Understanding the Poisoning, Aging and Degradation of Low ...

Understanding the Poisoning, Aging and Degradation of Low-temperature Fuel Cell Electrocatalysts using in situ X-ray Absorption Spectroscopy

by Badri Shyam

B.E. Chemical Engineering 2005, R.V. College of Engineering, Bangalore, India

A Dissertation submitted to

The Faculty of Columbian College of Arts and Sciences

of The George Washington University in partial satisfaction of the requirements for the degree of Doctor of Philosophy

May 16th, 2010

Dissertation directed by

David E. Ramaker

Columbian Professor of Chemistry

ii

The Columbian College of Arts and Sciences of The George Washington University

certifies that Badri Shyam has passed the Final Examination for the degree of Doctor of

Philosophy as of March 25th, 2010. This is the final and approved form of the

dissertation.


Badri Shyam

Dissertation Research Committee:

David E. Ramaker, Columbian Professor of Chemistry, Dissertation Director

Akos Vertes, Professor of Chemistry and of Biochemistry & Molecular

Biology, Committee member

Vladislav Sadtchenko, Associate Professor of Chemistry, Committee member

Stuart Licht, Professor of Chemistry, Committee member

iii

© Copyright 2010 by Badri Shyam

All rights reserved.

iv

Dedication

To Mum and Dad

v

Acknowledgments

First and foremost, I would like to take this opportunity to offer my sincerest

gratitude and thanks to Dr. Ramaker for being such a wonderful mentor and role-model. I

consider myself extremely fortunate to have worked closely with someone like him.

Always interested, invested, curious and in child-like wonder about the workings of the

world, he has inspired me to keep probing and keep asking questions, both big and small.

I would venture to say that our interactions represented the very best that academic life

anywhere has to offer. I was constantly reminded that it is indeed a beautiful thing to

pursue a passion with someone who cares deeply, to be together lost in thought and

completely absorbed in a problem, and finally, to always hold yourself to the highest

standards of integrity and quality of work. It is due to him that I can say that graduate

school at GW has been a most fulfilling experience. Needless to say, Dr. Ramaker has

left an indelible mark on the way I will teach and carry out research. For that and much

more, I am truly grateful.

I would like to thank the Chemistry Department and the Columbian College of Arts &

Sciences at GW for the fellowships and awards that funded my education here. None of it

would have been possible without their generous support. I would like to specially thank

the department and its’ members for providing a cheerful and spirited work environment

which made me feel so much at home in my lab and around the department. I think that

Shanna in particular, has single-handedly been responsible for most of the smooth

functioning of the chemistry department office. On good days or bad, she was always

there with a word of encouragement and a smile, and willing to help with any official

vi

paperwork at all times. As an international student, the amount of paperwork can be

overwhelming at times and I thank her for just being there whenever anything had to be

processed to ensure that it all went off smoothly. I will dearly miss her lively company.

I would like to thank the readers, Prof. Vertes and Prof. Sadtchenko for their careful

and critical proof-reading during the preparation of the final draft of the dissertation.

Thank you for your time and willingness to serve as readers amidst your busy schedules.

Within the Ramaker research group, I would like to thank Frances and Danny for

teaching me how to use the IFEFFIT suite to analyze XAS data. They both patiently

worked through a number of examples with me just as I was coming up the learning

curve during my first few semesters in school. Their expert assistance and companionship

is greatly appreciated. I would also like to acknowledge fellow current group members,

Anna and Maryam for their friendship, camaraderie and assistance in numerous ways

over the last few years. I have really enjoyed our time together and look forward to

keeping in touch in the years to come.

I have also been fortunate to work with a number of excellent researchers over the

last five years. In particular, I have benefited tremendously from our close collaboration

with Prof. Sanjeev Mukerjee’s group at Northeastern University. Tom and Joe taught me

most of what I know about running in situ XAS experiments even as navigating through

the maze of wires, equipment and software at the beamlines proved a real challenge

during my first few trips to the synchrotron facility. Their expertise was handed down to

me very generously. In particular, I have had the pleasure of working closely with Tom

on a number of research projects. Three of them are presented as complete chapters in

this dissertation. Our relationship has progressed from being colleagues to one of warm

vii

friendship and I really hope that we will create opportunities to collaborate on fun

research problems down the road. I would like to thank Sanjeev for inviting me to work

with his group in the summer of 2008. It was one of the most fruitful and fun summers I

have had during grad school – one outcome being that since then, I have collaborated

with nearly everyone working on fuel cells within their group on various projects that

really pushed me to learn more and take on more challenging problems. Another research

group we collaborated with over the last couple of years is Prof. Yu Ye Tong’s research

group at Georgetown University. Our research areas overlapped at a very exciting point

and led to an interesting finding in the electrochemistry of stabilized nanoparticles. I

thank Ceren, now Dr.Ceren Susut, for all her assistance with running the electrochemistry

and never failing to provide me with interesting catalyst samples for my trips up to

Brookhaven.

While teaching labs and doing research can be a daunting task, in my opinion, it is

also one of the most fulfilling and rewarding things a graduate student can experience.

After my first semester of teaching, I decided I would teach for the rest of graduate

school and I thank Dr. Ramaker and Dr. King for being simply and wholeheartedly

supportive of my decision. I have to thank, in particular, Dr. King for trusting an

engineer-turned-physical chemist to handle an organic chemistry lab session. I would like

to extend my deepest gratitude to him for all the support and encouragement he has

offered me in more ways than one. I have always admired his articulation and clarity of

expression that are hallmarks of all his lectures for the lab sessions. While I surely cannot

measure his impact on my pedagogical outlook, I learnt from him that if one can be

caring, fun, warm and always have time for your students, all of this while being chair of

viii

the department, it is surely the least you can expect of yourself as a graduate teaching

assistant. I would also like to mention that handling labs without Mr. Kingsbury or

Anthony (the man) would just not have been near as fun. Thanks for your competent

assistance and lively company during all those lab sessions I’ve been in and out of in the

last five years.

I have also been very lucky to have had some truly exceptional students in my lab

sessions. Owing to overlapping interests, perspectives and values, strong friendships with

some of them seemed inevitable and resulted in many memorable moments, both within

and outside of Corcoran Hall. In particular, I’d like to mention Stephen, with whom I

have had many intellectually refreshing conversations over coffee outside Starbucks, or

on random late-night walks around GW. Jeff and Alex – you are both wonderful people

and I feel very fortunate to have shared such a close friendship with both of you. Jeff

insisted that I stay at his childhood home in Watertown during my research stint at

Northeastern University two summers ago. I will always cherish our discussions and

walks around Harvard Square, our frequent trips to the Watertown Public Library and our

repeated playing of George Harrison’s ‘All Things Must Pass’ on the old record player

upstairs. Nan, Jeff’s grandma, cared for me as if I were a grandson myself and ensured

that I was eating and sleeping properly during my time there. In short, it was a perfect

summer and I couldn’t have planned it better if I tried. Thanks for everything.

I would also like to acknowledge a significant role my uncle, Govind Mama, played

in igniting my deep-seated interest for research. I was only eight or nine years old when

he took all the kids in our family to the Raman Research Institute in Bangalore for the

screening of a documentary. It was one of the several critically acclaimed National

ix

Geographic Specials shot by the legendary wildlife film-maker, Hugo Van Lawick; and it

completely blew my mind. The excitement and thrill of observing wild cats and big game

in their natural habitat, scenes of collecting, documenting and compiling volumes of data

in their tents, sweeping shots of the African Savannah and dramatic scenes of survival in

the wild, all of it presented larger than life on a giant screen in the auditorium made a

tremendous impact on me. Ever since that day, I would go to the Institute regularly and

watch many more such documentaries. And I had made up my mind on what I wanted to

do – study things. I thank him for opening up my little world.

I am grateful for the many powerful influences and examples I have had, both within

the family and among wonderful family friends; not the least among them were my

parents’ role in my education. They provided a great atmosphere of books and music and

over the years, let me read indiscriminately without once suggesting that I finish my

schoolwork first. I thank them both, and my dear sister, Niti, for being in constant support

and happiness for everything I choose to do.

Finally, I wish to acknowledge the unconditional friendships that have been extended

to me by many wonderful people I got to know at GW. It would not be too much of a

stretch to say that my quality of life here would have been severely impacted if it weren’t

for all of you. Five years and thousands of miles away from home can be a significant

challenge for anyone and you all made sure that I didn’t feel a thing. Kartik and Anton

were among my first close friends here and were largely responsible for easing my

transition to life in Washington. Anton’s parents (and Ellaine, of course) welcomed me

into their home and always made sure that I was well-fed and had a reading corner during

any breaks or extended holidays. I will truly miss their warm company and will always

x

have fond memories of all our times together. Nick, Erikka, Chris and Holly – three

holidays that stand out for me during grad school are the ones I’ve spent at your homes.

Your friendship has been invaluable in so many ways. Thanks for being there always and

for your constant reminders that there is a life outside the lab. As peers I look up to, Nick

and Karah have been two simply fantastic people and I was so lucky that their labs were

just down the hall from my own lab. Research interests aside, Nick and I share a deep

passion for teaching and I have learnt many a thing from him during our many

discussions and jam sessions over late night cups of tea. I am particularly grateful to him

for letting me stay over at his place during the crazy snowstorm as he ensured that I had

plenty to eat, giant amounts of orange juice to drink and no snow to plow while I was

working feverishly to finish my dissertation on time. Karah’s enthusiasm for research is

so visible that it is outright contagious. Her leadership during the planning of the

symposium at the ACS National Meeting at Salt Lake City was commendable and I thank

her for all that she has been for me, both as a colleague and as a friend. There are so

many more people I have to mention, but among them – Shelley, Oana, Jeff (the

Platonist), Benny, Shishir, Frank & Caroline, Zach, Neely, Joanna, Julie, Deep, Dinesh,

Baji … I thank them all for their constant friendship and company. I am especially

grateful for every one of them for being there for me when I had to be home one winter

break when Mom took really ill. Their love and support during my time in graduate

school here at GW have meant more to me than they will ever know. If I have left anyone

out here, please forgive me for it. I am sure you are aware of your role, just as anyone

else acknowledged here, in making my time in Washington a really special one.

xi

Abstract


In situ x-ray absorption spectroscopy (XAS) was employed in this dissertation to

probe the poisoning and degradation of platinum (and Pt-based) electrocatalysts under

realistic operating conditions. XAS is a high-energy spectroscopic technique able to

provide both structural and electronic information, and therefore is uniquely suited to

study electrocatalysts in operando. The conventional EXAFS analysis was combined

with the novel ∆µ-XANES method to provide both nanoparticle morphology and

adsorbate coverages on commercially available fuel cell catalysts. The spectroscopic data

were complemented with data from electrochemical techniques such as cyclic

voltammetry (CV) and chronoamperometry (CA), along with rotating disk electrode

(RDE) and copper underpotential deposition (Cu upd) experiments.

The loss of catalytic activity in a fuel cell with age occurs through two chief

processes: poisoning of active surface sites and loss of surface sites through particle

morphological changes, coalescence and aggregation. All of these processes were

investigated using spectroscopic and electrochemical techniques. The poisoning of Pt/C

electrocatalysts by chloride and ruthenium ions was studied using in situ XAS. RDE

experiments show unequivocally that adsorbed chloride drastically hinders the Pt

reactivity by blocking active surface sites and by increasing the overpotential for the

oxygen reduction reaction (ORR) by approximately 85 mV for every 10-fold increase in

chloride concentration. Through the use of the Δμ-XANES method, we were able to

provide direct spectroscopic evidence for site-specific adsorption of Cl- ions on the 3-fold

sites of the (111) planes of Pt nanoparticles, although some of the adsorbed chloride are

xii

forced into bridged or atop sites by strong lateral interactions at high Cl coverage.

