hard work,  there  would  have  been  many  important  questions  I  would  otherwise  be  unable  to understand  in  my  results.  I  also  appreciate  having  Dr.  Philip  Wong  in  UBC  Chemistry Department for his thoughtful inputs and for operating Auger electron spectrometer. Help from technical staff at Department of Chemistry was essential to my research. I am especially thankful to Brian, Des and Raz in mechanical shop not only for machining equipment for my setups but also for keeping me entertained with their conversations. I also thank Brian Ditchburn in glass shop for making electrochemical cells and the glass accessories. I would also like to acknowledge my colleagues in the lab: Dr. Emily Chung, Yanguo Yang, Dr. Jeff Shepherd, Dr. Robin Stoodley, Jannú Casanova Moreno, and Amanda Musgrove (Richer);  for  their  constant  help,  friendship,  useful  discussion,  and  entertainment.  I  would especially like to thank Robin for his friendship. It was pleasure to work with you, and I wish I had your patience for problem-solving. My friends  have  always  provided  me  comfort  and  a  little  break  from non-working experiments. I would especially like to acknowledge Ms. Chika Watanabe and Dr. Tak Ikeda for their constant mental support and for sharing their thoughts with me. Finally, I would like to send my special thanks to Jannú for his constant encouragement. Gracias por estar conmigo. Disfruto tu compañía y te quiero. xxviii 1.   Introduction Alternative energy resources that are free from the limited supply and the emission of toxic or greenhouse gases are in high demand in response to the current oil and climate crises. Whether it would be solar power, wind power, or hydro power, it does not necessarily replace fossil fuels or coal instantly, but certainly need to be available as the equivalent options to fossil fuels, and the community needs to adapt to more environmentally benign options quickly. One of such alternatives is a fuel cell, which is an electrochemical power supply device which converts chemical energy directly into an electrical energy at potentially high efficiency. There are several kinds of fuel cells, as shown in Table 1-1, and depending on their operating characteristics (i.e., operation temperature, types of fuels  etc), fuel cells can be used in a wide range of applications, including small portable devices, an automobile, or in a power plant. In this thesis, the focus will be on proton exchange membrane fuel cells (PEMFCs), which utilize air  (oxygen)  as  an oxidant  and hydrogen or  a low-molecular  weight  alcohol  as  a  fuel.  The diagram of  a  PEMFC is  shown in  Figure  1-1.  Energy conversion occurs  in  the  membrane electrode assembly (MEA), which consists of an acidic solid electrolyte membrane separating a cathode on one side and an anode on the other side. At the anode, hydrogen gas or an alcohol, such as methanol, is oxidized to form protons: H2  ⇄ 2H+ + 2e-       Eo = 0.0 V vs. SHE (1-1) H2O  + CH3OH  ⇄ CO2 + 6H+ + 6e-      Eo = 0.04 V vs. SHE (1-2) Electrons go through the external circuit as an electrical current available for generating useful work, and protons migrate to the cathode through the solid electrolyte. Oxygen is converted to water at the cathode with these protons: O2 + 4H+ + 4e- ⇄ 2H2O      Eo = 1.23 V vs. SHE (1-3) 1 This reaction is called an oxygen reduction reaction (ORR). The power output of fuel cells is calculated by multiplying the potential difference between the cathode and anode (Ecell) at the corresponding current, i: Power = Ecell × i (1-4) The  practical  energy conversion  efficiency of  fuel  cells  are  much  higher  than  that  of  heat engines as they operate at a much lower temperature, and the heat rejection process is much less severe. In theory, the efficiencies of a PEMFC could be as high as 94.5 % [1]. In reality, fuel cells do lose energy to heat and the practical efficiency becomes ~65 % which is still  more efficient than e.g., gas and steam turbines whose conversion efficiency is ~33% [2]. Although  the  MEA may  be  considered  the  heart  of  fuel  cells,  there  are  numerous components  or  conditions  that  affect  the  performance  of  fuel  cells.  For  example,  the  solid electrolyte must be fully humidified in order to carry protons from the anode to the cathode. However, water formed at the cathode also needs to exit the system to avoid flooding. Plates are placed on each side of MEA, called  flow field plates. They are responsible for exhausting the water, among other tasks, such as distribution of reactants into MEAs, carry current away from cells, and cool down MEAs [2]. Manufacturing reliable flow field plates is challenging in order to  satisfy all  demands.  Failure to do one or more of the tasks may drastically decrease the performance of the cell, and one must know the physical and chemical properties of materials used, the method of fabrication, and optimum operational conditions for a fuel cell to satisfy the energy  demand.  Therefore  the  fuel  cell  development  requires  expertise  not  only  in electrochmistry,  but also in material  science,  polymer science,  mechanical engineering,  fluid dynamics,  and computational modeling,  just  to name a few. Fuel production is also another challenging research.  Although the focus of this  thesis  is  in the field  of electrochemistry,  it should be emphasized that fuel cell research is very interdisciplinary, and the contribution from all fields of science is the key to the prompt development of commercially viable fuel cells [2]. 2 Table 1-1  Types of fuel cells and their characteristics [3] Types of fuel cells Ions exchanged in electrolyte Operating temperature (oC) Chemical – electrical conversion efficiency (%) Applications Alkaline fuel cell (AFC) OH- 60-80 40-60 Automobiles, space shuttles Phosphoric acid fuel cell (PAFC) H+ 200 55 Generation of distributed power Proton exchange membrane fuel cell (PEMFC) H+ 80 45-60 Portable devices, automobiles Molten carbon fuel cell (MCFC) CO32- 600 60-65 Distributed power generation Solid oxide fuel cell (SOFC) O2- 1000 55-65 Base load power generation Direct methanol fuel cell (DMFC) H+ 80 35 Portable devices, automobiles Heat engines 1000 ~33 Steam and gas turbines 3 Figure 1-1 The diagram of a PEM fuel cell with hydrogen as the  fuel. Hydrogen molecules are oxidized at the anode and oxygen molecules are reduced to water at the cathode.  Electrons go through the external circuit  to generate electricity.  The anode,  the solid electrolyte,  and the cathode layers are the parts  of a  membrane electrode assembly (MEA). 1.1.  Electrochemical reactions at the electrode Reactions  producing  electricity  in  low  temperature  MEAs  are  kinetically  slow  and require the presence of catalysts. Currently,  the most common catalyst  is Pt.  The amount of precious metals used in fuel cells has been drastically reduced over years, but such precious metals are expensive (currently 992 USD/oz or 15,872 USD/lb) [4] and, more problematically, rare. Dependence on precious metals alone as fuel cell catalysts will eventually lead to a similar catastrophe as  fossil fuels are currently causing. Learning from the current situation, it is wise to look for alternative catalysts while the utilization of fuel cells is still low. Another challenge with 4 H2 O2 e- e-e- + + H2O Solid electrolyte Catalyst CathodeAnode the fuel cell reactions is that the mechanisms of the reactions are not fully understood despite the decades of research effort. Knowledge of the active sites would be very useful when developing a catalyst, and the mechanistic research is actively pursued around the world.  An alternative catalyst to Pt is a Pt-based alloy using  the first row of transition metals. They have actually been shown to improve the reaction rates significantly. At the same time, however, they are not very stable under acidic conditions. Also, those alloys are often Pt-rich in composition and therefore still requires a substantial amount of Pt. Stable Pt-alloys that maintain an improved catalytic activity over time is a key technological step towards fuel cell distribution for practical applications. In this work, we examined a PtZn alloy as a possible candidate for an electrocatalyst for fuel cell reactions. There are three themes to the thesis: 1) electrochemical formation of PtZn and comparing its  catalytic  activity to  that  of pure Pt;  2)  changes  in surface characteristics before and after performing the reactions on the catalysts using electron spectroscopy; and 3) investigating the stability of PtZn under fuel cell operational conditions. The project has been initially started by previous students: Dr. Eduard Guerra first studied the Zn electrodeposition using Pt substrates in 2003. Together with Dr. Guerra, Mr. Yanguo Yang found that the PtZn alloy may be a possible candidate for ORR. Back then, and even now, there are numerous Pt- based alloys that showed an improved catalytic activity for fuel cell reactions, but PtZn had never been considered for the use in fuel cells, except the one reported by He et. al.,  in 2007 [5] probably because of the susceptible nature of metallic Zn in an acidic condition. Our results, however, did not seem to show a significant issue with the stability of Zn in the PtZn alloy, which makes the work described in this thesis unique and potentially significant when compared to the other Pt-based alloys in literature or those already in use. Usage of Zn as an alloying metal is also advantageous over other popular alloying metal,  such as Ni, because Zn is relatively inexpensive and abundantly available on the earth. The market price on Zn is 0.753 USD/lb; more than 9 times cheaper than Ni; 6.964 USD/lb [6]. Replacing even 30% of Pt by Zn could lower the cost for catalysts dramatically. The  second  theme,  surface  characterizations  of  the  catalyst,  is  important  for understanding the mechanisms of fuel cell reactions. Because catalysis happens at the surface (i.e.,  metal/solution  interface),  and  the  characteristics  of  the  metal  surface  are  often  quite 5 different from the bulk, tracking the state of the surface becomes crucial to understanding the reaction pathways. We were able to trace the changes in the surface of PtZn by electrochemistry and with further support using ex-situ electron spectroscopy. The  third  theme,  the  stability  of  the  PtZn  alloy,  is  crucial  in  a  practical  fuel  cell application. One of the challenges researchers have faced when using Pt-based alloys in a fuel cell is the leaching of alloying metal into the solid electrolyte. One obvious reason this is not favorable is because protons migrate from the anode to the cathode by an ion exchange process through the solid electrolyte. The leached metal ions will occupy the sites necessary for proton movement  which will  decrease the performance of the fuel  cells.  Stability of these Pt-alloy catalysts is therefore extremely important in the fuel cell development. A stability examination method that is unique to the current literature has been proposed and studied. Combining the three  themes  and the  results  of  the  research  presented  in  this  thesis,  PtZn appears  to  be  a reasonable candidate for use in a PEMFC. 1.2.  Outline of the thesis Chapter 2 provides a literature review of the current understanding of fuel cell reaction catalysis on Pt and Pt-based alloys. It also reviews the previous experimental and theoretical studies on a PtZn alloy,  which suggests  that  it  may be a promising candidate as a fuel cell catalyst. Chapter 3 describes experimental techniques used in this research. The work presented in this thesis employed some basic electrochemical techniques, as well as analytical methods for characterizing the chemical compositions of a sample surface or sample solutions. Chapter  4  describes  the  electrochemical  formation  of  PtZn  alloy  and  its  surface characterization  with  Auger  electron  spectroscopy  (AES).  We  then  compare  the  catalytic activities of Pt and PtZn for the ORR in an electrochemical setup. Kinetic data was deduced from the measurements. Results presented in this chapter show that PtZn is an effective catalyst candidate for ORR, one of the most sluggish fuel cell reactions. We also employed our PtZn in a single stack fuel cell and compared the performance to a Pt fuel cell. The PtZn alloy was also 6 prepared electrochemically directly onto a catalyst coated gas diffusion electrode. It was then pressed with the membrane and an anode to make a fuel cell.   The performance was again compared to Pt fuel cell tested under identical conditions. The results obtained were essentially the same as the one we obtained in the first section of this chapter, i.e., the PtZn fuel cell showed an improved performance compared to the Pt fuel cell. The bulk of the work described in this chapter has been published in Journal of Physical Chemistry B [7]. Chapter 5 reports on another catalytic reaction, carbon monoxide (CO) oxidation which is the rate determining step for the methanol oxidation, on Pt and PtZn. The surface roughness effects on the catalytic activities of fuel cell reactions were also examined. Although we were not able to prove PtZn improves the activity of the reaction, we observed changes in surface compositions of the metal alloy.  Depending on the molecules that are adsorbed on the alloy surface, the surface became enriched with one of the alloy components. The surface also became smoother  as  CO was  adsorbed  onto  the  metal  surfaces,  and  these  observations  were  made through  simple  voltammetry.  The  work  presented  in  this  chapter  was  published  in Electrochimica Acta [8]. Chapter 6 reports a new methodology to test the chemical stability of alloy metals and the results thus obtained. Previous to the new method, we had attempted two stability tests of PtZn in different methods, and the results are also presented. In the new method proposed, PtZn nanoparticles were first formed on glassy carbon substrates.  A fixed potential was applied to the substrate  under  a  condition  similar  to  an  operating  fuel  cell.  The  particle  surface  was then analyzed  using  ex-situ  AES.  The  particle  bulk  composition  were  further  analyzed  using inductively-coupled plasma – mass spectrometry. This method shows the surface composition of the PtZn alloy in relation to its bulk composition. Finally, the summary and conclusions are given in Chapter 7 reviewing the results given in different sections of the thesis. Suggestions for the future work are also presented. 7 2.  Literature Review 2.1.  Introduction This chapter has two main themes: The first is to review the current understanding of PEM fuel cell reaction catalysis and background theories of electrochemistry and heterogeneous catalysis. Two reactions are considered: Oxygen reduction reaction (ORR) and CO oxidation. The mechanisms of these reactions on a Pt surface on Pt-based alloys are summarized.  The second theme is  to  review the  characteristics  of  PtZn  alloy.  The  formation  of  this  alloy is facilitated by so-called an underpotential deposition (UPD) layer of Zn over the Pt surface. This alloy also exhibits characteristics distinct from its components. Alloying effects on Pt or Zn in PtZn are also compared to other alloys. 2.2.  Proton exchange membrane fuel cell 2.2.1.  Membrane electrode assembly Proton exchange membrane fuel cells (PEMFCs) used in practical applications consist of a number of single stack cells connected in series. Typically passenger vehicles require ~100 kW power output and the power density of fuel cells are ~2.5 kW/L. A single stack fuel cell consists of  a  proton-conducting  polymer  (or  solid  electrolyte),  sandwiched  by  two  catalyst  layers supported on carbon black. This sandwich unit  is the membrane-electrode assembly (MEA), which essentially acts like an electrochemical cell. The solid electrolyte is an ionomer;  i.e., a polymer  with  a  backbone containing  ionizable  functional  groups  attached at  the  end of  the chains. For example, the most commonly used solid electrolyte is NafionTM whose structure is shown in Figure 2-1 [2]. NafionTM consists of fluorinated carbon backbone with sulfonic acid attached at the end of the chains. When solid membranes are hydrated, the functional group ionizes and the counter ions (protons this case) will be readily replaced by incoming ions of the 8 same charge. On the catalyst layers of the MEA (refer to  Figure 1-1), either the hydrogen gas or a liquid methanol is fed as a fuel on the anode side. A PEMFC consuming the latter fuel is called a direct  methanol  fuel  cell  (DMFC).  Protons  are  produced  with  either  of  the  fuels,  and  they migrate through the electrolyte to the cathode side of the MEA, where the ORR takes place. For this reason, it is essential that the membrane is fully hydrated, and also the incoming gaseous fuel  or  oxidant  is  humidified.  The  most  commonly  employed  catalyst  material  is  Pt nanoparticles  (~3  nm)  dispersed  on  a  carbon  support  for  both  anode  and  cathode  in  low- temperature  PEMFCs  because  they  catalyze  the  reactions  reasonably  well.  The  chemical stability of Pt is benign in rather severe fuel cell operating conditions, and Pt can be recycled. For  these  reasons,  reaction  catalysis  is  extensively studied  on  Pt  as  a  benchmark  for  other catalyst candidates being developed or to help researchers understand the reaction mechanisms. The  studies  of  catalysis  on  a  Pt  surface  will  be  reviewed  after  describing  the  basic thermodynamics and kinetics of the electrode reactions. Figure  2-1 Structure  of  Nafion™.  Fluorinated  carbon  chains  with  sulfonic  acid terminals. When Nafion is hydrated, sulfonic acid becomes ionized, and the protons will be involved in the ion exchange process [2]. 9 2.3.  Electrochemistry Electrochemistry  is  the  study of  the  interactions  of  ions  or  molecules  either  among themselves or with the electrode material (i.e., metal) under an applied electric field. Because fuel  cells  are  electrochemical  devices,  understanding  the  fundamental  theories  of electrochemistry is essential in fuel cell research, especially in electrocatalysis – the focus of this thesis. This section reviews the theories of electrochemistry, covering the structure of the double layer and the theory of electron transfer. The interfacial region is where adsorption of molecules and electron  transfer  take  place,  and  is  best  described  by the  double  layer  theory.  Electron transfer kinetics is explained in relation to the observed current-potential relationship. 2.3.1.  Metal/liquid interface and electrical double layer An interface is formed when two distinct  media come into contact (e.g., liquid/solid, solid/gas). Assuming a metal (electrode) is in contact with a solution containing ionic species, and a potential difference (with respect to a reference or ground) is applied to the electrode, there are potential ranges where no redox reaction occurs at the metal/solution interface. Under such a circumstance,  the electrode is  said to be  ideally polarized and only charging current flows. In this  particular case,  the interface is called an electrical  double layer and acts  as a capacitor. It is important to note that the amount of charge residing on the metal side of the interface is counterbalanced by the amount of charge lining on the solution side of the interface. The structure of the double layer is described by the Gouy-Chapman-Stern model. Gouy-Chapman-Stern model considers the liquid/solid interface as two capacitors in series. The first “capacitor” extends from the metal surface into the solution to an imaginary plane, called the  Helmholtz  plane.  The  second  capacitor  extends  from the  Helmholtz  plane  to  the  bulk solution, and is called the diffuse layer. This is illustrated in Figure 2-2. The capacitance of the double layer or differential capacitance, Cd, of the interface can be described by, 10 1 C d = 1 C H  1 C D (2-1) where CH is the capacitance in the inner layer, and CD is the capacitance of a diffuse layer. The inner layer can further be divided into two planes, known as Inner Helmholtz plane (IHP) and outer  Helmholtz plane (OHP),  illustrated in  Figure 2-3.  At the IHP, either water molecules  lined  up  by pointing  their  dipole  moments  against  the  charged  metal  surface  or specifically adsorbed ions are found. Solvated ions close to the surface of the metal are found at the OHP. Since the charge accumulated in the vicinity to the metal surface lie along the OHP, the metal/solution interface up to the OHP acts as a simple capacitor (which is represented as a CH). Assuming no specific adsorption, the potential across the inner layer decreases linearly as no charged species are found in that  region,  and the potential  in diffuse layer decays as  a sinh function. The differential capacitance can be expressed as 1 C d = x2 0  1 20 zval 2e2 n0 /k B T  1/2cosh  zval e2/ 2 k B T  (2-2) where x2 is the distance from the metal to the OHP, ε is the dielectric constant of the medium, ε0 is permitivity of free space, zval is the valence on the ions in the solution,  e is the charge on electron  (1.602×10-19  C),  n0 is  the  total  number  of  an  ion  in  the  interfacial  region,  kB  is Boltzmann constant, T is the temperature, and φ2 is the electrostatic potential at the OHP. Noting the  first  term corresponds to  CH,  and  the  second to  CD,  this  equation shows that  the CH is independent  of  potential,  whereas  that  of  the  diffuse  layer  is  dependent  on  potential.  Cd is dominated by the smaller of the two effective capacitors in series. 11 Figure 2-2 The double layer structure based on Gouy-Chapman-Stern model. The double layer capacitance, Cd, is considered as two capacitors in series, a capacitor at the inner layer and the diffuse layer capacitor; CH and CD, respectively. The figure shows  the  potential  drop  across  the  double  layer,  calculated  for  a  0.01  M  1:1 electrolyte at 298 K, assuming a potential of 100 mV at the Outer Helmholtz Plane. The  compact  layer  is  an  alternative  name  for  the  inner  layer.  Reproduced  with permission from [9]. Copyright 2001 John Wiley & Sons Ltd. 12 Figure 2-3 Double layer structure of metal/solution interface. The line at x1 is inner- Helmholtz plane (IHP) and the line at x2 is the outer-Helmholtz plane (OHP). Only water and specifically adsorbed ions are found at IHP, and ions can only reach to the OHP in solvated form. Reproduced in part with permission from [9]. Copyright 2001 John Wiley & Sons, Ltd. 13 Specific adsorption of ions (e.g., SO42-  on Pt) may vary the potential gradient within the Helmholtz plane. Adsorbing ions with charge opposite to that on the electrode causes an increase in the gradient, while like charges result in a decrease [9]. 2.3.2.  Metal side of interface: surface dipole The previous sections described the charge distributions in the solution side of the metal/ solution interface. The metal side of interface is the focus next. The electron density profile with respect to distance from the metal surface is shown in Figure 2-4 [10]. The electron density close to the surface (i.e., distance = 0) is not confined to the surface, but is spread out into vacuum. This redistribution of charge over the surface is called surface dipole and assists to lower the surface energy of the metal. It also lowers the metal's work function. Work function is the minimum energy required to remove an electron from the Fermi level into vacuum [11]. The extent of the surface dipole increases with temperature and also with the roughness of the metal surface. Because of the electron spillage, the average work function on a rough surface is smaller than on a smoother surface [11]. 14 Figure 2-4 Electron density at a metal surface. Electrons “spill over” at the metal surface showing the deficit in the electron density below the Fermi level, creating the  net  surface  dipole.  Distance  is  shown  in  Fermi  wavelength,  which  is direction- and element-dependent but is approximately 0.5 nm. Reprinted with permission  from  [10].  Copyright  1970  by  American  Physical  Society. http://prola.aps.org/abstract/PRB/v1/i12/p4555_1 15 2.4.  Thermodynamics and kinetics of electrode reactions Thermodynamics and kinetics of electrode reactions, whether in an electrochemical cell or in a fuel cell, can be treated similarly. This section reviews basic electrochemistry relating potential (or “overpotential”) and current. 2.4.1.  Fuel cell power output Power output of fuel cells is calculated as, Power = Ecell × i (2-3) where Ecell is defined as the difference between the potential of cathode and anode, Ecell = Ecathode – Eanode, at a particular current, i. The current can be substituted by the current density, j, which is the current normalized for the surface area of the catalyst. Theoretically Ecathode and Eanode can be predicted by the Nernst equation, E Nernst = E 0 RT nF ln CO ∗ C R ∗  (For a cathodic reaction; Ox + ne- ⇄ Red) (2-4) ENernst is the potential of the cell, E0 is the standard potential of the redox reaction, R is the gas constant (8.3144 J/K mol), T is the temperature, F is Faraday's constant (96485 C/mol), n is the number  of  electrons  involved  in  the  electron  transfer,  and  C* is  the  bulk  concentrations  of species,  and  Ox  and  Red  which  stand  for  oxidized  and  reduced  forms  of  the  species, respectively.  The  Nernst  equation  relates  potential  to  concentrations  in  the  electrolyte  at thermodynamic equilibrium. In an electrochemical cell, however, reaction equilibrium may or may not be realized depending on the conditions. For example, the rates of reactions are very much dependent on the applied potential. Also, the theoretical equilibrium potential for the ORR is 1.23 V vs. SHE, but if Pt is used as a catalyst, an oxide film would form at potentials higher than 0.85 V vs. SHE in an acidic condition. The presence of oxide on the surface would interfere with the ORR reaction pathway, decreasing the current as compared to the oxide-free Pt surface 16 at the same applied conditions. 2.4.2.  Current-potential relationship in kinetics 2.4.2.1.  Overpotential When forward and backward reactions have considerably different rates, the reaction is said to be irreversible. The extent of irreversibility is often expressed by the  overpotential (or polarization), η: η = Eactual -ENernst (2-5) Eactual is the potential where the reaction produces net current, therefore overpotential is the extra potential needed to drive the reaction at a certain rate in addition to the Nernst potential  [9]. There are three mechanisms contributing to overpotentials: 1) activation overpotential; 2) ohmic overpotential; and 3) concentration overpotential.  Activation overpotential is due to the slow kinetics of the chemical reactions at an anode or a cathode, ohmic overpotential arises due to electrical  resistance  in  the  cell,  and  concentration  overpotential  is  due  to  mass  transfer limitation. These are illustrated in Figure 2-5 [9]. Ideally, the potential-current relation shows a straight horizontal line, shown at the top of the figure; however, potential constantly drops with increasing current due to one or more types of overpotentials. 17 Figure  2-5 A polarization  curve  showing the  three  mechanisms  of  overpotential observed in a fuel cell. The integral of the curve is essentially the power output from the fuel cell. The theoretical potential of the cell is calculated by the Nernst equation, and  the  potential  is  ideally  constant  regardless  the  current  flow.  However,  the potential is lost in overcoming the reaction irreversibility (activation polarization), the internal electrical resistance (ohmic polarization), and the reactant mass-transfer limitation to the catalyst surface (concentration polarization). 18 2.4.2.2.  Current-overpotential relationship The net current flow is simply the difference between forward and backward reaction current (if and  ib, respectively), and it is important to derive the equation relating potential (or overpotential) to the net current flow. Transition state theory is used in the derivation, which is detailed in  [9].  For  a  one  electron-transfer  reaction  with the  transition state  in  forward and backward  directions  shown in  Figure  2-6.  The  net  current,  inet, which  is  a  measure  of  the reaction rate, at a particular overpotential is calculated using the following equation, called the Butler-Volmer equation [12]: inet=i f −ib=i0[exp F RT − exp−1− F RT ] (2-6) The term α is a transfer coefficient; a term related to the symmetry of the activation barrier to the reaction. The notations F, R, and T are the same as the ones used for the Nernst equation. The term, io, is called the exchange current, which is the current at dynamic equilibrium flowing in both forward and backward directions (so the net current is zero). The exchange current is proportional to the standard rate constant (k0) of the reaction, i0 = FAk0Co*(1-α)CR*α (2-7) where A is an area of the electrode material. The exchange current is therefore a good measure of the intrinsic facility of the reaction [9]. When the overpotential  for the forward reaction is  large (i.e., |η|  > 118 mV), the contribution to the net current or current density from the backward reaction is minimal, hence the second term in Butler-Volmer expression can be neglected. Solving for overpotential, the equation becomes, =a  b log j  (2-8) where a = [(2.3RT/αF)*log  j0], b = -2.3RT/αF, and  j is the current density, which is the area normalized current. This expression is known as the Tafel equation. Tafel plot is an important diagnostic tool for determining thermodynamic parameters, i.e., α and j0, deduced by plotting η 19 (or  E)  against  log  j,  assuming  reaction  mechanism  does  not  change  within  the  range  of extrapolation [9]. The presence of a catalyst is required for most of the fuel cell reactions.  A catalyst is defined as a substance that increases the rate of a chemical reaction without being a part of its end  products  [13]. Most  of  the  fuel  cell  reactions  fall  into  the  category  of  heterogeneous catalysis,  i.e., a catalytic process with the reacting molecules and the catalyst in the different phases (e.g., gas phase molecules reacting on a solid catalyst). The concept of heterogeneous catalysis is described next. Because catalytic processes are sensitive to surface structures, a lot of work elucidating reaction mechanisms has been performed on Pt single crystals of low-Miller indices, as Pt is so far the best fuel cell catalyst. The structures of the crystals are also shown. 20 Figure 2-6 The transition state of the forward and backward reactions. ∆G0‡ shows the  standard  Gibbs  energy of  the  reaction  at  the  equilibrium potential.  When  a potential is applied externally, the potential curve shifts, as shown in the top figure. The transfer coefficients, α for the forward reaction or (1-α) for the reverse reaction, describes  the  contribution  to  the  net  current  from  either  side  of  the  reaction. Reproduced with permission from [9]. Copyright 2001 John Wiley & Sons Ltd. 21 2.4.3.  Heterogeneous catalysis Fuel cell catalysis or electrocatalysis is heterogeneous catalysis involving a metal surface interacting with the solvated gas molecules. The concept of catalysis is to increase the reaction kinetics by lowering the activation energy of the reaction that would otherwise be sluggish. This is achieved by adsorption of the reactant molecules onto the catalyst surface. Surface of solids generally  are  unsaturated  (i.e., have  fewer  neighboring  atoms  than  the  bulk),  thereby  the molecules in a gas phase can chemisorb onto the surface atoms. This process leads to weakening or even breaking the chemical bonds of the adsorbed molecules, which is called a dissociative chemisorption [11]. Figure 2-7 shows the Lennard-Jones energy diagram for such a process for a molecule, A2. The difference between the two potential curves is the energy of dissociation of the molecules (designated as Ediss).  ∆Eact is  the activation barrier  that  the molecules need to overcome in the presence of catalyst. The adsorption energy of the molecule is denoted as ∆Hads in the diagram. Because ∆Eact is smaller  than  ∆Ηads, A2 is stabilized on the surface while the dissociation of A-A bonds occurs [11]. Figure  2-7 Energy  diagram  showing  the  dissociative  chemisorption  process  in heterogeneous catalysis. The adsorption energy of A onto a site on a substrate is lowered to ∆Εact by the presence of catalyst. Adapted from [14]. 22 The activation energy of any reaction is calculated using Arrhenius equation, including the electrocatalytic reaction [15]: d log i0 d 1/T  = Eact 2.3RT (2-9) Adsorption of molecules onto the catalyst surface, particularly onto the electrode surface (i.e., metal/solution interface) is governed by the Gibbs adsorption isotherm: −d =∑ i i d i (2-10) The term, γ is the surface tension which is the energy required to create a new surface with an area A, and i is the electrochemical potential. Γi is the surface excess concentration, defined as the amount of species i in the interface (denoted as niσ;  σ is used to represent the interface) per area A [9]:  i= n i  A (2-11) Surface excess can also be expressed as exchange equilibrium of  i in  the interface and the solution (in terms of their activities, a) with the equilibrium constant, Keq [16]:  i=K eq a i sol ' n yi    Κeq = ai  a i sol ' n (2-12) The term yiσ is the activity coefficient of the adsorbed species and accounts for the adsorbate- adsorbate interactions in the interface. The mode of adsorption can be described by the term yσ, and it can be determined by fitting experimental data to the theoretical isotherms. For example, in  the  case  of  a  Langmuir  adsorption  isotherm,  where  adsorbed  particles  of  finite  size  are independent of one another, the adsorption isotherm becomes,  i max− i =K eq ai sol ' n   (2-13) where  Γmax is  the  maximum amount  of  i possibly adsorbed onto  the  surface  per  area  (i.e., saturation) [17]. Elucidation of adsorption isotherm in electrocatalysis is not straightforward as adsorbate-adsorbate interactions and/or the adsorption energy of molecules may be influenced 23 by the presence and the strength of electric field. On the other hand, understanding adsorption isotherms is critical  for reaction mechanism studies,  and this  is still  a challenge in fuel cell research. 2.5.  Platinum 2.5.1.  