It has been established in the literature that ruthenium ions are released into the

electrolyte as a result of degradation of PtRu anode catalysts. The electrochemistry,

electron-spin resonance (ESR) and XAS results reported in this thesis collectively

confirm earlier findings that these species travel through the polymer membrane and

deposit onto Pt/C cathodes, decreasing the ORR activity. ESR results show that the Ru

ions deposited in the membrane alter the hydration levels and transport properties of the

membrane. The deposition of Ru on the Pt cathode was found to be most severe at open

circuit potential (ca. 0.95 V vs. RHE), when the surface is partially covered with O

anions, which may induce a Coulombic attraction for the Ru cations. Comparisons

between the experimental Δμ-XANES results and full multiple scattering calculations

using the FEFF 8.0 code on Pt6 model clusters suggest that the Ru species adsorb

primarily in 3-fold sites on the Pt surface. Semi-quantitative estimates of the Ru coverage

on the Pt/C catalysts, the first such estimate using XAS, are shown to be in good

agreement with other studies in the literature.

An in situ XAS study at both the Pt L3 and Ru K edges on the stability of two

commercial PtRu catalysts aged through voltammetric cycling, along with

chronoamperometry results, reveals that the initial morphology of the PtRu nanoparticles

plays a major role in the catalysts long term stability. Δμ-XANES analysis was carried

out to follow the site number changes with aging, while EXAFS analysis provided

structural information on the changing composition and morphology of the catalysts. It

was found that the samples with larger, more oxidized Ru islands on the nanoparticle

surface are less susceptible to Ru dissolution than those with smaller, more metallic Ru

xiii

islands. Further, as expected, the smaller Ru islands grew faster than their more oxidized

counterparts. These findings and other insights provide an increased understanding of the

observed changes in the methanol oxidation CVs with aging.

Polyvinyl Pyrrolidone (PVP) is a widely used organic capping agent that is also used

to prevent coalescence and aggregation in nanoparticles, effectively slowing the aging

process that commonly occurs during catalysis. While it is generally accepted that a small

amount of PVP (ca. 5-10 wt. %) remains closely associated with the synthesized

nanoparticles to retain their shape, it has been generally assumed that the PVP itself does

not alter the catalytic activity of the Pt. We report that PVP-capped Pt/C nanoparticles

display a remarkable enhancement of their methanol oxidation activity over plain Pt/C

nanoparticles with the exact same size and nanostructure, corroborating a recent study on

Pt black catalysts. Thus the PVP capping agent not only stabilizes the nanoparticles

against aging, but also plays a role, presumably through a ligand effect, in enhancing the

Pt catalytic activity for certain reactions. An in situ XAS study aimed at directly probing

the PVP-Pt interaction reveals that this interaction is potential-dependent: a more neutral

PVP-Pt interaction (PVPN) exists at lower potentials (V < 0.60 V vs. RHE) and changes

to a stronger interaction at higher potentials, involving charge-transfer (PVPCT) from the

PVP to Pt. Theoretical FEFF 8.0 calculations modeling the PVPCT/Pt suggest that the

PVPCT bonds to platinum in atop sites, while the PVPN appears to be either more mobile

or not in registry on the surface. Further, CV data suggests that the PVPN preferentially

blocks H adsorption at sites on the (100) faces.

xiv

Table of Contents

Dedication .......................................................................................................................... iv

Acknowledgments............................................................................................................... v

Abstract .............................................................................................................................. xi

List of Figures ................................................................................................................ xviii

List of Tables .................................................................................................................. xxv

List of Abbreviations…………………………………………………………..............xxvi Chapter 1............................................................................................................................. 1

Introduction......................................................................................................................... 1

1.1 Fuel Cells .................................................................................................................. 2 1.1.1 A brief history of energy technologies prior to and leading to the fuel cell ...... 2 1.1.2 The historical development of the fuel cell........................................................ 5 1.1.3 The basic operating principles of the fuel cell ................................................. 12 1.1.4 Problems keeping the direct methanol fuel cell from commercialization ....... 14 1.1.5 Active areas of low-temperature fuel cell research.......................................... 16

1.2 Characterization of Fuel Cell Catalysts .................................................................. 19 1.2.1 The importance of in operando studies - bridging the structure and pressure gaps in heterogeneous catalysis ................................................................................ 19 1.2.2 Summary of Characterization Techniques...................................................... 22

1.3 Electrocatalyst Degradation .................................................................................... 27 1.3.1 Particle dissolution and growth........................................................................ 30

1.3.1.1 The thermodynamics of dissolution. ......................................................... 30 1.3.1.2 The mechanism for degradation. .............................................................. 32 1.3.1.3 Effects of metal alloying on degradation. ................................................. 38

1.3.2 Degradation of support .................................................................................... 39 1.3.2.1 Dissolution of carbon................................................................................ 39 1.3.1.2 Alternatives to carbon support.................................................................. 41

1.4 Organization of the Dissertation ............................................................................. 43 1.5 References............................................................................................................... 47

Chapter 2........................................................................................................................... 59

In situ X-ray Absorption Spectroscopy: Experiment, Theory and Analysis .................... 59

2.1 XAS – An overview................................................................................................ 59 2.2 Synchrotron Radiation and Experimental methods ................................................ 62

2.2.1 In situ spectroelectrochemical cell for XAS experiments: aspects of design and development ....................................................................................................... 71

xv

2.3 Principles of x-ray absorption spectroscopy........................................................... 76 2.3.1 Historical note on the development of the theory of the x-ray absorption spectrum.................................................................................................................... 82 2.3.2 A mathematical description of the EXAFS region .......................................... 86

2.4 EXAFS analysis ...................................................................................................... 91 2.5 XANES analysis ..................................................................................................... 97

2.5.1 The ∆μ-XANES method .................................................................................. 98 2.5.2 Data analysis .................................................................................................... 98

2.6 In situ vs. in operando XAS on electrodes and electrocatalysts – a literature review..................................................................................................................................... 104 2.7 References............................................................................................................. 111

Chapter 3......................................................................................................................... 123

An Investigation into the Competitive and Site-Specific Nature of Anion Adsorption on Pt Using In Situ X-ray Absorption Spectroscopy........................................................... 123

3.1 Introduction........................................................................................................... 123 3.2 Experimental ......................................................................................................... 128

3.2.1 Electrochemical Characterization .................................................................. 128 3.2.2 In Situ XAS Data Collection.......................................................................... 129 3.2.3 EXAFS and Δμ analysis ................................................................................ 130 3.2.4 Alignment and normalization of XAS data ................................................... 130

3.3 Results and Discussion ......................................................................................... 131 3.3.1 Electrochemical Characterization .................................................................. 131 3.3.2 EXAFS Results .............................................................................................. 136 3.3.3 Δμ-XANES Results ....................................................................................... 144 3.3.4 The 0.4 - 0.7 V region.................................................................................... 149 3.3.5 Chloride adsorption and rearrangement......................................................... 150 3.3.6 Water activation on low index Pt planes, corners and edges......................... 154 3.3.7 Interplay of bisulfate and halide ions on Pt ................................................... 157

3.4 Summary and Conclusions ................................................................................... 160 3.5 Acknowledgments................................................................................................. 164 3.6 References............................................................................................................. 165

Chapter 4......................................................................................................................... 170

Effect of RuOxHy Island Size on PtRu Particle Aging in Methanol ............................... 170

4.1 Introduction........................................................................................................... 170 4.2 Experimental Methods and Data Analysis............................................................ 175

4.2.1 Electrode preparation and XAS cell assembly............................................... 175 4.2.2 In Situ XAS measurements............................................................................ 176 4.2.3 Electrochemical Measurements .................................................................... 178 4.2.4 XANES and EXAFS analysis....................................................................... 179 4.2.5 FEFF 8.0 calculations ................................................................................... 181

4.3 Results.................................................................................................................. 182

xvi

4.3.1 Electrochemical Characterization .................................................................. 182 4.3.2 EXAFS.......................................................................................................... 187 4.3.3 Δµ-XANES Analysis .................................................................................... 190

4.4 Discussion ............................................................................................................ 198 4.4.1 Oxidation state of Ru islands ......................................................................... 198 4.4.2 Ru dissolution and agglomeration ................................................................. 199 4.4.3 Pt dissolution and agglomeration................................................................... 201 4.4.4 Interpretation of CO stripping curve changes................................................ 202

4.5 Conclusions........................................................................................................... 203 4.7 References............................................................................................................. 205

Chapter 5......................................................................................................................... 210

Fundamental Aspects of Spontaneous Cathodic Deposition of Ru onto Pt/C Electrocatalysts and Membranes under Direct Methanol Fuel Cell Operating Conditions: An In situ X-ray Absorption Spectroscopy and Electron Spin Resonance Study .......... 210

5.1 Introduction........................................................................................................... 210 5.2 Experimental Section ............................................................................................ 217

5.2.1 Electrochemical Characterization .................................................................. 217 5.2.2 Flow-through Cell Design.............................................................................. 219 5.2.4 EXAFS Analysis............................................................................................ 223 5.2.5 Δμ Analysis.................................................................................................... 224 5.2.6 Electron Spin Resonance ............................................................................... 225

5.3 Results and Discussion ......................................................................................... 227 5.3.1 Electrochemical Characterization .................................................................. 227 5.3.2 EXAFS Analysis............................................................................................ 231 5.3.3 Experimental Δμ Analysis ............................................................................. 237 5.3.4 FEFF Modeling.............................................................................................. 244 5.3.5 Deposition time dependence and coverage.................................................... 250 5.3.6 ESR results..................................................................................................... 252

5.4 Summary and Conclusions ................................................................................... 256 5.5 References............................................................................................................. 259

Chapter 6......................................................................................................................... 266

Probing the Influence of Polyvinyl Pyrrolidone (PVP) on Supported Platinum Electrocatalysts in 0.1M HClO4 Using in situ X ray Absorption Spectroscopy............. 266

6.1 Introduction........................................................................................................... 266 6.2 Experimental ......................................................................................................... 272

6.2.1 Catalyst synthesis........................................................................................... 272 6.2.2 XAS - sample preparation and data collection .............................................. 273 6.2.3 Electrochemical measurements...................................................................... 274

6.3 EXAFS analysis .................................................................................................... 275 6.4 XANES analysis ................................................................................................... 277

6.4.1 FEFF 8.0 modeling and theoretical calculations............................................ 278

xvii

6.5 Results................................................................................................................... 279 6.6 Discussion ............................................................................................................. 291

6.6.1 How does the PVP increase the Δμ-XANES signal above 0.5 V? ................ 291 6.6.2 The PVP-Pt binding site via FEFF calculations ............................................ 297 6.6.3 Further evidence for a change of PVP interaction from EXAFS results. ...... 299 6.6.4 Evidence for two forms of bonded PVP from other spectroscopic techniques................................................................................................................................. 301 6.6.5 Does the PVP-Pt interaction affect the metal nanoparticle?.......................... 303

6.6.5.1 Previously reported results:..................................................................... 303 6.6.5.2 Evidence from current work: .................................................................. 304

6.6.5.2.1 PVPN at lower potentials .................................................................. 305 6.6.5.2.2 PVPCT at higher potentials ............................................................... 306

6.6.6 Selective or preferential binding of PVP to certain low-index Pt planes (other reports) .................................................................................................................... 310

6.7 Conclusions........................................................................................................... 312 6.8 References............................................................................................................. 314

Chapter 7......................................................................................................................... 324

Conclusions..................................................................................................................... 324

7.1 Poisoning of Pt/C nanoparticles............................................................................ 328 7.1.1 Chloride poisoning......................................................................................... 328 7.1.2 Ru dissolution and poisoning......................................................................... 330

7.2 PtRu electrocatalyst degradation through morphological changes....................... 331 7.3 Probing the interaction between a stabilizing agent (PVP) and Pt/C electrocatalysts..................................................................................................................................... 333 7.4 Some limitations of XAS (experiment and theory) .............................................. 335 7.5 Looking ahead....................................................................................................... 338 7.5 References............................................................................................................. 341

Biography……………………………………………………………………………… 343 Publications……………………………………………………………………………..345 Conference Presentations……………………………………………………………….346

xviii

List of Figures

Figure 1.1 The potential of fuel cells as a power source when compared with other petroleum-based sources. Figure adapted from A.J. Appelby et al., Fuel Cell Handbook (Van Nostrand Reinhold, NY, USA 1989) ..................................... 9

Figure 1.2 A comparison of storage densities of various energy conversion systems. Assumptions: H2 fuel efficiency 40%; DMFC efficiency 25%. Data source: Samsung / SFC Smart Fuel Cell ..................................................................... 11

Figure 1.3 Schematic of a Proton Exchange Membrane (PEM) fuel cell. Adapted from Energy & Environment, MIT Newsletter March 2005 ................................... 13

Figure 1.4 Various mechanisms of catalyst deactivation (loss of active surface area) seen in low-temperature fuel cells. Figure adapted from Shao-Horn et al.62.......... 35

Figure 1.5 Mechanism of degradation in a carbon-supported platinum catalyst showing both, Ostwald ripening and role of carbon in serving as a channel for electronic transport. Figure adapted from a study by Virkar et al.61............... 42

Figure 2.1 The National Synchrotron Light Source located at Brookhaven National Lab, Long Island, N.Y. Picture credit: Courtesy of NSLS, Brookhaven National Laboratory....................................................................................................... 64

Figure 2.2 A schematic of a double-crystal monochromator commonly used to tune the energy of the photon beam.............................................................................. 65

Figure 2.3 Schematic of experimental setup at the beamline showing the two principal methods of collecting XAS data: transmission and fluorescence. .................. 68

Figure 2.4 A typical in situ XAS experiment setup. Shown here is a flow-through in situ XAS cell setup (center) at beamline X-3B at the NSLS. The gas ionization detectors are visible at the bottom of the picture and the cryostat-cooled, solid-state fluorescence detector is seen on the left. ................................................ 69

Figure 2.5 A schematic of an X-ray absorption spectrum over a large energy range showing the K, LI and LII edges. Note that the assignment of edge energies starts from the highest-energy transition......................................................... 77

Figure 2.6 Fundamental processes occurring during an x-ray absorption event a. Excitation of a core-level electron and b. backscattering of the ejected photoelectron due to neighboring atoms surrounding the absorber atom....... 79

xix

Figure 2.7 The various types of backscattering that occur during an absorption event.... 81

Figure 2.8 A reproduction of of an early x-ray absorption spectrum showing assignments of characteristic features that are seen in the EXAFS region of the spectrum………………………………………………………………….. ..84

Figure 2.9 Example of a muffin-tin potential (solid black line) that is frequently used to

compute a theoretical EXAFS spectrum......................................................... 90

Figure 2.10 Common adsorption site geometries seen for many small molecule adsorbates...................................................................................................... 102

Figure 2.11 Adsorbate-induced redistribution of electronic charge on substrate metal atoms closest to the adsorbate. Note that the adsorption reduces the electron density between the surface Pt atoms. .......................................................... 103

Figure 3.1 ORR polarization curves and Tafel plots. (a) ORR polarization curves (anodic sweep) for 30 wt% Pt/C on a glassy carbon disk in O2 saturated 1M HClO4 and 1M HClO4 + 10-3 M KCl at 20oC using a sweep rate of 20 mV s-1. The inset includes 1M HClO4 + 10-2 M KCl (dotted); (b) Mass transfer corrected Tafel plots taken at 900 RPM for 30 wt% Pt/C in 1M HClO4 (circles), 1M HClO4 + 10-3 M KCl (triangles), and 1M HClO4 + 10-2 M KCl (squares). All current densities utilize geometric surface area. ........................................... 133

Figure 3.2 Cyclic Voltammograms of 30 wt% Pt/C (E-TEK) in Ar-saturated 1M HClO4 (solid), 1M HClO4 + 10-3 M KCl (dashed) and 1M HClO4 + 10-2 M KCl (dotted) with a scan rate of 50 mV s-1 on a 5 mm glassy carbon RDE tip at 0 RPM. The vertical lines indicate potentials at which EXAFS measurements were made ..................................................................................................... 135

Figure 3.3 Fourier Transformed EXAFS for 30 wt% Pt/C in 1M HClO4 at 0.54 V vs. RHE measured in situ at the Pt-L3 edge. Phase and amplitude parameters were fit using those generated with IFEFFIT 1.2.9 and sample data. Single shell (Pt-Pt) fit, (1.5 < k < 15.8 Å-1, k2 weighted), performed in R-space. ........... 137

Figure 3.4 Variation of the ratio of population of various coordination sites on the surface of clusters and the total number of surface sites as a function of the particle size of the cluster. Calculations were performed using the methodology developed by Benfield.48 Also shown for comparison is the evolution of the total coordination number and those of the individual sites. All calculations were made using a cubo-octahedron model cluster. ..................................... 142

Figure 3.5 EXAFS results showing change in Pt-Pt coordination as a function of electrode potential for Pt/C electrocatalyst in (a) clean 1M HClO4, (b) 1M HClO4 + 10-3 M Cl and (c) 1M HClO4 + 10-2 M Cl. ................................... 143

Figure 3.6 Pt L3 edge Δμ = μ(V, xM Cl-) - μ(0.54 V clean) spectra for 30 wt% Pt/C in 1M HClO4 and the indicated KCl concentrations......................................... 146

xx

Figure 3.7 Theoretical Δμ = μ(Pt6-Cl) – μ(Pt6) signatures for atop (solid), bridged (dashed) and 3-fold fcc (dotted) chloride on Pt6 clusters.............................. 147

Figure 3.8 (a) Comparison of theoretical 3-fold O (solid) and 3-fold Cl (dotted) Δμ signatures. The dash-dot line shows the sum of the two curves. (b) Comparison of experimental Δμ in 10-2 M Cl- at 0.54 V (solid), 10-2 M Cl- at 1.00 V (dashed) and theoretical Δμ signature for 3-fold Cl-......................... 148

Figure 3.9 Plot of Br coverage55 and Δμ amplitudes representing Cl- coverage (this work) using left axis, and the Gibbs free energy for Cl- adsorption14 using the right axis. The Pt-Pt coordination numbers from Table 3.2 for the 10-2 M Cl- case are indicated with arbitrary units and the Δμ amplitude has been scaled so that it approximately represents coverage in ML. The small shaded lines indicate Cl- coverage at 0.25 and 0.70 V as estimated by Lucas et al.17 The vertical lines roughly separate the regions where Cl- adsorption, compression in the Cl- overlayer, more Cl- adsorption, and OH adsorption dominate as indicated. The symbols at the bottom indicate the dominant Δμ signatures from Figure 3.10 in each region................................................................................................ 151

Figure 3.10 Plot of Δμ= μ(V) - μ(0.40 V) for the indicated Cl- concentrations and comparison with FEFF 8.0 results from Figure 6. Vertical line separates the energy where below the atop Cl- Δμ signature dominates and above the O[H] dominates in magnitude. ............................................................................... 153

Figure 3.11 Cyclic voltammograms of 20 wt% Pt/C (E-TEK)in 0.5 M HClO4 and 0.5M HClO4 + 10-2M Cl- as reported by Schmidt et al.10 (50 mV s-1, 900 RPM, 7 μgPt cm-2). Also shown are fraction of H2O2 formed during ORR on these same samples (Ering = 1.2 V, 5 mV s-1, 1600 RPM) as reported by Schmidt et al.10 Finally the NPt-Pt data from Table 3.2 are plotted scaled and shifted as noted to fit on the right axis. Rectangle indicates region where O[H] from water activation occurs on the cluster corners/edges and on the Pt(100) planes........................................................................................................................ 155

Figure 3.12 Adsorbate mass change (from that at 0.0 V) with potential as estimated from EQCN data reported by Zolfaghari et al.45 in 0.5M H2SO4 and the indicated concentrations of Br- or Cl-. Arrows indicate anodic/cathodic potential direction. The data for Br have been shifted up by 20 g mol-1 Pt for clarity.158

Figure 4.1 CO stripping data24 for the Johnson Matthey (red) and Tanaka (blue) catalysts before and after an 8-hour chronoamperometic test at 500 mV. The data have not been normalized for surface area. ........................................................... 174

Figure 4.2 Summary of XAS data collected. .................................................................. 177

Figure 4.3 Cupric ion stripping voltammograms recorded in 1 M TFMSA + 2 mM CuSO4 taken at a sweep rate of 10 mV s-1. Cyclic voltammograms of a typical PtRu black catalyst in the presence (dot dashed) and absence of Cu ions

xxi

(dotted) showing the various underpotentially deposited regions on both Pt and Ru sites as well as the bulk Cu deposition region.................................. 183

Figure 4.4 Cyclic voltammograms for (a) Johnson Matthey, and (b) Tanaka catalysts after 5, 50 and 500 cycles showing differences in aging properties. The actual data is the same as in reference 24 except normalized to initial Cu stripping surface area for cycles 5 and 50, and normalized to Cu stripping area post cycle 500 for the 500th scans. The inset shows the detail of Cu upd data for the two catalysts.................................................................................................. 186

Figure 4.5 Representative k-space (top) and Fourier Transformed (bottom) EXAFS data and fit for Tanaka PtRu sample at the Ru-K edge taken at 0.54 V after 20 cycles. The 2 path (Ru-Ru and Ru-Pt) fit was performed in R-space (1.574 < k < 13.769 Å-1, k2 weighted. ............................................................................ 188

Figure 4.6 Changes in average Ru-Ru and Ru-Pt CNs with cycling for both the JM and Tanaka catalysts. Error bars of ±0.1 are representative of the relative error, but the absolute error is probably larger. ............................................................ 189

Figure 4.7 Representative CO and O(H) coverages for a PtRu anode in methanol as reported previously using the Δµ-XANES technique.60 ............................... 192

Figure 4.8 Comparison of Δμ(V_cycles) at the Ru K edge, using Equation 4.3b. Also shown are theoretical Δμ signatures denoted OH/Ru and CO/Ru. Note that the Δμ for the Tanaka sample has been scaled by a factor of 8 to place it on the same scale. .................................................................................................... 196

Figure 4.9 Comparison of Δμ(V_cycles) lineshapes at the Pt LIII edge using Eq. 4.3a. Also indicated are theoretical signatures for O(H)/Pt,27, 60 and CO/Pt. The three features in the OH/Pt signature correspond to OH/Pt near a Ru site, OH/Pt away from the Ru islands, and O/Pt. ................................................. 197

Figure 4.10 Schematic representation of the primary PtRu nanoparticle aging processes occurring in the (a) Johnson Matthey and (b) Tanaka catalyst. .................... 200

Figure 5.1 Schematic illustration of the specially designed flow-through style, spectro-electrochemical XAS cell ............................................................................. 221

Figure 5.2 Cyclic voltammograms of 30 wt. % Pt/C taken in Ar purged 1 M HClO4. The Pt/C was loaded onto a 5.56 mm diameter glassy carbon RDE tip with a rotation of 0 RPM, collected at a scan rate of 50 mV s-1 at 20 oC. (a) CV prior to contamination in 2.0 mM Run+ contaminated HClO4 (solid line) and after spontaneous Ru adsorption (OCP, 30 minutes), rinsing (DI H2O), and return to clean 1 M HClO4 (dashed). (b) clean catalyst CV (solid) overlaid with the CV after Ru cleaning step (dashed). The cleaning step involved performing 200 potential cycles between 0.05 – 1.2 V, followed by 50 cycles between 0.05 – 1.4 V clean 1 M HClO4 with a scan rate of 50 mV s-1.................................. 226

xxii

Figure 5.3 ORR polarization curves (anodic sweep) for 30 wt. % Pt/C on a 5.56 mm diameter glassy carbon disk in O2 saturated 1 M HClO4 with a 20 mV s-1 sweep rate at 900 RPM. The solid line represents the clean Pt/C prior to contamination, the dashed line has been exposed to 2.0 × 10-3 M Run+ and subsequently “cleaned” via the cycling procedure and the dash-dot line was collected in 1 M HClO4 + 2.0 × 10-3 M Run+................................................ 230

Figure 5.4 Mass transfer corrected Tafel plots shown at 900 RPM for the ORR polarization curves presented in Figure 5.3. Due to the changing active surface area, we utilize only geometric surface area for current density normalization........................................................................................................................ 232