Low Miller index Pt single crystals The crystal structure of bulk Pt is face centred cubic (fcc), but depending how the surface is made by exposing various Miller index planes, the surface energies will vary.  Figure 2-8 shows three different principle Pt crystal faces, namely (100), (110) and (111). Pt (111) surface is the most densely packed surface among the three single crystals, followed by (100) and (110). The surface energies of those surfaces are closely related to the packing density of the crystal faces  [11] as shown in the previous section. A larger surface dipole is induced by a rougher surface (see Figure 2-4), which tends to lower the work function. Because catalytic processes require  adsorption  of  molecules  on  the  surface,  and  a  rougher  surface  tends  to  offer  more adsorption sites, it is generally true that catalytic activity is better on a rougher surface. Polycrystalline Pt is a surface of Pt containing two or more single crystal faces separated by grain boundaries, thereby providing an average characteristics of those three low-index single crystals. For example, the charge due to the adsorption of hydrogen on Pt (111), (110), and (100) are  255  µC/cm2,  220 µC/cm2,  and  205  µC/cm2,  respectively [18,  19],  whereas  the  one  for polycrystalline  Pt  is  210  µC/cm2  [20].  Polycrystalline  Pt  is  also  commonly  used  to  study catalysis. The first  reaction of  interest  on the Pt surface is  ORR – the cathodic reaction in PEMFCs. 24 Figure 2-8 Face centrerd cube crystal (e.g., Pt) cut in three low Miller index planes. Pt (111) is the most densely packed, followed by Pt (100) and Pt (110). 25 2.5.2.  Oxygen reduction reaction 2.5.2.1.  Mechanisms of oxygen reduction reaction on Pt The ORR is a four-electron transfer electrocatalytic reaction where oxygen molecules are reduced to water. Because of its high cathodic standard potential (Eo = 1.229 V vs. SHE) and the formation of water as the reaction product, it is an ideal cathodic reaction for fuel cells. In an acidic medium, the ORR reaction is, O2 + 4e− + 4H+   ⇄ 2H2O       Eo = 1.229 V vs. SHE  (2-14) In a fuel cell, the ORR is coupled with e.g., the hydrogen oxidation reaction on the anode side; 2H2 ⇄   4e− + 4H+     Eo = 0.0 V vs. SHE  (2-15) The challenge with these reactions is that both of them require the presence of catalyst because the reaction rates are  slow.  Pt  is  a  good catalyst  for both reactions;  however,  the exchange current densities of ORR and hydrogen oxidation on the Pt surface are quite different;  10-10 A/cm2 and 10-3 A/cm2, respectively, in an acidic condition at room temperature [2]. The potential loss  by ORR is  much more severe;  therefore the catalytic  research on ORR is  much more actively pursued. ORR catalysis involves adsorption of oxygen onto the surface of catalyst. Oxygen lone pair electrons interact with the partially filled Pt  d orbital,  and electrons from Pt 5d orbitals back-donate into oxygen pi* orbitals, which weakens the O-O bond. The adsorption process is sensitive to the surface structure and easily affected by the environment in which the reaction proceeds. It is currently accepted that the ORR proceeds via two pathways; a direct path (direct conversion of O2 to  H2O involving four-electron transfer)  or a series pathway (two-electron transfer forming hydrogen peroxide which may or may not be further reduced to water) [21]. It is also well-accepted that oxygen molecules are adsorbed onto the electrode, forming reaction intermediates on the electrode surface. Numerous possible reaction pathways occurring during ORR are shown in Figure 2-9. In a series pathway, the reaction proceeds as [22], 26 Ο2 + 2e− +2Η+  ⇄ H2O2       Eo = 0.67 V vs. SHE (2-16) H2O2 + 2e− +2Η+  ⇄ 2H2O       Eo = 1.77 V vs. SHE (2-17) It has been proposed that different binding arrangements of oxygen molecules lead to either pathway. These arrangements are shown in  Figure 2-10.  The middle model (also known as Pauling  model)  is  believed  to  go  through both  direct  and  parallel  pathways  [23].  Peroxide formation in fuel cells is undesirable for two reasons: 1) hydrogen peroxide generation involves only two electrons, providing less current than four electron transfer; and 2) hydrogen peroxide oxidizes the membrane electrolyte in a PEMFC, shortening the lifetime of the fuel cell. Usually, the direct-pathway dominates on a Pt surface  [24, 25]  at potentials above 0.2 V vs. standard hydrogen electrode, SHE,  which is well below the operating potential of fuel cells. Indeed, the formation of H2O2 has been reported on a Pt surface, especially when the potential is below ~0.3 V vs. reversible hydrogen electrode, RHE [26, 27]. Figure 2-9 Possible reaction pathways for ORR. Final products are either water or hydrogen  peroxide.  Reprinted  from  [21].  Copyright  1987  with  permission  from Elsevier Sequoia S.A. 27 Figure 2-10 Three possible  binding arrangements of molecular  oxygen adsorbed onto a catalyst surface, Me. The end-on arrangement is believed to lead to the series pathway in ORR. The interactions between oxygen molecular orbitals and the Pt d orbitals  are  shown. Reprinted  from  [23].  Copyright  1986  with  permission  from Elsevier Sequoia. 28 The adsorption isotherm of oxygen on Pt was studied by Damjanovic  et. al., [28]. The authors  found  a  Temkin  type  adsorption  isotherm  by  fitting  their  experimental  data  to  a theoretical model, which assumes a linear decrease in the adsorption enthalpy with total surface coverage, θT. ∆Ηi = ∆H0,i - χiθT (2-18) ∆H0,i is the adsorption enthalpy of species i when the coverage is zero, and χi is the change in the enthalpy with respect to the coverage. The total coverage includes not only the species,  i, but also the reaction intermediates formed in the course of the reaction.  By further fitting their experimental data, the authors found the oxygen coverage was also linearly dependent on the applied potential and pH [22, 28]:  MAX = = F  E2.3RTF pH−E0 (2-19) Also, the enthalpy of activation of oxygen dissociation on the Pt (111), (110), and (100) surfaces were found to be 42 kJ/mol in sulfuric acid, according to the Arrhenius equation [15]. Once oxygen dissociates on the metal surface and forms a substrate–O bond, protons can bind to the adsorbed oxygen. On Pt single crystal electrodes, ORR catalytic activity increases in the order (111) < (110) < (100) in sulfuric acid electrolyte [29]. The rate determining step for ORR is still controversial. The following steps have been reported as possibly rate determining for the ORR: adsorption of oxygen on the surface of Pt [30], adsorption of oxygen simultaneous with the first- electron transfer  [31],  adsorption of oxygen simultaneous with the first-proton transfer  [32], simultaneous first-electron and first-proton transfer to oxygen  [33, 34], and the first-electron transfer [24]. The origin of the slow ORR kinetics is unknown, and it is extremely challenging to pin point an exact cause. However, Nørskov  et al., have used density functional theory (DFT) to calculate  the  free  energy of  candidate  intermediates  during  the  course  of  the  reaction.  The authors found that, adsorbed oxygen and hydroxyl species are quite stable and the electron or proton transfer to the adsorbed oxygen or hydroxyl may be responsible for the sluggish kinetics at potentials close to the equilibrium [35]. Pre-adsorbed species can also affect the kinetics of ORR. Damajanovic et al., found the 29 rate of ORR on the pre-reduced Pt surface is faster by a factor of 2 than the pre-oxidized Pt surface [36]. This is because the presence of oxide on the Pt surface can block the sites where molecular oxygen could adsorb [37]. Also, the specific adsorption of e.g., (bi)sulfate anions can influence the reaction kinetics. Bisulfate adsorption is particularly severe on Pt(111) surface – it occurs potentials above 0.3 V vs. RHE  [38]. (Bi)sulfate adsorption is less severe on Pt (100) surface although it surely occurs  [39] when the potentials are above 0.6 V [38]. On the other hand,  perchlorate  anions  do  not  significantly  adsorb  onto  Pt  surface.  The  rate  of  ORR determined in perchloric acid is therefore the fastest on Pt (111), followed by (110) > (100) surfaces; contrary to the rate measured in sulfuric acid [29]. The Tafel slope of ORR on the Pt (111) surface measured in sulfuric acid is  reported as 120 mV/decade throughout the whole potential region, whereas the one measured in perchloric acid at higher current region was 77 mV/decade  [15].  Despite  the  bisulfate  adsorption  and subsequent  variation  to  the  measured kinetics, sulfuric acid is commonly used for investigating ORR catalytic behavior because the most commonly available solid electrolyte,  NafionTM ,  is  a sulfonated ionomer,  as shown in Figure  2-1.  It  has  been  reported  that  although  specific  adsorption  of  (bi)sulfate  anions  is definitely present, it is not strong enough to cause a change in the ORR mechanism, i.e., change the direct pathway to the series pathway. Also water is believed to be co-adsorbed with the anion layer, which could relatively easily be displaced by incoming oxygen molecules [27]. 2.5.2.2.  ORR on rough Pt surface Oxygen dissociation catalysis on a rough surface is significantly different compared to a smooth surface. (Rough surfaces are often represented by the “terrace (i.e., flat) sites” and “step sites”.)  Adsorption and subsequent dissociation of oxygen molecules on Pt surfaces have been extensively studied [40, 41], showing that a Pt surface having steps and terraces exhibits more O2 adsorption/dissociation at step sites due to the stabilization of the dissociated transition state. This is due to the higher coordination number for Pt atoms on terrace sites than those on step sites  [42].  Work reported by Gee  et  al.,  showed that  the difference in the extent  of oxygen dissociation process between Pt(111) (no step sites) and Pt(533) (with step sites) was governed by the presence of defects or steps [42]. Maciá et al., also observed higher current on a Pt single 30 crystal electrode containing more steps and terraces than the one with a flatter surface  [27]. Topography of the catalyst surface is therefore important information for the catalysis research. 2.5.3.  CO oxidation reaction CO oxidation reaction is known as a rate determining step of the methanol oxidation reaction, the anodic reaction in a direct methanol fuel cell (DMFC) [43]. Similar to the ORR, the rate of the CO oxidation is very slow, and the presence of a catalyst is essential. CO oxidation also involves dissociative chemisorption on a catalyst surface. The CO oxidation mechanism is still  not  fully  understood.  In  this  section,  the  current  understanding  of  CO oxidation  on  Pt surfaces is presented. 2.5.3.1.  Mechanisms of CO oxidation on Pt The  mechanisms  of  CO oxidation  catalysis  is  still  under  debate,  but  the  Langmuir- Hinshelwood mechanism [44] is generally accepted: Pt + H2O → Pt-OHads + H+ + e- (2-20) Pt + CO → Pt-COads (2-21) Pt-COads + Pt-OHads → Pt-COOHads + Pt (2-22) Pt-COOHads → CO2 + H+ + e- + Pt (2-23) Pt + H2O + CO → CO2 + 2H+ + 2e- + Pt (2-24) OHads indicates the oxygen-containing species, which is formed by the dissociation of adsorbed water on the Pt surface. The standard potential of the net CO oxidation should be approximately 31 0.1  V vs. SHE [45] but it requires 600 mV overpotential on the most effective catalysts known. The sluggish kinetics of CO oxidation arises from the dissociation of adsorbed water on the Pt surface  to  give  -OHads [46].  Relatively high anodic potential  is  required  to  “activate”  water splitting. Alternatively, since CO and -OH are competitively adsorbed, there are many examples of the reaction rate being fit with the Langmuir-Hinshelwood mechanism [47]. Reaction rate=kCO N CON OH  = k CO N 2 K CO K OH pCO[OH ads] 1KCO pCOKOH [OH ads] 2 (2-25) The term θ is the coverage of the corresponding molecules on the surface, KCO and KOH are the exchange equilibrium constants for CO and OH, respectively, kCO is the rate constant of the CO oxidation, N is the total number of adsorption sites on the surface, and pCO is the pressure of CO. The activation barrier for the reactions (equations) 2-22 and 2-23 together was determined to be ~100 kJ/mol [48, 44]. It should be noted that Equation 2-25 does not necessarily mean the rate determining step is the reaction between CO and OH. Rather, the reaction mechanism on CO oxidation  is  still  not  fully understood as  pointed out  earlier.  The debate  for  this  apparently simple reaction has been going for decades. For example, -OHads appear in the reaction path may or may not be hydroxyl molecule  [49]. Also there are at least three proposed mechanisms for removal of COOHads at the last step [50]. CO oxidation is experimentally studied by anodic stripping voltammetry, which involves electrochemical preconcentration of adsorbed CO onto the electrode (catalyst) surface, followed by a potential sweep in an anodic direction, oxidizing the CO. The position (potential) of the stripping peak is examined and compared. 2.5.3.2.  CO oxidation on low Miller index Pt surfaces Although CO oxidation was studied on low Miller index Pt surfaces, the results differ among  laboratories.  Lamy et al., showed CO stripping voltammograms recorded on Pt (100), 32 (110), and (111) surfaces in a CO saturated 0.1 M HClO4 solution at room temperature at a sweeping rate of 50 mV/s [51]. The authors observed stripping potentials of 0.68 and 0.59 V vs. RHE on Pt (100) and (110) surfaces, respectively, and also observed two peaks at 0.694 and 0.95 V on Pt(111)  surface.  However,  the  results  shown by Beden  et  al. [43],  are  different  from Lamy's. They observed only a single peak on Pt (111) at potentials more negative than the peak recorded on Pt (100). The results reported by Herrero et al., [52] only showed single peaks for all surfaces, and the stripping peaks for Pt (111) and Pt (100) were observed at 0.7 V and 0.73 V vs. RHE, respectively. In a 0.5 M sulfuric acid solution, the authors found the stripping peaks at much more positive potentials;  0.83 V and 0.77 V for Pt (111) and Pt (100), respectively. Different shapes for the CO stripping voltammograms have been reported, just as Lamy found two peaks on the Pt (111) surface [51]. Two examples are given in Figure 2-11a) and b) which show multiple peaks and a single peak with a shoulder, respectively. The potential where the CO deposition was performed is shown as Edep. These voltammograms were, at least in part, explained by Cuesta  et al.,  who identified the shoulder as the oxidation of bridge-bonded CO and the peak as the oxidation of linearly bonded CO using a polycrystalline Pt electrode [53]. (This is the indication that Lamy's Pt single crystal electrodes had some defects on the surface.) Different bonding arrangements are shown in Figure 2-11b) They also found the appearance of the shoulder only when the CO Edep was lower than 0.35 V vs. RHE, where high CO coverage was achieved. Surface enhanced IR spectroscopy results reported by Miki et al., also indicated the slightly weaker bonding energy for bridge bonded CO  [54], which is consistent with the findings by Cuesta. Difficulties in understanding the CO oxidation electrocatalysis is, at least in part, due to the sensitive nature of adsorbate to its surroundings. It has been reported CO stripping potential is sensitive to temperature [52], specifically adsorbed anions present in electrolyte [43, 55], CO deposition  potentials [53], scan rates [56], and deposition time [43] but the most complex factor is the surface structure of a catalyst  [52, 51, 57, 58]. It is not only the type of crystal faces of catalysts, but also depends on the number of crystal defects. Even if a single crystal surface is prepared, defects are always present. Although CO oxidation appears as a very simple reaction, it  is  extremely sensitive  to  its  environment,  and  controlling  the  experimental  parameters  is necessary. 33 Figure 2-11 Examples  of different  CO stripping voltammograms recorded under identical conditions. a) Voltammogram with multiple peaks; b) A single peak with a shoulder.  Edep is  the  CO preconcentration  potential  applied  to  the  electrode.  The difference  is  probably  due  to  the  electrode  surface  characteristics  which  are dynamically changed. The shoulder in b) is possibly due to the oxidation of bridge bonded CO, whereas the main peak is due to the oxidation of linearly bonded CO [53]. 34 2.5.3.3.  CO oxidation on a rough Pt surface As shown in the previous section, when a catalyst surface is rough, its work function is lower than a smooth surface due to the surface dipole contribution [59]. Again, rough surfaces are  often  represented  by  the  “terrace  (i.e., flat)  sites”  and  “step  sites”,  but  hydrocarbon adsorbates often exhibit 10-20 % stronger binding energy than on a terrace site [60]. Xu et al., found that when CO and oxygen molecules are co-adsorbed on a surface containing both terrace and step sites, CO tends to be shifted to terrace sites, and oxygen tends to adsorb onto the step sites. They also found CO on terrace sites, oxidized to CO2 more easily than on step sites. The possible reason is the difference in bond strength for CO on terrace or step sites; 96 kJ/mol or 151 kJ/mol, respectively [59]. The binding energy of oxygen on step sites is higher than that on terrace sites. It is possible the weaker bonding of CO, and/or stronger binding of oxygen on steps favour the oxidation process. The stronger binding energy on step sites must be a result of electronic effects. However, the authors also suggest a geometrical effect. The distance between CO (terrace) to O2 (step) on Pt (335) (the surface containing both terraces and steps) is shorter than the distance between CO and O2 both on step sites, which may also contribute to more efficient catalysis  [59,  60].  An alternative explanation is  the surface structure-dependent CO adsorption, recognized by Mikita  et. al [58]. The authors realized CO adsorbs only on defect sites when the CO coverage is very low. This suggests the presence of crystal defects increases the surface concentration of CO, which would lead to more effective CO oxidation. 2.5.3.4.  Adsorbate-induced surface restructuring in the Pt surface When adsorbates are present on the surface, and if the interaction between the adsorbate and substrate is strong, it often breaks a metal-metal bond (in the substrate) to form a metal- adsorbate bond, leading to the shift in equilibrium positions in surface atoms [61]. When small molecules, such as hydrogen, CO and oxygen, are adsorbed onto a metal surface, and if the interaction between adsorbate and the surface atoms is stronger than the interaction between atoms  in  the  metal,  there  may be  a  surface  reordering  [62].  For  example,  when oxygen  is 35 adsorbed on the Ni(100) or Cu(100) surface, metal d and O 2p orbitals interact very strongly and the orbitals hybridize to form bonding and antibonding orbitals. Subsequently, Cu or Ni surface atoms reconstruct, which shifts the antibonding orbitals up higher than the Fermi level of the metal. This takes shared electrons away from antibonding orbital, which subsequently lowers the energy of  the  oxygen-adsorbed metal  surface  [63].  A Pt  surface  does  not  encounter  severe surface restructuring with oxygen adsorption during the ORR, but does with CO adsorption. Gritsch et al., observed the reconstruction of Pt(110) surface transforming from (1×2) to (1×1) after CO was adsorbed onto the surface [64]. Behm et al., has observed the phase transformation from Pt (100) hexagonal structure (Pt-hex) to Pt (100)-(1×1) when CO was adsorbed onto Pt- hex sites. The structures of Pt-hex and Pt (100)-(1×1) are shown in  Figure 2-12. Despite the CO-CO repulsive interactions, CO adsorbed onto Pt-hex sites formed small islands (~15 CO molecules/island)  even  at  low  coverage  (θ<0.5).  They  also  found  that  once  hex-(1×1) transformation occurred, CO molecules are trapped at (1×1) sites [65]. The reverse process also occurs; surface reconstruction of Pt(1×1) to Pt-hex is achieved by desorption of CO from the Pt (1×1) surface. The thermodynamics of this process has been presented by Thiel  et al [66]. An energy diagram of their findings is summarized in Figure 2-13. When a clean Pt (100) surface is present without CO adsorbed, Pt-hex is energetically more stable than (1×1) by “∆E” (shown at the upper part  of diagram).  The activation energy (E*) of transformation of hex to  (1×1) is 20~25 kcal/mol. However, when CO is adsorbed onto a Pt (100) surface, it lowers the energies of (1×1) and hex by θ2E2 and θ1E1, respectively, where E and θ correspond to the CO adsorption energy and CO coverage on each surface.  δE in the diagram therefore represents the energy difference between CO-covered Pt (1×1) and Pt-hex structures, so the phase transition occurs when δE is negative and also smaller than ∆E; i.e., θ2E2 – θ1E1 < ∆E. In their contribution, they also found that the trigger for this transformation is purely the adsorption of CO, but because E2 is greater than E1,  θ1 must be greater than  q2 at equilibrium, thereby the energies required to transform the hex structure to (1×1) by CO adsorption or to transform (1×1) to the hex structure by CO desorption are different; i.e., A hysteresis loop is observed [66]. Although this particular thermodynamic solution may not apply to all other phase transition phenomena, their findings do show the approach one may need to take in order to study the mechanisms and energetics of such  transformations.  This  is  important  for  electrocatalysis  as  many catalytic  processes  are dependent on the detailed surface structure. 36 Figure 2-12 Atomic arrangements of Pt (100)-hex and Pt (100)-(1×1) structures. Pt (100)-hex is formed over Pt (100) to minimize the surface energy of the crystal; however, when CO is adsorbed onto the hex structure, surface reconstruction occurs, transforming the overlayer into (1×1) structure. 37 Figure 2-13 Energy diagram showing the surface reconstruction of Pt(1×1) ↔  Pt- hex structure  with the  CO adsorption.  When a  clean  Pt  (100)  surface is  present without CO adsorbed, Pt-hex is energetically more stable than (1×1) by “∆E” and the transformation to Pt-hex requires the activation energy, E*. However, when CO molecules are adsorbed onto the Pt (1×1) surface, its energy is lowered significantly that  the  Pt  (1×1)  exhibits  a  lower  energy.  Reprinted  with  permission  from  [66]. Copyright 1983 American Institute of Physics. Understanding the CO oxidation mechanism is probably made more complicated by the adsorbate-induced surface reorganization. The surface always attempts to attain the minimum energy possible. In fact, it is suggested by Somorjai et al., that adsorbate-induced restructuring may change the active sites of catalytic reactions (especially the rate determining steps) and therefore, it is crucial to study active sites in the presence of adsorbates, rather than on a clean catalyst surface [14]. 38 2.6.  Pt-based alloys Assume the PEMFC is operated at E=0.6V with a current density of 500 mA/cm2. The power output from the single cell, according to Equation 2-3 gives 0.3 W/cm2 (the power output calculated with a current density,  j). Typical Pt loading of one catalyst layer is ~ 1 mg/cm2; therefore anode and cathode together, one single stack fuel cell contains 2 mg/cm2  of Pt. If an automobile requires a minimum power of 50 kW, the price of Pt required (assuming Pt costs 992 USD/oz [4]) is [2], 50,000 W 0.3W /cm2 × 2 mg Pt cm2 × US $ 992 oz × 1 oz 28,349 mg = US $ 13,164 (2-26) The price of platinum alone is too expensive for the cost of automobiles, and research seeking to reduce the usage of Pt (aiming for the Pt loading of 0.1 mg/cm2) has been actively pursued. Diluting Pt with other metals – the first row of transition metals in particular – is an excellent way to reduce the Pt loading in fuel cells and also the alloys are known to improve the kinetics of the fuel cell reactions as compared to Pt. This section reviews the current understanding of Pt- based alloys for fuel cell reactions. 2.6.1.  Fuel cell reactions on Pt bimetallic alloy catalysts Fuel  cell  reactions  have also been extensively studied on Pt-based alloys.  Numerous alloys  of  transition metals  in  the  first  row of  the periodic  table  have  shown more  efficient catalysis on ORR compared to Pt. Fewer numbers of Pt-alloys have shown improved kinetics for CO oxidation. Nonetheless, the improvement appears to be due to the changes in the electronic structure and the geometrical parameters in either component of the alloy compared to its pure form. This section describes the ORR and CO oxidation catalysis measured on Pt-based alloys and  summarizes  the  alloying  effects  and  their  influence  to  the  fuel  cell  reaction  catalysis. Adsorbate-induced surface reorganization observed on alloys is also briefly presented. 39 2.6.1.1.  ORR on Pt bimetallic alloy catalyst There are a number of Pt-transition metal alloys reported as more efficient catalysts than Pt itself. In this thesis, the focus is limited to bimetallic Pt alloys, although trimetallic Pt alloys have also been tested for fuel cell applications [67, 68]. Metals that have been alloyed with Pt and showed improvements in measured kinetics for ORR include Co [69, 26], Ni [26, 69], Cr [70], Au  [71], Ru, Ti, V, Fe, and Ir  [67]. PtRu, PtCo, PtCr, and PtNi have particularly shown excellent catalytic activities; 2~4 fold enhancement in kinetics of the reaction. There are two changes that alloying seem to induce in Pt: 1) increase in Pt 5d vacancies, and 2) decrease in Pt- Pt  bond  distance.  An  increase  in  the  Pt  5d orbital  vacancy  concentration  reduces  2pi* backdonation into O2, modifying O-Pt or destabilizing O-O bonds, aiding oxygen dissociation. The shorter Pt-Pt bond may promote ORR by a geometrical effect. Compressive strain, caused by  shortening  Pt-Pt  bond  distance,  is  also  said  to  be  related  to  the  increase  in  Pt  d band vacancies; the d-orbital overlaps are increased, resulting in a broadening of the average d-band width and lowering the average energy [72]. Activation  energies  for  the  ORR  on  Pt-Ni,  Pt-Cr,  and  Pt-Co  in  an  actual  fuel  cell operation  condition  have  been  measured  and  compared  to  pure  Pt  [70,  73].  The  activation energies of alloys  were 1.5~3 times less than that  of Pt,  accounting for the enhanced ORR activities. H2O2 production on these alloys was also investigated [74], but no significant amount of H2O2 was detected. Also, the calculated Tafel slopes on Pt-alloys were similar to the one on Pt surface, indicating alloying did not influence the reaction mechanism  [70]. The authors were also able to show an increase in the exchange current density on PtCo by a factor of 3 comparing to pure Pt in their fuel cell  setup.  Other Pt alloys such as -Ni or -Cr also showed 2~4-fold increase in the exchange current densities. Figure  2-14 shows  the  activity  of  O2 dissociation  (i.e.,  log-scaled  rate  constant  for oxygen dissociation) with respect to oxygen binding energy [35]. Activity is also related to the occupancy of the Pt 5d orbital [75]. This plot is called a volcano plot, and the closer the element to the summit, the more efficient the catalyst. Pt is the best catalyst for oxygen dissociation for a single element on the plot shown. Xu  et al., showed that by alloying Pt with Co, Pt3Co was placed at the higher position than Pt on the volcano plot [75]. 40 Figure 2-14 The volcano plot (theoretically determined) showing the ORR activity with respect to the oxygen binding energy.  The ORR activity corresponds to the energy of activation barrier of the rate determining step calculated. The closer the element  is  to  the  summit  of  the  plot,  the  more  efficient  the  ORR  catalysis  is. Reproduced in part with permission from [35]. Copyright 2004 American Chemical Society. 2.6.1.2.  Pt skin formation Although Pt alloys have been formed, the very top layer of the alloy often consists of only pure Pt atoms [76, 26, 77, 64]. This enrichment of Pt on the surface is often called a “Pt skin” structure  [26]. The second layer, on the other hand, becomes Pt deficient and therefore enriched with the other component of the alloy. Such phenomena are the result of dissolution of transition  metal  that  was  initially  present  on  the  surface,  as  well  as  three  other  factors:  1) Differences  in  surface  energies  between  alloy  components  [78,  79];  2)  Differences  in  size between the alloy components; and 3) the energy of mixing [79]. Pt-skin can also be achieved by 41 annealing Pt-alloys, such as PtCo or PtNi [26, 80]. Bardi et al., reported the formation of Pt-skin on Pt3Co alloy,  i.e.,  the topmost layer of the alloy was covered only by Pt atoms,  but  also reported  the  electronic  structure  of  the  top  Pt  layer  observed  with  the  X-ray photoelectron spectroscopy (XPS) was not the same as that of pure Pt [76], suggesting the electronic disruption due to the underlying Co. ORR catalytic activities have also been investigated on the Pt-skin.  An example was shown  with Pt3Ni alloy compared to Pt [80]. The polarization curves comparing polycrystalline Pt and as-prepared-Pt3Ni clearly showed the positive shift in onset potential of the ORR on the Pt3Ni  surface;  however,  annealed  Pt3Ni  showed  even  more  positive  onset  potential  for  the reaction. The reason for the further enhancement in the ORR catalysis by Pt-skin was the weak Pt-OHads interaction due to a modification of the electronic structure of Pt by the underlayer Ni. Interestingly, though, the activation energy for the ORR on the Pt-skin was not significantly different from the one on the as-prepared-Pt3Ni. The authors also detected no H2O2  formation [80]. 2.6.1.3.  CO-oxidation on Pt-alloy surfaces There are a few Pt-alloys that enhance CO oxidation over pure Pt. These include PtRu [81-84, 20, 85] and PtSn [81, 86]. Other alloying metals, such as Mo [87, 88], Re [89], and Rh [90] have also been reported, although the Pt alloyed with these metals did not influence the kinetics of the reaction significantly. There  are  two theories  explaining  the  enhanced mechanism of  CO oxidation  on  Pt- alloys: Bifunctional theory  [85] and the ligand effect  [91]. Bifunctional theory states that the alloying metal provides nucleation sites for OHads at more negative potential than on Pt, whereas Pt serves to adsorb CO. Fast formation of -OHads leads to  more facile CO oxidation process. The ligand effect arises when the electronic structure of Pt is altered due to the presence of alloying metals, which in turn modifies Pt-CO binding sites. Gasteiger and coworkers observed 2-3 fold increase in the CO oxidation kinetics on a PtRu alloy compared to the one measured on Pt [83]. The  contribution  from  the  ligand  effect  is  under  debate.  Numerous  reports  have 42 suggested that it is less important than the bifunctional effect [92-95]. For example, placing Ru atoms in  the  vicinity of  Pt  atoms,  instead  of  forming a  PtRu alloy,  also improved the  CO oxidation kinetics [96]  In summary, Pt alloyed with the first row of transition metals increase Pt 5d vacancies and decrease the Pt-Pt bond distance.  These electrical  and geometrical  changes enhance the catalytic activities of ORR on alloy surfaces, compared to Pt. Pt-skin formation on Pt-alloys further improve the catalytic activity by changing the Pt electronic structure, modifying the Pt- OHads bond energies. Pt alloys reported in the literature are usually Pt1M1 or Pt3M, where M is one of the transition metals. The use of Pt-alloys, instead of pure Pt, replaces 30~50 % of Pt loading in fuel cells.  However, Pt-alloys face a challenge, especially in the fuel cell's  acidic condition  – chemical stability [97, 26]. 2.6.1.4.  Stability of Pt-based alloys under a fuel cell operation conditions The stability of catalysts is an important issue upon developing fuel cells. During the fuel cell  operation,  the  non-precious  metals  of  Pt-alloys  leach  out  and  contaminate  the  solid electrolyte, which may cause four unfavourable outcomes: 1) decrease in the ionic conductivity of  membrane  due  to  dehydration;  2)  increase  in  the  ionic  resistivity  in  the  cathode  due  to blocking the cation exchange sites; 3) decrease in oxygen transport to the catalyst layer; and 4) damaging the membrane if a certain type of metals, such as Ti or Fe, are released [97]. Research seeking  more  stable  Pt-alloys  is  important,  as  this  issue  is  strongly related  to  the  fuel  cell lifetime. In this section, however, only the chemical stabilities of alloys will be discussed. There are two reasons why metal components of alloys leach out, i.e., either incomplete alloying during the manufacture or insufficient enthalpies of mixing of Pt-alloys under fuel cell operating  conditions  [97].  Pt-alloys  are  vulnerable  to  acidic  conditions.  Stamenkovic  et  al., prepared 3:1 PtCo and PtNi alloys to find their catalytic activity compared to pure Pt; however, they  found  the  transition  metals  leached  out  in  an  anodic  (oxidative)  scans  of  cyclic voltammograms recorded in a 0.5 M acidic solution [26]. 43 Colon-Mercado and Popov studied commercially available alloys, i.e., Pt-Co, Pt-Ni, Pt- Fe, and Pt-V, and measured the loss of transition metals from the alloys. The authors immersed the alloys (dispersed on GDL) in an acidic liquid electrolyte, applied a constant potential at 0.8 V vs. NHE  over a prolonged time, and tracked the metal dissolution by analyzing the solution by atomic absorption spectroscopy  [69]. The results indicated the dissolution of a significant amount of metal. Though they studied for over 70 hours, most of the metal dissolved in the first couple of hours of polarization. The alloy that showed the least metal leaching was Pt3Ni, but it still showed 5 % metal dissolution within the first ~3 hours. Considering this was 75% Pt, these alloys are not stable in an acidic environment. The worst alloy was 1:1 PtCo which showed 40% metal loss in the first 1 or 2 hours of polarization. Gasteiger has also found PtCo exhibits a high degree of  leaching;  35 % Co loss  after  24 hours of immersing the alloy into the alloy (no potential applied).  