Figure 5.5 (a) Pt-L3 edge EXAFS spectrum (Kaiser-Bessel window 2.0 < k < 15 Å-1, k2 weighted) and corresponding least-squares fit for 30 wt. % Pt/C in 1 M HClO4 + 2.0 × 10-3 M Run+ fixed at 0.80 V. (b) Fourier transformed EXAFS, fitting was performed in R space using a single shell Pt-Pt scattering path and a Kaiser-Bessel window (1.0 < R < 3.5 Å, k2)................................................. 233

Figure 5.6 Plot of NPt-Pt (solid lines, left axis) for Pt/C in 1 M HClO4 plotted as a function of potential. Also shown are the Ru Δμ magnitudes (Equation 5.1) for Ru deposition on Pt (dashed line, right axis). The dominant Ru adsorption site (n-fold or atop) as indicated by the Δμ spectral line-shape is also given. ......... 236

Figure 5.7 Pt L3 edge Δμ = μ(2.0 × 10-3 M Run+, OCP) – μ(clean, 0.50 V) spectra for 30 wt. % Pt/C using the μ obtained in 2.0 × 10-3 M Run+ in 1 M HClO4 at open circuit, and 0.50 V in clean HClO4. .............................................................. 238

Figure 5.8 Pt L3 edge Δμ = μ(2.0 × 10-3 M Run+, xV) – μ(no Run+, 0.5 V) spectra for 30 wt. % Pt/C taken after 60 minutes exposure to Run+ contaminated 1 M HClO4. 240

Figure 5.9 (a) Pt L3 edge O-adsorption Δμ = μ(2.0 × 10-3 M Run+, xV) – μ(2.0 × 10-3 M Run+, 0.5 V) spectra for 30 wt. % Pt/C taken after 60 minutes exposure to Run+ contaminated 1 M HClO4. (b) Maximum magnitude of similar O Δµ vs. potential under 3 different indicated conditions; i.e. when the 1 M HClO4 electrolyte de-oxygenated with Ar, when saturated with O2, and when saturated with O2 after 60 minutes of Run+ exposure. The shaded arrows indicate the dominant adsorbate as reflected in the Δµ spectral line-shape and discussed in the text. ..................................................................................... 242

Figure 5.10 FEFF 8.0 generated Δμ = μ(Pt6-Ru, site) – μ(Pt6) theoretical spectra for the indicated Ru adsorption sites. The Pt-Ru bond distances used were ~ 2.6 Å........................................................................................................................ 245

Figure 5.11 Comparison of Δμ spectra obtained after 60 minutes exposure in Run+ contaminated HClO4 with the theoretical 3-fold fcc adsorbed Pt6-Ru cluster………………………........................................................................ 246

xxiii

Figure 5.12 Relative coverage of Ru on Pt at 0.5V vs. OCP (ca. 0.9V) by comparison of experimental Δμ-magnitudes at the two potentials....................................... 249

Figure 5.13 Plot of gravimetrically measured water uptake versus extent of Ru exchange in Nafion membranes. Data are fit with a linear trend with a slope of -4.3 and y-intercept 11.5. ............................................................................................ 253

Figure 5.14 Plot of correlation time, τc, versus extent of Ru exchange in Nafion membranes calculated from the rotational diffusion of Tempone spin probe measured using X-Band ESR spectroscopy. Data are fit with a linear trend with slope 1.0711 × 10-9 and y-intercept 1.4037 × 10-9. ............................... 254

Figure 6.1 Chemical formula for PVP polymer (a), illustration of PVP carbonyl-Pt interaction (b) and (c); illustration of PVP polymer on Pt (d). Models after Borodko et al.66 ............................................................................................. 269

Figure 6.2 Illustration of EXAFS and Δµ-XANES analysis procedure, with pre-edge background removal, normalization, and then isolation of the EXAFS signal and fit to model functional in EXAFS, and isolation of the adsorbate effect on the XANES by taking the difference, Δµ. After Roth et al.72....................... 271

Figure 6.3 Comparison of the CV curves for water activation on Pt/C and PVP-Pt/C taken in a) 0.5M sulfuric acid and b) 0.1M perchloric acid. The data were collected at a scan rate of 50 mV/s. .............................................................. 280

Figure 6.4 Comparison of CO stripping curves for Pt/C and PVP-Pt/C taken in a) 0.5M sulfuric acid and b) 0.1M perchloric acid ..................................................... 281

Figure 6.5 Comparison of the methanol oxidation data for Pt/C and PVP-Pt/C with 0.5M methanol in a) 0.5M sulfuric acid and b) 0.1M perchloric acid. .................. 282

Figure 6.6 Chronoamperometry data for Pt/C and PVP-Pt/C in 0.5M methanol and a) 0.5M sulfuric acid and b) 0.1M perchloric acid............................................ 284

Figure 6.7a Experimental ∆µ = µ (Vi) - µ (0.54V) curves for Pt/C at potentials below 0.40 V (vs. RHE) showing adsorbed upd hydrogen. .................................... 285

Figure 6.8 Experimental delmu XANES curves for a) Hupd region (below 0.40 V) and b) oxidation region (above 0.60 V) for PVP/Pt/C. Note the absence of a shift in peak energy for data below 1.1 V. ................................................................ 287

Figure 6.9 A model single-shell EXAFS fit to the PVP-Pt/C data collected at 0.70 V vs. RHE. The data was collected at the Pt L3 edge............................................. 289

Figure 6.10 ∆µ-XANES magnitudes for positive features of lineshapes seen in Figures 6.7 and 6.8. Note the marked increase in an atop OH-like feature between 0.50 and 1.00 V for PVP/Pt/C............................................................................... 292

xxiv

Figure 6.11 Theoretical FEFF 8.0 calculations for atop (1-fold) and 3-fold bonded PVP-Pt/C. .............................................................................................................. 298

Figure 6.12 EXAFS fit results showing changes in Pt-Pt coordination number,NPt-Pt, as a function of applied potential. Also shown (on right) for comparison are the delmu magnitudes originally shown in Figure 6.10...................................... 307

xxv

List of Tables

Table 1.1 Major fuel cell types, operating range, their advantages and limitations. Table adapted from reference 12. ............................................................................... 8

Table 1.2 A brief summary of various experimental techniques and their capabilities.... 23

Table 1.3 Failure modes and their possible reasons for various components of a fuel cell stack. Table adapted from reference 64. ......................................................... 29

Table 3.1 Summary of EXAFS Resultsa ......................................................................... 138

Table 3.2 Distribution of surface sites in a cubo-octahedral Pt cluster as a function of particle size ................................................................................................... 140

Table 3.3 Summary of results at different faces and corner/edge sites on Pt particles... 162

Table 4.1 Summary of Cu stripping results for surface area analysis ............................ 185

Table 4.2 Summary of coordination numbers obtained from Pt-L III edge data.*........... 191

Table 4.3 Summary of results from electrochemical, Cu upd and x-ray absorption data194

Table 5.1 Electrochemically active surface area determination results .......................... 228

Table 5.2 Summary of EXAFS parameters derived from first-shell fits ........................ 235

Table 5.3 Summary of estimates of Ru adsorption coverage on various Pt catalysts .... 247

Table 6.1 Summary of EXAFS results for the Pt/C and PVP-Pt/C catalyst samples ..... 290

Table 6.2 Summary of literature showing two different types of PVP binding ............. 302

xxvi

List of Abbreviations

AES Auger Electron Spectroscopy ATR Attenuated Total internal Reflection CA Chronoamperometry CE Counter Electrode Cuupd Copper underpotential deposition CV Cyclic Voltammetry DEMS Differential Electrochemical Mass Spectrometry DMFC Direct Methanol Fuel Cell DRIFTS Diffuse Reflectance Infra-Red Fourier-Transform Spectroscopy EC-NMR Electrochemical Nuclear Magnetic Resonance ECSA Electrochemically active Surface Area EMIRS Electrochemically Modulated Infra-Red Spectrsocopy EQCM Electron Quartz Crystal Microbalance ESR Electron Spin Resonance EXAFS Extended X-ray Absorption Fine Structure FTIR Fourier-Transform Infra-Red HOR Hydrogen Oxidation Reaction HREELS High Resolution Electron Energy Loss Spectroscopy Hupd Hydrogen Underpotential Deposition IRAS Infra-Red Absorption Spectroscopy ISS Ion Scattering Spectroscopy LEED Low Energy Electron Diffraction ML Monolayer MOR Methanol Oxidation Reaction OCP Open Circuit Potential ORR Oxygen Reduction Reaction PEMFC Proton Exchange Membrane Fuel Cell PIPS Passivated Implanted Planar Silicon ppm Parts per million RAIRS Reflection Absorption Infra-Red Spectroscopy RDE Rotating Disk Electrode RHE Reversible Hydrogen Electrode RRDE Rotating ring disk electrode SEIRAS Surface Enehance Infra-Red Absorption Spectroscopy SFG Sum Frequency Generation SHG Second harmonic generation SIMS Secondary Ion Mass Spectroscopy SNIFTRS Subtractively Normalized Infra-Red Absorption

Spectroscopy SXS Surface X-ay Scattering TFMSA Trifluoromethanesulfonic Acid UHV Ultra-high Vacuum UPD Underpotential Deposition

xxvii

UV Ultraviolet Vis Visible WE Working Electrode XANES X-ray Absorption Near Edge Structure XAS X-ray Absorption Spectroscopy XRD X-ray Diffraction XPS X-ray Photoelectron Spectroscopy

1

Chapter 1

Introduction

This thesis aims to understand some of the different forms of degradation that occur

on low-temperature Pt and Pt-based fuel cell electrocatalysts using in situ x-ray

absorption spectroscopy. In this chapter, a broad albeit brief history of the predominant

energy technologies that were extant before the invention of the fuel cell and a short

history of fuel cell technologies are first presented. This is followed by a section on the

operating principles of low-temperature fuel cells and challenges facing their

commercialization, highlighting electrocatalyst degradation as one such major challenge.

A discussion of the experimental techniques used in heterogeneous catalysis that are

bridging surface science and catalytic studies is then presented. The chapter concludes

with a literature review summarizing our current state-of-knowledge on electrocatalyst

degradation in low-temperature fuel cells. All topics relating to x-ray absorption

spectroscopy, the primary technique utilized in this work, and its application to study

electrocatalysts will be taken up in chapter 2.

2

1.1 Fuel Cells

1.1.1 A brief history of energy technologies prior to and leading to the fuel cell

The energy demands of modern-day society are met chiefly through the strategic

utilization of energy-rich fossil fuels, such as coal, petroleum and natural gas, or from

nuclear fuels, and more recently through the harnessing of alternate renewable sources

such as that obtained from solar, wind, geo-thermal, and tidal waves. While early man’s

energy sources were primarily those occurring in nature, such as wind, running water,

and muscle power of domesticated livestock (even human muscle power), a paradigm

shift in the use of energy occurred in the late 17th century in Britain, when the expansive

power of steam was discovered and harnessed mainly through the pioneering works of

Savery, Newcomen, Watt and Trevithick, 1 paving the way for the Industrial Age. Life

was transformed completely, and ever since modern society has depended heavily on

powered machinery and engines for all its needs.

Somewhere around the same time, the Italian biologist Luigi Galvani discovered that

an electric current was obtained when two different metals were inserted into the muscles

of a frog, and speculated that it was due to an ‘animal electric fluid’ that was present in

living creatures. Following up on Galvani’s discovery, Alessandro Volta invented the

‘Voltaic Pile’, made of an alternating stack of Zinc and Copper disks separated by

cardboard disks soaked in vinegar, 2 and in 1800 demonstrated the nature of the newly

discovered effect to the French Academy of Sciences. He termed the strange ‘force’

Electromotive Force (emf), a term in use even in the current literature to indicate voltage

3

or more correctly, a potential difference. The Voltaic pile is widely acknowledged as the

first battery.∗ (see footnote)

The nature of the emf and its chemical effects were carefully investigated by Michael

Faraday, the protégé of one of the giants of 19th century chemistry, Sir Humphrey Davy.