Multiple pre-leaching steps have been suggested by Mukerjee et al. [98],  to purposely dissolve  any  transition  metals  before  incorporating  the  alloys  (on  MEAs)  into  a  fuel  cell assembly. In this way, only a few per cent of the transition metals leached out of the alloy during fuel cell testing, and the catalytic activity was still better than that on Pt. However, this is an extra step that industry must perform in the fabrication process, that can only increase the cost. 2.6.1.5.  Adsorbate induced surface segregation Adsorption of atoms or small molecules on alloy surfaces may induce a large change in the surface composition.  When one component  of  multi-metallic  alloys  has  a very different adsorption energy from the other component(s) of the alloy, atomic diffusion perpendicular to the surface may occur by breaking metal-metal bonds. Several examples are found in ref. [99]; for example, surface of Cu-Mn or Ag-Mn films become enriched with Mn upon adsorption of oxygen, or the surface of AgPd alloy becomes Pd-enriched upon CO adsorption. The authors explain such phenomena purely due to thermochemical effect. 44 2.7.  PtZn This section provides a general overview of the characteristics of the PtZn alloy. The PtZn alloy utilized in this thesis was facilitated by so called the underpotential deposition of Zn onto the Pt surface. Subsequent alloying effects are presented and compared to those observed in other alloys. Although the stability of a PtZn alloy in a fuel cell operating condition has never been reported prior to our work, theoretical and experimental data encompasses a reasonable chemical stability in an acidic condition, and this will also be reviewed in detail. 2.7.1.  Pt, Zn, and bulk PtZn “alloy” The crystal structures of pure Pt and Zn are fcc with the lattice parameter of 2.77 Å, and hcp with the lattice parameters of 2.6649 and 4.9468 Å at room temperatures, respectively [100]. When  Pt  and  Zn are  combined,  they form compounds  at  various  compositions.  The  phase diagram of PtZn (some regions are still in question) is shown in Figure 2-15. The regions where the compositions are known are intermetallic compounds (i.e.,  Pt-Zn compounds at the given stoichiometry with  ordered  crystal  structures).  For  example,  Pt1Zn1 and  Pt3Zn1 intermetallic compounds with L10 (AuCu) and L12 (AuCu3) structures  [101]. Pure Pt and Zn in the phase diagram and possibly some regions of the phase diagram (especially at higher temperatures) are solid solutions (i.e., disordered crystal phases with various compositions). An “alloy” is defined as either a solid solution or intermetallic compounds with some disorders in its lattice atomic arrangements. In the current fuel cell literature, the term, “alloy” is interchangeably  used  for  a  solid  solution  or  for  an  intermetallic  compound,  although “intermetallic  compounds”  are,  strictly  speaking,  not  an  alloy  because  they  have  definite stoichiometry and ordered lattice structures that exhibit a characteristic diffraction pattern. In Chapter 4~6, we will show an electrochemically prepared mixed metal containing Pt and Zn. This metal showed  orbital mixing between Pt and Zn atoms; however, the crystal structure and the atomic arrangements of this metal are still in question. Also the metal surface structure is generally very different from the bulk structure. Therefore in this thesis, we call our mixed metal 45 an “alloy” which may be a solid solution or an ordered intermetallic compound. Figure 2-15 PtZn phase diagram. Reproduced in part from [102]. Copyright 1991 with kind permission of Springer Science and Business Media. 46 2.7.2.  PtZn alloy formation 2.7.2.1.  Zn underpotential deposition on Pt Bulk metal deposition from a solution onto another metal can occur electrochemically, and the onset potential can be predicted by the Nernst equation. The standard potential of Zn2+ reduction (i.e., Zn2+ + 2e-  Zn(s)) is -0.763 V vs. ⇄ SHE [103]. Deposition of Zn on Pt in a highly acidic medium therefore is predicted not to occur as hydrogen evolution from Pt (Eo = 0.0 Vvs. SHE) at such a low potential prevents Zn2+ from approaching to the electrode surface. However, this is not the case, as Zn forms a so-called underpotential deposition (UPD) layer onto Pt before hydrogen evolution starts. UPD is an electrochemical phenomenon where deposition of metal occurs  at  a  potential  more  positive  than  predicted  by  the  Nernst  equation  [104].  UPD  of electrodeposited metal thus formed grows epitaxially, i.e., the growth is two-dimensional and the morphology of the film follows that of the substrate. There are two criteria for metal to form a UPD onto a substrate: 1) The metal-substrate interaction is stronger than metal-metal interaction, and  2)  the  sizes  of  atoms  of  deposited  metal  and  substrate  are  similar  (i.e., small  lattice mismatch) [104]. Experimentally, the formation of a Zn UPD layer was reported by Aramata, via cyclic voltammograms [105]. The epitaxial growth of Zn onto Pt has been confirmed [106], and Pt and Zn together satisfy both of the criteria for UPD [105, 107]. Zn UPD on a polycrystalline Pt in a H2SO4 solution  occurs  at  potential  1.11  V  more  positive  than  bulk  Zn  deposition  [108]. According to Kolb [104], the difference in the UPD and bulk deposition potentials represents the difference  in  chemical  potentials  due  to  the  interaction  among  Zn  atoms  in  bulk  and  the interaction  between  Zn  and  Pt.  The  potential  of  a  UPD layer,  EUPD, relative  to  the  Nernst potential is calculated by the following equation [109]: EUPD = ENernst − H S−M nF CN S (2-27) In this equation,  θ is the coverage of the UPD layer,  ∆HS-M is the enthalpy of bond formation between metal-substrate, and CNS is the lattice coordination number of the substrate. A Zn UPD 47 layer at a potential negative of EUPD is stable, and usually the difference between EUPD and ENernst for the Pt-Zn system is ~1 V [105]. Guerra  et al., also calculated the 1:1 Pt/Zn lattice mismatch to be only 4% (using  the interatomic bond distance of 2.775 Å for Pt, d0,Pt; and of 2.665 Å for Zn, d0,Zn) [107], f = d 0, Zn−d 0, Pt d 0, Pt =2.665−2.775 2.775 =−0.04 (2-28) Also the diffusion coefficient of Zn into Pt is relatively large; 107 times greater than for Zn into Cu at 100  0C  [110]; indicating the possibility of the rapid alloy formation without significant distortion of the crystal lattice. 2.7.2.2.  Transformation of Zn UPD into Pt-Zn alloy Transformation of Zn UPD layer into a Pt-Zn alloy was first realized by Despic  et al. [106], and later well-characterized by Guerra et al. [107]. The cyclic voltammograms recorded by Despic et al., in a 1.5 M ZnSO4 solution (pH=3.5) on a Pt electrode is shown in Figure 2-16. Peak  III'  were  identified  as  hydrogen  desorption.  Bulk  Zn  deposition  (or  overpotential deposition; OPD) occurs at the long tail at the end of cathodic scan, and peak I' was assigned as the stripping of the bulk Zn thus deposited. Zn UPD layer formation was assigned to region II which occurs before hydrogen adsorption, in agreement with another literature report [105]. The authors  gradually  extended  the  cathodic  limits  of  voltammograms  and  observed  dramatic changes in the reverse scan. Peak II' was identified also as stripping of Zn; that was more tightly bounded to the substrate than Zn in Peak I'. Two remarkable features are Peak IV' and V'. Peak IV' appears only when the potential limit was made very cathodic. The authors concluded this was the oxidation of PtZn alloy, which might have formed in the potential region of  III. Peak V' was also assigned as stripping of Zn that had been so strongly bounded, and it could only be removed by the oxidation of the substrate. This experiment shows the rapid transformation of Zn UPD into an alloy. Similar behavior was observed by Guerra  et al. When the electrode was polarized in a 48 0.1M ZnSO4 solution at the Zn UPD potential for a long time, an increase in measured current, just like Peak IV' and V' of Despic's, in the anodic scan was also observed. This increase was limited to the first 600 seconds of polarization. Once formed, PtZn alloy appears to be very stable as it could only be removed by oxidation of the substrate or by immersion into aqua regia [107]. Figure 2-16 The cyclic voltammograms recorded by Despic et al., in 1.5 M ZnSO4 solution  (pH=3.5)  on  a  Pt  electrode.  Peak  potential  region  assignments  are  as follows: Peak I – Pt oxide reduction; Region II – The Zn UPD formation on the Pt; Region III – Post-UPD potential region reflecting the Pt-Zn alloy formation; Peak I' – Zn stripping from the bulk Zn; Peak II' – UPD Zn stripping; Peak III' – hydrogen desorption; Peak IV' – Oxidation of PtZn alloy; Peak V' – Zn stripping from the PtZn  alloy.  The  PtZn  alloy  was  formed  when  the  cathodic  potential  limit  was extended  in  Region  III  while  the  CVs  were  recorded,  indicating  a  rapid transformation  of  the  UPD  Zn.  Reprinted  from  [106].  Copyright  1982  with permission from Elsevier (Pergamon Press Ltd). Taguchi  et al., investigated Zn UPD in 0.1 M sulfuric acid solution onto Pt (111), Pt (110), and Pt (100) surfaces. They observed no significant difference between (111) and (100) surfaces with or without the presence of Zn2+ in the electrolyte, whereas they clearly observed 49 the formation of a Zn UPD layer on the Pt (110)-(1×2) surface  [111]. This work suggests Zn preferentially deposits onto particular crystal  faces of Pt,  and therefore,  polycrystalline PtZn surfaces, such as the one prepared by Guerra, may be heterogeneous. 2.7.2.3.  Stability of PtZn – Theoretical point of view The majority of work reviewed in this section was done by Rodriguez et al., reported in 1990s. A lot of their work involved investigating metal-metal bonds in bimetallic alloy surfaces mostly by utilizing CO- thermal desorption spectroscopy (TDS), XPS, and density functional theory (DFT).  When Pt  and Zn form an alloy,  electrons  are  delocalized extensively occurs, which in turn lowers the energy of Pt in the alloy relative to the pure metal. It is also intriguing that Pt acts as an electron acceptor and Zn acts as an electron donor in the alloy formation; contrary to other Pt-transition alloys. This in turn appears to give the PtZn alloy its chemical stability which is unique compared to other Pt-transition metal alloys. This section reviews the electronic and geometric structural changes observed in both Pt and Zn due to alloying that may account for the chemical stability of the alloy, which may be very useful for the applications in fuel cells. 2.7.2.4.  Pt alloying effects Theoretical calculations have predicted several electronic perturbations upon forming a bimetallic bond between Pt and Zn, which was also, to some extent, supported by experimental data. As  mentioned  in  the  previous  section,  one  of  the  changes  due  to  alloying  Pt  with transition metals  is a decrease in Pt 5d orbital population. Molecular oxygen is then assumed to be adsorbed more readily to Pt in the alloy, resulting in the enhancement of the ORR kinetics as compared to Pt.  Prior to our work, transition metals on the right hand side of the periodic table relative  to  Pt  had  never  been  considered  as  alloying  metals  in  the  field  of  fuel  cell 50 electrocatalysis. This was probably because many predicted that electron transfer from Pt 5d to the more-filled  d-orbitals of metals, like Cu or Zn, was unlikely.  However, according to the molecular orbital theory calculations by Rodriguez,  et al., an increase in Pt 5d orbital vacancy was predicted and then further supported by experimented CO thermal desorption spectrometry [112]. The desorption temperatures for adsorbed CO from the PtZn surface were always less than that for pure Pt, indicating that CO is less tightly bonded compared to pure Pt. The ability of the metal to adsorb CO depends on its  d-band structure. Alloying Pt with Zn creates more vacancies in the Pt 5d states, and the extent of 2pi* back-donation from the metal to the CO decreases.  In  their  experimental  results,  the  strength  of  Pt-CO  bonding  was  reduced  by approximately 2~6 kcal/mol  [112].  This  result  clearly  showed  a  decrease  in  Pt  5d electron density after PtZn alloy formation. The decrease in Pt 5d electron density in PtZn is not due to a simple electron transfer from the Pt 5d to the Zn (4s,p) orbitals, but rather involves an extensive electron delocalization and Pt orbital rehybridization, which eventually moves Pt 5d electrons into the Pt-Zn bimetallic bonding  orbitals.  Rodriguez's  prediction  was  further  supported  by  Chen  et  al.  [101] The separation between Pt-Pt and Zn-Zn in their pure forms are  2.77 (fcc structure) and 2.6649 Å (hcp structure), respectively. However, cluster calculations predict that after alloying, Pt-Pt and Zn-Zn bond distances increase to 2.818 and 2.890 Å, respectively, but Pt-Zn bond distance is 2.696 Å; much shorter than either Pt-Pt or Zn-Zn bonds. This indicates that there is a substantial electron transfer from both Pt and Zn orbitals into the bimetallic bonding orbitals, which is an indication  of  the  strength  of  the  PtZn bimetallic  bond.  This  characteristic  was  observed by Despic;  the  stable  Pt-Zn alloy could  not  be  removed unless  a  high  oxidative  potential  was applied  in  an  acidic  solution.  This  may  mean  PtZn  alloys  are  very  stable  under  fuel  cell operating condition, which could contribute to an increase in PEMFC lifetime. 51 2.7.2.5.  Comparison of PtZn to other alloys 2.7.2.5.1.  Pt-transition metal alloys The stabilities of alloys are very important in the fuel cells. This statement applies to all other Pt-based alloys, such as Pt3Co and Pt3Ni, reviewed earlier  [26]. When oxygen interacts with a bimetallic surface,  generally surface segregation occurs in order for one of the alloy constituents to form a stable oxide on the surface. Zn forms oxide films very easily with a large negative heat of oxide formation; ZnO heat of formation: -83.17 Kcal/g mol as opposed to the heat  of  oxide  formation  of  e.g.,  RuO2:  -52.5  Kcal/g  mol  [113].  This  may be  some  of  the reasoning that many avoid using Zn as an admetal [114]. However, alloying effects similar to the one  just  introduced in  the  previous  section  have  also  been  observed with  popular  Pt-based alloys,  such  as  Pt3Co  [76],  Pt3Ti  [115],  PtMo  [116],  PtRu,  and  PtW  [117].  All  showed  an increase in Pt d-orbital vacancies and also showed the decrease in the desorption temperatures in CO-TDS studies. The difference between such Pt-based alloys and PtZn is that Pt acts as an electron donor when alloyed to most of the transition metals, whereas Pt acts as an electron acceptor when alloyed to Zn. This may explain the difference in the trend observed with the Pt- Pt  bond distance;  i.e.,  Pt-Pt  bond distance in PtZn runs counter  to  other  Pt-transition metal alloys; it increases in PtZn, but decreases in other Pt alloys [112]. It has been suggested that the contraction  of  the  Pt-Pt  bond  distance  is  the  key  to  the  improved  catalytic  activity  [118]. However, the elongation of the Pt-Pt bonds in PtZn is the result of shifting electrons from Pt-d orbital into the bimetallic bonding, contributing to the strength of Pt-Zn bonds. Shorter Pt-Pt bonds may thus be responsible for the instability of Pt-based alloys in fuel cells. 2.7.2.5.2.  RuZn RuZn and PtZn show a number of similarities in their effects of alloying. When Ru (001) and Zn form an alloy, similar changes in the electronic structure of both Ru and Zn have been observed  as  compared  to  the  changes  resulted  by  the  formation  of  the  Pt/Zn  alloy  [119]. According to ref. [120], the desorption temperature of Zn from the Ru surface increased by 250 52 K after alloy formation, as compared to the desorption temperature of Zn from Zn surface. This is  a  clear  indication  that  the  interaction  of  Zn-Ru is  stronger  than  the  Zn-Zn interaction  – analogous to the Zn UPD layer formation on Pt. The core orbitals of Zn, namely Zn (3d) and Zn (3s), have also shown negative shifts in the binding energies when alloyed with Ru; both orbitals showed ~0.5 eV shift towards lower binding energies  [121]. A CO-TDS study, carried out in UHV conditions, has also been reported [119, 120]. A sharp peak was observed on CO-TDS at 150 K which does not appear if CO is desorbed from Ru (001) surface. (For Ru (001), two broad peaks  appear  around  the  desorption  temperatures  of  400  and  500  K.)  The  peak  at  150  K corresponds to CO desorbed from Zn. It should be noted that pure Zn does not adsorb CO under UHV conditions  [112, 119]. Also, the broad peaks of CO desorbing from Ru shifts to lower desorption temperatures when Ru is alloyed to Zn. Also evidence of Zn-O bonding was not observed using  Zn TDS, XPS for Zn (2p1/2), and AES when oxygen and Zn (≤ 1 monolayer) are coadsorbed  [119].  The  authors  carried  out  theoretical  calculations  and  showed  that  when adatoms are present on Ru, it induces mixing of s and p orbitals in the adatoms (Zn (4s,p) or O (2s,p)), producing a hybrid orbital that are oriented towards Ru atoms. This rehybridization in turn prevents Zn-O interactions [119]. We indeed observed the same behavior on our PtZn alloy, which will be shown in Chapter 4. This  example  shows  bimetallic  bonding  may  result  in  significantly  different characteristics from those of the pure components, which is principally the result of achieving the minimum energy for the bimetallic system. Predicting such changes is not straightforward, making bimetallic alloy research very interesting and challenging. 2.8.  Summary This chapter reviewed the current understanding of fuel cell catalysis, ORR and CO oxidation reaction mechanisms on Pt and Pt-based alloys and their adsorbate-induced surface reorganization, and PtZn alloy characteristics. Fuel cells convert chemical energy into electrical energy in the MEA. The fuel cell reactions are kinetically sluggish and therefore catalysts are essential for its efficient operation. 53 The best catalyst for PEMFCs is Pt, which reduces oxygen in a direct pathway or in a series pathway.  CO oxidation occurs by adsorption of CO and water on the Pt surface. Pt-alloys are often employed to enhance the catalytic activities of fuel cell reactions and also to reduce the usage of Pt in MEA. They improve the kinetics of ORR by creating more vacancies in Pt 5d orbitals and by shortening the Pt-Pt bond distance in the alloy. CO oxidation is also enhanced for Pt alloys via a bifunctional mechanism and/or ligand effects. Chemical stability of the Pt alloys in an acidic environment of fuel cells is, however, still a serious issue. Adsorbates on metal surfaces often induce surface reorganization or segregation. The presence of adsorbate changes the surface energy, leading to restructuring of atoms to restore the lowest energy. Segregation of one component of the alloy may occur due to a thermochemical effect. PtZn alloys are prepared after the formation of a Zn UPD layer onto Pt, followed by spontaneous transformation to an alloy. The alloy also showed an increase in Pt-5d vacancy, but Pt-Pt bond distance increased. Instead, Pt-Zn bond distance is much shorter than the Pt-Pt or Zn- Zn bond, suggesting the strong bimetallic bonds. This may indicate good chemical stability in acidic conditions such as found in an operating fuel cell. 54 3.  Materials And Methods 3.1.  Introduction This chapter describes  experimental  methods and techniques employed in this  thesis. Since  fuel  cells  are  electrochemical  devices,  and  electrochemistry  is  a  surface  phenomena, voltammetry and instrumental surface analysis were the two major methods used to characterize the catalyst surface. Fundamental theories of the electrochemical double layer, electron transfer and  adsorption  isotherm  have  been  described  in Chapter  2. This  chapter  details  the experimental methods used for the experiments. Analytical instruments used in the experiments are:  Auger  electron  spectroscopy  (AES),  scanning  electron  microscopy  (SEM),  energy dispersive X-ray spectroscopy (EDX), and inductively-coupled plasma mass spectrometry (ICP- MS). The principles of operation for each of the instruments are briefly described. 3.2.  Electrochemistry In  this  section,  the  electrochemical  technique  and  general  instrumental  setup  are described. Voltammetry was extensively used. The method for the electrochemical formation of PtZn electrodes is outlined. They were made on bulk Pt substrates and used for kinetic data analysis. When polarization curves were recorded for kinetic analysis, rotating disc electrodes (RDEs) were used. The principle of the RDE operation is summarized, and the data analysis method is also presented. 55 3.2.1.  Rotating disc electrodes Rotating disc electrodes consist of a disk of metal Pt attached to a cylindrical conductive material (e.g., stainless steel) shielded in an insulating material. A schematic view of the RDE is shown in Figure 3-1. The RDE is connected to a motor and rotates at a specific speed to actively transport reactants towards the surface of the electrode. Flow of the electrolyte caused by the rotation is shown in Figure 3-2. In this figure, r is the lateral axis and z is the longitudinal axis [122].  The rotation speed is often expressed in  rotations per minute (rpm), which could be converted into the rotational frequency, ω (in sec-1): =  rpm×2 60 sec (3-1) The  thickness  of  the  stagnant  solution  near  the  surface  of  the  disc  electrode  is  called  the hydrodynamic  boundary layer  thickness,  and  when the  electrode  rotates  at  2000 rpm in  an aqueous solution, it is roughly 250 µm thick (in z direction) at 25 oC. Therefore, the current is determined  by  the  flux  at  the  electrode  surface  due  to  the  diffusion  of  reactant,  o,  to  the metal/solution interface through the boundary layer: i = nAFDo∂C o∂ z z=0 (3-2) where Do is the diffusion coefficient of species o in cm2/s, A is the area of the electrode, and Co is the concentration of o. The partial derivative of Co with respect to z is the flux of species o very close  to  the  electrode surface.  When the  current  is  limited by mass  transfer,  which will  be denoted as il, il = 0.62nFA Do 2/31/2−1 /6C o * (3-3) known as the Levich equation. The term, ν, is a kinematic viscosity (ratio between viscosity and density of a solution) [9]. 56 Figure 3-1 Schematic view of a rotating disk electrode (bottom view on the right and the cross section on the left). A disk of metal (e.g., Pt) is connected to a stainless steel  shaft  and  shielded  in  a  Teflon  or  plastic  insulating  material.  The  RDEs employed in this  thesis  had the Pt disc diameter of 3 mm (total  diameter of the electrode is 10 mm) 57 Figure  3-2 Hydrodynamics  at  the  surface  of  an  RDE.  Reprinted  in  part  with permission from [122]. Copyright 1971 Oxford University Press. 3.2.2.  Kinetic and mass transport-limited current relationship for RDE Current observed in polarization curves can be classified into two categories:  kinetic current, ik, and mass-transfer-limiting current, il. Kinetic current is the current determined by  the kinetics of the reaction at the applied potential whereas the mass transfer-limiting current is the current determined by the flux of species to the electrode. The total current is obtained by, 1 i = 1 i k 1 i l = 1 ik  1 0.62nFADo 2 /31/2−1/6Co * (3-4) 58 This equation is known as Koutecky-Levich equation. The limiting current is readily determined from experimental data, and the kinetic current is deduced by plotting i-1 vs. ω-0.5 . Rearranging the equation, kinetic current can also be determined by, ik=i il i l−i (3-5) With this equation, the kinetic current at a certain rotational speed is deduced [123]. In Chapter 4, we will use the kinetic current to construct the ORR Tafel plot. Kinetic current was calculated using in-house developed LabView program. In this calculation, the values of 1.97 × 10-5 cm2/s and 0.01014 cm2/s were used for the diffusion coefficient of oxygen in water at 25 oC and the kinematic viscosity of the 0.1 M H2SO4 solution at 20 oC [113] 3.2.3.  Cyclic voltammetry Cyclic voltammetry is an electrochemical technique where the potential of the electrode is  linearly scanned between two fixed potentials at a certain sweep rate (in V/sec) while the current is continuously monitored. This is very commonly used to characterize the redox couple of an electrode/electrolyte  system. Current  measured with respect  to the applied potential  is plotted, and this graph is called cyclic voltammogram (CV). 3.2.4.  Anodic stripping voltammetry Anodic  stripping  voltammetry  (ASV)  is  an  electrochemical  technique  involving preconcentration of substance in electrolyte by applying a reductive potential to the electrolyte, followed by stripping (or re-dissolution) of the substance by sweeping the potential to the anodic (oxidative) direction. The scheme of anodic stripping voltammetry is shown in  Figure 3-3. In the figure, ip denotes the peak current where the preconcentrated substances are oxidized and Ep denotes the corresponding potential. Ep could be predicted by the Nernst equation if the redox couple  is  reversible,  and  this  potential  is  unique  to  each  substance.  It  is  also  possible  to 59 determine  the  concentration  of  the  corresponding  substance  in  solution  by  constructing  a calibration curve. In this work, ASV will be used in Chapter 5 in order to investigate the carbon monoxide (CO) oxidation catalysis. Ep of CO stripping voltammetry is greatly affected by the type of catalyst (electrode material), and the characteristics of the surface, thereby the efficiency of the catalyst is compared by measuring the peak potential, Ep [9]. In Chapter 6 ASV was also used to determine the concentration of Zn in the solution where prolonged polarization of PtZn was performed.  3.2.5.  Differential pulse voltammetry The  differential  pulse  method  is  commonly  used  with  anodic  stripping  voltammetry where a trace amount of analyte metal ions is present in the test solution (typically sub-micro molar level). The potential profile with respect to time is shown in  Figure 3-4.  A deposition potential is applied to the working electrode, and after the species of interest is concentrated on the surface of the electrode, the potential pulse (modulation amplitude) is applied as shown in the diagram. The current is recorded at S1 and S2. The difference between the two is recorded as a  function  of  the  average  applied  potential.  If  both  potentials  are  far  from  the  reversible potential, ENernst, the currents measured are similar, giving the difference in measured current at S1 and S2 approximately zero. Only when the potential approaches ENernst, the difference between the  faradaic  current  at  the  two  points  becomes  significant  which  results  in  a  peak  in  the voltammogram ([i(S2)-i(S1)] vs. applied potential). This technique is more sensitive and useful in trace elemental analysis compared to linear sweep voltammetry [9]. The differential pulse method, together with ASV, is used in Chapter 6 to determine the concentration of Zn in an acidic solution. 60 Figure 3-3 Potential and current profiles with respect to time in anodic stripping voltammetry. Analyte in a solution is first preconcentrated at a deposition potential, followed by stripping (oxidation) of the deposited film by scanning the potential in the anodic direction, resulting in a peak in a voltammogram. The peak potential, Ep can be calculated by the Nernst equation, and the current, ip is measured. Because of the  preconcentration  step,  the  sensitivity  of  detection  is  enhanced,  making  this method suitable for trace metal analysis (< 1 µM). Figure 3-4 Potential profile with respect to the time in differential pulse method. Potential is applied to the electrode for a certain period (“interval time”; typically 0.5~4 seconds), a potential step (“modulation amplitude”; typically 10~100 mV) is applied to the electrode and fixed at the stepped potential for some period of time (“modulation time”; typically 5-100 milliseconds). Current is measured just before the step  and just  before  the potential  is  stepped back to  the base (at  S1 and S2, respectively), and the difference is plotted as a function of potential. 61 3.2.6.  Electrochemistry – experimental consensus Electrochemistry  generally  requires  stringent  experimental  conditions.  This  section explains  the  general  aspects  of  electrochemical  setups  that  apply  to  all  electrochemical experiments  performed  in  this  thesis.  Cleanliness  of  the  experimental  systems  was  very important,  and  ensuring  no  organic  or  inorganic  contaminants  was  crucial  for  meaningful results. Glassware used in all the experiments throughout this thesis were first soaked in a warm 1:1 (by volume) concentrated sulfuric acid and nitric acid for at  least  2 hours,  followed by rigorous rinsing with Milli-Q water (Millipore, Billerica, MA). The Milli-Q system uses reverse osmosis and ion exchange cartridges to deionize the water, followed by UV light exposure to photolyze organic contaminants, providing clean water with high resistivity (>18.2 MΩ cm), low total  oxidizable  carbon (<  1  ppb),  and  pH of  5.  The  quality  of  water  was  ensured  by following the set schedule for replacing filters provided by Millipore. There are mainly two types of electrolyte used in this thesis: 0.1 M sulfuric acid (SeaStar Baseline® 93-98% purity with impurities in ppt levels) and 0.1 M zinc sulfate solution (Fluka, puriss p.a. ACS grade, low Cl), both made with Milli-Q water. The reference electrode (RE) was a home-made reversible hydrogen electrode (RHE; 0 V vs. standard hydrogen electrode, SHE), unless otherwise stated.  Potentials reported in this thesis are measured with respect to RHE. A drawing of the RHE is shown in Figure 3-5. A small compartment at the bottom of the electrode contains a coiled Pt wire which is soldered to an extended tungsten wire. The compartment is filled  with  an  acidic  solution  from the  opening  at  the  tip  of  the  electrode.  The  tip  is  then immersed into an acid solution together with a gold wire. The electrode and the wire can then be connected to a 9 V alkaline battery to evolve hydrogen at the Pt wire in the compartment to create  a bubble of hydrogen gas,  which establishes equilibrium between protons and H2 (g) around  the  Pt  wire,  acting  essentially  like  a  standard  hydrogen  electrode.  The  top  of  the compartment is well-sealed to avoid a hydrogen gas leakage. Hydrogen gas bubble was freshly formed at the beginning of every experiment, but the electrode was usually stable for 2 to 3 days once  generated.  RHE was  connected  to  an  electrochemical  cell  through  a  salt  bridge.  The counter electrode (CE) was a coiled Pt wire, cleaned by heating in a butane-air flame for 1 minute followed by rinsing with water. The working electrode (WE) was either an RDE or a Pt bead. A diagram of the electrochemical cell is shown in Figure 3-6. The RDEs employed in this 62 work  had  the  disk  diameter  of  3  mm (geometrical  area  of  0.071  cm2),  with  the  electrode diameter of 10 mm. The RDEs were cleaned in a mixture of diluted HNO3 and 30 % H2O2, and the Pt bead was cleaned by heating it in a butane-air flame, followed by rinsing with Milli-Q water. Approximately 70 to 80 mL of electrolyte was placed in an electrochemical cell, and the solution was first deaerated using Ar gas (99.9995 % purity; Praxair). Solution cleanliness in the sulfuric acid solution was assessed by recording a CV with a Pt electrode between potentials of 0.03 and 1.3 V at sweep rate of 50 mV/s, which is shown in Figure 3-7. Hereafter, CVs recorded in this manner is designated as “Ar saturated  CV” or CV, unless otherwise stated. Hydrogen adsorption and desorption peaks (the potential region of 0~0.3 V) are particularly sensitive to the presence of contaminants. CVs would appear distorted or show diminished or extra peaks if they are present. The characteristic shape of a clean Pt CV was always ensured before proceeding, otherwise the cell was brought back to the acid bath for a complete cleaning. 63 Figure 3-5 Home-made reversible hydrogen electrode (RHE). A small compartment at the bottom of the electrode contains a coiled Pt wire which is connected to the extended tungsten wire. The compartment is filled with an acid solution from the opening at the tip of the electrode. Hydrogen gas is evolved in the compartment, resulting in the Pt wire in equilibrium with hydrogen gas and protons in solution. This acts as a hydrogen electrode. 64 Figure 3-6 Schematic view of electrochemical cell. The counter electrode (CE) was a  coiled  Pt  wire  and  the  reference  electrode  (RE)  was  typically  the  RHE.  The working electrode (WE) and other electrodes were connected to potentiostat which was computer controlled. 65 Figure 3-7 A cyclic voltammogram of polycrystalline Pt recorded in 0.1 M H2SO4 deaerated solution at the scan rate of 50 mV/s is presented. Integration of hydrogen adsorption or desorption peaks give approximately 210  µC/cm2 of charge density. This is used as a benchmark CV recorded in a clean H2SO4 solution. 3.2.7.  PtZn alloy formation PtZn  electrodes  were  prepared  by  electrochemically  depositing  bulk  Zn  onto interchangeable Pt RDE or Pt bead electrodes. As mentioned in Chapter 2, Zn-modified Pt can quickly transform into a PtZn alloy at a room temperature without a rigorous treatment. The formation of PtZn on a  bulk Pt substrate  (i.e.,  a Pt rotating disc  or  a  Pt bead electrode) is described in this section. 66 3.2.7.1.  Zn deposition and PtZn alloy formation The Zn electrodeposition (0.1 M ZnSO4·7H2O) was carried out in an electrochemical cell on a Pt RDE (Metrohm, Switzerland) or on Pt bead electrodes (99.99 % GoodFellow). The Pt beads were prepared by melting one end of a Pt rod with an acetylene torch, then polishing, starting with 1200 grit sandpaper, then successive grades of diamond slurry ending with 1 µm diamond  slurry  (Leco,  MI).  The  electrodes  were  connected  to  a  potentiostat  (µAutolab, EcoChemie), and electrochemical measurements were performed at room temperature (22 ± 2 oC). Initially, a CV of the Pt electrode in the ZnSO4 electrolyte was recorded and the bulk Zn desorption  potential  was  determined  (i.e.,  the  equilibrium  potential  for  Zn  reduction).  