Starting in 1832, he discovered the fundamental laws of electrolysis viz. that the quantity

of an element electrodeposited under a given potential is directly proportional to both, the

quantity of electric charge passed through it as well as the atomic mass of the element

itself. The work of Volta and Faraday paved the way for the invention of the first primary

cell in 1836: the Daniell Cell. It had two liquid electrolytes and produced a steadier

current than the Voltaic Pile. The lead-acid battery, the first secondary battery, was

invented in 1859 by the French physicist Gaston Planté.2 Lead acid batteries are still

widely used in automobiles and back-up power systems.

It is remarkable that little could be done to improve upon the simple chemistry of the

robust lead acid battery, and apart from a few minor additions and improvements, it has

stayed virtually unchanged to this day. While a wide variety of batteries, capable of

delivering power across the complete spectrum of applications, are now used

ubiquitously, the development of this fascinating field is not discussed any further here;

instead, as warranted in a dissertation concerning fuel cells, we will briefly outline the

historical development of the fuel cell. (see Section 1.1.2)

∗ Some historians claim that a primitive version of a battery already existed in the Baghdad Battery found in 1936 by archaeologist Wilhelm Konig. The clay jar, dating back to the Parthian period (250 B.C. - 250 A.D.), contained two different metal ‘electrodes’ and space for a liquid, which was possibly a fruit juice, wine or even vinegar. It is speculated that it was used for either medical reasons or even to electroplate gold onto other metals

4

It was around the same period as development of the battery that Sir William Grove

invented the Fuel Cell (1839),3 a device that would produce electricity from hydrogen

and oxygen, which were combined to produce an electric current; essentially a ‘reverse

hydrolysis’ process. For close to a century after, there was little progress on fuel cells, if

any, and this can be attributed to the rapid and widespread availability of electricity

produced through an application of the principles of electromagnetic induction (also due

to Faraday, ca. 1831), the discovery of petroleum, and the subsequent development of the

Internal Combustion (I.C.) engine during the latter half of the 19th century. The

generation and widespread distribution of electricity was due to the efforts of Nikola

Tesla, George Westinghouse and Thomas Alva Edison, amongst whom bitter legal

disputes concerning priority of ideas in the generation and distribution of direct and

alternating current ensued. Notwithstanding all of this, in 1896, the city of Buffalo, N.Y.

was powered by electricity transmitted from a hydroelectric generator installed at Niagara

Falls, N.Y., 26 miles away, heralding the electric age.

Meanwhile, rapid developments in the I.C. engine were also taking place following

major theoretical and engineering advances. Frenchman Nicolas Sadi Carnot delineated

the thermodynamic principles of a cyclic process operating between two temperatures in

his seminal work Reflexions on the motive power of fire published in 1824, a work which

impacted literally all future progress in thermodynamics. After some engineering

advances by a number of people, Jean-Joseph Étienne Lenoir in 1860, built the first mass-

produced I.C. engine. 4 In what was a definitive turning point for the development of I.C.

engines, the four-stroke and two-stroke cycles for an I.C. engine were designed and

refined in the last three decades of the 19th century by Nikolaus Otto, Gottlieb Daimler,

5

Wilhelm Maybach and Karl Benz; 4, 5 finally, the compression-ignition engine (more

commonly known as the Diesel engine) was invented by Rudolph Diesel in 1893, which

was improved upon to its near-present form by inventor Charles F. Kettering. 6

The development of engines running on either gasoline or diesel secured a demand

for petroleum that has only risen exponentially over the past century. However, this is

clearly unsustainable as not only are the Earths’ reserves of fossil fuel limited, but more

importantly, the combustion of coal and petroleum releases large amounts of CO and CO2

into the atmosphere, upsetting the natural balance of our delicate ecosystem that supports

all life on earth. It is hoped that clean battery technology and fuel cells of one kind or

another, powered through a sustainable energy grid (which will most definitely involve

solar energy) will one day replace most of current technology running on various forms

of coal and petroleum.

1.1.2 The historical development of the fuel cell

The first resurgence in fuel cell research after Grove’s invention in 1839 came with

the work of Ludwig Mond and Charles Langer (1889), who was using coal to develop a

Direct Coal Fuel Cell (DCFC). Here, a gas was derived from coke and coal which could

be used as a fuel.7 However, impurities in the gas quickly poisoned the platinum catalyst

and made the process prohibitively expensive, as the catalyst had to be replaced often or

the loading had to be increased substantially to derive any power from the device. DCFCs

were also investigated by W.W. Jacques in the U.S. and Prof. Baur and his students at the

Swiss Federal Institute of Technology (ETH), Zurich, where considerable progress was

6

made on various kinds of electrolytes and cell-designs.8 Their effort was eventually

abandoned due to a host of practical problems.

In 1933, Sir Francis Bacon developed the first practical fuel cell – the Alkaline Fuel

Cell (AFC).9 However, corrosion of the components of this fuel cell was a major problem

associated with the AFC. After subsequent improvements in both design and choice of

materials, the technology was licensed to Pratt & Whitney in 1959; this became the

precursor to the AFCs developed for the first Apollo mission (and all following space

missions) in the early 1960s. The space mission led to rapid advances in both AFC and

Phosphoric Acid Fuel Cell (PAFC) technology.

The PAFC was invented by William Grubb of General Electric in 1955. 10, 11

However, the two major drawbacks of the PAFC were a) the limited lifetime of the

polymer electrolyte membrane, which was susceptible to acid attack and oxidative

degradation, and b) the high Pt loadings required. The AFC did not have the latter

problem, as it was operated at higher temperatures, and due to increased reaction kinetics,

required lower Pt catalyst loadings. It is noteworthy that for a long time, i.e. until the

early part of this decade, PAFCs were the only fuel cells that were commercially

available and in use for stand-alone power generation. Many commercial PAFCs were

used, for instance, as critical back-up power systems in large factories, banks etc. where a

power outage on the grid could mean serious consequences in cost and liability. Also, it

is remarkable that some of the fuel cells developed in the 1960s for the space missions

are still in operation today, more than four decades later! A resurgence in research on fuel

cell technology, which had peaked during the ‘space age’, along with research into

sustainable and renewable energy, came in the wake of the drop in the worldwide supply

7

of oil in the 1970s due to the OPEC oil embargo. However, research momentum soon

plummeted again following the drop in government tax incentives for such companies,

and due to the widespread availability and low-cost of oil and natural gas through the

1980s. Over the last couple of decades however, there has been a strong, concerted

worldwide research effort to help bring fuel cells (along with solar and wind energy) to

the market. Governments, especially in the U.S. and Europe working with national labs,

research institutes and universities have established both, short-term and long-term goals

for the target costs of materials and the performance level of catalysts (for e.g. loading)

required to make fuel cells a cost-effective, viable alternate technology to meet our

growing energy demands, as well as to decrease our dependence on petroleum, and set

our sights on a cleaner and greener energy future.

Over the years, many different fuel cell types have been developed. A summary of the

major fuel cell types available today, their operating range, advantages and disadvantages

etc. are shown in Table 1.1.12 From the table, it is quite apparent that low-temperature

fuel cells (20-120 ºC) hold significant promise for widespread application, especially the

portable energy market. Shown in Figure 1.1 are the efficiencies of various fuel cell types

against I.C. engines.

Even considering losses at the system level, the operating efficiency of many fuel

cells are well above the 50-60% mark and as such, may be twice as efficient as even the

best I.C. engines, and thus hold considerable promise as alternate or even mainstream

energy conversion sources. Fuel cells, unlike I.C. engines, also have no moving parts

which makes them less susceptible to mechanical wear and tear. Further note that while

the performance of the DMFC is only around a-half to a-third that of the PEMFC, for

8

Table 1.1 Major fuel cell types, operating range, their advantages and limitations. Table adapted from reference 12.

9

100 101 102 103 104 105

10

20

30

40

50

60

70

Gasoline Electric

Steam and Gas Turbines

Diesel Electric

Phosphoric Acid FCClean Fossil-Fueled Fuel Cells

Molten Carbonate FC with Internal Reforming

Fuel Cells with Hydrogen Fuel

Power Output, kW

Effic

ienc

y, %

LH

V

100 101 102 103 104 105100 101 102 103 104 105

10

20

30

40

50

60

70

Gasoline Electric

Steam and Gas Turbines

Diesel Electric

Phosphoric Acid FCClean Fossil-Fueled Fuel Cells

Molten Carbonate FC with Internal Reforming

Fuel Cells with Hydrogen Fuel

Power Output, kW

Effic

ienc

y, %

LH

V

Figure 1.1 The potential of various types of fuel cells when compared with other widely used power sources. Figure adapted from A.J. Appelby et al., Fuel Cell Handbook (Van Nostrand Reinhold, NY, USA 1989)

10

portable applications, it still offers considerable advantages. The fact that methanol is a

liquid under typical operating conditions (room temperature, atmospheric pressure) is

advantageous from a systemic standpoint and more importantly, is readily compatible

with the existing gasoline distribution infrastructure.

Four of the six atoms in the methanol molecule are hydrogen, which makes it quite an

energy-rich source of fuel; it is also easily synthesized with available industrial

processes.13 Some other advantages over, for e.g. the state-of-the-art lithium ion batteries,

include it being environmentally-clean, allows fast recharge-cycles, and has a long

lifetime.14 Shown in Figure 1.2 is a comparison of the storage densities of the different

battery technologies alongside those of the PEMFC and DMFC. If these numbers are

achievable on a large-scale, it is seen that low-temperature fuel cells clearly possess an

edge over existing battery technologies.

In short, the performance of most fuel cells as they stand is more than competitive

when compared with many battery types and I.C. engines. Only the durability and

economic issues of low-temperature fuel cells keep them from entering the market. They

have to compete with the usual lifetimes of 10,000-15,000 hours commonly seen in I.C.

engines used in cars and bikes all over the world. Some mature fuel cell technologies

have already met the stringent demands placed on operating lifetimes. For instance, some

200 kW PAFC systems built by UTC Power routinely operate for over 20,000 hours,15

but such impressive limits are still to be realized for the principal low-temperature fuel

cells, viz., PEMFCs and DMFCs.

As this dissertation will focus on efforts to understand at a molecular-level, the aging

and degradation of electrocatalysts in DMFCs, the following section will aim to provide a

11

30 40 100 150

300 350

1200

0

200

400

600

800

1000

1200

Stor

age

dens

ity, W

h/kg

Lead-Acid

battery

NiCdbattery

NiMeHbattery

Li-ionbattery

Li-polymerbattery

HydrogenFC

DMFC

Figure 1.2 A comparison of storage densities of various energy conversion systems. Assumptions: H2 fuel efficiency 40%; DMFC efficiency 25%. Data source: Samsung / SFC Smart Fuel Cell.

12

more detailed picture into the basic operating principles of low-temperature PEMFCs

followed by a brief introduction to the Direct Methanol Fuel Cell, and finally, some of

the main challenges and prospects faced by this technology.

1.1.3 The basic operating principles of the fuel cell

In its most basic form, the fuel cell is an electrochemical energy conversion device

which ‘burns’ a fuel (hydrogen-based) electrochemically in a controlled manner using an

oxidant (usually oxygen or air). Some of the common hydrogen-rich fuels being explored

for use in low-temperature fuel cells include methanol (CH3OH), ethanol (C2H5OH) and

sodium borohydride (NaBH4). However, for sake of brevity and relevance, we will

discuss only the H2/O2 PEM fuel cell (PEMFC) and the Direct Methanol Fuel Cell

(DMFC). The functional center is much the same for both types of fuel cells and is called

the membrane electrode assembly (MEA), which consists of an anode, a polymer

electrolyte membrane (also known as proton exchange membrane) and a cathode. A

schematic of a typical fuel cell with MEA architecture is shown in Figure 1.3. The

primary reactions of interest occurring at the anode and cathode are as follows - 16

Anode: H2 2H+ + 2e- (PEMFC) E0 = 0.00 V

CH3OH +H2O CO2 + 6H+ + 6e- (DMFC) E0 = 0.046 V

Cathode: 3/2 O2 + 6H+ + 6e- 3H2O E0 = 1.23 V

Overall: 2H2 + O2 2H2O (PEMFC) E0 = 1.23 V

CH3OH + 3/2O2 CO2 + 2H2O (DMFC) E0 = 1.18 V

13

Figure 1.3 Schematic of a Proton Exchange Membrane (PEM) fuel cell. Adapted from Energy & Environment, MIT Newsletter March 2005

14

Thus, at the anode, hydrogen is dissociated into protons on a catalyst (typically a Pt-M

alloy catalyst) which can then travel through the electrolyte known as the polymer

electrolyte membrane (PEM) onto the cathode side of the cell. The electrons released

from the oxidation reaction on the anode are conducted away electronically out of the cell

and directed into the desired application (electric load). The electric circuit is closed by

allowing these electrons to recombine with the incoming protons and the supplied oxygen

on the cathode, reducing oxygen (on a Pt catalyst) effectively to pure water. If there is

carbon in the fuel (methanol, ethanol), an additional oxidation product viz. CO2 is also

formed. The CO2 produced is rejected by the acidic membrane and remains as a product

on the anode side and is typically removed by the circulating methanol.17 Note that in

case of the DMFC, six electrons are released on oxidation of one molecule of methanol

when compared to only two electrons in case of H2 in a PEMFC.

1.1.4 Problems keeping the direct methanol fuel cell from commercialization

Currently, some of the major challenges in the development of DMFCs include:

a) Poor anode kinetics: The poisoning of the catalyst surface by intermediates

and by-products of methanol dissociation 18 causes severe overpotential losses relative to

the Hydrogen Oxidation Reaction (HOR), of ca. 200-300 mV. 19 On pure platinum,

which is still the best catalyst for methanol dissociation, the anode potential may have to

be raised to as much as 0.60 V to completely oxidize the impurities and byproducts of

methanol dissociation resulting in dramatic loss in cell potential. While significant

progress has been made over the last 15-20 years to develop more efficient

electrocatalysts to reduce the overpotential losses on the anode, the strategy has nearly

always been to alloy the Pt catalyst with more oxophilic elements such as Ru, Sn, Ti, Pb

15

etc. 20 whereby the adsorbed O(H) groups on these elements assist in the oxidation of the

adsorbed species (chiefly CO) through a synergistic mechanism. This effect, termed the

bi-functional mechanism, was first studied in the context of CO oxidation on platinum by

Bockris and Wroblowa 21, most notably by Watanabe and Motoo 22, 23 and, Janssen and

Moolhuysen. 24 To date, the most effective anode catalyst remains a 1:1 alloy of platinum

and ruthenium. 25

b) Methanol crossover: A second problem is the slow but definite ‘crossover’ of

methanol to the cathode, leading to a drastic reduction in operating cell potential and

lifetime. This phenomenon is widely known as ‘methanol crossover’. A similar reaction

to that of the anodic reaction occurs on the cathode side, wherein methanol is oxidized on

the Pt catalyst to CO2 and H2O alongside the oxygen reduction reaction, leading to a

‘mixed potential’. The overpotential loss and rate of methanol crossover has been

estimated by several researchers; 26-28 the overpotential loss is estimated to be of the order

of 100-120 mV. This is in addition to the losses already incurred on the cathode due to

sluggish ORR kinetics. In practice, the cathode potential rarely exceeds 1.0 V (theoretical

ORR potential is 1.23 V) and is chiefly due to the poisoning of available Pt sites by OHads

and the associated water-dipole layer that cannot be easily penetrated by molecular O2.

c) Electrode stability: The stability of both the anode and cathode catalysts and

the reduced methanol oxidation activity are major concerns with the DMFC. Further

degradation studies and advances in our understanding, as well as the development of

novel materials with enhanced stability, will be discussed in greater detail in a separate

section (see Electrocatalyst degradation below)

16

d).Cost: While DMFCs are fairly rugged, quite clean and efficient, the costs are

still prohibitively expensive for widespread application (also true of the PEMFC).

1.1.5 Active areas of low-temperature fuel cell research

Several advances are yet to made on many fronts if low-temperature fuel cells are to

be commercialized and remain competitive with existing energy technologies. Some key

areas include: 12

1. Membranes: Some of the major improvements required of membrane technology

are the development of rugged, high-temperature membranes, membranes with

sufficiently high proton conductivity but lower permeability to reduce methanol

crossover and finally, reducing costs.

2. Catalyst loading: While dramatic reductions in electrocatalyst loadings have

been achieved over the past decade, the cost (and availability) of platinum is an

ever- present concern in the fuel cell community.∗ (see footnote) Attempts to

∗ “It has sometimes been suggested that the full exploitation of low temperature fuel cells may be limited by the availability of the platinum-group metals. Mike Steel (Johnson Matthey) posed the question of how much platinum is likely to be required, and whether the increased demand can be met. Bill Ford of the Ford Motor Company has forecast that by 2025, one quarter of all light vehicles will be powered by hydrogen. Assuming that each car will require about 75 kW of fuel cell power, and using the U.S. Department of Energy target of 0.2 g kW−1 of platinum, Mike Steel estimated that platinum demand for fuel cell cars could be 150–300 tons per year by 2025. This compares with a production rate of 180 tons per year in 2000, and proven reserves of 5000 tons, with inferred reserves of 30,000 tons of platinum, but does not include platinum recovered and recycled, a practice already developed for automotive emissions control catalysts in the advanced economies of the world. He concluded that platinum is a key catalyst for PEMFC development, and that there should be sufficient resources available to meet the needs for the foreseeable future.”

Source: The Eighth Grove Fuel Cell Symposium Developing a fuel cell manufacturing industry;

D. S. Cameron, Platinum Metals Review, 48 (1), 2004, 32-37

17

continually derive acceptable performance with lower loadings is a continuing

research effort. There have also been a number of groups working on the

development of non-precious metal catalysts. While some of them have

comparable performance to that of platinum and its alloys, none of them have the

requisite stability to operate for long hours in a fuel cell environment.

3. CO-tolerant anode catalysts: PtRu black is still the best catalyst as far as CO-

tolerance is concerned. However, research into developing more efficient anode

catalysts to better the benchmark set by PtRu alloy catalysts has and continues to

be an active research area.

4. Bipolar plate materials: Non-porous graphite is the material of choice for the

bipolar plates in fuel cell stacks owing to their excellent conductivity, reasonable

chemical resistance and mechanical properties. However, machining graphite is

both expensive and time-intensive. Further, while they have good chemical

stability compared to most materials, they are still not completely immune to

chemical attack from the acid in the environment, which leads to slow dissolution

of parts of the plates and reduces the operating lifetime of the fuel cell stack.

5. Engineering concerns: Bipolar plate design to reduce mass-transfer limitations

and system design, which includes standard fuel cell stack components such as

fuel cell processor, pumps, membrane humidifiers, pressure controllers, gas

membrane separators (for e.g., to separate CO from H2 in the feed stream), are

also active fields of engineering research.

6. Durability and lifetime of fuel cells: It was mentioned earlier that fuel cells are

immune to any mechanical stress and degradation. While this is true, owing to the

18

very nature of its activity, they are certainly affected by various kinds of chemical

degradation, which manifests as a loss in activity due to irreversible changes

happening on the surfaces of these electrocatalysts. As mentioned earlier, a

proven operating lifetime of 1000-2000 hours is a minimum requirement if

DMFCs have to be competitive in the portable applications market. In some cases,

the requirement may exceed 8000-10,000 hours.

Thus, all aspects of degradation in membranes, stack components and electrocatalysts

(and their supports) need to be thoroughly explored and significant gains have to be

made before any of these low-temperature fuel cells are to hold their position in the

portable power sector. Many companies have invested significant resources and have

directed some of their efforts into developing DMFCs for commercial applications.

Some of them are listed below-13

1. Ball Aerospace & Technologies 2. Direct Methanol Fuel Cell Corp.

3. Giner Electrochemical systems 4. Plug Power

5. Fideris (formerly Lynntech Industries) 6. Manhattan Scientifics

7. MTI Micro Fuel Cells 8. Medis Technologies

9. Motorola Labs 10. NEC

11. PolyFuel 12. Yuasa Corp.

13. Samsung (AIT) 14. Toshiba

19

1.2 Characterization of Fuel Cell Catalysts

1.2.1 The importance of in operando studies - bridging the structure and

pressure gaps in heterogeneous catalysis

The catalyst is arguably the most critical component in the fuel cell. The word

‘catalysis’ was first coined by the great Swedish chemist Jons Jakob Berzelius in 1836

(from the Greek word ‘katalein’ meaning ‘to dissolve’ or ‘to break down’) when he stated

“Catalytic power means that substances are able to awake affinities that are asleep at a

particular temperature by their mere presence and not by their own affinity”.29, 30 Thus,

the word is used to describe the property by which the rate of a particular reaction is

accelerated by a substance, while it itself is not consumed (we will see later that given

our current understanding of catalytic processes, this is not entirely correct). Most

catalysts are impressive in what they can do to speed up reactions. Some catalysts are

known to increase the rates of particular reactions by more than a factor of 1010, i.e. 10

billion-fold. It is even more fascinating that catalysts favor a certain reaction pathway

over another, a property called its ‘selectivity’. They are indispensible in society today as

they are used in virtually all industries to aid in the manufacture of important chemicals,

many of them which are essential to modern life. These chemicals are eventually used to

make products ranging from cosmetics and medicinal compounds to polymers such as

plastics and foams. Some widely used catalysts include iron powder in the Haber-Bosch

process to produce ammonia, Pt to convert saturated alkanes to aromatic and branched

hydrocarbons, Co, Ni, Fe and Ru to hydrogenate CO to produce methane and other

valuable organic compounds (Fischer-Tropsch synthesis) such as alcohols, aldehydes and

20

acids; Co and Mo compounds in the hydro-desulfurization of petroleum and coal, and

zeolites for petroleum cracking etc. Many processes occurring in nature, including

photosynthesis, and most of the biological processes in our own bodies (chiefly as

enzymes) are catalytic. Catalysts exist in myriad chemical and physical forms such as

metals (pure or as alloys), oxides, nitrides, and sulfides; they may be single-phase or

multi-phase. They can be amorphous or crystalline. Catalysts can be metallic,

semiconducting or insulating; they may be in the form of a fine gauze, or powder-like or

even a suspension of particles in a slurry. Catalysts also play key roles in energy

production, transformation, storage and utilization.31

The catalysis field may be broadly categorized into two main types of catalysis:

homogeneous catalysis and heterogeneous catalysis. In homogeneous catalysis, the

reactants and catalyst are in the same phase (or form a uniform phase when mixed

together) whereas in heterogeneous catalysis, reactions occur on a solid surface. Due to

the difficulty in separating and eventually removing the reagents and products from the

catalysts, most industrial processes are designed to employ heterogeneous catalysts. It has

been reported that products obtained via some form of heterogeneous catalysis accounts

for more than 25% of the world’s total GDP.32

Every heterogeneous catalytic reaction, by definition involves a surface, be it a solid-

liquid interface or a solid-gas interface. However, even the simplest of such reactions can

be very complex at the molecular level and a full understanding of a reaction mechanism

is arrived often, only through the use of a number of techniques which are complimentary

in nature. Such studies would ideally include both experimental and theoretical methods.

Further, information at the macroscopic level, such as physical properties (viscosity,

21

density, tensile strength etc.) and reaction rates, have to be measured and correlated with

microscopic/molecular phenomena – indeed, this is the very goal of physical chemistry

itself. The realization of this objective within the field of catalysis will also naturally lead

to what many consider the ‘Holy Grail’ of catalysis i.e. the rational design of catalysts to

catalyze any given reaction, with a 100 % selectivity and maximum theoretically possible

yield.

Given that heterogeneous catalysis is the field of study of reactions at interfaces and

surfaces, one would imagine that surface science can and should provide substantial

insights into the way catalysts work. However, heterogeneous catalysis is largely an

empirical science and for various reasons, progress in the discipline and our fundamental

understanding of catalytic processes on surfaces lagged developments in industry by

years.