The potential of the Pt electrode was set 50 mV negative of the onset of the Zn deposition, and bulk Zn was deposited, during which the current transient was recorded. The electrode was removed from the electrolyte after 300 seconds of Zn deposition, and the Pt/Zn alloy was allowed to form under Ar atmosphere at room temperature for 1 hour. Excess (nonalloyed) Zn was stripped off the Pt electrode by immersing the electrode back into the Zn electrolyte and polarizing to a potential positive of the Zn stripping potential measured in the initial CV. The electrode was then dipped into a diluted HNO3/H2SO4 solution to ensure all unalloyed Zn (in either metallic or hydroxide form) was removed. However, when the amount of charge passed during the stripping process was not measured, unalloyed Zn was simply removed by immersing into a concentrated 1:1 acid mixture of HNO3 and H2SO4. An Ar saturated CV for PtZn was then recorded between 0.03 V and 1.0 V at 50 mV/s and compared to that for Pt. The kinetic current analysis was then performed and compared to the pure Pt. It is important to note that the PtZn electrode was used as prepared without polishing the surface as the alloy formed is in the surface region, therefore polishing would likely remove it. However,  the  deposition  of  Zn  and  subsequent  alloy  formation  substantially  roughens  the surface of the Pt substrate. As shown in Chapter 2, roughness of the catalyst surface influences the measured activity. Surface roughness of the both the Pt and PtZn electrode was determined for comparison of the electrochemical activity in Chapter 5. 67 3.3.  Analytical instrumentation Several  analytical  instruments  were  used  for  surface  characterization  of  selected electrode  surfaces  or  for  the  determination  of  trace  metal  concentration  in  a  solution  to supplement the electrochemical observations. Surface and/or elemental analysis instrumentation used were Auger electron spectroscopy, scanning electron microscopy, and energy dispersive X- ray spectroscopy. Trace metal analysis was performed on inductively coupled plasma – mass spectroscopy. The principle of operation for each instrument is briefly described in this section. 3.3.1.  Auger electron spectroscopy 3.3.1.1.  Operation principles Auger electron spectroscopy (AES) is a powerful surface analytical technique based on the Auger process discovered by Pierre Auger in 1923  [124]. It probes the first 2~10 atomic layers (i.e., up to 10 Å) of conducting samples in a ultra high vacuum (UHV; typical pressure of 10-9  ~ 10-11 mbar) environment. The sampling depth of the Auger process into the surface is depicted  in  Figure  3-8 [124].  AES  is  used  in  this  thesis  to  identify  the  surface  chemical composition of electrode surfaces. The Auger process is shown schematically in Figure 3-9. The core electron of an atom is first ejected by the incoming electron beam (Process 1; the typical energies of the incident beam is 3~30 keV), which creates a hole in the inner orbital of energy, EW.  This hole is  filled by an electron in an upper orbital  with energy,  EX.  This process can transfer the energy to valence electrons, whose energy is EY, which can then be emitted from the atom. This electron is called a secondary electron or an Auger electron. The kinetic energy of the Auger electron (generally from 50~3000 eV) is therefore determined by, EWXY = EW – EX – EY - Φe (3-6) where Φe is the work function of the sampled atoms. The energies of EX and EY may be a little 68 more negative than their original values as removing electrons takes place after a core electron has been ejected. Since EWXY measured is unique to each element, an Auger spectrum provides a fingerprint of atoms present on the sample surface. Quantitative  analysis  of  relative  atomic  composition  of  surfaces  is  also  possible  by integrating the Auger spectral peaks. The intensity contribution (I) for Auger electrons emitted from the sample is, I = m f  Aeject T inst  (3-7) where m is the number of atoms per unit volume of the sample,  f is the flux of the incident electrons, σ is the cross section of the Auger process, λ is the inelastic mean free path (i.e., the distance an electron can travel without losing its energy),  Aeject is the area of the sample where the electrons are ejected, and Tinst is the instrumental constant describing transmission efficiency of  the  electron  analyzer.  Contributions  from  f,   σ,   Aeject,  and  Tinst can  be  grouped  into  an instrumental coefficient, B, for given instrumental conditions and a given peak. This term is called a sensitivity factor. Atomic ratio for two elements in the sample (i.e., m1/m2) is determined by, m1 m2 = [  I 1/B1 I 2/B2 ]21   (3-8) The ratio of the inelastic mean free paths is often assumed to be unity, therefore the atomic ratio of two elements can be estimated by the intensities of Auger peaks and the sensitivity factors [125]. Elemental sensitivity factors are often measured relative to clean Ag, but the method often introduces  instrumental  uncertainties  to  the  measured  quantities [124].  However,  we  were fortunate to measure our own sensitivity factor for Zn relative to Pt (BZn=3.625, BPt=1). We obtained clean Pt and Zn metals present on the same sample substrate.  This will be detailed in Chapter 4. 69 Figure  3-8 Distribution  of  electrons  in  various  analytical  techniques.  Auger electrons, backscattered electrons, X-ray (fluorescence in EDX) were analyzed in this thesis. Adapted  from  [124]. Incident electrons (beam diameter of ~ 10 nm) impinge on the solid surface and penetrate causing various types of excitation of the atoms in the sample. Typical path lengths for the excitation phenomena are shown in the figure. 70 Figure 3-9 Auger electron transition. Incident beam hits the core electron in Ew in (1 and 2) followed by the relaxation of an electron in the upper orbital, Ex (3). This relaxation provides enough energy to the electron in EY to be ejected as an Auger electron. Φe is the work function and EWXY is the energy of Auger electron, which is typically 50~3000 eV. 71 3.3.1.2.  Nomenclature of Auger transition Auger transitions and the corresponding peaks in the spectrum are expressed as,  e.g., “KLL”  or  “LM2”.  These  notations  are  often  used  in  X-ray  spectroscopy,  and  each  letter corresponds to an electronic state of the atom. The first of three letters in the former example denotes the core level, the middle one corresponds to the orbital from where an electron falls to fill the core hole, and the last describes the orbital from where the Auger electron is emitted. The notations are correlated to quantum numbers, i.e., principal quantum number, n, orbital, l, spin, s, and spin-orbit coupling, j. Examples of j-j coupling notations in X-ray spectroscopy is given in  Table 2-1 [126].  The middle energy level is  often dropped, and the orbital  where Auger electron is ejected is specified (the latter example). Table 2-1 Auger notations. Different electronic states are expressed in terms of their quantum numbers, n, l, and j [126]. Quantum numbers n l j Suffix X-ray level Spectroscopic level 1 0 1/2 1 K 1s1/2 2 0 1/2 1 L1 2s1/2 2 1 1/2 2 L2 2p1/2 2 1 3/2 3 L3 2p3/2 3 0 1/2 1 M1 3s1/2 3 1 1/2 2 M2 3p1/2 3 1 3/2 3 M3 3p3/2 3 2 3/2 4 M4 3d3/2 3 2 5/2 5 M5 3d5/2 4 0 1/2 1 N1 4s1/2 4 1 1/2 2 N2 4p1/2 ...etc. 72 3.3.2.  Secondary (Scanning) electron microscopy 3.3.2.1.  Scanning electron microscopy (SEM) Often surface analytical  instruments  are  equipped with a  secondary electron detector which  detects  the  electrons  with  kinetic  energy less  than  50  eV.  Because  these  low-energy electrons emerge from the surface, they are used to form a micrograph (known as secondary or scanning electron micrograph; SEM) to give morphological images of the sample. Secondary electrons are collected in the photomultiplier detector as the incident electron beam scans the sample surface. The brightness of the image depends on the nature of the atoms on the surface; if the sample is grounded properly, a steeper surface gives brighter contrast than a flat surface. Also, heavier atoms emit secondary electrons more readily than lighter ones. The advantage of having SEM together with AES or EDX is the ability of obtaining Auger spectra from regions of different morphologies. This technique was particularly useful as shown in Chapter 4 when we found three distinguishable regions on our Zn-modified Pt electrode [127]. 3.3.2.2.  Backscattered electron (BSE) imaging When  an  incident  electron  beam  hits  a  sample  surface,  some  of  electrons  are backscattered without losing significant amount of energy. The number of such electrons varies with atomic number of the atoms near the surface, as well as topography. The penetration depth of backscattered electrons is much deeper relative to the Auger electrons, according to Figure 3- 8. Heavier atoms on the surface backscatters more electrons, hence the resultant image appears brighter, and vice versa. Backscattered electrons travel with high velocities to the four quadrants (solid-state silicon-based detectors) that are located above the sample. Electrons collected at two quadrant pairs, A and B, can either be processed as an A+B mode or an A-B mode. The former sums all signals collected in A and B to generate an image based on the atomic number contrast present  on  the  surface.  The  latter  mode  subtracts  the  signals  in  B  from  A to  generate  a topographical contrast of the surface where contour features are better enhanced [127]. 73 BSE was  useful  in  Chapter 5 where  the  distribution  of  Zn in  the  Pt  substrate  was examined. The presented image was recorded using the A+B mode. AES  used  in  this  thesis  was  Microlab  350  (Thermo  Electron  Co.),  equipped  with secondary  electron  detector,  concentric  hemispherical  analyzer  (separate  electrons  by  their kinetic energies), and a backscattered electron detector. All Auger spectra, some SEM images, and all BSE images presented in this thesis were recorded in this instrument. Incoming electron beam was 10 keV with the primary electron beam diameter less than 10 nm. The detection limit of the instrument is 0.2 atomic %. Resolution factor (constant retardation rate) of 4 was used, unless otherwise stated.  This instrument was operated by Dr. Philip Wong in Department of Chemistry, University of British Columbia. 3.3.3.  Energy dispersive X-ray spectroscopy When the process (3) in  Figure 3-9 occurs, atoms can also lose energy by emitting a photon in a process called X-ray fluorescence, and observing these photons provides the basis of another chemical analysis technique, known as energy dispersive X-ray spectroscopy (EDX). Because the inelastic mean free path (i.e., the distance electrons or photons can travel before losing their energies) is relatively long (typically ~0.5 µm), EDX is used for the bulk elemental analysis. Auger electron emission and X-ray fluorescence are competing processes; the former dominates when the initial vacancy created by the incoming electron beam has a binding energy of less than 2000 eV, which is often the case for lighter atoms. The latter dominates otherwise. We also used EDX for PtZn compositional analysis. Quantitative analysis is also possible. This is done by determining the area under the spectral peaks after correcting for a non-linear background. Compositions of elements in the analyzed metal were reported relative to one another [128].  EDX employed in this thesis was a S3000N SEM (Hitachi; available in Department of 74 Materials Engineering at University of British Columbia). The instrument was equipped with a tungsten hairpin source that produces electrons by thermoionic emission. Incident electron beam was 20 keV. All EDX spectra and some SEM images were recorded on this instrument. The detection limit of the instrument is 0.1 weight %. 3.3.4.  Inductively coupled plasma – Mass spectrometry Inductively-coupled plasma (ICP) is a commonly used ionization source, generated by passing noble gases, typically Ar, through a high frequency electromagnetic field. When a spark is  introduced to  the  stream of  Ar  gas,  electrons  are  stripped from some of  the  atoms.  The electrons collide with other Ar atoms producing further ionization. The energetic collisions are also responsible for the high temperature of  the plasma (6000 ~ 10,000 K).  The sample solution is injected into the flow of plasma where it can be desolvated, vaporized, atomized, and partially ionized. Charged elements are readily analyzed in a mass spectrometer with the detection limit of sub-ppb level.  The first ionization energies of Pt and Zn are 9.0 and 9.391 eV, respectively [113] which can be partially (40~60 %) achieved by Ar plasma [129, 130]. In Chapter 6, we created PtZn nanoparticles on a glassy carbon substrate. To determine the bulk composition of the particles, we dissolved them into aqua regia (1:3 HNO3 : HCl), and the solution was analyzed in the ICP – mass spectrometry (ICP-MS) available in the Department of Chemistry undergraduate physical/analytical teaching lab at University of British Columbia (Elan model 6000; mass range of 3-270 amu with resolution of 0.05 amu). The stoichiometric ratio between Zn to Pt was obtained. The instrument employed in this thesis consists of a quartz ICP torch with three-stage pumping system coupled to a quadrupole mass analyzer and electron multiplier tube as a detector. The instrument is controlled by the commercial software. 75 3.4.  Single fuel cell The PtZn alloy was tested both in an electrochemical setup and in an actual fuel cell setup in Chapter 4. PtZn membrane electrode assembly (MEA) was created just as PtZn RDEs were formed; by electrochemical Zn deposition over Pt-dispersed MEA. This is a rather unique method to  create  Pt-alloy MEAs and is  also readily performed in a laboratory.  This section describes the experimental method of preparing a PtZn MEA and hot-pressing the MEA into a single stack fuel cell. 3.4.1.  MEA fabrication The MEA consists of solid electrolyte (50~175 µm) in the middle, sandwiched by two catalyst layers (5~50 µm) which are applied onto gas diffusion layer (GDL) backings (100~300 µm) [2]. Catalyst layers usually consist of nanoparticles of catalyst, typically Pt, supported on carbon particles, which are often denoted as Pt/C. Schematics of the MEA are shown in Figure 3-10. At both ends of the GDLs, flow field plates are attached to complete the unit. Flow field plates have numerous grooves, called flow channels, where gases (either a fuel or an oxidant) flow and reach to the electrode surfaces. The solid electrolyte and catalyst layers and/or catalyst layers with the GDLs are often hot pressed at elevated temperature and pressure (typical values of  120-160  oC  at  5000~15000  kPa,  respectively  [131]).  Hot  pressing  is  another  important process that impacts fuel cell performance as forming poor bonds between the catalyst layer and a membrane electrolyte or GDL backings lead to high contact resistance which would act to increase any type of polarization. In Chapter 4, we describe the preparation of an MEA by hot pressing commercially available carbon supported Pt already attached to a GDL with NafionTM membrane.  Although  MEA fabrication  methods  are  not  the  focus  of  this  thesis,  different preparation modes of MEA significantly alters the performance of PEMFCs [132, 131]. In that chapter, we used NafionTM solution sprayed between GDL and electrolyte membrane before hot pressing  to  ensure  good bonding  and  ionic  conductivity.  This  would  create  a  thin  layer  of membrane over Pt/C which has shown better performance than  e.g., Pt/C directly attached to 76 membrane without the coating layer [2]. The cross-section of the MEA thus fabricated is shown in Figure 3-11 [133]. In  Chapter  4,  the  MEAs  were  prepared  with  and  without  the  Zn  treatment  of  the dispersed Pt cathode. The gas diffusion cathode consisted of a 5 cm2 ELAT carbon cloth backing layer (Electrochem Inc.) with 1 mg cm-2 Pt supported on Vulcan XC-72 (i.e., 20 wt % Pt on the carbon support).  The Zn treatment of the gas diffusion cathode was similar to the procedure employed for the Pt RDE described previously.  Zinc modification of the cathode layer  was accomplished by immersion of the cathode in 0.1 M ZnSO4 in the main compartment of a two- compartment  glass  cell  along  with  the  stainless  steel  counter  electrode  and  a  Hg/Hg2SO4 reference electrode (-0.64 V vs. RHE in saturated KCl). A CV was performed measuring the Zn deposition and stripping currents. The picture showing the electrochemical Zn deposition on the GDL is shown in Figure 3-12. The potential of the gas diffusion cathode was set to -0.95 V vs. RHE for 5 min and the current transient recorded. The cell was turned off and the alloy was allowed to form for 20 min. The cell potential was then set to -0.3 V and the bulk Zn was stripped off as indicated by the anodic current recorded. The modified cathode was rinsed with Milli-Q water and was stored in air until the MEA was pressed. 77 Figure 3-10 The components of MEA are shown. Solid electrolyte is sandwiched between two catalyst layers that are attached to GDLs. These units are hot pressed to form a MEA. It is further sandwiched by field flow plates which are responsible for delivering reactant gases to the catalyst layers. 78 Figure 3-11 Cross section of solid electrolyte, catalyst layer, and GDL hot pressed together with Nafion™ bonding (“proton conducting media” in the figure). Better ionic conduction and electron transfer are achieved by having the conducting media supporting the catalyst layer. Cross section of solid electrolyte, catalyst layer, and GDL hot pressed together with Nafion™ bonding (“proton conducting media” in the figure).  Better  ionic conduction and electron transfer are achieved by having the conducting  media supporting the catalyst  layer.  Reprinted  from  [133].  Copyright 2004 with permission from Elsevier Ltd. 79 Figure  3-12 Electrochemical  setup  to  perform Zn  modification  onto  a  catalyst- attached  GDL.  GDL was  connected  to  potentiostat  and  immersed  into  a  ZnSO4 solution. The counter electrode was stainless steel mesh and the reference electrode was Hg/Hg2SO4. Following the  Zn modification  of  the  cathode,  the  full  MEA was  assembled  by hot pressing the gas diffusion cathode to the NafionTM 115 membrane (127 µm dry thickness). The anode, consists of 5 cm2 ELAT carbon cloth with 1 mg cm-2 Pt supported on Vulcan XC-72, was commercially available with a pre-pressed membrane on one side (Electrochem Inc.). To ensure good bonding between the cathode and the membrane and to impart  ionic conductivity to the cathode catalyst layer, a few drops of NafionTM solution (5 wt % in lower alcohols,  Aldrich  Inc.)  were  uniformly  dispersed  on  the  side  of  the  cathode  facing  the membrane, giving a cathode NafionTM load of about 1 mg cm-2. The hot pressing of the cathode to the half-MEA was performed at 115 oC, under a weight of 800 lbs applied for 60 sec. Single cells were assembled using Au-coated end plates (9 cm × 9 cm) with serpentine 80 flow field pattern (Electrochem Inc.). Fuel cell tests were carried out employing the Greenlight test station at the Institute for Fuel Cell Innovation of the National Research Council of Canada, located in Vancouver, BC. 3.5.  Summary This chapter describes the method and materials used to carry out the experiments and to process  the  data  thus  acquired.  Electrochemistry  of  RDEs  was  reviewed.  Most  of  the electrochemical work presented in this thesis depends on voltammetry, and the techniques used were described. Also the method to form PtZn electrodes are presented. Bulk Zn layers were deposited over Pt electrode surface, and the alloy was spontaneously formed. Analytical instruments, namely AES, SEM, EDX, and ICP-MS, were also extensively used for elemental analysis and the quantification in order to supplement the electrochemical data. The principle of operation in each technique was also presented. Finally, the method to create a single-stack fuel cell was described. PtZn MEAs were formed in the same manner as PtZn electrodes were prepared. The resultant MEAs are hot- pressed into a complete fuel cell unit. 81 4.  Oxygen Reduction Reactions on Pt and PtZn 4.1.  Introduction The  oxygen  reduction  reaction  (ORR)  occurs  on  the  cathode  in  a  proton  exchange membrane fuel cell (PEMFC). Its sluggish kinetics is the source of a large fraction of the loss of power in fuel cell operation. Significant research efforts are actively pursuing a better catalyst for the ORR, or trying to reduce the amount of Pt catalyst needed for fuel cell operation. The biggest challenges that face the fuel cell industry are to increase the activity per mass of the dispersed Pt in current state of the art fuel cells by an order of magnitude through improved utilization of the dispersed Pt, or by improving the kinetics of the ORR. Increasing the kinetic rate of ORR is observed through alloying of the Pt metal with various transition metals, which borrows from the experience with alkaline fuel cells. The modification of the Pt catalyst through alloying has met with varying degrees of success in terms of decreasing the ORR overpotential as introduced in Chapter 2. Many transition metals have been alloyed with Pt, and the kinetics of the ORR has been measured using the rotating disk electrode (RDE) method or through the comparison of fuel cell polarization curves. As pointed out in Gasteiger's review [97], a lack of standard testing procedures has created difficulties in comparing various published data. It was suggested that all the data be compared using a set benchmark. The authors suggest that a good practice for all  measurements (with an RDE and a fuel cell)  is  to report  the kinetic current measured at 0.9 V vs. reversible hydrogen electrode (RHE), albeit biased by a requirement for a high cell voltage, making this benchmark appropriate for comparing Pt-based catalysts. In general, the various Pt alloy catalysts (e.g., alloyed with Fe, Co, Cr, Ni, Mn, Ir, Ti, or combinations thereof (reviewed in ref. [134]) have been observed to decrease the overpotential for the ORR by 20 to 40 mV. The creation of a bimetallic surface that is more reactive toward ORR has  been  explained  through changes  in  the  electronic  nature  of  the  surface,  inhibited adsorption of species that block the surface, or changes in the structure of the surface atoms. As summarized in Gasteiger's review, [97] these possible explanations have not been conclusively 82 proven. Fundamental studies using an RDE is a method that allows for precise modification of the Pt metal as well as accurate area determination. Of the various combinations, PtCr, PtCo, and PtNi are the best performing of the alloys. This group displays a 3-4-fold increase in the ORR rate (measured at 0.8 V vs. RHE), and the alloys have been tested in fuel cells with similar results. The electrochemical behavior of these alloys was very similar to that of pure Pt. The current understanding is that the Pt atoms segregate to the surface through exchange with the Co and form a Pt skin over the surface. The modification of the electronic character of this "Pt skin" due  to  the underlying transition  metals  resulted  in  the  increased  ORR kinetics.  In  addition, dilution of the Pt with the alloying element increases the mass activity of the Pt. As Gasteiger points out, the application of bulk alloy results to a high area catalyst environment is difficult, so fuel cell testing is an important component of the evaluation process. Also important,  but not  explicitly stated in the review, is  the ease of creation of the modified catalyst. In many cases, the alloy is created through high temperature alloying and the fine  dispersion  needed can  be  difficult  to  realize  [135].  Another  issue  is  the  change in  the hydrophilic nature of the catalyst particles, making them more susceptible to flooding due to the increased hydroxide content on the surface. It is certainly true that the move toward Pt alloys as the catalyst for ORR is necessary, but many issues still need to be resolved. Prior to the start of our project, Zn modified Pt alloy had never been investigated as a candidate for ORR in literature,  presumably because people have predicted  electron transfer from Pt 5d  to Zn 3d is not possible as the Zn 3d orbital is already full. Also, Fernandez et al., [114] has considered Zn as a possible alloying material for Pt, but concluded Zn-alloy is not favorable for catalyzing ORR as Zn would form stable Zn-O bonds rather than aid breaking O-O bonds.  However,  Rodriguez's  PtZn  work  described  in  Chapter  2 contradicts  with  these predictions. In this chapter, we will describe the characterization of an electrodeposited layer of Zn on  Pt  and  the  resulting  surface  alloy  formation  as  demonstrated  by  surface  analytical measurements. The analysis was performed by Dr. Ed Guerra and Mr. Yanguo Yang (under the supervision of Dr. Dan Bizzotto). Through this analytical effort, the realization of the ease which the Zn/Pt alloy interacts with oxygen led us to investigate its catalytic properties for ORR. The ORR electrochemistry for Pt and the electrochemistry for Zn-modified Pt surfaces are explored 83 using RDE electrochemical measurements. The method for creating the Zn-modified Pt surface and the resulting improvement in the ORR overpotential will be described, and comments will be made on the stability of the Zn/Pt alloy. The results from these fundamental studies are used in the creation of a modified cathode gas diffusion electrode (GDE), which is tested in a single fuel  cell.  This  section  was  performed  by  Mr.  Winton  Li  in  the  Department  of  Chemical Engineering (under supervision of Dr. Elod Gyenge and Dr. Dan Bizzotto) as a 4th year project. Comparison  of  the  polarization  curves  for  a  single  membrane  electrode  assembly  (MEA) composed of either a Pt or a Zn-modified Pt cathode GDE will be presented. 4.2.  Experimental section 4.2.1.  General electrochemistry and the PtZn alloy preparation Electrochemical setups, as well as Zn deposition and subsequent alloy formation was performed in a same manner described in  Chapter 3. Briefly, Zn was electrodeposited on Pt RDE from a ZnSO4 solution, followed by a spontaneous alloy formation at the room temperature under Ar atmosphere. Unalloyed Zn was stripped off electrochemically and the electrode surface was  washed  with  a  dilute  H2SO4 acid  solution.  Potentials  reported  are  referenced  to  the reversible hydrogen electrode (RHE). 4.2.2.  Oxygen reduction reaction and the data processing After the solution cleanliness was assessed using the Pt Ar saturated CV, the electrolyte (0.1 M H2SO4) was saturated with oxygen (Praxair) and the ORR measurements were performed using a smooth Pt and PtZn electrodes at 22  oC. While the ORR CVs were recorded, oxygen flow  was  kept  on  the  electrolyte  surface.  The  kinetic  current  was  calculated  using  two approaches. The anodic and cathodic polarization curves were recorded at 10 mV/s with rotation 84 speeds varying from 500 to 3000 rpm. The potential range was restricted to 0~1.00 V. The CV for 2000 rpm rotation rate was used to calculate the kinetic current using  Equation  3-5. This data analysis was limited to potentials where the measured current was less than half the limiting current, typically above 0.8 V. The Koutecky-Levich (KL) analysis (Equation  3-4) was also performed using the data obtained at all rotation frequencies. In-house built LabView software was  used  to  perform the  calculation.  The  program first  reads  the  ORR curves  recorded  at different frequencies on a particular electrode. The intercept of the KL plot (j-1 vs. ω-1/2) over a restricted potential range was used to calculate the kinetic region, which leads into the Tafel analysis (E vs. log ik). Regression analysis on the Tafel plot is also the part of the program, and the potential region can be manually controlled to optimize for the R2-value as close to 1 as possible. These kinetic currents were plotted as a function of potential in the usual Tafel plot. 4.2.3.  Auger electron spectroscopy on Zn modified electrode Auger electron spectrometry (AES) analysis was carried out in a MICROLAB 350 Auger electron spectrometer (ThermoElectron Corp.) with a field emission electron gun (10 keV, 3 nA) under  vacuum  (~10-10 mbar).  This  instrument  also  allowed  simultaneous  measurements  of scanning  electron  microscopy (SEM)  images.  Survey Auger  electron  spectra  were  acquired using the hemispherical energy analyzer for a constant resolution of 4, while the higher energy resolution  spectra,  to  show  Auger  fine  structure,  were  measured  with  the  resolution  at  40 (corresponding to approximately 0.5% energy resolution). 4.2.4.  Membrane electrode assembly preparation and testing The preparation method of the PtZn or Pt MEA is detailed in  Chapter 3. All fuel cell experiments were performed by Mr. Winton Li. Briefly, the Zn modification onto the 20 wt% Pt/ C (5 cm2) was performed electrochemically, using a 0.1 M ZnSO4 solution. Zn was deposited to Pt/C at -0.95 V for 5 minutes, followed by the 20 minutes of a spontaneous alloy formation and 85 the stripping of unreacted in the ZnSO4 solution at -0.3 V. The full MEA was then assembled by hot pressing the Zn-modified gas diffusion layer (GDL) with a NafionTM membrane at 115 oC under a weight of 800 lbs for 60 seconds. Single cells were then assembled using Au-coated end plates (9 cm  × 9 cm) with serpentine flow field pattern (Electrochem Inc.). Typical fuel cell operating parameters were as follows: cell temperature 74  oC, anode and cathode gas stream humidity characterized by a dew point of 75 oC, hydrogen and oxygen flow rates 500 and 400 mL min-1 respectively,  and stoichiometric coefficients  equal to 1.1 for hydrogen and 1.2 for oxygen. The absolute pressure on the cathode side was kept constant at 350 kPa, while the anode side pressure was either 350 or 101 kPa. Before recording the  polarization  curves,  all  MEAs were conditioned in  an identical fashion  ensuring  proper  and stable  membrane  hydration.  Conditioning  was performed using three different constant-voltage steps, starting with 0.5 V for 3-4 h, until the superficial current density reached about 400 mA cm-2, followed by operation at 0.3 V and finally 0.1 V for 1 h each. At the end of the conditioning stage, the polarization curve was recorded by applying successive constant-current steps between 0.1 and 3 A for 5 min duration at each step. 4.3.  Results and discussion The  Zn-modified  Pt  electrode  was  prepared  electrochemically  relying  on  previously reported work dealing with the underpotential deposition (UPD) of Zn onto Pt for the study of the  kinetics  of  Zn  electrodeposition  in  highly  acidic  electrolyte  [107].  In  that  paper,  we confirmed the creation of a UPD Zn layer on Pt and its eventual transformation into an alloy originally reported in ref [110]. The deposition of a monolayer of Zn onto the Pt electrode was recognized through its influence on the electrochemical response where the overpotential for H2 evolution increased, a new peak was generated in the double-layer region, and the CV showed a change in the Pt oxide region. These changes were all thought to be due to the presence of Zn. The  existence  of  a  PtZn  alloy  was  indirectly  shown through  electrochemical  investigations which agreed with other reports in the literature [106]. It was important to confirm the presence of Zn on the Pt electrode and if possible to show the presence of the PtZn alloy. Initial surface 86 analytical  measurements  were  performed  on  a  Pt  wire  since  introduction  of  the  polymer sheathed RDE electrodes into the vacuum chamber was prohibited due to outgassing and space constraints. 4.3.1.  Formation of surface PtZn alloy characterized by AES A Pt  wire  was  electrochemically  modified  with  an  electrodeposited  layer  of  Zn  and introduced  into  the  ultrahigh  vacuum  (UHV)  chamber  for  analysis  by  Auger  electron spectroscopy (AES). A significant amount of oxygen was detected, suggesting that the electrode surface was oxidized. In an attempt to protect the UPD layer from air oxidation, a thick layer of bulk Zn was electrochemically deposited onto the Pt surface. The thick Zn coated electrode was then carefully etched by Ar+ sputtering in the UHV environment until oxygen was no longer detected. Figure 4-1 shows the overall SEM image taken from the Ar+ etched electrode surface. The image indicates three distinguishable regions, labeled "I", "II", and "III" in Figure 4-1 with more detailed secondary electron images of these regions shown in Figure 4-2. Region I has a plateau-like  topography;  in  contrast,  region  II  is  composed  of  conelike  structures,  whereas region  III  has  a  rugged topography.  Low-energy-resolution  survey-scan  Auger  spectra  were measured for each region (Figure 4-3). These spectra clearly establish that regions I and II are pure metallic Pt and Zn, respectively, whereas region III has characteristic Auger peaks of both Zn and Pt. 87 Figure 4-1 SEM micrograph after Ar+ etching of a thick electrodeposited Zn layer on a Pt electrode. Three regions are indicated on the SEM which correspond to Pt (I), Zn (II), and Zn/Pt alloy (III). Reproduced with permission from [7]. Copyright 2006 American Chemical Society. 88 Figure 4-2 Higher magnification SEM images of (a) region I (Pt), (b) region II (Zn), and (c) region III (Zn/Pt alloy) identified in Figure 4-1. Reproduced with permission from [7]. Copyright 2006 American Chemical Society. 89 Figure 4-3 Auger  electron survey spectra  for  the regions  I  (Pt),  II  (Zn),  and III (Zn/Pt  alloy)  (bottom,  middle,  top,  respectively)  identified  in  Figure  4-1. Reproduced with permission from [7]. Copyright 2006 American Chemical Society. 