Two widely acknowledged barriers to using surface science methods to understand

catalytic behavior still exist in one form or another, viz. the ‘structure-gap’ and the

‘pressure-gap’.31, 33-36 They are discussed briefly here in order to fully appreciate how

close researchers have come towards nearly eliminating them and in the process,

unraveling the secrets of heterogeneous catalytic processes.

Industrial catalysts are often supported and highly dispersed, containing particles a

few nanometers across and having any number of defects, kinks, terraces, corners and

edges etc., while model studies in surface science are carried out on clean, single-crystal

surfaces having a preferred orientation (e.g. Pt(111), Ni(100) etc.) and are of the order of

several microns to a few millimeters in size. This forms the basis for the ‘structure-gap’

that exists between the two disciplines. As for the ‘pressure-gap’, surface science studies

22

traditionally involve the use of clean surfaces under ultra-high vacuum (UHV) systems

wherein pressures lower than 10-10 Torr are routinely used; most industrial processes are

designed to take place at high pressures, or under atmospheric pressure (760 Torr), at the

very least. A cursory glance at these numbers immediately reveals that this pressure gap

exceeds a factor of 1012. Thus the conditions for traditional ‘surface science’ and catalysis

in industry could not be further apart. Yet, significant advances have been made to

correlate molecular-level detail gleaned chiefly from spectroscopic studies with kinetic

and thermodynamic data. It must be stressed here that these studies did achieve what they

set out to do – observe species on surfaces and understand how they behave or react

under the conditions of experiment. Our understanding of catalysis on various catalysts

at a molecular level, albeit at low pressures and model surfaces, was put on solid ground

by these studies.

Attempts to bridge these gaps have involved the classic ‘before-after’ approach which

continues to prove important in many studies. Careful comparisons of the two data sets

were made in order to extrapolate the results to the catalytic process, often leading to

proposed reaction mechanism(s) that nevertheless proved hard to validate. Numerous

analytical techniques were developed to overcome this problem, some of which are taken

up in greater detail in the following section.

1.2.2 Summary of Characterization Techniques

Very few techniques are capable of providing molecular-level information on

electrocatalysts under operating conditions, as they often require specially-prepared

surfaces, vacuum conditions, or other special environments, such as those summarized in

Table 1.2. In situ studies of model systems have been carried out using newly- developed

23

Table 1.2 A brief summary of various experimental techniques and their capabilities

Technique Technique Type

Information obtained

Advantages/disadvantages Conditions

HREELS Vibrational Spectroscopy

Adsorbate information

Can be coupled with TEM; sub-monolayer resolution

UHV required

LEED Diffraction Surface structure (only geometry)

Surface reconstructions can be observed

UHV required

XPS/Auger Electronic Spectroscopy

Surf. composition, oxidation state

Qualitative and quantitative information; imaging/mapping possible; 0.01-1 at. % detection limit

UHV required

ISS Ion Scattering Spectroscopy

Surf. composition, structure

Complex analysis; surface imaging/mapping possible

UHV required

SIMS Mass Spectrometry

Surface composition Destructive technique; difficult to get quantitative information

UHV required

TDS/TPD Thermal Spectroscopy


Number, binding energy of species can be obtained; sub-monolayer resolution

Vacuum required

SNIFTIRS, SHG/SFG

Vibrational Spectroscopy


Difference technique possible but noisy; selection rules makes 3-fold adsorption sites invisible

Ambient conditions

IRAS/ATR DRIFTS RAIRS/ EMIRS

Vibrational Spectroscopy


Internal bonds of adsorbate detectable; only species with bonds perpendicular to the surface are detected

Ambient conditions

XRD Diffraction Bulk structure Synchrotron required for in-situ XRD studies in a condensed phase.

Cond. phase possible

GI-XRD Diffraction Surface structure, adlayer information

Synchrotron source required due to low surface area of exposed crystal


UV/Vis, Laser, ESR

Electronic Spectroscopy

Atomic/molecular species

Not a surface sensitive technique; ultrafast spectroscopy possible.


EC-NMR Magnetic resonance

Electronic landscape of interface based on Ef-LDOS; dynamics of surface processes

Adsorbate specific information possible with isotopes; surface diffusion parameters of adsorbates can be estimated

Condensed phase possible

Neutron Scattering

Diffraction Bulk structure Sensitive to detect hydrogens in a structure; However, a neutron source is required; not surface sensitive


EQCM Gravimetry Net adsorbate mass changes

Possible to simultaneously obtain a voltammetric response


DEMS Mass spectrometry

Detection of oxidation products

Quantitative technique for CO2 measurement (a main oxidation product from all hydrocarbon fuels)


Radiotracer Radio-labeling

Identification of specific adsorption

Possible to study co-adsorption and reversibility of adsorbed species; can track concentration, determine rates of reactions and coverage. Equipment and materials can be quite expensive.


XAS Electronic Spectroscopy

Oxidation state, local symmetry around absorber, structural information (bond distance, coordination number), adsorbate information.

Synchrotron source required; analysis can be complex to carry out. Element specific; Site-specific adsorbate information possible; in situ electrochemical/XAS setup can be sophisticated.


24

x-ray spectroscopies, such as thermal desorption spectroscopy and molecular beam

studies, along with optical and magnetic spectroscopic techniques at high pressures.35

Valuable information on catalytic activity and the role of steps, edges and corners,

defect, adatoms (promoters and inhibitors), and ensemble effects etc. have been

established using some of these techniques. It can be seen from Table 1.2 (a compilation

from various sources) that while many techniques like XPS, LEED, HREELS and ion

beam scattering techniques are very powerful techniques to understand surface

phenomena, they can only be used under ultra-high vacuum conditions. Other vibrational

techniques like SFG, SNIFTRS, FTIR, SEIRAS etc. may be applied to study catalysts at

normal pressures and even in solution, but owing to the selection rules, are quite

insensitive to adsorbates that are not oriented in a specific way on the surface. Further,

except for a few surface-sensitive IR techniques, most of them will also probe many ionic

species in the liquid electrolyte, making it impossible to separate out the contribution

from adsorbed vs. non-adsorbed ions. Also, adsorbed OH, a chemically important species

in many electrocatalytic reactions, cannot be discerned with IR techniques as there is also

a significant number of these species in any aqueous electrolyte layer around the catalyst.

Diffraction techniques provide primarily structural information and are not as effective

on nanoparticle catalysts, which do not possess long range order. Gravimetric techniques,

like EQCN, can provide some information on the extent of adsorption but unlike

spectroscopic techniques, cannot be used to probe the nature of the interaction between

the surface and adsorbate. Thus, XAS, as a high-energy spectroscopic technique that can

provide both electronic and structural information, seems uniquely positioned to probe

electrocatalysts.

25

The study of chemical processes on surfaces has had a long and distinguished history.

Some of the major contributions to heterogeneous catalysis that have been awarded the

Nobel Prize include Paul Sabatier’s contribution for developing catalysts for the

hydrogenation of organic compounds (1912), Fritz Haber for “the synthesis of ammonia

from its elements” (1918) and Irving Langmuir (1932) for his seminal contributions to a

number of areas within surface science.37 He is also credited with having laid the

foundations for the very field itself. The next award for research in surface science only

came in 2007, after a period of 75 years, with the recognition going to Gerhard Ertl “for

his studies of chemical processes on solid surfaces”.37 Major advances in vacuum science

and semiconductor technology led to new experimental techniques in the 1950s and

1960s; Ertl took advantage of many of these techniques (primarily LEED) and even

developed many methods to systematically and thoroughly study important reactions

such as the adsorption of hydrogen on metal surfaces,38-41 the production of NH3 from N2

and H2, and the oxidation of CO on platinum,37 (and references therein) to name a few.

His theoretical and experimental investigations have greatly contributed to our

understanding of catalytic reactions on surfaces.37 Gabor Somorjai’s research group at

UC Berkeley has also contributed significantly to the field. A sampling of some research

‘firsts’ from his group include the first observation of a surface reconstruction,42 the first

observation of a catalytic reaction on a single crystal surface at atmospheric pressure,43

the first determination of an absolute turnover rate in heterogeneous catalysis etc. 44 Over

the years, they have also developed a number of surface-science techniques 45-50 which

are widely used in many groups around the world today.

26

It must be mentioned here that the majority of this research, albeit important, has been

mainly around the solid-gas interface while inroads into understanding catalysis in the

condensed phase has remained quite intractable as only a few techniques (SEIRAS and

XAS being notable exceptions) can provide sufficient element specific, molecular level

information to elucidate a surface reaction mechanism. However, the increased

availability of resources at advanced synchrotron sources, and thus access to extremely

high intensity photon beams, has greatly facilitated in situ catalysis experiments and

research in the last decade has yielded unprecedented insight into the nature of catalysis

during operation. Many groups, including our own, are carrying out studies on the

middle-ground between surface science and heterogeneous catalysis, working to

completely bridge the two disciplines together. It is hoped that these studies will

eventually provide an understanding of catalysis like never before; an understanding that

will prove indispensible to solving our energy and environmental concerns.

The work described in this dissertation is an attempt to make a small addition to the

body of information that is steadily being accumulated in this fascinating area of

chemistry viz., in situ and in operando spectroscopic studies in electrocatalysis. We use

in situ x-ray absorption spectroscopy (XAS) to probe electrocatalysts under conditions

typically encountered during catalysis. We shall see in the next (and subsequent chapters)

how XAS is uniquely positioned as a technique to provide molecular-level insights into

the nature of catalytic activity, the experimental and theoretical methods used to collect

and analyze XAS data, and describe some of our findings on the nature of poisoning,

degradation and aging of electrocatalysts used in state-of-the-art fuel cells today.

27

1.3 Electrocatalyst Degradation

It was stated earlier that a catalyst is defined as a substance that accelerates a

chemical reaction without itself being consumed. This statement is not entirely true as it

is well-known that catalysts do not remain unaffected. They have definite lifetimes which

may vary from minutes to over a few years.31, 35 Many processes are designed so as to be

able to reuse some or all of the catalyst used through a separate ‘regeneration’ step. Given

that all catalytic activity lies in a catalysts’ ability to aid in the breaking and forming of

strong bonds between reactants, it is not surprising that over extended periods of time,

they are susceptible to some form of degradation or deterioration on an atomic scale. For

instance, at high temperatures, platinum particles tend to grow, decreasing the effective

surface area and hence reducing the mass specific activity (activity normalized to mass)

of the catalyst.33 Surfaces studied using techniques like Low Energy Electron Diffraction

(LEED), High Resolution Electron Energy Loss Spectroscopy (HREELS), Scanning

Tunneling Microscopy (STM) and Transmission Electron Microscopy (TEM) reveal that

atoms near the surface of metals undergo some form of rearrangement on exposure to

certain adsorbates, and also suggest that this process is reversible.35 This phenomenon is

known as ‘reconstruction’. This could be an additional factor in the aging or roughening

of metal particles, especially in solid-gas catalytic systems. Several studies of the aging

of platinum and other metals in gas phase catalysis exist in the literature.51-55 Our

interests lie chiefly in electrocatalysis, i.e. catalysis at the solid-liquid interface under the

application of an electric field or external potential. We will thus focus on aging

phenomena insofar as it applies to electrocatalysis in fuel cell electrodes.

28

Platinum and its alloys (commonly dispersed on high-surface area carbon supports)

are widely used as electrocatalysts in fuel cells. Substantial reductions in the loading used

for fuel cell electrodes have been achieved through the careful optimization of the surface

properties and microstructure of the electrodes. These electrodes have comparable

performance to unsupported catalysts of much higher Pt loading but nevertheless suffer a

loss in activity over time. The extent of loss is dependent on a number of factors

including electrolyte, operating conditions or temperature, humidity, poisons and

impurities in fuel and oxidant, potential and current conditions, and even whether it

operates continuously or intermittently. Specifically, loss in the electrochemically active

surface area (ECSA) has been observed under both, steady state and potential cycling

conditions and has been reported in a number of studies.56-59 Most of the loss in activity

(or ECSA) has been attributed chiefly to Pt metal dissolution, especially at higher

potentials (V > 0.80 V vs. RHE).60-63 This dissolved Pt (or alloyed metal, M) can then

redeposit on other particles of the same metal, or deposit on other metal particles

eventually increasing the average particle size distribution on the electrode/catalyst layer;

it can also diffuse into the membrane, forming crystallites on being reduced with the

traveling protons, or even traveling all the way through and depositing onto the cathode.