90 Changes in both structure and composition of surface alloys may alter their Auger line shape  [136, 137] implying that  Auger  line shape analysis  may provide further  evidence for identifying whether the deposited zinc interacts chemically with the platinum to form a Pt/Zn surface alloy. Figure 4-4a compares the high-resolution Auger electron spectra for Pt in region III with that from pure Pt (region I). Auger line shape analysis indicates that Pt AES peaks at 1960.5 (MN3) and 2040.8 eV (MN2) from region I give an area ratio of 1.86, whereas region III has a peak area ratio of 1.14. This change appears to result from interactions between Zn and Pt altering the Pt MNN Auger transition probability. Changes are also seen in the Zn Auger spectra (Figure 4-4b), where the shoulder at about 980.6 eV decreases markedly. This evidence shows that both the Pt and Zn Auger electron spectra obtained from region III differ from those of pure metallic Pt and Zn. Therefore, it appears reasonable to assign region III as a Zn/Pt alloy region. Comparison of Auger peak areas for region III reveals the alloy has a chemical composition close to Zn0.73Pt (42 at. % Zn), where the relative sensitivity factors for Pt and Zn used in the quantitative analysis of the Auger spectra were self-generated from Figure 4-1 measured for the pure Pt and Zn regions. This approach should minimize most of the uncertainties involved in routine quantitative Auger analysis. The possibility is still open that Zn may have been removed preferentially during the Ar+ sputtering, but considering the prolonged sputtering did not change the spectral features, we concluded that the composition of region III is satisfactorily close to the 1:1 Pt:Zn alloy, which has a large stability region in the bulk Pt-Zn phase diagram [102]. The top panel in  Figure 4-5 shows AES spectra measured from the electrode surface after sputtering and leaving in 2.5 × 10-9 mbar UHV environment for 2 h. The spectra show the presence of more oxygen (peak at 509 eV) at the PtZn alloy (region III) than at pure Zn in region II, whereas no oxygen was found at the Pt region. This observation suggests that the PtZn alloy interacts more readily with oxygen than does pure Zn. Interestingly, however, Zn atoms in the alloy do not interact strongly with oxygen.  Zn Auger spectra for both metallic Zn and the PtZn, before and after the oxidation of the surface, are shown in Figure 4-6a and b, respectively. When oxygen binds to metallic Zn, the Auger transitions (peaks at 990 eV and 1020 eV) shifts to  lower  kinetic  energy by ~10 eV.  The  negative  shift  is  even  observed  for  the  secondary electrons ejected from the inner level (i.e. the peak at 1200 eV), indicating the formation of stable  ZnO. On the other  hand,  alloyed Zn did not  show such shifts  after  oxidation,  which suggests  alloyed  Zn  does  not  interact  with  oxygen  strongly.  Such  behavior  has  also  been 91 observed with Zn in Ru(001)/Zn alloy as mentioned in Chapter 2 [119]. As shown earlier, PtZn exhibited stronger affinity for oxygen than pure Zn, but this indicates the interaction of oxygen may only occur on Pt atoms in the alloy, or only with the outer most orbitals of Zn. The latter may be a likely case, as the relative Auger transition probabilities of Zn LM2 and LM3 before and after oxidation change. After exposing the surface to the same UHV conditions for further 10 h, the amount of oxygen detected increased in both the Zn/Pt and Zn regions (Figure 4-5, bottom panel) whereas the Pt region was still not oxidized. Interestingly, a trace amount of Zn is now detectable in the Pt region. In this case some Zn/Pt alloy may be formed via zinc migration across the surface. The room temperature diffusion of Zn into Pt is well-known and is characterized by a very large diffusion coefficient  [110]. This apparently results from a small crystallographic misfit so that Zn atoms can occupy the Pt atom positions with little change in the lattice parameter [107]. 92 Figure 4-4 High-resolution Auger electron spectra of (a) Pt from regions I and III and (b) Zn from regions II and III. Pt Auger transition peak heights at 1960.5 and at 2040.8 eV from region I give an area ratio of 1.86, whereas region III has a peak area ratio of 1.14. The metallic Zn Auger transition spectrum shows a shoulder by the peak at 980.6 eV but the shoulder diminished when Pt was modified by Zn. Reproduced with permission from [7]. Copyright 2006 American Chemical Society. 93 Figure 4-5 Auger electron survey spectra of regions I (Pt), II (Zn), and III (PtZn) after leaving in a vacuum (2.5 × 10-9 mbar) for (a) 2 and (b) 10 h. After 2 hours, the amount of oxygen on the PtZn surface was greater than the amount on the Pt surface. Reproduced with permission from [7]. Copyright 2006 American Chemical Society. 94 Figure 4-6 Auger electron spectra of a) metallic Zn and b) PtZn; before and after the surfaces interact with oxygen in the UHV chamber. The peaks observed at 500 eV on the spectra in the top panels are the indication of the presence of oxygen. Auger transition peaks of Zn in 900 ~ 1100 eV region are shown in the bottom. The peak positions shift to lower kinetic energy with the presence of oxygen on the metallic Zn, indicating oxygen is strongly interacting with the Zn surface. On the other hand, the signals  from the  PtZn surface were  not  affected by the  presence of  oxygen, indicating  very weak or no interaction between Zn in PtZn and oxygen. 95 4.3.2.  Preparation of the Zn-modified Pt RDE. The Zn-modified Pt RDE was prepared via electrodeposition from ZnSO4 solution at 22 oC.  The  Pt  CV  in  this  solution  is  shown  in  Figure  4-7a.  The  potential  is  referenced  to Hg/Hg2SO4 in the plot. Bulk deposition of Zn begins at -0.85 V, although the first monolayer of Zn  is  deposited  at  potentials ca. 1  V more  positive.  Zn  was  electrodeposited  onto  the  Pt electrode by setting the Pt potential to -1.06 V for 300 seconds. A significant quantity of Zn was deposited onto the Pt surface (2.43 C cm-2). The electrode with deposited Zn was removed from solution,  and the alloy was allowed to form for 60 min in an Ar atmosphere. The modified surface  was  then  placed  back  into  the  electrolyte  solution  at  a  potential  where  Zn oxidation/stripping occurs (-0.3 V), showing a characteristic current transient that indicates a limited amount of Zn was removed (2.27 C cm-2).  The deposition and stripping currents are shown in Figure 4-7b. Ignoring the contribution from residual oxygen reduction that is present in both curves, the difference between the two charges would be equivalent to the amount left on the surface, or alloyed with the Pt. The difference accounted to 0.167 C cm-2. Approximately 300 µC cm-2 has been reported for a UPD layer of Zn on Pt single crystal surface [138, 139] giving a net deposition of ~500 monolayers of Zn in the present case, but our own studies showed only 4% of  Zn  deposited  was  alloyed  to  Pt  (determined by gravimetry which  will  be  shown in Chapter 6),  which leads  to  the formation  of  a  1:1 surface  alloy in  the  first  ~40 layers. In addition, it is important to note that the Zn stripping potential was not made very anodic so as to avoid  oxidation  of  the  Pt  surface  which  would  remove  the  Zn/Pt  alloy.  As  demonstrated previously [107] the alloy formed in this way is stable at all potentials below 1.2 V. In fact, the Ar  saturated  CVs  and  ORR  curves  discussed  next  can  be  reproducibly  obtained  from  the modified electrode surface many days afterward. In addition, because the diffusion of Zn into the bulk of the Pt electrode occurs very quickly, one Pt RDE has irreversibly changed into the Zn/Pt  alloy form after  many deposition/alloying cycles.  The same ORR results  can be  also obtained from this now permanent Zn/Pt electrode as those reported. 96 Figure 4-7 (a) Cyclic voltammetry (100 mV/s, 22 C) of Pt in 0.1 M ZnSO4 showing deposition and stripping processes. (b) Current transients for the deposition (dashed line) of Zn onto the Pt electrode and the resulting current transient for the stripping (solid line) of the Zn remaining on the Pt electrode after allowing the formation of the Pt/Zn alloy.  Reproduced with permission from  [7]. Copyright 2006 American Chemical Society. 97 4.3.2.1.  Ar saturated CVs of Pt and PtZn Ar saturated CVs recorded on Pt and PtZn (i.e., in an Ar saturated 0.1 M H2SO4 solution between 0.03 and 1.3 V or between 0.03 V and 1.0 V, respectively, at the sweep rate of 50 mV/s; 22 oC) are compared in Figure 4-8. Hydrogen adsorption/desorption peaks, as well as hydrogen evolution, on PtZn are diminished compared to Pt, which is consistent with the literature  [90, 107].  Solution  cleanliness  was  assessed  by  the  Pt  CV before  and  after  the  PtZn  CV was recorded, indicating the difference between the Pt and PtZn CV's was due to the presence of Zn, rather than solution contamination. The PtZn-oxide formation is also shifted anodically by ~80 mV at 20 µA/cm2 which indicates the oxide layer formed on PtZn is more weakly adsorbed than on Pt. Such a layer is readily reducible, thereby improving the ORR catalysis  [69, 140]. The kinetics of ORR catalysis on Pt and PtZn is compared next. Figure 4-8 Comparison of Ar saturated CVs recorded on the Pt (dashed line) and the PtZn (solid  line) electrodes at  22  oC. Hydrogen adsorption/ desorption peaks are disappeared  in  the  PtZn  CV,  and  the  Pt  (or  PtZn)  oxide  formation  is  shifted anodically. Reprinted from [8]. Copyright 2009 with permission from Elsevier Ltd. 98 4.3.2.2.  ORR kinetics for Pt and Zn-modified Pt electrodes An anodic voltammogram of  the smooth Pt  RDE in a  solution of 0.1 M H2SO4 (22 oC)saturated with O2 rotating at  2000 rpm is  shown in  Figure 4-9.  The ORR kinetics  were measured using the Koutecky-Levich (KL) methodology (Equation  3-4).  The mass  transfer corrected  current  was  determined  using  the  rotating  disk  electrodes  from the  anodic  sweep direction. A Tafel plot generated from this voltammogram properly corrected for mass transport is  shown  in  Figure  4-10 compared  to  a  KL analysis  of  a  series  of  these  voltammograms measured at various rotation rates (500-3000 rpm). The kinetic current calculated from the KL analysis  (Equation  3-4) and  the  mass  transport  corrected  current  (Equation  3-5) are  very similar. The slope of the lines in the KL plot is 3.7 cm2 s-1/2 mA-1 for the ORR on smooth Pt, consistent with values reported in the literature [27]. The Tafel slope in the potential region around 0.80 V was calculated to be 98.4  ± 5.9 mV/decade for the anodic scan direction as expected for ORR in H2SO4. The ORR performance of the Pt/Zn electrode is compared to that of the smooth Pt RDE through a voltammogram measured in the same electrolyte (0.1 M H2SO4 saturated with O2; 22 oC) at the same rotation rate (2000 rpm); both are shown in Figure 4-9. A shift in the potential ( approximately +30 mV) of ORR is observed for the Pt/Zn electrode. It is important to note the ORR curve on PtZn was recorded without polishing the surface,  i.e., the electrode surface is rough comparing to the starting Pt.  As shown in  Chapter 2, this has a significant impact on catalysis. However, the positive shift for PtZn in the recorded voltammogram was reproducible. This topic will be discussed in detail in Chapter 5.3.4. The KL analysis for the Zn-modified Pt RDE gave slopes that were similar to that measured for the smooth Pt RDE, about 3.8 cm2 s-1/2 mA-1. The  kinetic  currents  are  plotted  with  respect  to  potential  in  Figure  4-10 for  both electrodes using the two methods of determining kinetic currents. Both methods give similar curves, though the KL analysis is slightly shifted to larger currents when the kinetic current is calculated via Equation 3-5. The shift is consistent within both sets of analyses, which suggests a slight underrepresentation of the kinetic current determined by  Equation  3-5.  The shift in potential (~30 mV) due to Zn modification is demonstrated in these Tafel plots. For example, at 1  mA cm-2,  the  Pt  electrode  is  at  0.890 V/RHE while  at  the  same current  the  Zn-modified 99 electrode is at 0.920 V/RHE. The Tafel slope at 0.8 V for the Zn-modified Pt RDE (96.3 ± 4.1 mV/decade) is slightly smaller than for smooth Pt. The number of electrons exchanged during the reaction was calculated using Levich equation (i.e., Equation 3-3). The kinematic viscosity of 0.1 M H2SO4 at 20 oC is 0.01014 cm2/s [113], and the concentration of oxygen was (8.35 ± 1.97) × 10-4 mol/L, the  number of  electrons  was calculated to  be 3.97.  Our Tafel  plot  data suggests a possible modification in the mechanism for the ORR on the modified surface, yet maintaining the direct pathway rather than a parallel pathway (i.e., water as a product rather than hydrogen peroxide). PEM fuel cells are usually operated at 70~80 oC, and it would have been appropriate to conduct the RDE experiments at elevated temperatures. However, the RDE shaft is made of Teflon, which expands upon heating, but does not contract to the same extent upon cooling. This could cause leakage of electrolyte into the interior of the RDE. Therefore we limited our ORR investigation only at the temperature of 22  oC, which may not represent a fuel cell operating conditions, but should be sufficient for the comparative studies. The  ORR  results  shown  above  were  observed  under  ideal  conditions  in  an electrochemical cell, but it is important to test these results in a fuel cell. The modification of commercially available gas diffusion electrodes with electrodeposited Zn is described next along with  creation  of  a  modified  MEA and testing  under  fuel  cell  conditions.  The  conventional method to prepare the carbon-supported Pt-alloy catalyst layers is to deposit a transition metal onto a preformed Pt/C by chemical vapor deposition, followed by a heat treatment at 700-1000 oC under an inert or reducing condition [69, 134, 141, 26, 140]. The method we used was the electrochemical deposition of Zn onto the Pt/C, which is unique to this work. 100 Figure 4-9 Comparison of the ORR activity measured at 2000 rpm (22 C) with an anodic voltage sweep rate of 10 mV/s for a smooth Pt RDE (solid line) and a Zn- modified  Pt  RDE  (dashed  line).  Approximately  30  mV  improvement  in  the overpotential  with  the  PtZn  alloy  compared  to  with  the  Pt.  Reproduced  with permission from [7]. Copyright 2006 American Chemical Society. 101 Figure 4-10 Calculated kinetic currents for the ORR on smooth Pt (solid line, o) and on Zn-modified Pt surface (dashed line,  ) for an anodic scan (22  oC). The data points were determined from the Koutecky-Levich analysis. The solid and dashed lines were mass transport corrected currents measured at 2000 rpm shown in Figure 4-9.  Reproduced  with  permission  from  [7].  Copyright  2006  American  Chemical Society. 102 4.3.3.  Zn modification of a gas diffusion electrode Commercially  available  gas  diffusion  electrodes  (GDEs)  were  modified  with  Zn  as described  in  the  Experimental  Section.  These  GDEs  gave  deposition  and  stripping  curves (Figure 4-11) very similar to those seen for the Pt RDE. We assumed that only the Pt catalyst particles were modified with the electrochemically deposited Zn since this metal does not alloy to a significant extent with the carbon support and Zn is easily removed from the carbon surface [106]. The cyclic voltammogram of the GDE in ZnSO4 shows the characteristic deposition and stripping regions at  the same potentials  as observed for the Pt RDE. Deposition of Zn was accomplished potentiostatically (-1.0 V for 5 min). The cell was disconnected and the alloy was allowed to form ex-situ. The potential was stepped to -0.3 V to strip the unalloyed Zn from the GDE. The difference between the deposition and stripping charge shows that 0.135 C cm-2 (~0.1 mg cm-2) Zn remained, in the GDE. The Pt loading of this GDE was 1 mg cm-2, suggesting that a 10 % loading of Zn on Pt was realized. This is a small modification, but alloying behavior may be significantly different on bulk Pt or on Pt nanoparticles on carbon (3~5 nm in size). This issue will  be  revisited  in  Chapter  6. This  simple  electrochemical  deposition  of  Zn  onto  a commercially available GDE was sufficient to change the nature of the ORR kinetics, which suggests that such a process should be cost-effective and industrially feasible. The Zn-modified GDEs were used as cathodes and pressed with an unmodified GDE as the anode. The resulting Zn-modified MEA was tested under two different fuel cell operating conditions and compared to a similarly prepared MEA that did not undergo the Zn treatment of the  cathode.  The  treated  and  untreated  fuel  cells  were  tested  at  equal  anode  and  cathode pressures (350 kPa), and the measured polarization curves are shown in Figure 4-12a. The open circuit voltage of the fuel cell containing the Zn-modified MEA (0.98 V) was slightly higher when compared to that of untreated Pt cathode (0.95 V), which is in good agreement with the RDE results.  This  shift  to  higher  potentials  was  also  observed in  the  kinetic  region  of  the polarization curve (Figure 4-12a, inset) and at the higher superficial current densities. The influence of Zn treatment of the cathode was also tested under operating conditions that maintained a pressure difference between the cathode and the anode inlet gas streams, i.e., 350 and 101 kPa, respectively (Figure 4-12b). Operating at a higher cathode pressure increased 103 the  superficial  current  density range  of  the  fuel  cells  for  both  the  Zn-treated  and untreated MEAs. This effect is most likely due to the improved membrane humidification as a result of water back flow created by the 249 kPa pressure difference between the cathode and anode. The open circuit voltage for the Zn-treated MEA was higher by 30 mV. This shift was also evident in the kinetic region of the polarization curve shown in the inset. The increase in potential was ~70 mV at 200 mA cm-2 under these operating conditions as compared to the polarization curve measured for  similar  anode and cathode inlet  pressures.  Interestingly,  for superficial  current densities higher than 200 mA cm-2 the voltage differential between the treated and untreated MEAs diminished to about 30 mV, which was maintained up to 600 mA cm-2. This was likely due to the larger ohmic drop observed in the case of the Zn-treated MEA which could arise in two ways. First, the Zn-treated cathode was hot pressed and bonded in-house to the cathode side of the MEA. Therefore, it is likely that the ionomer content and proton-conductive network of the cathode catalyst layer might be different as compared to the unmodified MEA, which could affect its performance. Second, tests have not yet been done to rule out the possibility of Zn2+ leaching from the catalyst, although the Zn/Pt alloy is stable in acidic environments as long as the  cathode  potential  never  exceeds  1.2  V (which  did  not  occur  in  these  tests  [106]).  The presence of Zn2+ in the ionomer would increase the ohmic drop. The stability of PtZn alloy will be discussed in  Chapter 6. However, the improvement in the polarization curve is clear. The measured curves do not show optimal performance for potentials in the ohmic drop region (E = 0.6-0.2 V). It has been our experience that the performance of the untreated, commercial MEA from Electrochem Inc. is somewhat inferior to MEAs produced by other manufacturers when tested under similar conditions  [142]. Nevertheless, it is important to emphasize that over the entire range of superficial current densities, the MEA with the Zn-modified Pt cathode gave a cell voltage between 30 and 70 mV greater than for the commercial MEA. When comparing polarization curves, reproducibility is important. This issue was tested using in-house pressed unmodified MEAs. The conditioning method used in this work produced MEAs  that  gave  reproducible  polarization  curves  over  an  extended  period  of  testing.  The variability was found to be roughly ±20 mV throughout the polarization curve. The observed reproducibility  of  the  polarization  curves  following  the  conditioning  procedure  helps substantiate the new conclusion that modification of the Pt MEA through Zn deposition leads to an improved activity for ORR compared to the current Pt cathode. 104 Figure 4-11 (a) Cyclic voltammogram (100 mV/s, 22 oC) of a 1 mg cm-2 Pt GDE in 0.1 M ZnSO4 demonstrating the deposition and oxidation of the deposited Zn (solid line,  cathodic  limit  of  -1.6  V/Hg/Hg2SO4;  dashed  line,  cathodic  limit  of  -2.0 V/Hg/Hg2SO4). (b) The deposition and stripping current transients following a step to - 1.6 and -1.0 V vs Hg/Hg2SO4, respectively. Reproduced with permission from [7]. Copyright 2006 American Chemical Society. 105 Figure 4-12 MEA polarization curves (74 oC) for Zn-modified (●) and unmodified () cathodes: (a)  measured using the same inlet  gas pressures for the anode and cathode; (b) measured using the higher cathode inlet gas pressure (350 kPa for the cathode and 101 kPa for the anode). In each case, the inset shows the low current kinetic  region  of  the  polarization  curve.  Reproduced  with  permission  from  [7]. Copyright 2006 American Chemical Society. 106 4.3.4.  Effect on the electronic nature of the surface due to alloying Zn with Pt The decrease in the ORR overpotential  on the PtZn alloy surface as compared to Pt suggests a change in the manner in which O2 interacts with the PtZn alloy. This implies a change in the electronic nature of the catalyst surface as a result of alloying with Zn. The inclusion of Zn as an alloy component with Pt for use as an ORR catalyst has not been reported prior to our finding. However, Zn is used as an alloying component in transition metal surfaces resulting in catalysts that are more selective or specific in their activity [112, 143, 144]. These changes in the surface properties have been related to a ligand effect, which changes the electronic properties of the surface,  or to an ensemble effect,  which modifies the number of active sites  [145]. The addition of Zn to Rh, Ni, Pt, and Pd shifted the valence d band toward higher binding energies (thereby creating d-band vacancies) along with a shift to lower binding energies for the Zn 2p core level and Zn 3d band. For Pt, these changes result from a rehybridization that shifts electron density from the Pt 5d band toward the bimetallic  bond with Zn which involves  Pt 6(s,  p) orbitals. These differences in the electronic structure of the PtZn compared to Pt will affect the surface-molecule interaction,  which is  important when considering the decrease in the ORR overpotential measured experimentally. Further insight into these surface-molecular interactions and the chemical nature of the surface can be realized through the thermal desorption studies of adsorbed CO [144, 143]. The addition of Zn to Pt decreases the CO desorption temperature, suggesting a reduction in the ability of Pt to back-bond, which decreases the strength of the Pt-CO interaction [143] resulting from the Pt 5d to 6(s, p) rehybridization. A recent density functional theory (DFT) analysis of the bonding in a PtZn alloy [101] (L10, tetragonal lattice) described the changes in the band structure due to alloying with Zn. As  mentioned  in  Chapter  1,  a  significant  amount  of  research  has  focused  on  the improvement of ORR kinetics by alloying Pt with various transition metals. A systematic study using in situ extended X-ray absorption fine structure of dispersed binary Pt alloy catalysts in a fuel  cell  MEA arrangement  [117,  70] indicated  specific  changes  in  bond  distances  and  in electronic structure due to the presence of the transition metal. The alloys were primarily of the Pt3M1 structure. A decrease in the Pt-Pt bond distance which, correlated with an increase in Pt 107 5d orbital vacancies, was indicated as the alloying metal was systematically changed from Mn to Ni.  These changes  were apparently related to  the effect  of the bulk electronegativity of  the alloying metal, where an increase in electronegativity leads to an increase in Pt d-band vacancy. However, the use of bulk electronegativity for predicting or interpreting the characteristics of a bimetallic surface alloy has been criticized by Rodriguez  [121, 120] with experimental results suggesting that orbital rehybridization at the surface can change the effective electronegativity of the alloying metal as compared to the bulk. The rate of ORR, measured at 0.9 V/SHE, was found to  be dependent  upon the  Pt-Pt  bond distance  and Pt  d-band vacancies,  exhibiting  a maximum for the Pt/Cr alloy. A similar explanation can be invoked for the current Pt/Zn alloy studies. The formation of a Pt-Zn alloy results in orbital rehybridization which creates  d-band vacancies, resulting in an improvement in ORR kinetics similar to that seen for transition metal- Pt alloys  which are  currently used.  Moreover,  the strength of the Pt-Zn bond may result  in enhanced stability for the Pt-Zn alloy under operating fuel cell conditions which may mitigate the leaching problem present in many of the Pt alloys currently under study. An alternative explanation for the improved ORR kinetics on the PtZn surface is  by borrowing the ORR mechanisms proposed by Fernandez et al [114]. The authors suggested that the adsorption of oxygen molecules occurs on the transition metal  component of the alloys, followed by transferring the oxygen to the neighboring Pt sites. This is shown schematically in Figure 4-13. Considering 1) the PtZn attracts oxygen more than Zn, but the Zn atoms in the PtZn do not bind to oxygen too strongly, and 2) Zn modification of Pt increases the catalytic activity of the ORR, Fernandez's model may be the explanation for the ORR mechanism on our PtZn surface. 108 Figure 4-13 ORR mechanisms suggested by Fernandez,  et al. The oxygen is first adsorbed onto transition metal sites on the surface, followed by a rapid transferring of the oxygen. The gray circles in lattice represent the transition metals, and the dark circles in (b) represent oxygen atoms. The reaction of ORR is believed to improve. Reprinted in part with permission from [114]. Copyright 2005 American Chemical Society. 109 4.4.  Summary and conclusions This  chapter  investigated  the  ORR catalysis  on  the  electrochemically  prepared  PtZn alloy. The electrochemical deposition of Zn onto a Pt substrate results in the formation of a PtZn alloy. This was confirmed through a combination of electrochemical and AES analytical studies. Interestingly, the oxygen uptake on the PtZn region appeared more efficient than on the pure Zn region. The ORR catalytic activities were measured on both smooth Pt and PtZn surfaces at 22 oC,  and  the  PtZn  showed  30  mV improvement  in  the  onset  potential  of  the  reaction.  The resulting  analysis  showed that  the  Tafel  slope  was  slightly   smaller  than  that  measured  for smooth Pt. This alloy was also studied in a single fuel cell,  where the PtZn GDE was also prepared by the electrochemical deposition of Zn onto a Pt-dispersed GDE. It was found that the MEAs with the Zn-treated cathode gave a 30-70 mV higher cell voltage over the entire range of explored superficial current densities up to 800 mA cm-2. A possible  explanation was presented that  changes  in the metal's  electronic  structure through alloying may account for the improved performance. Specifically, addition of Zn may cause a decrease in Pt 5d-band occupancy with a 5d to 6 (s,p) rehybridization and an increase in electron density at the heteronuclear Zn-Pt bond [112]. Our assumption is that this enables an enhanced stability in acidic environments. In fact,  the alloy could only be removed through oxidation of the Pt substrate. Such effects may provide a great potential benefit for fuel cell applications. 110 5.  CO Oxidation Catalysis, Surface Roughness And Restructuring on PtZn 5.1.  Introduction The  previous  chapter  showed  that  electrochemically  prepared  PtZn  is  a  promising catalyst  candidate  for  the  oxygen  reduction  reaction  (ORR).  The  ORR  catalysis  requires adsorption of oxygen onto the catalyst surface as does CO oxidation; an anodic reaction known as a rate determining step for methanol oxidation which is employed in a direct methanol fuel cell (DMFC) [43]. Currently PtRu catalyst is employed in a DMFC which is also an effective catalyst for CO oxidation. PtRu has some similarities to PtZn. According to the work done by Buatier et al. [146], PtRu also showed a decrease of Pt (5d) electron density and a decrease in CO desorption temperature in a UHV environment; similar to what Rodriguez showed for PtZn [143, 112].  This chapter first  describes CO oxidation electrocatalysis  on the PtZn. Using an as- prepared PtZn electrode, we observed a significant improvement in the CO oxidation catalysis. However,  as shown in Chapter 2, surface structures and roughness impact catalytic activity in general. We noticed that the formation of the PtZn alloy significantly roughens the electrode surface  compared  to  the  original  substrate.  Also  in  the  previous  chapter,  the  measured overpotential  on PtZn was 30 mV less than on the polished Pt surface,  but again,  the PtZn surface was much rougher. This roughness has to be compensated. Because we were not able to polish PtZn to a smooth finish as it would remove the surface alloy (approximately 40 layers on the surface; estimated in  Chapter 4),  we deposited Pt particles on a smooth Pt electrode to create an equivalently rough surface to that of PtZn. The comparison between the rough and smooth electrodes is presented. Also shown is the adsorbate-induced surface restructuring which was electrochemically realized.  In  Chapter  2,  the  literature  review  describing  the  surface  atoms  rearranging 111 themselves according to the surface energy changes before and after the adsorption of small molecules, such as CO or oxygen, was reviewed. The evolution of the characteristics of the PtZn surface  (roughness,  composition)  during  the  various  electrochemical  treatments  (i.e.,  CO oxidation and ORR) is also presented in this chapter. 5.2.  Experimental 5.2.1.  Electrochemistry and PtZn alloy formation Electrochemistry was carried out in a regular three-electrode cell setup similar to that described in Chapter 3. The potentiostat (HEKA PG 590) was connected to a data acquisition board (National Instrument;  PCI-6052E),  and data recorded using LabView in-house written program (National Instrument).  The electrolyte was 0.1 M H2SO4 prepared with the Milli-Q water.  All  the  potentials  are  reported  with  respect  to  RHE,  and  the  electrochemical  data presented in this chapter was recorded at the room temperature (22 oC). The working electrodes were either Pt rotating disc electrodes (RDEs) or Pt beads. One set  of  electrodes  was  used  without  Zn  modification  resulting  in  Pt  electrodes  for  baseline comparison. For the polished Pt (designated as a “smooth Pt” hereafter), the areas of electrodes were determined by integrating the area under hydrogen adsorption region of CV, assuming the monolayer  adsorption  charge  of  210  μC/cm2 [147].  For  PtZn,  the  area  was  determined  by comparing the double layer capacitance of its CVs to double layer capacitance determined from the smooth Pt CVs for which the area is determined using the hydrogen adsorption (Hads) charge (as shown in the example using a smooth Pt and a rough Pt surfaces in Figure 5-1). This was required since as we will show, the area under the Hads region is strongly modified by PtZn alloy formation. The CVs for the PtZn alloy were recorded between 0.03 and 1.0 V at 50 mV/s. The upper limit for PtZn electrode was restricted so as to protect the alloy from oxidation which occurs above 1.2 V vs. RHE [106]. Pt and PtZn samples were analyzed in EDX-equipped SEM. SEM images were analyzed using ImageJ (National Institute of Health). Backscattered electron images were obtained in a Microlab 350 using a field emission source of 10 keV, 3 nA, and a 112 hemispherical analyzer. Figure 5-1 The electrode surface area can be determined by measuring the double layer capacitance and comparing it to the one with the known area. The electrode area on the smooth Pt was determined by integrating Hads peaks. 5.2.2.  CO stripping voltammetry The electrochemistry of CO oxidation on both smooth Pt and PtZn surfaces was studied. The 0.1 M H2SO4 electrolyte was bubbled with carbon monoxide (Praxair) for approximately 20 minutes after measuring a CV in an Ar saturated solution. This was sufficient time to saturate the solution with CO. Over the electrolyte, Ar flow was always maintained to avoid any oxygen 113 ingress, important as a trace amount of oxygen will react with CO to form CO2 [83]. While the experiments  were  carried  out,  CO  was  constantly  bubbled  into  the  solution  to  ensure  CO saturation. The smooth Pt or PtZn electrode was then immersed into the solution and polarized at 0.3 V for 1 minute to create a CO saturated surface. The potential was scanned from 0.3 to 1.1 V at 50 mV/s to record the CO stripping voltammograms. All the processes were done under the ambient conditions (Temperature = 22 oC). 5.2.3.  Oxygen reduction reaction ORR was performed using smooth and rough Pt and PtZn RDEs at 2000 rpm at 22 oC. The solution was 0.1 M H2SO4 saturated with oxygen (Praxair; bubbling for approximately 20 minutes with oxygen surface flow). The electrode potential was swept from 0.3 to 1.05 V at 20 mV/s for both Pt and PtZn. Analysis was performed on anodic scans of ORR curves. Current densities were calculated using the electrochemical  surface area,  rather  than the geometrical areas of electrodes. 5.2.4.  Pt electrodeposition A solution of 0.1 M H2SO4 containing 1 mM of H2PtCl6 (Aldrich) was prepared with Millipore water. For this section of experiments, the potentiostat used was a USB-interfaced µAutolab  (Metrohm,  Switzerland).  Deposition  potential  of  0.5  V  was  first  applied  for  15 minutes  to  the  Pt  RDE electrode  and  then  the  surface  was  further  roughened  by a  second deposition for 30 minutes. The solution was previously deaerated with Ar to reduce the current due to reduction of oxygen. For consistency, we express the roughness of these Pt electrodes based on the capacitance determined in the double layer region (assuming that the capacitance is similar for a rough Pt and the Zn modified Pt surfaces) as compared to that measured for the smooth Pt electrode. Pt electrodeposition was carried out at 22 oC. 114 5.3.  Results and discussion 5.3.1.  The nature of the PtZn surface after electrodeposition Pt and Zn-modified Pt surfaces were studied using SEM as shown in Figure 5-2a and b, respectively. It is worth mentioning the samples were left in an ambient condition at least for one week before the measurements. It is obvious from the figure that the Zn-modified Pt surface is covered with white particles, whereas such particles are not seen on the Pt surface.  EDX analysis was performed on the Zn-modified surface. On the surface of the particles, we found both Pt and Zn (Zn:Pt atomic ratio of 5:100).  A Zn depleted surface indicates that Zn diffuses into the bulk. This issue will be revisited in Chapter 6 but it appears Zn migrates into Pt with time under ambient conditions. On flat regions of the surface only Pt was detected; Zn was not detected above the instrumental detection limit (0.2 wt %).  Also we did not detect oxygen above our detection limit from any part of the surface, indicating ZnO is not present on the surface. On the particles, Zn was present before and after electrochemical analysis at a similar concentration, therefore  the  electrochemical  data  recorded  on  the  PtZn  electrode  presented  in  this  article, whether directly or indirectly, shows a contribution from Zn. It is very interesting to examine the left hand side of the  Figure 5-2b. There are three regions covered with different particle densities. On average, the PtZn particle coverage is about 30 %, but on regions i) and ii), the coverages were approximately 80 % and 40 %, respectively. However, Zn was not detected in the bare region iii) above the detection limit. PtZn particles are preferentially found decorating grain boundaries. The SEM image also has two large scratches that cross each other, and they are also decorated more extensively with PtZn, similar to grain boundaries. Taguchi  et al., investigated Zn underpotential deposition (UPD) in 0.1 M sulfuric acid solution onto Pt (111), Pt (110), and Pt (100) surfaces. They observed no significant difference observed on (111) and (100) surfaces with or without the presence of Zn2+ in the electrolyte, whereas they clearly observed the formation of Zn UPD layer on a Pt (110)-(1×2) surface [106]. This work suggests Zn preferentially deposits onto particular crystal faces of Pt. Also, a number of  articles  have  shown electrodeposited  films  with  preferred  orientations  depending  on  the deposition conditions [148-150] or on the crystallinity of substrates [151]. When the interactions 115 between substrate atoms and atoms in a deposited film are relatively strong, the crystal structure of substrate affects the resulting structure of deposits [151]. Figure 5-2 SEM images of a) smooth Pt and b) PtZn surfaces. EDX analysis showed both Pt and Zn on the white particles in b), whereas the flat gray part contains only Pt.  PtZn  particle  densities  are  different  and  seem to  depend  on  the  different  Pt surfaces. Region i) and ii) have approximately 40% and 80% coverage, respectively, whereas no particles were seen in Region iii). Reprinted from [8]. Copyright 2009 with permission from Elsevier Ltd. In Chapter 4, we observed a migration of Zn to Pt after leaving the PtZn sample in the UHV chamber for  10 hours.  Diffusion of Zn into Pt was  further  confirmed on the Pt bead surface by backscattered electron detection. Heavy atoms backscatter electrons more than light atoms. The regions containing heavy atoms appear bright in the image. Figure 5-3 shows both the  regular  SEM  (a)  and  backscattered  SEM  images  (b)  of  the  same  regions  of  the  PtZn 116 electrode. It was noted the middle white patch in backscattered image showed dark “shadow” around the grain boundaries which gradually fades away from the grain boundary. This indicates Zn is migrating into Pt region from a more Zn rich region resulting in a depleted region of about 1~1.5 µm. Because the PtZn particles were present only on the surface, and polishing the surface would  easily  remove  them,  we  carried  out  our  experiments  on  the  alloy  without  further modification. Therefore, the electrochemical data presented in this contribution was recorded on a rough surface which was considered and compensated for in all electrochemical measurements reported in this chapter. 5.3.2.  Ar saturated cyclic voltammograms of Pt and PtZn Ar saturated CVs recorded on smooth Pt and freshly prepared PtZn in an Ar saturated 0.1 M H2SO4 solution  are  shown in  Figure  5-4 (Also  shown in  Chapter 4).  Again,  hydrogen adsorption/ desorption (Hads/des) regions diminished on PtZn, and an anodic shift in Pt (or PtZn alloy) oxidation potential is also realized. As mentioned in  Chapter 2, the Pt electrochemical surface area can be determined by integrating the Hads/des peaks; however, the surface area of PtZn could not be reliably determined using the same method; instead, the surface area was estimated by comparing the double layer capacitance of the CV recorded for the PtZn electrode to that of the smooth Pt. This method can estimate the electrochemical surface area within 7.5 % error relative to the surface area determined based on the charge under  hydrogen adsorption peaks (confirmed by using the Pt electrode with a varieties of surface roughnesses). Generally, when  the  PtZn  alloy  was  created,  the  resulting  surface  was  considerably  rougher  than  the original Pt substrate. The surface area of PtZn (as-prepared) was generally about 10 times larger when  comparing  the  double  layer  capacitance  to  that  obtained  with  a  smooth  polished  Pt electrode in the same solution. 117 Figure 5-3 a) SEM image of the PtZn bead electrode, and b) backscattered SEM of the same region of the electrode. Backscattered SEM shows heavier atoms in darker colors relative to lighter atoms. The white particles in the regular SEM appear in a darker color in the backscattered SEM, indicating the white particles contain Zn. The grain boundaries appear to be decorated with PtZn. Also the middle bright patch in b) shows in backscattered image showed dark “shadow” around the grain boundaries which gradually fades away from the grain boundary. This indicates Zn is migrating from either the grain boundaries or from the neighboring Zn-rich region into Pt. 118 Figure  5-4 Ar  saturated  CVs  of  smooth  Pt  (dotted  line)  and  PtZn  (solid  line) recorded at 50 mV/s in Ar saturated 0.1 M H2SO4  at 22 oC. PtZn electrode shows a positive shift in Pt-oxide formation. Also hydrogen adsorption/desorption peaks on PtZn are  diminished.  Reprinted  from  [8].  Copyright  2009 with  permission  from Elsevier Ltd. 119 5.3.3.  CO oxidation and Pt enrichment of the alloy surface CO oxidation  voltammograms  for  a  smooth  Pt  surface  and a  freshly prepared  PtZn electrode are shown in Figure 5-5a. The CO stripping peaks for the Pt and PtZn were at 0.91 V and  0.75  V,  respectively.  A 150  mV negative  shift  was  observed  with  the  PtZn  electrode. However,  when  the  electrode  surface  is  rough  or  contains  defect  sites,  the  nature  of  CO adsorption  and the  oxidation  are  quite  different  from a smooth surface,  as  has  been shown theoretically and experimentally [59], and reviewed by Yates [60]. Briefly, Xu et al., found that CO oxidation proceeds preferentially on a metal surface containing both step and terrace sites due to both electronic and geometric effects [59]. To investigate if the negative shift was due to the effect of alloying or to the effect of surface roughness, CO stripping voltammetry was also performed on Pt electrodes that were purposely roughened by electrodeposition of Pt onto a smooth Pt surface. Electrodes with two roughnesses were prepared: one is slightly less, and the other is slightly more rough than the surface  of  the  freshly prepared  PtZn surface.  Relative  roughness  of  each  of  electrodes  was assigned by taking the ratio of the double layer capacitance of roughened surface to that of the smooth Pt. The ratios were 7.89, 6.0, and 8.45 for the PtZn electrode and the two roughened Pt electrodes, respectively. CO stripping voltammetry was then performed on those surfaces. The results shown in Figure 5-5b indicate that the shift observed with PtZn was most likely due to surface roughness. However, an interesting observation was made after either the Pt or PtZn electrodes were immersed into a CO saturated acidic solution and their potentials were scanned between 0.03 and 1.05 V at 50 mV/s over prolonged time (~3 hours; this procedure is designated as “CO treatment” hereafter).  Figure 5-6a and b show the Ar saturated CVs recorded before (curve 1) and after CO treatment (curve 2 to 3 or to 4) on Pt and PtZn, respectively. Each CV presented in Figure 5-6 was recorded in fresh electrolyte, thereby differences observed among the CVs were not due to the solution contamination. Curve 3 (or 4) experienced more CO treatment than curve 2 (or 3). Both electrodes were RDEs and had well-defined geometric areas. This suggests that the surfaces of both electrodes became smoother after CO oxidation. Further CO treatment on PtZn did not change the electrochemical surface area very much, as measured by the double 120 layer capacitance (shown in  Figure 5-6c); however, the hydrogen adsorption/desorption peaks of CVs (E = 0.3 V) after the CO treatment became greatly enhanced, suggesting the surface was becoming Pt-rich, further confirmed by the cathodic shift in the Pt-oxide formation. (shown with arrows in the figure.) Figure 5-5 a) CO stripping voltammograms recorded at 50 mV/s in CO saturated 0.1 M H2SO4 on a smooth Pt (dotted line) and on a PtZn (solid line) electrodes recorded at 22 oC. A 200 mV negative shift was observed on the PtZn electrode relative to Pt; b)  CO  stripping  voltammograms  on  Pt  and  PtZn  electrodes  of  comparable roughnesses (recorded in the same manner as [a]). “Roughness” were calculated by comparing the double layer capacitances measured with Ar saturated CVs to that of smooth Pt. The negative shift in the CO stripping peak observed with PtZn in a) was purely due to roughness. Reprinted from [8]. Copyright 2009 with permission from Elsevier Ltd. 121 Figure 5-6 Ar saturated  CVs recorded with a)  smooth Pt and b) PtZn.  All  CVs presented in here were recorded at 50 mV/s in fresh electrolyte (Ar saturated 0.1 M H2SO4  ).  On both  electrodes,  the  charge  under  the  CVs decreased  as  more  CO- treatment was applied, suggesting the surface became smoother.; c)  CVs recorded (in  fresh  electrolyte)  on  PtZn  after  further  CO-treatment  was  applied.  Pt  oxide formation was shifted negatively, and hydrogen adsorption/desorption peaks were enhanced (indicated by arrows),  suggesting the surface became Pt-rich after  CO- treatment. Reprinted from [8]. Copyright 2009 with permission from Elsevier Ltd 122 Figure 5-7 shows the CO stripping voltammograms on the PtZn electrode recorded on the same days as curves 1 - 4 in Figure 5-6b were obtained (designated as Day 1 - 4). A dotted line at 0.92 V corresponds to the potential of the stripping peak recorded on a smooth Pt surface. The voltammogram for Day 1 was recorded with a freshly prepared PtZn electrode before CO treatment was applied. Each successive day (after CO treatment for ~ 3 hours), the main CO stripping  peaks  shifted  to  more  positive  potentials,  with  the  stripping  curves  also  changing shape. As expected from the CVs shown in Figure 5-6b, the charge under the voltammograms also diminished with time. On Day 1, there is only one peak observed at E = 0.77 V, but on Day 2, CO stripping occurred over a wider potential range, with the main peak at about 0.83 V with a shoulder at 0.78 V. On Day 3, no shoulder was observed in the stripping curve, but the main peak appeared at a potential between the peak and the shoulder observed on Day 2. On Day 4, the main peak was observed at a more positive potential than that of Day 3. This also shows that the  surface  is  changing  with  CO-treatment,  and  it  appeared  that  the  surface  of  PtZn  was becoming smoother and more Pt-rich – consistent with our observations from Figure 5-6. It was also noted that the charge density during CO stripping experiments generally increased as the electrode was exposed more to CO. The charge passed during CO oxidation was determined by integrating  the  CO  stripping  peaks  and  shoulders,  if  applicable,  and  Table  5-1 shows  the electrochemical  surface  areas  and  charge  densities  determined  by  the  method  outlined  in reference [55]. As mentioned in Chapter 2, Rodriguez et al., performed CO temperature programmed desorption studies on PtZn surface and found CO desorption temperature was less than that measured for Pt, suggesting that incorporation of Zn into the surface of the metal decreased substrate-CO  interactions  [112].  This  supports  our  conclusion  that  with  CO-treatment,  the surface becomes more like Pt.  Also,  the surface reorganization may occur  if  the interaction between the adsorbates and surface atoms are stronger than the interaction between the atoms in the surface  [152]. For example, Gritsch  et al., observed the reorganization of Pt(110) surface from (1×2) to (1×1) after CO was adsorbed onto the surface  [64]. It  is likely that the PtZn surface is  also going through a surface reorganization similar to  these reports.  Arenz  et  al., investigated the CO oxidation catalytic  behavior  on Pt  nanoparticles  [152].  As part  of  their electrode  cleaning  procedure,  they  compared  two  methods:  “Oxide  annealing”  and  “CO annealing”. The potential of the electrode was scanned in an Ar saturated acidic solution for the 123 “oxide annealing procedure”, whereas the potential was cycled in the CO saturated solution in the  “CO  annealing  procedure”.  They  previously  found  CO  annealing  removed  the  surface irregularities  similar  to  flame  annealing.  They  recorded  current  transient  curves  for  CO oxidation with the electrodes pretreated by the two different annealing methods and found that the nucleation and growth process for OHads occurred more slowly on CO annealed electrode than on oxide annealed electrode. This indicated that the CO annealed sample had a lower active site  density  than  the  oxide  annealed  sample  due  to  a  smoother  surface  after  the  annealing process. Figure 5-7 CO stripping voltammograms recorded at 50 mV/s in CO saturated 0.1 M H2SO4 (22 oC) on PtZn electrode over four days. CO-treatment was applied for approximately 3  hours  on  the  electrode  on each  day.  The  dotted  line  at  0.92  V corresponds to the CO stripping peak on smooth Pt.  As more CO-treatment  was applied  the  stripping  peaks  recorded  on  PtZn  were  moving  towards  the  peak potential  measured  on  smooth  Pt.  Reprinted  from  [8].  Copyright  2009  with permission from Elsevier Ltd. 124 Shoulders on the CO stripping peaks shown in Figure 5-7 have been observed. Cuesta et al., reported  a  CO  stripping  voltammogram  similar  to  the  curve  recorded  on  Day  2  and concluded that the shoulder corresponds to the oxidation of bridge-bonded CO, whereas the main peak corresponds to the oxidation of linearly-bonded CO [53]. Therefore it is possible that as the PtZn surface changed with CO-treatment and more sites  for bridge-bonded CO were created. Table  5-1 Electrochemical  surface  areas  and  the  charge  densities  of  CO  stripping voltammograms on a smooth Pt electrode and a CO-treated PtZn electrode (over 4 days). As PtZn electrode was treated with CO, the electrochemical surface areas decreased and the charge densities increased. Reprinted from [8]. Copyright 2009 with permission from Elsevier Ltd. * Calculated based on the double layer capacitance. Type of electrode Electrochemical surface area* (cm2) Charge density (µC cm-2) Smooth Pt 0.097 ± 0.007 6.17×102 PtZn (Day 1) 1.04 ± 0.08 4.44×102 PtZn (Day2) 0.85 ±  0.06 4.56×102 PtZn (Day3) 0.79 ± 0.06 5.12×102 PtZn (Day 4) 0.52 ± 0.04 6.08×102 125 5.3.4.  ORR and Zn enrichment of the alloy surface Pt enrichment of the PtZn surface was observed after CO-treatment. In Figure 5-6c, Hads/ Hdes peaks were enhanced after the treatment. We also realized Pt-oxide formation on PtZn was shifting cathodically as more CO-treatment was applied, similar to what we observed in Figure 5-4 when CVs of Pt and PtZn were compared.  In the previous chapter, we observed improved kinetics in ORR with our PtZn alloy as a catalyst compared to Pt. This improvement may be correlated to the potential shift in the oxide formation. Stamenkovic  et al., showed the same trend for Pt3Co and Pt3Ni  [26]. As shown, however, the PtZn surface became rough once the alloy  was  formed.  It  is  known  that  oxygen  dissociation  catalysis  on  a  rough  surface  is significantly  different  than  on  a  smooth  surface.  Adsorption  and subsequent  dissociation  of oxygen molecules on Pt surfaces have been extensively studied  [41,  40],  showing that  a Pt surface having steps and terraces exhibits more O2 adsorption/dissociation at step sites due to the stabilization of the dissociated transition state. This is due to the higher coordination number for Pt atoms on terrace sites than those on step sites [42]. Work reported by Gee et al., showed that the difference in the extent of oxygen dissociation process between Pt (111) (no step sites) and Pt (533) (with step sites) was governed by the presence of defects or steps  [41]. Maciá  et al., also observed higher current on Pt(n+1,  n-1,  n-1) surface (i.e.,, contains steps) than on Pt(2n- 1,11) surface (i.e., contains more terraces) during ORR  [27]. They also showed Pt electrodes with different crystal structures exhibited significantly different kinetics for ORR. The comparison made on the ORR curves, presented in  Figure 4-9 was between PtZn (with a rough surface) and a smooth Pt surface. To ensure a reasonable comparison, the ORR on the PtZn electrode was compared to the roughened Pt electrode which had a slightly rougher surface than the PtZn electrode.  The ORR anodic scans for the roughened Pt and the PtZn electrodes are shown in Figure 5-8 including an ORR curve recorded for a smooth Pt electrode. The rough Pt surface showed a +15 mV shift at the current density of -1.0 mA/cm2 compared to smooth Pt. On the other hand, the ORR onset potential for PtZn was approximately 35 mV more positive  at  the  current  density  of  -1.0  mA/cm2 compared  to  the  smooth  Pt  electrode. Gambardella et al., reported that having a Pt surface with step and terrace sites show much more pronounced ORR kinetics than on a Pt surface with only terrace sites. However, the authors also 126 discovered  that  when  the  stepped  sites  were  decorated  with  Ag,  the  reactivity  of  the  sites towards  oxygen  dissociation  decreased.  They  found  that  the  amount  of  oxygen  adsorption decreased even though Ag did not physically block Pt atoms at the edge, indicating an electronic perturbation caused by Ag adjacent to Pt atoms  [42]. Together with Gambardella's work, our observation strongly suggests that the shift is due to both roughness and electronic perturbation, which  resulted  from the  formation  of  a  PtZn  alloy.  Once  compensated  for  roughness,  it  is convenient to use the shift in onset potential for ORR as an indication of the the presence of PtZn alloy on the surface of the electrode. Figure 5-8 ORR curves recorded at 20 mV/s with the rotation speed of 2000 rpm in oxygen saturated 0.1 M H2SO4 (22 oC)on smooth Pt, rough Pt, and PtZn electrodes. Roughnesses of PtZn and “rough Pt” are equivalent. The rough Pt showed a 15 mV positive shift, whereas PtZn showed a 35 mV positive shift,  compared to the smooth Pt electrode at current density of -1.0 mA/cm2. Reprinted from [8]. Copyright 2009 with permission from Elsevier Ltd. 127 ORR voltammograms were recorded before and after CO treatment.  Figure 5-9 shows ORR curves  recorded on Pt and PtZn RDEs before (solid  line)  and after  (dashed line)  CO treatment. Polarization curves for Pt were identical before and after CO treatment, whereas PtZn showed a slight negative shift after the electrode was treated with CO. This negative shift also suggests Pt enrichment of the alloy surface after CO treatment. Figure 5-9 ORR curves  in kinetic-limited regions  recorded at  20 mV/s with the rotation speed of 2000 rpm in oxygen saturated 0.1 M H2SO4  (22 oC) on smooth Pt and PtZn electrodes before (solid line) and after (dotted line) CO-treatment. ORR curve  recorded  on  the  smooth  Pt  surface  is  plotted  on  the  PtZn  graph  for comparison. There is a slight negative shift observed after CO-treatment was applied to  the  PtZn  surface.  Reprinted  from  [8].  Copyright  2009  with  permission  from Elsevier Ltd. 128 The apparent Pt enrichment of the surface may be due to loss of Zn from the surface, or by changing the surface composition through segregation. Shen et al., observed an adsorption- induced  site  exchange  phenomenon  on  Al  alloyed  with  Pd(001)  [153].  During  temperature annealing above 800 K (as a part of the alloy-formation process), the authors found that Pd formed a “skin” over the surface, and it remained on the surface when the temperature of the substrate  was  dropped  to  room  temperature.  However,  when  the  alloy  was  exposed  to  a significant  amount of oxygen, Al segregated to  the surface.  This process was reversible;  Al migrated back into the bulk when oxygen was desorbed from the surface. They did not observe the  same  phenomenon  when  CO  adsorbed  onto  the  alloy  surface.  They  explained  this observation realizing that the formation of Al-O bonding is energetically favoured over Pd-O bonds and exposure to oxygen resulted in a site exchange [153]. According to this report, site exchange  between  atoms  in  the  bulk  and  on  the  surface  is  certainly  possible,  and  the composition of the surface is determined by the strength of metal-adsorbate interaction. We have already  demonstrated  Pt  enrichment  on  the  surface  due  to  CO  adsorption,  but  it  is  also reasonable to predict oxygen adsorption may enrich Zn at the surface. The influence of ORR on the behavior  of a PtZn surface was studied  after  various treatment regimes.  Figure 5-10 shows the ORR curves at low current regions recorded on the PtZn electrode previously exposed to CO for an extended period of time (labeled PtZn 1 in Figure 5-10). The same electrode was then exposed to an oxygen saturated solution and the potential was scanned between -0.3 and 1.05 V at 20 mV/s for several hours. Another ORR polarization curve was measured (labeled PtZn 2). The electrode was immediately re-exposed to CO saturated solution,  and the potential  was swept between 0.03 and 1.1 V at 50 mV/s for several  hours.  At the end of this  experiment,  another  ORR polarization curve was recorded (labeled PtZn 3). The ORR onset potentials on the PtZn electrodes were slightly different, with PtZn 2 showing the lowest overpotential. From our previous observations, the surface of an electrode becomes Pt-rich when it is CO-treated. The same is observed in Figure 5-10; previously CO treated PtZn electrode shows the  onset  of  ORR closer  to  that  of  smooth  Pt.  On the  other  hand,  when the electrode was exposed to oxygen for prolonged time, the onset of ORR occurred at a higher potential. Han et al., used density functional theory and Monte Carlo simulations to study the surface state of PtRu(111) as oxygen adsorbed onto the alloy surface. Under a vacuum environment, Pt migrates 129 towards the surface forming a Pt-skin. The authors found that though the atomic arrangements of Pt, Ru, and O are different depending on the chemical potentials of each atom, Ru generally migrates  towards  surface  when  oxygen  is  adsorbed  onto  the  alloy surface.  As  in  the  same argument made by Shen et al., the authors concluded that the strong interaction between Ru and O overcame the Ru surface energy which by itself is higher than that of Pt [154].  Shen et al., reported Al migration towards surface as the PdAl alloy surface was exposed to oxygen. Their explanation was simply that the enthalpy of formation of Al2O3 was much more negative than the corresponding constant of PdO [153]. We can make the same argument in our PtZn surface reorganization.  ∆Hoformation of ZnO is -350.5 kJ/mol  [155], whereas  ∆Hoformation of PtO2 is -80 kJ/mol [156] showing that Zn will be preferentially segregated to the surface due to a stronger interaction with oxygen. 130 Figure 5-10 ORR polarization curves recorded at 20 mV/s with the rotation speed of 2000 rpm in oxygen saturated 0.1 M H2SO4 (22  oC) on PtZn that has been treated with CO (PtZn 1), then with oxygen (PtZn 2), and then retreated with CO (PtZn 3). Negative  shift  in  the  onset  potential  of  ORR  on  PtZn  after  CO  treatment  was reproducible. ORR curve recorded on a smooth Pt surface is shown for comparison. Reprinted from [8]. Copyright 2009 with permission from Elsevier Ltd. 131 Another possible explanation for the decrease in the onset potential after CO-treatment is surface smoothing. Molecular oxygen adsorption and subsequent dissociation is more facile on a rough surface than on a smooth surface as discussed earlier. Since CO-treatment makes the alloy smoother, this might have decreased the amount of oxygen adsorbed on the surface of the PtZn alloy. In fact, Lebedeva et al., has studied the CO oxidation on a Pt surface containing both steps and  terraces  and  concluded  that  oxidation  reactivity  is  enhanced  at  the  step  sites  because oxygen-containing species adsorption preferentially occurs at step sites rather than at terrace sites [157]. This suggests concentration of oxygen adsorbed at step sites is higher than at terrace sites, which may be related to the fact oxygen dissociation preferentially happens at step sites [41].  Losing  these  step  sites  by  CO-treatment  will  decrease  the  ORR  onset  potential.  Our observation in Figure 5-10 also suggests that the PtZn surface becomes rougher after oxygen- treatment.  Nevertheless,  the extent  of shifts  observed in  onset  potentials  of  ORR after  CO- treatment or oxygen treatment was very reproducible, indicating the site exchange between Pt and Zn occurred within the metal. Therefore, we believe it is unlikely that the negative shift after CO treatment was due to irreversible loss of Zn from the electrode surface. 5.3.5.  Surface reorganization and the CO oxidation reaction mechanisms Both  CO-  and  oxygen-treatments  of  the  PtZn  alloy  have  altered  its  surface,  which subsequently led to a change in the adsorption of oxygen or CO. As mentioned in  Chapter 2, Buatier et al., and Rodriguez et al., showed a decrease in Pt(5d) electron density and a decrease in CO desorption temperature on PtRu and PtZn alloys, respectively, compared to pure Pt [112, 146]. In fact, these changes have also been observed with Pt3Co [76], PtTi [115, 77], and PtMo [116]. These Pt-alloys have been reported to enhance ORR catalysis compared to pure Pt, but only PtRu has been experimentally well-known to improve CO oxidation catalysis relative to Pt. This may be related to the site exchange process. The adsorption energy of CO onto Pt(111) surface is determined to be -1.39 ~ -1.43 eV[81]. On pure Zn, however, no adsorption of CO was observed [119]. On pure Ru(0001), CO adsorption energy was reported as -1.70 eV [81]; similar to that on Pt. For PtZn exposed to CO, Pt migrated to the surface, leaving Zn in the bulk. In the case of PtRu, because CO adsorption energies on both atoms are very similar, the surface 132 composition does not change dramatically. Bifunctional theory suggests that Ru provides -OHads to neighboring Pt sites  [82] which is reasonable since the binding energy of -OH onto Ru is approximately 5 times greater than on Pt  [81]. The observed site exchange occurring on PtZn may add to the developing picture addressing the mechanism of ORR (or CO oxidation) on Pt alloys,  which  is  currently  explained  by  a  decrease  in  Pt  (5d)  electron  density,  the electronegativity of admetals, and decrease in the Pt-Pt bond distance [74]. 5.4.  Summary and conclusions The work described in this chapter details the roughness-compensated CO oxidation and the ORR electrocatalysis on the Pt and PtZn surfaces. PtZn showed no improvement on the CO oxidation  catalysis  as  compared  to  Pt.  Apparent  enhancement  in  catalytic  activity  on  PtZn compared to smooth Pt was purely due to the roughness of the PtZn surface. On the other hand, PtZn showed a better catalytic activity for ORR compared to Pt surface of equivalent roughness. Moreover, we observed Pt-enrichment and smoothing on the PtZn surface after extended CO- oxidation,  whereas we observed Zn-enrichment and possible roughening of the surface after extended oxygen-reduction.  Comparing Ar saturated CVs of Pt and PtZn in sulfuric acid, there was approximately 80 mV anodic shift at current density of 20  µA/cm2 in the onset of Pt-oxide formation for PtZn. SEM-EDX  confirmed  the  presence  of  Zn  on  the  Zn-modified  Pt  surface  before  and  after electrochemical treatments were applied. SEM also showed that Zn modification of Pt produces a rougher surface and the preferential formation of PtZn on various surface regions. Segregation of selected atoms to the surface was explained by simple thermochemical effects. More detailed studies on the surface segregation effect, composition of Zn-modified Pt surface, and the stability of PtZn are presented in the next chapter. 133 6.  Stability of PtZn Alloy in Acidic Aqueous Environment 6.1.  Introduction A number of Pt-alloy catalysts have demonstrated an improvement in kinetics of ORR. Many of them have been tested in fuel cells and also shown promising results. However, one of the challenges researchers have faced upon developing fuel cells is the chemical stability of catalysts  [26, 97].  During fuel cell  operation,  the non-precious metal  components of the Pt- alloys typically leach out into the electrolyte, leading to five unfavourable outcomes: 1) decrease in ionic conductivity of membrane due to dehydration as cation contaminants take water away from membrane [158]; 2) increase in ionic resistivity in the cathode due to blocking of cation exchange sites; 3) decrease in oxygen transport to the catalyst layer; 4) damaging the membrane if a certain metals, such as Ti or Fe, are released [97]; and 5) decrease in catalytic activity [159]. As shown in Chapter 2, the stability of PtZn alloy in an acidic condition is theoretically quite  promising  and,  if  so,  this  is  an  advantage  for  the  fuel  cell  application.  This  chapter investigates the chemical stability of PtZn under conditions similar to an operating fuel cell. There are three sections to this chapter; the first two are the results of preliminary stability tests, with  more  concrete  experimental  results  at  the  end.  They all  involve  applying  a  prolonged polarization (i.e.,  constant  potential)  to  PtZn in  acidic  electrolyte.  The first  experiment  was performed by Ms. Olivia Ye as the fourth year Honours project which was supervised by Dr. Bizzotto and myself. Stability of the alloy was tested by weighing the alloy before and after polarization, looking for evidences of weight loss due to leaching. The second part involved determination of the concentration of Zn in the electrolyte after the polarization using anodic stripping  voltammetry.  Unfortunately  these  two  sets  of  experiments  did  not  provide  us convincing results on the stability of PtZn. The  last  section  of  this  chapter  investigates  the  chemical  stability  of  a  PtZn  alloy particles dispersed onto a glassy carbon (GC) substrate. This section builds upon our findings in 134 Chapter  5.  Those  results  indicated  that  the  PtZn  surface  became  enriched  with  Pt  upon interaction  with  CO,  while  Zn segregated  to  the  surface  upon the  interaction  with  oxygen. Previous work was performed on large Pt samples which resulted in uncertainty in the surface concentration  of  Zn;  both  the  concentration  of  Zn  and  the  possible  Zn  dissolution.  A GC substrate will be used  instead of bulk Pt in this chapter. When we analytically observed the decrease in Zn concentration in the alloy in Chapter 5, we were unsure if Zn was leached out into solution or diffusing deeper into the Pt substrate. However, Zn does not appear to react or form an  alloy  with  carbon  under  the  conditions  used  [106],  so  Zn  will  not  persist  on  the substrate. The use of small nanoparticles of Pt on GC would improve our ability to determine the fate of the Zn atoms in the alloy under electrochemical environments. We predicted that, for small particles at a fixed applied potential, the surface concentration of Zn would increase until it reached an equilibrium concentration. If the alloy is stable, the surface concentration would eventually plateau. If the alloy is unstable, the surface concentration would initially increase but quickly decay due to constant Zn dissolution, leaving only Pt on GC substrate.  The surface concentration  of  Zn  was  determined  by  Auger  electron  spectroscopy  (AES)  and  the  bulk composition was determined by dissolution of the remaining metal and analysis by inductively coupled plasma mass spectroscopy (ICP-MS). 6.2.  Experimental method: 6.2.1.  Preliminary test 1 – Gravimetric determination of Zn loss This section investigates the stability of PtZn by measuring the weight loss before and after the stability test,  i.e., polarizing PtZn foils. This set of experiment was performed by Ms. Olivia Ye as her 4th year Honours project  at Department of Chemistry,  University of British Columbia, supervised by Dr. Dan Bizzotto and myself. 135 6.2.1.1.  Preparation of PtZn foil Three Pt foils (25  µm 99.99% Alfa Aesar) were cut into 2 cm by 0.4 cm strips. The weight of each foil was measured using a microbalance (Mettler, Toledo) with accuracy of 0.1 µg. To ensure accuracy, the foils were heated at 150  oC overnight, followed by cooling in a desiccator to eliminate moisture. PtZn alloy was formed on the foils as outlined in Chapter 3. After Zn deposition, the foil was removed from the solution, rinsed with Millipore water, dried with Ar, and was left under Ar atmosphere for 1 hour. Then the foil was either immersed into the acidic  solution  to  the  strip  remaining  Zn  from the  surface  or  placed  in  a  300  oC  furnace overnight. The latter step was introduced to deposit a large amount of Zn into the Pt foil. After heat treatment, unreacted Zn was then removed with warm acid mixture of HNO3 and H2SO4. The sample was then dried in 150  oC oven overnight.  The foils were again weighed on the microbalance (after cooling in a desiccator) and the  difference in mass before and after Zn deposition was determined. 6.2.1.2.  Stability test The PtZn foils thus prepared were then immersed into a 0.1 M H2SO4 solution. It was necessary first to remove un-alloyed Zn that was driven into the bulk of the foil. All the foils were first polarized at 0.5 V for 36 hours. In a fresh electrolyte, the foils were then polarized at 0.5, 0.75, and 1.0 V for 2 h and 24 h in an oxygen saturated 0.1 M H2SO4 solution at 22 oC. The foils were then removed from the solution, washed with the acid mixture and water, dried in a 150  oC oven overnight,  and weighed on a  microbalance.  The amount  of  Zn lost  during the polarization process was determined by comparing the weights before and after polarization. 136 6.2.2.  Preliminary test 2 – Determination of Zn loss by anodic stripping method Platinum foil (polycrystalline) of 25 µm thickness (Alfa Aesar, MA 99.99%) was cut into three strips of 1 by 5 cm sheets. The electrochemical deposition of Zn onto these Pt foils was performed as outlined in Chapter 3. The potentiostat used was the μAutolab (EcoChemie). Zn deposition/stripping processes were repeated for ten times to ensure a copious amount of PtZn alloy was formed on the foil. The amount of Zn that presumably alloyed with Pt was calculated by taking the difference between the charge measured for deposition and stripping in the current transient curves. The stability test was carried out in a similar manner to the Preliminary test 1 for 12 hours, but after the polarization, the foils remained in the same solution for another 12 hours  under  ambient  conditions  without  any applied  potential.  The  Zn concentration  in  the electrolyte was examined. 6.2.2.1.  Anodic Stripping- differential pulse voltammetry Anodic  stripping  voltammetry  (ASV)  analysis  was  performed  on  a  commercially available  glassy  carbon  rotating  disc  electrode  (GC  RDE)  (Metrohm,  Switzerland).  The electrode was first polished slightly with 4000 grit sandpaper to ensure a fresh surface, followed by 6 µm and 1 µm diamond slurry. A 50 mL of an acetate (Fisher Scientific) buffer solution (pH 4.5, 0.15 M), containing 2.29410-3 M Hg2+ (Bethlehem Apparatus Co., PA) dissolved in nitric acid, was measured into the electrochemical cell with 5.00 mL of the electrolyte solution in which the stability test was performed and 30 μL Cd2+ solution (7.9110-4 M, Cd(NO3)2·4H2O, Fluka) as an internal standard. First, the GC electrode was rotated at 1000 rpm and polarized at – 0.2V for 60 seconds in the acetate buffer/Hg solution (deaerated with Ar gas) in order to deposit only Hg onto the electrode. The Zn2+ was reduced into the Hg film using a 50-second deposition at –0.9 V and a 7 second equilibration (i.e., stopping the rotation). The potential was then swept positively (to -0.1 V) at a sweep rate of 20 mV/s to oxidize Zn and Cd from the Hg thin film. The differential pulse method was carried out using a modulation time of 0.005 sec, the interval time of 0.1 sec, and the modulation amplitude of 10 mV. At the end of each run, the electrode 137 was polarized at 1.0 V to strip off deposited Hg from the GC surface. Also, after one ASV run, the electrode was then removed from the solution and sonicated in a methanol and water mixture to ensure the surface of the electrode was clean before the next deposition of a fresh Hg film. The concentration of Zn was determined by the standard addition method. 6.2.3.  PtZn nanoparticle stability test: 6.2.3.1.  Pt particle deposition The working electrode was a GC (Tokai Carbon) substrate machined in a “top hat shape” with a surface area of 1.23 cm2, as shown in Figure 6-1. Each substrate was polished with 4000 grid sandpaper, followed by 6 µm and 1 µm diamond slurry. It was then accommodated into a home-made electrode holder, also shown in  Figure 6-1. The substrate was placed on a shaft where the bottom face contacts an aluminum plate. The substrate was screwed tight by a Teflon holder between which flexible Teflon o-ring was placed to ensure a good seal. Pt deposition onto the substrate was carried out in a 0.1 M H2SO4 (93-98 %; SeaStar; Baseline) solution containing 1 mM of H2PtCl6 (Aldrich) prepared with Millipore water (22 oC). Solution was deaerated with Ar (Praxair) before deposition. The potentiostat used was USB- interfaced µAutolab (Metrohm, Switzerland). Nucleation potential of 0.0 V was applied for 10 millisecond followed by a growth potential of 0.16 V for 10 sec. 6.2.3.2.  PtZn particle preparation PtZn particles  were prepared by depositing Zn onto Pt particles  previously prepared, followed by a heat treatment. A thin film of Zn was first deposited onto the Pt-dispersed GC substrate by potential cycling, and then bulk Zn was deposited on top by applying the potential of -0.81 V for 10 minutes. This created a thick layer of Zn covering the whole surface of the substrate. The substrate was placed in a 120 oC preheated furnace for an hour to form a PtZn alloy. After the substrate was cooled to room temperature, unreacted Zn was dissolved in diluted 138 sulfuric acid. As-prepared PtZn particles were analyzed by AES giving surface compositional analysis.  Also,  the  particles  were  dissolved  into  aqua  regia  (1:3  HNO3:HCl),  the  diluted  to 100.00 mL, and the solution was analyzed in ICP-MS for bulk compositional analysis. Figure 6-1 Home-made working electrode used for a PtZn nanoparticle stability test. A sheet of glassy carbon was machined in a top hat shape with a surface area of 1.23 cm2.  The  substrate  was  then  accommodated  into  an  electrode  holder  where  the bottom face of the substrate  is  held tight to  an aluminum plate.  Teflon o-ring is placed in between the substrate and the electrode holder to ensure a tight seal. 139 6.2.3.3.  PtZn stability test 6.2.3.3.1.  Polarization of PtZn particles PtZn particles thus prepared were polarized at potentials of 0.75, 1.0, 0.8, or 1.25 V for various amount of time in an oxygen saturated 0.1 M H2SO4 solution at 22 oC. The polarization time was up to 14, 20, 40, and 2.5 hours, respectively. After the polarization, the electrode was immediately taken out, rinsed with Millipore water, dried with argon, and placed into a UHV chamber for AES analysis. 6.2.3.3.2.  Auger electron spectroscopy Auger  electron  spectroscopy  (AES)  instrumentation  and  operating  conditions  are outlined in Chapter 4. Point analysis was performed on PtZn particles. Spectra were recorded at kinetic energy ranges of 700-1200 eV for Zn and 1800-2100 eV for Pt. 6.2.3.3.3.   Inductively coupled plasma – mass spectroscopy measurement PtZn particles  (as  prepared  and  treated  with  potential)  were  dissolved  in  aqua  regia (HNO3 + HCl; Fisher Scientific) and the solution was analyzed by inductively coupled plasma- mass spectroscopy (ICP-MS; ELAN 6000) in order to determine the bulk composition of the metal remaining on the GC. The instrument was calibrated using a standard solution containing Ba, Be, Ce, Co, In, Li, Mg, Pb, Rh, Tl, U, and Y. After polarization, the particles were also dissolved in aqua regia to determine the bulk Pt and Zn content from a calibration curve, and Zn loss was calculated. 140 6.3.  Results and discussion: 6.3.1.  Preliminary test 1 – Determination of Zn loss by gravimetry PtZn foils were created by deposition of bulk Zn onto a Pt foil, followed by stripping of unreacted Zn. According to deposition /stripping current transient curves, the amount of Zn that alloyed to Pt after a spontaneous alloying process (i.e., leaving the sample under Ar atmosphere) was only ~4 % of the amount initially deposited. Therefore it required prolonged deposition numerous times to detect weight changes before and after alloying. Heat treatment was therefore introduced to encourage a sufficient amount of Zn to diffuse into the foil. This method appeared to drive bulk amount of Zn into the Pt foil very rapidly without letting most of the Zn to form bonds with the Pt atoms, possibly because the temperature was too high. When the foil was polarized at non-oxidative potential  (i.e., at 0.5 V) or at oxidative potential (i.e.,  1.0 V), we observed a significant amount of Zn loss at both potentials by the same extent. Pre-polarization leaching was therefore necessary, and this was done by applying the potential of 0.5 V to the PtZn foils in 0.1 M H2SO4 until no further weight loss was observed. This process took about 36 hours. The foils were then polarized at 0.5, 0.75, or 1.0V up to 24 hours in an oxygen saturated acidic solution. The percent loss of Zn (the total amount of weight lost compared to the initial amount of Zn deposited) after the polarization tests are shown in Table 6-1. Polarization at 1.0 V showed the most Zn loss from the PtZn foil, although the other foils also showed similar loss within 7 %, and overall the amount of Zn (presumably) leached out was not as significant as compared to  those reported in  literature  [69]. However,  it  is  not  clear  whether  this  Zn loss simply  came  from  the  residual  un-alloyed  Zn  in  the  bulk.  Due  to  time  constraints,  the experiment was performed only once, and a better experimental design was necessary. Therefore further reproducibility for this experiment was not tested. 141 Table 6-1 Stability of the PtZn alloy at various fixed potential in a 0.1 M H2SO4 solution for 24 hours determined by gravimetry. Polarization potential (V vs. RHE) PtZn foil Weight lost (µg/cm2) % loss of Zn (wt/wt)* 0.5 3 3 0.75 7 4 1.0 5 7 6.3.2.  Preliminary test 2 – Determination of Zn loss by anodic stripping voltammetry PtZn alloy formation was performed as outlined in the experimental section. The foils prepared were polarized at 0.75, 1.0, and 1.3 V in a 0.1 M H2SO4 solution for 12 hours at 22 oC. After the polarization, the foils remained in the same solution for another 12 hours under the ambient conditions. The initial amount of Zn deposited onto the Pt foil was predicted according to the previous preliminary test 1 showing that only 4% of Zn deposited on Pt remains as an alloy. The resultant solution was analyzed using the ASV technique with standard addition and internal  standard  methods.  No significant  loss  was  observed with  0.75 and 1.0 V polarized samples (only 1~3 % loss), whereas 1.2V one showed 78.1 ± 10.0 % loss of Zn leached out of the  foil.  Although  the  error  associated  with  these  measurements  were  high,  the  result  was consistent with Despic's work  [106]. However, the measurements were not always consistent, possibly due to contamination of solutions. Therefore we proposed the next stability test which involves dispersion of PtZn particles on glassy carbon substrates. 142 6.3.3.  Pt particle deposition Pt was  electrochemically  deposited  onto  a  GC substrate  as  outlined  in  Experimental section. A deposition current transient curve is shown in Figure 6-2 and an SEM image of the surface of the GC substrate is shown in Figure 6-3. The typical particle size was 50 ~ 100 nm with a round shape. AES was recorded on those particles, and they showed the Auger transition lines at kinetic energies of 1700, 1780, 1960, and 2400 eV, in agreement with the results in Chapter 4.  During recording, however, it  was noted that the beam current measured on the particles was approximately five times less  than the current  measured on the GC substrate, indicating the poor adhesion of Pt particles or small contact areas between the particles and the GC substrate. 143 Figure 6-2 Current transient curve showing the deposition of Pt particles onto a glassy  carbon  substrate.  The  working  solution  was  deaerated  0.1  M  H2SO4 containing 1 mM H2PtCl6. Nucleation of Pt particle was performed by applying the nucleation  potential  of  0.0  V for  10  seconds,  followed  by  applying  the  growth potential of 0.16 V for 10 seconds. 144 Figure 6-3 SEM image of Pt particles dispersed on a glassy carbon substrate. The particles are spherical with diameters typically 50 ~ 100 nm. 145 6.3.3.1.  Zn deposition on the Pt particles Once Pt particles were formed on the GC substrate, Zn was also electrodeposited onto the substrate. The deposition potential was determined similar to the one introduced in Chapter 4;  i.e., ~50  mV more  negative  than  the  reversible  potential.  However,  before  the  bulk  Zn deposition, it was necessary to form a thin coating of Zn over the GC substrate because the bulk Zn deposition requires a large negative potential causing extensive hydrogen evolution on Pt particles. As a result, Zn would not be deposited uniformly. The thin coating of Zn was formed by  20  potential  sweep  cycles  between  -0.8  V  and  -0.63  V at  50  mV/s,  where  Zn  is  not completely oxidized from the GC surface, as shown in a dotted line in Figure 6-4. Once the thin film was formed, bulk Zn was deposited by applying a constant potential. However, when the alloy formation was attempted under an Ar atmosphere, we were not able to detect any Zn in AES. This was probably because of the poor adhesion of Pt onto the GC substrate.  Although the alloy formation occurs very quickly,  it  requires a reductive potential applied to Pt and Zn atoms. Because of the poor electrical contact between the Pt and carbon, sufficient current did not flow to form the alloy. In fact, when the deposition time was increased to 1 hour, alloyed Zn was detected in AES. However, this is time consuming and also presents a high risk of contaminating the surface. Therefore, the PtZn preparation procedure was modified. Zn was electrodeposited,  and then a  heat  treatment  at  120  oC was applied for  1  hour.  This temperature is below the boiling point of surface Zn, which is ~460 K [144] and also, as shown previously, heating at 300 oC may send bulk Zn deep into the Pt particles without forming Pt-Zn bonds. After the heat treatment, Zn was successfully alloyed to Pt. In Chapter 4, we deposited Zn onto a Pt-dispersed GDE. We described that the PtZn was spontaneously formed after the Zn deposition was performed on bulk Pt; however, considering that Zn did not spontaneously alloy with Pt particles of size 50~100 nm, it is reasonable to assume the Pt nanoparticles in the GDE also did not naturally form alloys after the Zn deposition. Rather, they formed alloys when the GDE was hot-pressed to the MEA, which requires elevated temperature and pressure. 146 Figure 6-4 Zn thin film deposition over the Pt dispersed glassy carbon substrate was performed by sweeping the potential  between -0.8 V and -0.63 V 20 times in  a deaerated 0.1 M zinc sulfate solution at  the sweep rate of 50 mV/s (shown in a dotted line). The CV in a solid line shows a complete oxidation of deposited Zn, whereas partial oxidation of deposited Zn leaves a thin film of Zn over the glassy carbon substrate. This will be useful when a bulk Zn deposition is applied in the following experimental step as hydrogen evolution from Pt particles are suppressed. 147 6.3.3.2.  Zn-modified Pt particles Figure 6-5 shows the AES of Zn and Pt recorded from a single particle. Only the Zn and Pt kinetic energy ranges were selectively scanned because the scale of a spectrum in general is set according to the highest peak of the recorded spectrum. If a survey scan is recorded for the entire or even a small part of the substrate, the spectrum is adjusted to the large carbon peak, scaling both Pt and Zn down.  Integrating the Zn LM2  (at 990 eV) and Pt MN3 (at 1960 eV) peaks, the average composition of Zn and Pt were determined to be  4.2, and 95.8 ± 1.6 atomic %, respectively. It was noted in Chapter 4 that peak heights of Pt MN3 and MN2 (at 2040 eV) were almost 1:1 for most of particles, suggesting chemical interaction between Pt and Zn. The particles on the GC substrate were dissolved in aqua regia, and the solution was analyzed  for  Pt  and  Zn  by ICP-MS.  This  was  to  confirm the  bulk  PtZn  composition.  The maximum  Zn content in the alloy particles was determined to be approximately 40 atomic %, assuming all Pt reacted with Zn. This is consistent with the data obtained in  Chapter 4; the composition of the bulk alloy was determined to be 42 atomic % Zn. 148 Figure 6-5 Auger electron spectra of Zn (left) and Pt (right) recorded from a single Zn-modified Pt particle after the heat treatment. The average % amount of Zn on the surface on as-prepared particles was 4.2 %. 6.3.3.3.  Polarization of PtZn particles The Pt and PtZn particles on GC substrates were then polarized at potentials of 0.75, 0.8, 1.0, and 1.25 V for various durations to test stability. These potential values were chosen for the following reasons: (1) Potential of 0.75 V is where both Pt and PtZn do not participate in any reaction, according to the Ar saturated CVs, and therefore the metals are expected to be stable at this potential. (2) Potential of 1.0 V (or 0.8 V) is where oxidation of Pt or PtZn starts to occur and we could not predict if the metals were stable. (3) Potential of 1.25 V is where PtZn is known to oxidize, and therefore we would be able to monitor the Zn dissolution behavior. The substrates were then rinsed with water, dried with Ar, and immediately transferred into the UHV 149 chamber for AES analysis. SEM images showed that after any polarization process, the particles became larger by the factor of 3 to 5 than those on unpolarized substrates due to the sintering process. Sintering is the agglomeration of Pt particles during the fuel cell operation, decreasing the active catalytic surface area. Sintering mainly occurs either by dissolution/ re-precipitation of particles or by the surface diffusion [160, 161]. The particles tend to agglomerate during the polarization, making them larger. Also, some of the larger particles appeared to be spherical. AES on such particles showed a trace amount of Zn, but with the surface mainly Pt, possibly because Zn diffused into the bulk. Also, one of the difficulties in AES analysis was to obtain a number of scans from a single particle because when illuminated by the electron beam, the particles moved due to the poor particle-substrate adhesion. However, there also were particles that had irregular shapes and also appeared to adhere very well to the substrate, as shown in Figure  6-6.  These  particles  were  not  affected  by  sintering  and  seemed  to  be  a  good representative of PtZn particles. For these reasons, the Zn concentrations were measured on such irregular shaped particles. Such strong adhesion of a Pt-alloy onto a carbon substrate (known as the anchoring effect) has also been reported for the PtFe  [162] and for the PtNi nanoparticles [163] Although the reasoning behind the effect was not clarified by these reports,  they both observed the suppression in particle sintering and strong adhesion to carbon with these Pt alloys. The PtZn alloy we prepared on the GC substrate shows the same phenomenon. This anchoring effect is certainly beneficial in a fuel cell application since the sintering decreases the active surface area for catalysis and results in an uneven distribution of catalyst particles in an MEA [131]. Figure 6-7 shows the surface Zn concentration (in atomic %) with respect to the duration of  polarization  measured  at  the  applied  potential.  The  first  hour  of  polarization  at  1.0  V increased the Zn surface concentration to 6.5 atomic %, and it further increased to 19.3 atomic % after 20 hours of polarization. We did not observe a decrease in the surface concentration, meaning either the particles are stable, or some Zn may have leached out, yet the particles may contain sufficient Zn to maintain an equilibrium surface concentration.  It was also noted no oxygen was detected in AES, indicating no metallic Zn was left on the surface. The polarization potential  of 0.75 V is, according to the Ar saturated CV, the region where no obvious redox reaction occurs in the absence of oxygen, and therefore we predicted PtZn is probably stable at this potential. In fact the open circuit potential of the electrode was 150 0.95  V;  much  more  oxidative  than  the  applied  potential.  After  1  hour  of  polarization,  the concentration of Zn did not seem to change from unpolarized sample (4 atomic %), but after 5 hours,  the  concentration  increased  to  7.13  atomic  %.  Further  polarization  did  not  seem to dramatically change the surface concentration; 7.76 atomic % after 12 hours. This indicates the equilibrium surface composition at the applied potential of 0.75 V is approximately 8 atomic % Zn. No further polarization was applied as PtZn is considered stable at this potential. Polarization at 1.25 V provided quite different results. Only 30 minutes of polarization made the  surface concentration of  Zn 3.22 atomic  %; less  than  the unconditioned one (4.2 atomic %). After 2.5 hours of polarization, AES detected Zn just above the detection limit of the instrument, which is 0.2 atomic %. It is clear from this plot that at 1.25 V, PtZn is unstable and Zn is completely leached out of the alloy. 151 Figure 6-6 SEM image of PtZn particles on the glassy carbon substrate. Unlike Pt particles that appeared round and showed poor adhesion, PtZn particles appeared in irregular shapes and adhered to the substrate well. Generally, Pt particles grew in size as longer polarization was applied due to sintering, but this was not an issue with PtZn particles. 152 Figure 6-7 The change in surface concentration of Zn with respect to duration of polarization at the potential of 0.75 V (filled circle), 1.0 V (square), or 1.25 V (empty circle).  Polarization  was  performed  in  an  oxygen  saturated  0.1  M  sulfuric  acid solution at 22 oC. Zn concentration on the surface increases as the duration increased for the samples polarized at 0.75 and 1.0 V. They even reached a plateau, but it never decreased during the polarization period, indicating the surface became Zn-saturated with or without having more Zn leached out into the solution. On the other hand, the samples polarized at 1.25 V showed a rapid decrease in the surface Zn concentration indicating PtZn is unstable at this potential and Zn immediately leaches out from the bulk into the solution. PtZn particles polarized at 1.0 V for 12 h and 20 hours were further analyzed by ICP-MS after the AES analysis. The results are tabulated in  Table 6-2. The particles were dissolved in aqua regia, and the solutions were analyzed. After 12 h polarization, the sample did not show a drastic  difference  compared  to  the  unpolarized  sample;  only  0.77  ±  0.09  %  loss.  The measurements  after  20  h  of  polarization  were  duplicated,  and  the  average  Zn:Pt  after  the polarization  was determined to  be  0.59:1.  Polarization  of  PtZn for  30 h at  1.0  V was  also attempted, but this treatment oxidized and damaged the surface of GC so that it delaminated 153 from the surface. Therefore, we performed the polarization tests for 30 h and 40 h at 0.8 V and then dissolved the remaining particles  in aqua regia,  and analyzed the solution by ICP-MS. Auger analysis on these samples were not recorded due to technical constraints. After 30 h, the amount of Zn lost was 16.1 ± 1.9 %, and after 40 hours, 18.8 ± 2.3 %. It should be noted that only  one  sample  (except  for  1.0  V,  20  h  polarization  samples)  was  available  from  each polarization condition. Reproducibility of these measurements needs to be assessed in the future. Although we observed approximately 20% loss of Zn after 40 hour polarization at 0.8 V, the composition of PtZn was about 1:0.53 (almost 3:1); still retaining a high Zn content. The Pt-Zn phase diagram (Figure 2-15) shows Pt3Zn, as well as Pt1Zn1, is stable as a bulk intermetallic compound. Colon-Mercado and Popov studied the stability of 20 wt% PtCo and PtNi MEAs in a similar manner to ours (alloy compositions of 1:1 and 3:1 for both alloys). The authors reported 40 % and 30 % of the transition metal dissolution from Pt1Co1 and Pt1Ni1,, respectively, after only a few hours of polarization. Stabilities of Pt3Co1 and Pt3Ni1 were, on the other hand, much more  improved;  less  than  10  %  metal  dissolution  over  60  hours  of  polarization  [163]. Considering Zn loss between 30 and 40 hours was only about 2 %, Pt3Zn may be significantly more stable than other Pt alloys. In the literature, Pt3M1 alloys (where M is to a transition metal) show a better stability; less than 10% transition metal dissolution over 80 hours of polarization, whereas Pt1M1 alloys showed almost 40% dissolution within an hour of polarization in an acid solution [69]. 154 Table 6-2 Relative Zn loss from an initial concentration on the nanoparticles calculated from ICP-MS measurements (Zn:Pt was 0.653 :1 before polarization) Polarization potential (V) Duration (hour) Zn loss (%) Zn:Pt ratio 1.0 12 0.77 ± 0.09 0.64:1 20 8.9 ± 1.1 0.59:1 0.8 30 16.1 ± 1.9 0.54:1 40 18.8 ± 2.3 0.53:1 The apparent PtZn stability relative to other Pt alloys may be explained by considering the Pt-Pt and M-M (where M is a transition metal) bond lengths. Various types of bond lengths in Pt-transition metal alloys and in the PtZn alloy are tabulated in  Table 6-3 [101, 70]. In the table, percent changes in Pt-Pt and M-M bond lengths are also shown (The presented values were not obtained in an equivalent method, but the trends in the changes would be apparent by calculating  the  percent  differences).  As  described  in  Chapter  2,  the  Pt-Pt  bond  distance generally decreases when a Pt-M alloy is formed, which is evident from -Cr, -Co, and -Ni alloys, where  the  Pt-Pt  bond distances  decrease  by 3  %  [70].  In  contrast,  M-M distances  tend  to increase after Pt-alloys are formed; 5.2 and 1.6 % changes in -Co and -Ni alloys, respectively. This may suggest that only the transition metals contribute to heteronuclear bonding in Pt-M alloys.  The heteronuclear bond distances in these alloys  are slightly smaller  than their  Pt-Pt distances. In the PtZn alloy, on the other hand, the Pt-Pt distances increased to 2.81 Å in PtZn from 2.77 Å in bulk Pt (+1.44 % change). The Zn-Zn bond lengths changed significantly from the value for bulk Zn (2.66 to 2.85 Å corresponding to + 8.64 % change). Interestingly, the Pt-Zn bond  length  was  much  shorter  (2.65  Å)  than  the  homonuclear  bond lengths,  reflecting  the increase in electron density in the bimetallic bond which makes the Pt-Zn bond strong. In fact, the Pd-Zn and Pt-Zn surfaces displayed the largest changes in the CO desorption temperature 155 and were the first and second strongest metal-Zn bimetallic bonds respectively in the DFT study [101]. This suggests that these alloys may be significantly stable, which is important for a fuel cell catalyst. The changes observed for CO adsorption on the Pt/Zn alloy surface, as compared to Pt, is similar to that reported for Pt alloyed with the early transition metals [112]. Table 6-3 Pt-Pt and M-M bond lengths  (M denotes the transition metal) in the pure forms or in the Pt alloys; percent change in bond lengths before and after alloying. The values for Pt, PtCr, PtCo, and PtNi were taken from [70], whereas the ones for PtZn were taken from [101]. PtCr, PtCo, and PtNi show approximately 3 % decrease in Pt-Pt bond distance after alloying, although the Pt-Pt bond distance in PtZn increased by 1.4 %. On contrary, M-M bond distances in PtCr, PtNi, and PtZn increased. This may be because only transition metal electrons contribute to the bimetallic bonding. Pt-M bond distance in PtZn is considerably shorter in PtZn than Pt-Pt or Zn- Zn.  Pt appears to contribute  to  the bimetallic bonding in  PtZn,  which may be important in designing a stable catalyst in fuel cells. M Pt-Pt (Å) Pt-M (Å) M-Mpure (Å) M-Malloy (Å) % change (Pt-Pt) % change (M-M) Pt 2.77 2.77 PtCr 2.71 2.69 2.49 -2.16 PtCo 2.68 2.63 2.50 2.63 -3.25 +5.20 PtNi 2.68 2.61 2.49 2.53 -3.25 +1.61 PtZn 2.81 2.696 2.66 2.89 +1.44 +8.64 156 Another  interesting observation was the AES peak height  ratio  between Pt MN3 and MN2. Figure 6-8 shows the typical AES Pt signals from the PtZn particles polarized at 1.25 V for 1 hours (a), 1.0 V for 20 hours (b), and 0.75 V for 14.5 hour (c). In Chapter 4, we observed that the MN3 and MN2 peak heights become similar when Pt interacts with Zn. As shown in the figure, peak heights at polarization at 0.75 V and 1.0 V show almost 1:1 peak heights (the ratio of 1.30 and 1.28, respectively), whereas it is no longer the case for the 1.25 V polarized sample (the ratio of 1.52). This is another indication that polarization at 0.75 V and 1.0 V retains the chemical interaction between Pt and Zn for a prolonged period, in contrast to 1.25 V. The peak ratio for the pure Pt determined in Chapter 4 was 1.86, therefore the polarized sample at 1.25 V may be retaining a little interaction between Pt and Zn. 157 Figure 6-8 Typical Pt Auger spectra from the PtZn particles polarized at a) 1.25 V for 1 hours, b) 1.0 V for 20 hours, and c) 0.75 V for 14.5 hour. The ratios of Pt MN3 to MN2 peak heights are shown on the right hand side of the spectra. As shown in Chapter 4,  1:1  ratio  of  those  peaks  implies  the  interaction  between Pt  and Zn. Spectra b) and c) show almost 1:1 peak height ratios whereas the spectrum a) (the sample polarized at 1.25 V) shows an obvious difference in the peak heights. This indicates  that  the  samples  polarized  at  0.75  and 1.0  V after  14.5  and 20  hours, respectively, still maintain Pt-Zn bimetallic bonds, whereas the sample polarized at 1.25 V for 1 hour does not. 158 6.4.  Summary and conclusions In this chapter, the chemical stability of PtZn was investigated by forming PtZn particles on GC substrates and applying constant potentials of 0.75, 1.0, 0.8, or 1.25 V for a variety of durations  under  ambient  conditions.  Pt60Zn40 particles  (bulk  composition)  were  successfully formed after  a  heat  treatment.  Initially,  the  surface  concentration  of  Zn was 4.2 atomic  %; however, it increased and plateaued at 7.7 % after 14.5 hours of polarization at 0.75 V or at 19 % after 20 hours at 1.0 V. This supports the increase in the Zn surface concentration as described in Chapter 5. Samples polarized at 1.25 V showed almost no Zn at the surface concentration after 2.5 hours of polarization, showing the alloy was not stable at this potential. PtZn particles polarized at 1.0 V for 12 and 20 hours, as well as those polarized at 0.8 V for 30 and 40 hours, were dissolved in aqua regia and the bulk concentrations were determined by ICP-MS. Approximately 9 % of Zn was lost up to 20 hours of polarization at 1.0 V compared to the unpolarized particle. After 30 and 40 hour polarization at 0.8 V, 16 and 19 atomic % of Zn was lost from the original composition. Although more prolonged investigation is necessary for assessing the leaching behavior, Pt60Zn40 appears more stable than other 1:1 Pt alloys reported in literature. Moreover, the remaining PtZn after 40 hour polarization showed 3:1 composition, which is also a stable form of the alloy according to the phase diagram. Considering the amount of Zn lost over 30 to 40 hour polarization was by 2 %, it can be predicted Zn dissolution will not be severe after 40 hours. Further investigations are necessary to confirm the even longer term stability. Stability of the PtZn alloy after 0.75 and 1.0 V polarization was further confirmed by comparing  Pt  MN3 and  MN2 peak  heights.  Particles  treated  at  those  polarization  potentials retained almost 1:1 peak ratios, whereas the ones polarized at 1.25 V showed a ratio of 3:1, indicating a minimal interaction between Pt and Zn. Although the experimental data need to be confirmed, the results indicate the chemical stability of PtZn may be quite promising for use in fuel cells. 159 7.  Conclusions And Future Work The fuel cell will be one of the key power sources in the future for its environmentally benign operation, the high energy conversion efficiency, and the wide applicability in electrical devices.  It  is  also  a  device  that  encompasses  knowledge  in  numerous  branches  of  science stacked together.  The fuel cell is certainly a state of the art device, yet with a large number of challenges.  This  project  involved  a  number  of  people  assisting  us  either  to  support  our experimental data or to expand our knowledge with their expertise. That is essential to the fuel cell research as it is such an interdisciplinary field of science with a lot of correlations from one branch to another. This thesis focused on the electrocatalysis of the fuel cell reactions on an electrochemically-prepared PtZn alloy as the catalyst and outlined new findings which create experimental opportunities for future work.  This  chapter  first  summarizes  the  results  presented  in  Chapter 4 to 6 followed by suggestions for future work. 7.1.  Summary of the thesis In Chapter 4 the formation of the PtZn alloy on a bulk Pt substrate was first confirmed by scanning  electron microscopy (SEM) and Auger  electron  spectroscopy (AES).  Our PtZn alloy also revealed more facile oxygen uptake than metallic Zn which gave us the impetus to apply this alloy as an ORR catalyst. Improved catalytic activity (by 30 mV) towards the ORR was observed both in an electrochemical cell setup and in a single stack fuel cell. The Tafel slopes measured on smooth Pt and PtZn were both very similar, and the number of electrons exchanged during the ORR on the PtZn surface was 3.97. The work presented in this chapter envisages the prospect of the PtZn in the PEMFC application. The catalytic activity of the CO oxidation reaction on the PtZn surface was investigated in  Chapter  5.  When  compensated  for  the  surface  roughness  induced  by  Zn  deposition/ 160 modification, our PtZn alloy did not show improved activity for the CO oxidation. However, more facile reaction kinetics for the ORR was still observed after the roughness correction. Also the alloy surface exhibited a structural reorganization after the adsorption of oxygen or CO. This was  observed  in  electrochemical  measurements.  The  oxygen  adsorption  induced  more  Zn segregation to the surface, whereas CO adsorption induced Pt to segregate to the surface. The findings in this chapter could be important for reaction mechanism studies. The enhancement in the ORR or the CO oxidation catalysis is currently explained by an electronic structure change induced in Pt  atoms or  by the decrease in the Pt-Pt bond distance due to  alloying,  but the observed  site  exchange  process  may  add  another  element  to  the  explanation  of  the  ORR mechanism  or the CO oxidation reaction. The PtZn surfaces presented throughout Chapter 4 and Chapter 5 are prepared by the Zn electrodeposition followed by spontaneous alloy formation on bulk Pt, but this methodology raises a question – what is the amount of Zn remaining in the Pt or on the surface? Currently we do not have a clear answer to this somewhat obvious question because (1) Zn electrodeposition efficiency  is  limited  by  hydrogen  evolution,  and  it  is  a  challenge  to  quantify  the  current efficiency; (2) the resultant PtZn surface is heterogeneous as presented in  Chapter 5 and the reproducibility of PtZn surfaces depend on the character of the Pt substrate; and (3) the diffusion of Zn into Pt occurs quickly as presented in  Chapter 4, therefore the amount of Zn diffusing into Pt during the alloy formation is difficult to estimate. This question may not have been an issue if we made a bulk PtZn alloy by sputtering, but the facility was not available, and it would not meet the original motivation of the project. This is still an open question which requires a further investigation. Chapter 6 explored different methods to investigate the chemical stability of the PtZn alloys using electrochemistry and AES. All stability studies involved the polarization of the alloy in an oxygen saturated acidic solution at the room temperature. The preliminary studies using gravimetry and anodic stripping voltammetry showed that the alloy was stable under the fuel cell  operating  conditions.  To further  study the  stability  of  the  PtZn alloy,  we created  PtZn particles on glassy carbon (GC) substrates and then performed the polarization study. Because Zn does not form an alloy with carbon, Zn remaining on the GC substrates must be in the form of the Pt-alloy. During the polarization process, Zn was expected to migrate to the surface of the alloy, based on the results presented in Chapter 5. The surface Zn concentration measured by 161 AES showed an increase initially, but eventually reached a plateau at the polarization potentials of  0.75 and 1.0  V (the  surface  concentrations  of  6.5  and 19.3  atomic  % Zn,  respectively), whereas the surface concentration constantly decreased for the polarization potential at 1.25 V, showing the instability of the alloy. The polarized samples at 0.75 and 1.0 V may still suffer from Zn dissolution from the alloy, as the bulk concentrations of the PtZn particles constantly decreased during a prolonged polarization was applied (8.9 % of Zn was lost over 20 hours of polarization at 1.0 V); however, the amount of Zn leached out from the alloy is quite small relative to other transition metals, such as PtCo or PtNi, dissolving from their Pt-alloys. The work presented in this chapter is still incomplete and more data relating the surface and bulk Zn concentrations within the alloy particles needs to be obtained to fully describe the stability of the PtZn alloy. However, our data so far suggests reasonable stability of the PtZn alloy under a rather severe fuel cell condition over the duration of  the stability studies. Another  very interesting  finding  from the  work  shown in  Chapter 6 was  that  PtZn particles adhere to the carbon substrate more strongly than Pt. Pt-rich particles, especially after the prolonged polarization process, exhibited sintering (i.e., coagulation of smaller particles), but this was not the case for the PtZn particles. This is very important for the fuel cell application because sintering decreases surface area-to-volume ratio of the Pt particles on the catalyst layer which subsequently leads to the uneven distribution of Pt particles in an MEA and subsequent decrease in current density. If we could successfully prepare a catalyst layer with the catalysts well-adhered to the support, it would be very beneficial for fuel cell durability. The experimental results presented in this work lead to four main findings: 1) The PtZn alloy improves the ORR kinetics compared to Pt; 2) the adsorption of oxygen or CO molecules induced surface restructuring on the alloy surface – the adsorption of oxygen makes the alloy surface enriched with Zn, whereas the adsorption of CO makes the surface enriched with Pt; and 3) the stability of PtZn; and 4) the PtZn alloy may be free from sintering process as  it adheres well to carbon. Since there still are numerous uncertainties regarding fuel cell reactions, such new findings may significantly aid understanding the fundamental questions, thus developing the fuel cell technology. Also, the finding (2) is an interesting one for surface analysis in general, as  the adsorbate-induced segregation of one component  of an alloy has been observed with 162 various systems, but not with a PtZn alloy, to our best knowledge. Although these findings are unique, there are several experiments need to be performed to further clarify the fuel cell reaction process on the PtZn surface. Suggested experiments are outlined next. 7.2.  Future work 7.2.1.  Hydrogen peroxide production on the PtZn surface and the reaction mechanisms The first experiment one could perform is the measurement of hydrogen peroxide (H2O2) production during ORR on the PtZn alloy. The presence of H2O2 is, as mentioned in Chapter 2, unfavorable as the oxygen molecule reduction to H2O2 only involves two electrons and damages the solid electrolyte in the MEA. The number of electrons transferred calculated in Chapter 4 was 3.97, but this still  does not eliminate the possible H2O2  production at a lower potential. There are a number of literature reports showing the H2O2 production both on a pure Pt surface and the Pt-based alloy surfaces, confirmed by rotating ring disk electrodes (RRDEs) [164, 29, 27, 26]. An RRDE is similarly constructed to an RDE, except there is a ring electrode (usually Pt) around the disk electrode, separated by an insulating material. As the electrode rotates, the reactants in the solution are forced to come to the disk surface, the redox reaction occurs, and the reaction products depart from the surface. If an appropriate potential is applied to the ring, the products could either be reduced or oxidized at the ring, resulting in a current flow. The investigation of H2O2 formation on our PtZn surface can also be done using an RRDE; or alternatively, it could also be performed by placing a microelectrode (the electrode with a small but well defined area) in the vicinity of the RDE. The microelectrode essentially acts as the ring on the RRDE. The Bizzotto laboratory at University of British Columbia has 163 also been developing the method to fabricate the in-house microelectrode using a pipet puller and  a  coax  cable  [165].  The  challenge  with  this  method  is  the  collection  of  H2O2 by  the microelectrode  surface,  but  could  be  an  alternative  way  to  investigate  not  only  H2O2 but products in any reactions and spatial distribution of the production. An alternative method to the RRDE is scanning electrochemical microscopy (SECM) [166, 167]. This technique uses a microelectrode to probe the surface of interest, in this case it would be the PtZn alloy.  There are  different  modes of operations,  but the one employed in determining H2O2 generation is called the substrate generation/tip collection (SG/TC). This is shown  in  Figure  7-1 [168] The  microelectrode  is  connected  to  piezo-controller  to  allow rastering of the electrode just above the substrate at a constant height in a controlled manner. The microelectrode is held at the H2O2 oxidation potential while that of the substrate is changed. Once  H2O2 is produced at the substrate, the microelectrode oxidizes the H2O2 and the current is recorded. The operating principle is equivalent to the RRDE method, but SECM is more versatile [169]. For example, it can also be used to study the ORR mechanism. Assume the substrate is immersed into an oxygen saturated electrolyte, and the tip (microelectrode) is placed above the substrate  [170,  171].  When  the  tip  is  far  enough  from  the  substrate  so  that  all  oxygen hemispherically diffuses onto the tip (i.e., nothing hinders the oxygen molecules to diffuse onto the surface), the current measured at the tip, iT,∞, is constant and determined by the equation below: iT,∞ = 4nFDoCo *a (7-1) In  this  equation,  n is  a  number  of  electrons  transferred,  F is  Faraday  Constant,  Do is  the diffusion coefficient of oxygen, Co* is the bulk concentration of oxygen, and a is the radius of the microelectrode. When the tip is brought closer to the substrate, the current measured at the tip, iT, will decrease because the oxygen diffusion is now hindered. The tip is then approached to the substrate so that measured iT,∞ is, e.g., 90% of iT,∞. Keeping the same distance between the tip and the substrate, the microelectrode is then moved across the substrate so the measured 164 current  is  constant  (i.e.,  90%  iT,∞).  During  this  measurement,  the  substrate  potential  is maintained in the kinetic region for ORR. The tip is scanned across the sample. The measured current will be constant (i.e., at 90% iT,∞) above an inactive region. However, when the tip is brought above the active site, some oxygen is consumed at the site, so the microelectrode senses a  lower  concentration  of  oxygen,  resulting  in  decrease  in  measured  current  at  the microelectrode. Therefore, by knowing the position of the tip and by monitoring the tip current, the location of active sites can be approximated. Once the position of the active sites are known, in-situ spectroscopic analysis, such as Raman spectroscopy, can be employed to determine the chemical properties of the active site during ORR. 165 Figure 7-1 Substrate generation/ tip collection mode in SECM. The microelectrode is connected to piezo-controller to allow rastering of the electrode just above the substrate at a constant height in a controlled manner. The microelectrode is held at the  appropriate  oxidation  potential  while  that  of  the  substrate  is  changed.  Once oxidizable  products  are  produced  at  the  substrate,  they  are  detected  at  the microelectrode, generating the oxidative current. Reprinted from  [168]. Copyright 2007 with permission from Elsevier Ltd. 166 7.2.2.  Density functional theory calculations Previously reported density functional theory (DFT) calculations on small clusters for PtZn were published by Rodriguez et al., [112, 143, 119]] and Chen et al. [101]. These authors came to similar conclusions: When Pt and Zn were alloyed, both the Pt-Pt and Zn-Zn bond distances increased relative to the lengths in their pure forms, but the Pt-Zn bond distance was shorter  than  either  of  bond  lengths.  Also  the  Pt  5d orbital  population  decreased  after  the alloying. In addition, Rodriguez used ab initio calculations and showed Pt acts as an electron acceptor, whereas Zn acts as an electron donor [112]. Currently DFT calculations on the Zn modified Pt surface and the interaction of oxygen with  the  resultant  surface  have  been  investigated  in  collaboration  with  Dr.  Eben  Dy  in Department of Chemical Engineering at  University of British Columbia.  Dr Dy was able to reproduce the results obtained by Rodriguez and Chen listed above using the DFT method. He was also able to calculate the distance of an oxygen molecule approaching to the Pt(111) surface in “end-on” or “Pauling arrangement” (see Figure 2-10 in Chapter 2.), and his data agreed well with the previously reported experimental result [172]. Theoretical calculations for fuel cell reactions on Pt-based alloys have extensively been carried out, especially by Nørskov and coworkers, e.g., [173, 35, 174]. For example, they have successfully  reproduced  the  volcano  plot.  They  also  used  density  functional  theory  (DFT) calculations to model the oxygen reduction reaction pathway on Pt and calculated the energies of  intermediates  as  a  function  of  applied  potentials.  Based  on  the  calculated  energies,  they developed a kinetic model that allowed them to predict the reaction rate at a given potential [35]. This  is  certainly  useful  for  the  determination  of  the  rate  determining  step  for  ORR.  DFT methods are powerful for simulating complex systems and for confirming if the theories and experimental data agree each other. DFT can also be useful in understanding the PtZn system, such as the surface energy of the  alloy  with  or  without  adsorbates.  When  oxygen  molecules  are  adsorbed onto  the  PtZn surface, Zn migrates to the surface, and this appears to enhance the ORR kinetics. The driving force is assumed to be a thermochemical effect, but the electronic state of Zn in the alloy is obviously  different  from its  pure  form.  The  DFT calculations  would  help  determining  the 167 segregation energy of Zn from which the characteristics of the alloy surface could be better understood. 7.2.3.  Alternative PtZn alloy formation – A bulk alloy or alloy particles Our method of the PtZn preparation is simple and cost-effective for catalyst layer mass production;  it  requires  an  electrochemical  Zn  deposition  onto  Pt,  and  the  alloying  would subsequently occur at ambient conditions. However, the formation of surface alloys introduces severe  surface  roughening.  This  is  unfavorable  if  one  needs  to  conduct  the  kinetic  studies because defects on the surface play an important role in the kinetics, as introduced in Chapter 2 and 5. In Chapter 5, we prepared an equivalently “rough” Pt surface to the alloy surface by Pt deposition, but this does not mean we have perfectly equivalent surfaces. Preparation of bulk PtZn followed by polishing would provide a better comparison to the equivalently polished Pt surface. There are several preparation methods introduced in the literature, such as arc melting, chemical vapor deposition of Pt onto a zinc oxide substrate or a sputtering method, followed by the heat treatment,  [119, 175, 5, 176]. The method introduced by Huang and coworkers uses ZnCl2-1-ethyl-3-methelimidazolium chloride (EMIC) ionic neat solution containing Pt (II) to electrochemically codeposit both Pt and Zn onto a tungsten substrate. Ionic liquids are aprotic, thereby hydrogen evolution from the Pt surface is not an issue [177]. Also, the composition of the alloy can be controlled by tuning the deposition potential. The PtZn alloy formation was confirmed after applying a fixed reductive potential for as short as 20 seconds. Also observed by Huang and coworkers  was  the formation of  particles  on the PtZn deposited film.  The SEM images  recorded by the authors in  the ZnCl2 -EMIC is  shown in Figure  7-2.  The  middle  image  shows  the  appearance  of  irregular  shaped  flakes  which  are presumably PtZn alloys  [177]. Once any unalloyed Zn is eliminated from the deposit (as the authors claimed these films are ~90 atomic % Zn prepared at the conditions they applied), these particles may anchor well to the carbon support in the catalyst layer in an MEA, just as we observed in  Chapter 6, but without the bulk Pt particles which will, if present in the MEA, 168 aggregate during the course of the fuel cell operation. Figure 7-2 SEM images of the PtZn alloy deposited on a tungsten substrate from the ZnCl2-EMIC ionic solution at a) E=-0.2 V vs. Zn (Pt atomic % = 15.69) ; b)  E=- 0.22 V vs. Zn (Pt atomic % = 7.71); and c) E=-0.25 V vs. Zn (Pt atomic % = 5.67). The SEM image b) shows irregular-shaped particles, similar to the PtZn particles on glassy carbon reported in  Chapter 6. Reprinted from  [177]. Copyright 2004 with permission from  Elsevier Ltd. 169 7.2.4.  Application of PtZn in the PEMFC A PtZn GDL was prepared and used as a fuel cell  catalyst  in  Chapter 4.  Our alloy exhibited enhanced performance in the single stack fuel cell  compared to the one with a Pt MEA. However, the power output of the fuel cell was significantly worse than the one reported in  literature.  Experimental  optimization  for  e.g.,  the  Zn  deposition  process  or  hot-pressing conditions, is necessary to achieve adequate power output. It is also important to investigate the stability of the PtZn alloy under fuel cell operating conditions. In Chapter 6, we examined the chemical stability of PtZn in an electrochemical cell at room temperature, but this method may not sufficiently mimic the fuel cell conditions. It is also important  to  test  the durability of  components,  i.e.  membrane and GDL, in  a  fuel  cell containing a PtZn catalyst. 7.2.5.  In-situ Neutron reflectometry Neutron reflectometry (NR) is a powerful technique used to investigate the metal/liquid, metal/gas, or the thin film formed on a metal surface with the probing depth from a few to thousands of  angstroms in the direction normal to the sample surface. A collimated neutron beam is directed to a surface of interest at an extreme grazing angle. The neutrons interact with the nuclei of the atoms, and the reflectivity of neutron is measured as a function of momentum transfer, which is related to the angle of incidence, the wavelength of the neutron beam, and the scattering-length density, ρ, which is defined as, =∑ j b j n j (7-2) In this  equation,  bj is  the number density of  nuclei  j, and  nj is  nucleus-dependent  coherent neutron scattering length analogous to the refractive indices of materials for neutrons [178]. In practice, the angle of incidence or the wavelength of the neutron beam is changed (thus the 170 penetration depth of the beam into the sample) and the scattering-length density is measured to probe the metal surface or the corresponding interface. NR also allows us to measure the metal surface while electrochemistry is performed on the sample [179-181]. In Chapter 5, we observed surface segregation of Pt or Zn, depending on different molecules adsorbed onto the surface of the PtZn. Surface enrichment of one component of the alloy may be observed by NR if the scattering-length density of Pt and Zn is sufficiently different. According to Noel  et al.,  [180], two materials with their scattering-length densities differ by 1×10-6 Å-2 are clearly distinguishable. According to NIST Centre for Neutron Research [182], the values for nPt and nZn are 9.60 and 5.682 fm, respectively. Assuming their b values are equal or similar, as our PtZn is 1:1 in atomic composition and the lattice mismatch is small, we should be able to see a reasonable contrast between Pt and Zn. Also, we can monitor the layering of Pt and Zn and watch the movement of Zn or Pt to the surface when ORR or CO oxidation is performed. The NR facility is available at Canadian Neutron Beam Centre in National Research Council of Canada, located in Chalk River, Ontario. 171 8.  References [1] H. Hassanzadeh, S. Mansouri, Proc IME J Power Energ, 219 (2005) 245-254. [2] X. Li, Principles of Fuel Cells, Taylor & Francis Group, NY, 2006. [3] M. Al-Baghdadi, in: Fuel Cell Research Trends. Vasquez, L.O. ed., Nova Science Publisher Inc., New York, 2007, p. 273 Ch. 7. [4] Kitco Preious Metals. Accessed October 23, 2008. <http://www.kitco.com/charts/liveplatinum.html>. [5] T. He, E. Kreidler, L. Xiong, E. Ding, J. Power Sources, 165 (2007) 87-91. [6] Kitco Base Metals. Accessed October 23, 2008. <http://www.kitcometals.com/>. [7] A. Sode, W. Li, Y. Yang, P. Wong, E. Gyenge, K. Mitchell, D. Bizzotto, J. Phys. Chem. B, 110 (2006) 8715-8722. [8] A. Sode, D. Bizzotto, Electrochimica Acta, 54 (2009) 1095-1101. [9] A. Bard, L. Faulkner, Electrochemical Methods: Fundamentals and Applications., 2nd ed., John Wiley & Sons Inc., New York, 2001. [10] N. Lang, W. Kohn, Phys. Rev. B, 1 (1970) 4555-4568. [11] G. Somorjai, Chemistry in Two Dimentions, Cornell University Press, USA, 1981. [12] S. Trasatti, in: Handbook of Fuel Cells - Fundamentals, Technology, and Applications. Woulf Vielstich, Hubert A. Gasteiger, and Arnold Lamm eds., John Wiley & Sons Ltd., New York, 2003, pp. 81, Ch. 9. [13] G. Ertl,  Angew. Chem. Int. Ed., 47 (2008) 3524-3535. [14] G. Somorjai, in: Chemistry in Two Dimentions, Cornell University Press, USA, 1981, pp. 381-413. [15] N. Markovic, P.J. Ross, in: Interfacial Electrochemistry: Theory, Experiment, and Applications. A. Wieckowski ed., Marcel Dekker, Inc., New York, 1999, p. 821 Ch. 46. [16] S. Trasatti, in: Handbook of Fuel Cells - Fundamentals, Technology, and Applications. Woulf Vielstich, Hubert A. Gasteiger, and Arnold Lamm eds., John Wiley & Sons Ltd., New York, 2003, pp. 71-78 Ch.8. 172 [17] S. Trasatti, in: Handbook of Fuel Cells. Wolf Vielstich, Arnold Lamm, Hubert A. Gasteiger (Eds.), John Wiley & Sons Ltd., New York, 2003, p. 88 (Ch. 10). [18] J. Clavilier, R. Faure, R. Durand, J. Electroanal. Chem., 107 (1980) 205-209. [19] J. Clavilier, D. Armanda, B. Wu, J. Electroanal. Chem., 135 (1982) 159-166. [20] C. Green, A. Kucernak, J. Phys. Chem. B, 106 (2002) 1036-1047. [21] N. Anastasijevic, V. Vesovic, R. Adzic, J. Electroanal. Chem., 229 (1987) 305. [22] M. Gattrell, B. MacDougall, in: Handbook of Fuel Cells - Fundamentals, Technology, and Applications. Woulf Vielstich, Hubert A. Gasteiger, and Arnold Lamm eds., John Wiley & Sons Ltd., New York, 2003, pp. 443-464. [23] E. Yeager, J. Mol. Catal, 38 (1986) 5. [24] S. Chen, A. Kucernak, J. Phys. Chem. B, 108 (2004) 3262-3272. [25] A. Damjanovic, D. Sepa, M. Vojnovic, Electrochim. Acta, 24 (1979) 887-889. [26] V. Stamenkovic, T. Schmidt, P. Ross, N. Markovic, J. Phys. Chem. B, 106 (2002) 11970- 11979. [27] M.D. Macia, J.M. Campiña, E. Herrero, J.M. Feliu, J. Electroanal. Chem., 564 (2004) 141- 150. [28] A. Damjanovic, V. Brusic, Electrochim. Acta, 12 (1967) 615-628. [29] N. Markovic, H. Gasteiger, P.N. Ross, J. Electrochem. Soc., 144 (1997) 1591-1597. [30] R. Adzic, in: Electrocatalysis. Lipkowski, J. and Ross, P.N. eds., Wiley-VCH, New York, 1998, pp. 197, Ch.5. [31] E. Yeager, Electrochim. Acta, 29 (1984) 1527-1537. [32] O. Antorine, Y. Baltel, R. Durand, J. Electroanal. Chem., 499 (2001) 85-94. [33] D. Sepa, M. Vojnovic, L. Vracar, A. Damjanovic, Electrochim. Acta, 32 (1987) 129-134. [34] S. Mukerjee, J. Appl. Electrochem., 20 (1990) 537-548. [35] J. Norskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J. Kitchin, T. Bligaard, H. Jonsson, J. Phys. Chem. B, 108 (2004) 17886-17892. [36] A. Damjanovic, J.O. Bockris, Electrochim. Acta, 11 (1966) 376-377. 173 [37] N. Markovioc, H. Gasteiger, P. Ross, J. Phys. Chem., 100 (1996) 6715-6721. [38] P. Faguy, N.M. Markovic, P.J. Ross, J. Electrochem. Soc., 140 (1993) 1638-1641. [39] J. Clavilier, R. Durand, G. Guinet, R. Faure, J. Electroanal. Chem., 127 (1981) 281-287. [40] A.C. Luntz, M.D. Williams, D.S. Bethune, J. Chem. Phys., 89 (1988) 4381-4395. [41] A.T. Gee, B.E. Hayden, J. Chem. Phys., 113 (2000) 10333-10343. [42] P. Gambardella, Ž. Šljivančanin, B. Hammer, M. Blanc, K. Kuhnke, K. Kern, Phys. Rev. Lett., 87 (2001) 056103. [43] B. Beden, F. Hahn, J. Leger, C. Lamy, J. Electroanal. Chem., 258 (1989) 463-467. [44] J. Wintterlin, S. Volkening, T.V. Janssens, T. Zambelli, G. Ertl, Science, 278 (1997) 1931- 1934. [45] S. Enback, G. Lindbergh, J. Electrochem. Soc., 152 (2004) A23-A31. [46] M. Liao, C.R. Cabrera, Y. Ishikawa, Surf. Sci., 445 (2000) 267-282. [47] Laidler, K.J, Meiser, John H, Physical Chemistry, 3rd ed., Houghton Mifflin College Div, MA, 1999. [48] C.T. Campbell, G. Ertl, H. Kuipers, J. Segner, J. Chem. Phys., 73 (1980) 5862-5873. [49] G. Jerkiewicz, G. Vatankhah, J. Lessard, M.P. Soriaga, Y. Park, Electrochim. Acta, 49 (2004) 1451-1459. [50] B. Andreaus, F. Maillard, J. Kocylo, E. Savinova, M. Eikerling, J. Phys. Chem. B, 110 (2006) 21028-21040. [51] C. Lamy, J. Leger, J. Electroanal. Chem., 150 (1983) 71-77. [52] E. Herrero, B. Álvarez, J.M. Feliu, S. Blais, Z. Radovic-Hrapovic, G. Jerkiewicz, J. Electroanal. Chem., 567 (2004) 139-149. [53] A. Cuesta, A. Couto, A. Rincón, M. Pérez, A. López-Cudero, C. Gutiérrez, J. Electroanal. Chem., 586 (2006) 184-195. [54] A. Miki, S. Ye, T. Senzaki, M. Osawa, J. Electroanal. Chem., 563 (2004) 23-31. [55] A. López-Cudero, A. Cuesta, C. Gutiérrez, J. Electroanal. Chem., 548 (2003) 109-119. [56] K. Chou, N. Markovic, J. Kim, P. Ross, G. Somorjai, J. Phys. Chem. B, 107 (2003) 1840- 174 1844. [57] C. McCallum, D. Pletcher, J. Electroanal. Chem., 70 (1976) 277-290. [58] K. Mikita, M. Nakamura, N. Hoshi, Langmuir, 23 (2007) 9092-9097. [59] J. Xu, J. Yates, J. Chem. Phys., 99 (1993) 725-732. [60] J. Yates, Jr., J. Vac. Sci. Technol. A., 13 (1995) 1359-1367. [61] F. Besenbacher, I. Stensgaard, in: The Chemical Physics of Solid Surfaces. King, D.A. and Woodruff, D.P. eds., Elsevier, Amsterdam, 1994, pp. 573-629 Ch.14. [62] N.M. Markovic, P.N. Ross, Surf. Sci. Rep., 45 (2002) 117-229. [63] D. Woodruff, in: The Chemical Physics of Solid Surfaces. King, D.A. and Woodruff, D.P. eds., Elsevier, Amsterdam, 1994, pp. 465-499. Ch.12. [64] T. Gritsch, D. Coulman, R.J. Behm, G. Ertl, Phys. Rev. Lett., 63 (1989) 1086. [65] R.J. Behm, P.A. Thiel, P.R. Norton, G. Ertl, J. Chem. Phys., 78 (1983) 7437-7447. [66] P.A. Thiel, R.J. Behm, P.R. Norton, G. Ertl, J. Chem. Phys., 78 (1983) 7448-7458. [67] G. Chen, D.A. Delafuente, S. Sarangapani, T.E. Mallouk, Catal. Today, 67 (2001) 341-355. [68] J. Whitacre, T. Valdez, S. Narayanan, Electrochim.Acta, 53 (2008) 3680-3689. [69] H.R. Colón-Mercado, B.N. Popov, J. Power Sources, 155 (2006) 253-263. [70] S. Mukerjee, S. Srinivasan, M.P. Soriaga, J. McBreen, J. Phys. Chem., 99 (1995) 4577- 4589. [71] B. Brown, S.D. Wolter, B.R. Stoner, J.T. Glass, J. Electrochem. Soc., 155 (2008) B852- B859. [72] M. Mavrikakis, B. Hammer, J.K. Nørskov, Phys. Rev. Lett., 81 (1998) 2819. [73] S. Mukerjee, S. Srinivasan, J. Electroanal. Chem., 357 (1993) 201-224. [74] N. Markovic, T. Schmidt, P. Ross, Fuel Cells, 1 (2001) 105-116. [75] Y. Xu, A. Ruban, M. Mavrikakis, J. Am. Chem. Soc., 126 (2004) 4717-4725. [76] U. Bardi, B. Beard, P. Ross, J. Catal., 124 (1990) 22-29. [77] S. Hsieh, T. Matsumoto, M. Batzill, B. Koel, Phys. Rev. B, 68 (2003) 205417/1-205417/11. 175 [78] J.G. Chen, C.A. Menning, M.B. Zellner, Surf. Sci. Rep., 63 (2008) 201-254. [79] G. Treglia, B. Legrand, F. Ducastelle, A. Saul, C. Gallis, I. Meunier, C. Mottet, A. Senhaji, Comput. Mater. Sci., 15 (1999) 196-235. [80] V. Stamenkovic, T.J. Schmidt, P.N. Ross, N.M. Markovic, J. Electroanal. Chem., 554-555 (2003) 191-199. [81] P. Liu, A. Logadottir, J.K. Nørskov, Electrochim. Acta, 48 (2003) 3731-3742. [82] C. Roth, A. Papworth, I. Hussain, R. Nichols, D. Schiffrin, J. Electroanal. Chem., 581 (2005) 79-85. [83] H.A. Gasteiger, N. Markovic, P.N. Ross, Jr., E.J. Cairns, J. Phys. Chem., 98 (1994) 617- 625. [84] H.A. Gasteiger, N.M. Markovic, P.N. Ross, J. Phys. Chem., 99 (1995) 8290-8301. [85] M. Watanabe, S. Motoo, J. Electroanal. Chem., 60 (1975) 267-273. [86] M. Arenz, V. Stamenkovic, B. Blizanac, K. Mayrhofer, N. Markovic, P. Ross,  J. Catal., 232 (2005) 402-410. [87] S. Sen Gupta, J. Datta, J. Electroanal. Chem., 594 (2006) 65-72. [88] A. Kalinkin, A. Pashis, V. Bukhtiyarov,  Kinet. Catal., 48 (2007) 298-304. [89] B.N. Grgur, N.M. Markovic, P.N. Ross, Electrochim. Acta, 43 (1998) 3631-3635. [90] A. Aramata, I. Toyoshima, M. Enyo, Electrochim. Acta, 37 (1992) 1317-1320. [91] M. Janssen, J. Moolhuysen, J. Catal., 46 (1977) 289-296. [92] C. Lu, C. Rice, R. Masel, P. Babu, P. Waszczuk, H. Kim, E. Oldfield, A. Wieckowski, J. Phys. Chem. B, 106 (2002) 9581-9589. [93] T. Frelink, W. Visscher, A.P. Cox, J.A.R. van Veen, Electrochim. Acta, 40 (1995) 1537- 1543. [94] C. Roth, N. Benker, T. Buhrmester, M. Mazurek, M. Loster, H. Fuess, D. Koningsberger, D. Ramaker, J. Am. Chem. Soc., 127 (2005) 14607-14615. [95] P. Waszczuk, G.-. Lu, A. Wieckowski, C. Lu, C. Rice, R.I. Masel, Electrochim. Acta, 47 (2002) 3637-3652. [96] J. Spendelow, P. Babu, A. Wieckowski,  Curr. Opin. Solid State Mater. Sci., 9 (2005) 37-48. 176 [97] H.A. Gasteiger, S.S. Kocha, B. Sompalli, F.T. Wagner,  Appl. Catal., B, 56 (2005) 9-35. [98] S. Mukerjee, S. Srinivasan, in: Handbook of Fuel Cells - Fundamentals, Technology, and Applications. Woulf Vielstich, Hubert A. Gasteiger, and Arnold Lamm eds., John Wiley & Sons Ltd., New York, 2003, pp. 502-519 Ch. 34. [99] G. Somorjai, M. Van Hove, Prog. in Sur. Sci., 29-30 (1989) 201-231. [100] A. West, Basic Solid State Chemistry, 2nd ed., John Wiley & Sons Inc., West Sussex, UK, 1999. [101] Z. Chen, K.M. Neyman, A.B. Gordienko, N. Rösch, Phys. Rev. B, 68 (2003) 075417. [102] Z. Moser, J. Phase. Equilib, 12 (1991) 439-443. [103] S. Taguchi, A. Aramata, J. Electroanal. Chem., 457 (1998) 73-81. [104] D. Kolb, Adv. in Electrochem. and Electrochem. Eng., 11 (1978) 125. [105] A. Aramata, M. Quaiyyum, W.A. Balais, T. Atoguchi, J. Electroanal. Chem., 338 (1992) 367-372. [106] A. Despic, M. Pavlovic, Electrochim. Acta, 27 (1982) 1539-1549. [107] E. Guerra, G.H. Kelsall, M. Bestetti, D. Dreisinger, K. Wong, K.A.R. Mitchell, D. Bizzotto, J. Electrochem. Soc., 151 (2004) E1-E6. [108] M. Quaiyyum, A. Aramata, S. Moniwa, S. Taguchi, M. Enyo, J. Electroanal. Chem., 373 (1994) 61-66. [109] V. Sudha, M. Sangaranarayanan, J. Chem. Sci., 117 (2005) 207-218. [110] H. Uhlig, J. MacNairn, D. Vaughn, Acta Metall. , 3 (1955) 302-304. [111] S. Taguchi, A. Aramata, J. Electroanal. Chem., 374 (1994) 275-277. [112] J.A. Rodriguez, M. Kuhn, J. Chem. Phys., 102 (1995) 4279-4289. [113] Handbook of Chemistry and Physics. Accessed on September 19, 2008. <http://www.hbcpnetbase.com/>. [114] J. Fernandez, D. Walsh, A. Bard, J. Am. Chem. Soc., 127 (2005) 357-365. [115] S. Cameron, D. Dwyer, Surf. Sci., 176 (1986) L857-L862. [116] A. Linsebigler, G. Lu, J. Yates, Surf. Sci., 294 (1993) 284-296. 177 [117] R. Demmin, S. Shivaprasad, T. Madey, Langmuir, 4 (1988) 1104-1108. [118] S. Mukerjee, S. Srinivasan, M.P. Soriaga, J. McBreen, J. Electrochem. Soc., 142 (1995) 1409-1422. [119] Rodriguez, J.A., Surf. Sci., 289 (1993) L584-L590. [120] J.A. Rodriguez, J. Hrbek, J. Vac. Sci. Technol. A., 12 (1994) 2153-2158. [121] J. Rodriguez, Science (Washington, D.C.), 257 (1992) 897-903. [122] W. Albery, M. Hitchman, Ring-disc Electrodes, Oxford University Press, London, UK, 1971. [123] V. Filinovsky, Y. Pleskov, in: Comprehensive Treatise of Electrochemistry., J.O. Bockris ed., Plenum Press, New York, 1980. [124] H.J. Mathieu, in: Surface Analysis - The Principal Technique, John Wiley & Sons Ltd., 1997, p. 99. [125] M.P. Seah, W.A. Dench,  Surf. Interface Anal., 1 (1979) 2-11. [126] J. Riviere, Surface Analytical Techniques, Claresdon Press, Oxford, 1990. [127] L. Reimer, Scanning Electron Microscopy: Physics of Image Formation and Microanalysis, Springer-Verlag, Berlin, 1985. [128] J. Bozzola, L. Russell, Electron Microscopy: Principles and Techniques for Biologist, 2nd ed., Jones & Bartlett Publishers, Sudbury, 1998. [129] S.J. Hill, Inductively Coupled Plasma Spectroscopy And Its Application, 1st ed., CRC Press, Florida, 1999. [130] S.M. Nelms, ICP Mass Spetrometry Handbook, 1st ed., Blackwell Publishing, Florida, 2005. [131] S. Kocha, in: Handbook of Fuel Cells - Fundamentals, Technology, and Applications. Woulf Vielstich, Hubert A. Gasteiger, and Arnold Lamm eds., John Wiley & Sons Ltd., New York, 2003, pp. 538-565 Ch.43. [132] L. Xiong, A. Manthiram, Electrochim. Acta, 50 (2005) 3200-3204. [133] S. Litster, G. McLean, J. Power Sources, 130 (2004) 61-76. [134] T. Ralph, M. Hogarth, Platinum Met. Rev., 46 (2002) 3-14. 178 [135] X. Li, H. Colón-Mercado, G. Wu, J. Lee, B. Popov, ECS Trans., 11 (2007) 1259-1266. [136] M. Vos, D. Van den Marel, G. Sawatzky, Phys. Rev. B, 29 (1984) 3073-3084. [137] P.A. Bennett, J.C. Fuggle, F.U. Hillebrecht, A. Lenselink, G.A. Sawatzky, Phys. Rev. B, 27 (1983) 2194. [138] A. Aramata, S. Taguchi, T. Fukuda, M. Nakamura, G. Horányi, Electrochim. Acta, 44 (1998) 999-1007. [139] E. Guerra, Ph.D. dissertation, UBC, Vancouver, (2003). [140] U. Paulus, A. Wokaun, G. Scherer, T. Schmidt, V. Stamenkovic, V. Radmilovic, N. Markovic, P. Ross, J. Phys. Chem. B, 106 (2002) 4181-4191. [141] S.J. Seo, H. Joh, H.T. Kim, S.H. Moon, J. Power Sources, 163 (2006) 403-408. [142] R. Connel, M.SC. Dissertation, UBC, Vancouver, (2005). [143] J. Rodriguez, M. Kuhn, J. Phys. Chem., 100 (1996) 381-389. [144] J. Rodriguez, Surf. Sci. Rep., 24 (1996) 223-287. [145] Taft, C. A., Guimaraes, T. C., Pavao, A. C., Lester, W. A., Int. Rev. Phys. Chem., 18 (1999) 163-233. [146] F. Buatier de Mongeot, M. Scherer, B. Gleich, E. Kopatzki, R. Behm, Surf. Sci., 411 (1998) 249-262. [147] T. Biegler, D. Rand, R. Woods, J. Electroanal. Chem., 29 (1971) 269. [148] P.L. Cavallotti, M. Bestetti, S. Franz, Electrochim. Acta, 48 (2003) 3013-3020. [149] S. Itoh, N. Yamazoe, T. Seiyama, Surf. Technol., 5 (1977) 27-42. [150] K. Raeissi, A. Saatchi, M. Golozar, J. Szpunar, J. Appl. Electrochem., 34 (2004) 1249- 1258. [151] V.M. Kozlov, L. Peraldo Bicelli, Mater. Chem. Phys., 77 (2003) 289-293. [152] M. Arenz, K. Mayrhofer, V. Stamenkovic, B. Blizanac, T. Tada, P. Ross, N. Markovic, J. Am. Chem. Soc., 127 (2005) 6819-6829. [153] Y.G. Shen, D.J. O'Connor, J. Yao, Appl. Surf. Sci., 125 (1998) 300-312. [154] B.C. Han, A. Van der Ven, G. Ceder, B. Hwang, Phys. Rev. B, 72 (2005) 205409-9. 179 [155] S. Le, A. Navrotsky, J. Solid State Chem., 180 (2007) 2443-2451. [156] Y. Nagano, J. Therm. Anal. Cal., 69 (2002) 831-839. [157] N. Lebedeva, M. Koper, J. Feliu, R. van Santen, J. Phys. Chem. B, 106 (2002) 12938- 12947. [158] A. LaConti, R. McDonald, in: Handbook of Fuel Cells - Fundamentals, Technology, and Applications. Woulf Vielstich, Hubert A. Gasteiger, and Arnold Lamm eds., John Wiley & Sons Ltd., New York, 2003, pp. 647-677. Ch.49. [159] Y. Shao, G. Yin, Y. Gao, J. Power Sources, 171 (2007) 558-566. [160] A. Arico, A. Stassi, E. Modica, R. Ornelas, I. Gatto, E. Passalacqua, V. Antonucci, J. Power Sources, 178 (2008) 525-536. [161] E. Antolini, J.R. Salgado, E.R. Gonzalez, J. Power Sources, 160 (2006) 957-968. [162] Z. Wei, H. Guo, Z. Tang, J. Power Sources, 62 (1996) 233-236. [163] H.R. Colón-Mercado, H. Kim, B.N. Popov,  Electrochem. Commun., 6 (2004) 795-799. [164] U.A. Paulus, T.J. Schmidt, H.A. Gasteiger, R.J. Behm, J. Electroanal. Chem., 495 (2001) 134-145. [165] H. Zietzman, B.Sc. Dissertation, UBC, Vancouver, (2007). [166] A. Bard, M.E. Mirkin, Scanning Electrochemical Microscopy, Marcel Dekker, Inc., New York, 2001. [167] J. Kwak, A. Bard, Anal. Chem., 61 1221-1227. [168] X. Lu, Q. Wang, X. Liu,  Anal. Chim. Acta, 601 (2007) 10-25. [169] M.V. Mirkin, B.R. Horrocks, Anal. Chim. Acta, 406 (2000) 119-146. [170] J. Fernandez, A. Bard, Anal. Chem., 75 (2003) 2967-2974. [171] J. Fernandez, A. Bard, Anal. Chem., 76 (2004) 2281-2289. [172] C. Puglia, A. Nilsson, B. Hernnäs, O. Karis, P. Bennich, N. Mårtensson, Surf. Sci., 342 (1995) 119-133. [173] V. Stamenkovic, B.S. Mun, K.J. Mayrhofer, P.N. Ross, N.M. Markovic, J. Rossmeisl, J. Greeley, J.K. Nørskov,  Angew. Chem. Int. Ed., 45 (2006) 2897-2901. [174] J. Greeley, J. Nørskov, Electrochimica Acta, 52 (2007) 5829-5836. 180 [175] Y. Okano, A. Katagiri, J. Electrochem. Soc., 144 (1997) 1927-1932. [176] W. Harding, Surf. Finish., 65 (1978) 30. [177] J. Huang, I. Sun, Electrochim.Acta, 49 (2004) 3251-3258. [178] R. Thomas, Annu. Rev. Phys. Chem., 55 (2004) 391-426. [179] J.J. Noel, H.L. Jensen, Z. Tun, D.W. Shoesmith, Electrochem. Solid-State Lett., 3 (2000) 473-476. [180] J.J. Noel, D.W. Shoesmith, Z. Tun, J. Electrochem. Soc., 155 (2008) C444-C454. [181] V. Zamlynny, I. Burgess, G. Szymanski, J. Lipkowski, J. Majewski, G. Smith, S. Satija, R. Ivkov, Langmuir, 16 (2000) 9861-9870. [182] NIST Center for Neutron Research, Accessed October 25, 2008. <http://www.ncnr.nist.gov/resources/n-lengths/list.html>. 181

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.24.1-0061704/manifest

Comment

Related Items