Many of these causes of electrocatalyst degradation will be taken up in greater detail in a

separate section. Chapters 4 and 5 in this thesis deal exclusively with these issues as

investigated using commercial PtRu catalysts and lend further confirmatory evidence to

studies in the literature.

While we will be focusing on the nature and causes of electrocatalyst degradation, it

is nevertheless instructive to be acquainted with all possible forms of degradation and

29

Table 1.3 Failure modes and their possible reasons for various components of a fuel cell stack. Table adapted from reference 64.

Corrosion; mechanical stressMechanical failureGaskets

Mechanical stressFracture/deformationBipolar Plates

Degradation of backing material; mechanical stress

Corrosion

Change in hydrophobicity of materials

Decrease in rate of mass transport of reactants

Conductivity loss

Water management control

Gas Diffusion Layer

Aging and sintering; Corrosion

Mechanical stress; change in hydrophobicity of materials

Activation loss

Decrease in mass transport rate and water management control

Active Layer

Non-uniform distribution of reactants, membrane drying

Thermal or Mechanical stressUneven ‘pinch’

Chemical attack

Conductivity loss; delamination

Membrane

Possible CausesFailure ModeComponent

Corrosion; mechanical stressMechanical failureGaskets

Mechanical stressFracture/deformationBipolar Plates

Degradation of backing material; mechanical stress

Corrosion

Change in hydrophobicity of materials

Decrease in rate of mass transport of reactants

Conductivity loss

Water management control

Gas Diffusion Layer

Aging and sintering; Corrosion

Mechanical stress; change in hydrophobicity of materials

Activation loss

Decrease in mass transport rate and water management control

Active Layer

Non-uniform distribution of reactants, membrane drying

Thermal or Mechanical stressUneven ‘pinch’

Chemical attack

Conductivity loss; delamination

Membrane

Possible CausesFailure ModeComponent

30

reasons for failure leading to limited lifetimes in fuel cell systems. Some of the most

important causes are listed in Table 1.3.

1.3.1 Particle dissolution and growth

1.3.1.1 The thermodynamics of dissolution.

A Pourbaix diagram indicates the thermodynamically stable species (at 25 ºC) at

equilibrium, of a given element in an aqueous environment under various conditions of

pH and electrode potential as determined by the well-known Nernst equation65, 66, E = Eº

- RT/nF ln [H+], where Eº is the cell potential under standard conditions, and the other

symbols have their usual meanings. Thus, in order to determine the tendency for a metal

to go into solution (or remain as is), one need only look up the Pourbaix atlas for the

element concerned. Several studies on the solubility of Pt under equilibrium conditions in

aqueous electrolyte exist in the literature.56, 67-72

A brief survey of studies on Pt dissolution under various conditions of pH,

temperature, electrolyte and potential leads to the following key findings –

1. The solubility of Pt increases with pH, suggesting a dissolution mechanism that is

basic in nature.

2. The metal solubility increases with temperature, also indicating that it is an

endothermic process.

3. The solubility increases with electrode potential up until around 1.10 V (vs. RHE)

after which it levels off or decreases, consistent with the formation of surface

oxides which leads to passivisation of the surface, thereby slowing down the

dissolution.

31

The equations governing this oxide formation on platinum are shown below.65, 66

Pt + H2O PtO + 2H+ + 2e- E0 = 0.980 – 0.0591 pH

PtO + H2O PtO2 + 2H+ + 2e- E0 = 1.045 – 0.0591 pH

Pt Pt2+ + 2e- E0 = 1.188 – 0.0295 Log [Pt2+]

All of these dissolution studies reveal that there is no consensus on the nature of

dissolved Pt species under equilibrium conditions. The Pourbaix diagrams suggest a Pt2+

species, which would likely be an aquo-complex of Pt2+. Azaroul et al. have suggested

that a PtOH+ exists in mildly acidic and basic aqueous conditions 68, 73 while Kim et al.,

using the ‘dithizone-benzene’ method, suggest that a Pt4+ species is produced from the

dissolution process in sulfuric acid.71 However, all these studies have been performed

under equilibrium conditions. Other studies under ‘non-equilibrium’ conditions, using

various methods such as potential cycling,74-77 constant current74, 78 and potential67, 74

methods and even square-wave potential steps76, 79-81 also reveal either a Pt2+ or a Pt4+

species in solution. There is general agreement that rapid dissolution of Pt is found

around 1.0 V (vs. RHE) and that the dissolution rate is of the order of 10-9 – 10-11 g cm-2

s-1 82. However we note that nearly all the aforementioned studies were carried out on

polycrystalline or bulk Pt metal in the form of disks, sheets, foil or even rods. The rates of

dissolution can be expected to be much greater in nanoparticle catalysts. The latest

studies on electrocatalyst degradation in PEMFCs and DMFCs are discussed in the

following section.

32

1.3.1.2 The mechanism for degradation.

Over the last 4-5 years, significant interest has been generated within the fuel cell

community towards lifetime studies and fundamental aspects of electrocatalyst

degradation. There is a growing body of information that strongly suggests that Pt

dissolution (or alloyed metal dissolution) is one of the primary causes of degradation of

PEM fuel cells and is largely responsible for their limited lifetime. However, a

mechanistic understanding of the dissolution process still eludes the community, or more

precisely, more fundamental studies are called for in this area before any general

conclusions can be drawn and solutions proposed. A survey of the current body of

literature in this area follows.

One of the earliest studies calling attention to the aging and degradation in low-

temperature fuel cell electrodes was a paper by Wendt et al. in 1996. While the paper

chiefly addressed optimization and modeling of catalyst utilization, a section on the

“ageing and ageing prevention in low-temperature fuel cells” already recognized that

platinum dissolution and agglomeration, as well as oxidative loss of anodic supports

would be a primary means of catalyst degradation. Citing a number of previous studies

on binary and ternary Pt-M alloys, they suggested that alloying of platinum

electrocatalysts would be an efficient way to mitigate the effects of platinum

dissolution.83 However, this approach has now been widely discussed in the literature

(see Section 1.3.1.3). An in situ XAS study by Hwang and Strehblow showed clearly that

platinum is coordinated to more oxygen atoms at higher potentials, in good agreement

with anodic oxidation currents seen in cyclic voltammetric experiments. They also

concluded that platinum crystallization must have occurred as the particle size,

33

determined from NPt-Pt values in EXAFS analysis, showed a clear increase. However, this

conclusion may be in error as it would be highly unlikely that the average particle size

nearly doubles after just a single voltammetric cycle.84 We have carried out similar

experiments on two commercial PtRu catalysts (see chapter 4) and find that the particle

size does increase, but only after 20-40 cycles and not any sooner.

In 2004, a study by Wilkinson’s group at Ballard Power Systems reported the effect

of various operating conditions on aging and degradation in PEMFCs and DMFCs. Some

conditions of operation included varying reactant flow, conditions of temperature and

humidity. They found that the lifetime of the fuel cells may vary from 1000 hours to as

much as 13,000 hours depending on just operating conditions, a very significant finding

85. Various studies on the durability of different kinds of fuel cells carried out before 2003

were summarized and presented in a book chapter entitled, ‘Durability’ in the Handbook

of Fuel Cell Technology and Applications. 86

Key evidence for the existence of ruthenium dissolution losses from a PtRu anode

catalyst was first provided by Piela et al. 87 Darling and Meyers provided the first detailed

theoretical model of the aging process as modeled using dissolution and re-deposition

processes. They found that platinum oxidation has a significant effect on the stability in a

PEMFC environment and also found that start-up and shut-down events accelerate the

metal loss from electrodes. 60 Ferreira et al. carried out studies on a Pt/C catalyst in 0.5M

H2SO4 at 80 ºC and found that Pt dissolution increased with potential from 0.8-1.1 V (vs.

RHE) and that loss in ECSA was much higher at OCP (ca.0.95 V) than under potential

control. They also found that coarsening of the catalyst surface and new Pt crystallites in

the membrane occur through Ostwald ripening and H+ ion encounters in the membrane,

34

respectively.56, 62, 63 Our study on the spontaneous deposition of Ru onto Pt under DMFC

operating conditions (see chapter 5) corroborate their findings, as it was found that

ruthenium ions deposit onto a Pt surface much more readily at OCP than under potential

control, thereby accelerating the degradation process on cathodes. These and other

commonly observed degradation phenomena leading to deactivation in supported metal

catalysts are shown in Figure 1.4.

The stability of low-index and nanofaceted single crystal Pt surfaces has been studied

using cyclic voltammetry (CV), atomic force microscopy (AFM) and inductively-coupled

plasma mass spectrometry (ICP-MS) by Komanicky, Markovic, Myers and others. In

agreement with previous studies mentioned above, it was found that an oxide layer

passivates the surface at higher potentials, lowering dissolution whereas the nanofaceted

surfaces underwent increased dissolution at high potentials. Among the low-index single

crystal faces studied, increased dissolution was found at both 0.65 V and 1.15 V when

compared with 0.95 V for both the Pt(111) and Pt(110) surface. However, while the

dissolution is lowest at 0.95 V for the Pt(111) surface, the rate of dissolution on the

Pt(100) surface decreases uniformly with increasing potential. The authors attribute this

to increased passivisation due to the larger affinity for oxygen on the Pt(100) face over

the Pt(111) surface.67

Place-exchange is the phenomenon by which at high potentials, oxygen goes

subsurface i.e. beneath the surface layer of platinum exposing a fresh layer of Pt atoms to

the surface. Not surprisingly, this is believed to accelerate the rate of Pt dissolution with

cycling as the fresh surface layer of Pt atoms are now susceptible to further oxidation and

dissolution.57

35

Figure 1.4 Various mechanisms of catalyst deactivation (loss of active surface area) seen in low-temperature fuel cells. Figure adapted from Shao-Horn et al.62

36

Yasuda et al. published a number of papers on the subject of Pt and Ru dissolution from

catalysts in PEM fuel cells. They presented compelling evidence for Pt dissolution on the

anode, diffusion through the membrane and deposition at the membrane-cathode

interface by reduction with dissolved hydrogen.88-90

In another study on anode catalysts, various commercially available PtRu catalyst of

varying Ru content were investigated for aging signs in 0.5 M H2SO4 through potential

cycling. They found that the rate of degradation increased with increasing the upper limit

of potential between which the material was cycled. Also, the different catalysts showed

different aging properties and they proposed that the aging differences were due to levels

of crystallinity and surface state but were unable to provide any conclusive evidence for

the same.91

In a similar study on PtRu black anode catalysts (see Chapter 4), we too found that

the aging properties on two different commercial PtRu catalysts were quite different and

using XAS, CO stripping data, Copper under-potential deposition (UPD) measurements

and CVs, were able to attribute the difference in aging mechanisms to the Ru island size

on the catalysts: they were quite large in one case (Tanaka TEC90110) while smaller and

more uniformly alloyed in the other (Johnson-Matthey HiSpec6000).92

Molecular dynamic modeling on Pt catalysts revealed that Pt nanocrystallites can

undergo additional dissolution through electric interaction between the crystallites and

the polarized polymer electrolyte under operating conditions. It was found that this would

lead to a local increase in temperature, increasing the rate of oxidative attack on the

membranes.93 The long term effects of exposure to heat, contact with methanol solution

and CO2 were also reported to cause degradation due to ‘delamination’, a process by

37

which the membrane electrode assembly comes apart when the Nafion ™ membrane

detaches from either of the electrodes, thereby losing contact with the catalyst layer and

causing serious, often irreversible damage.94 Note that this form of degradation is

completely different from many of the other forms of degradation discussed thus far.

Studies using x-ray photoelectron spectroscopy (XPS) and time-of-flight secondary

ion mass spectrometry (TOF-SIMS) showed that loss of Ru is more apparent and

damaging to anodes than Pt dissolution. The Ru species migrated to the cathode side

through the membrane and was found to be deposited chiefly as a RuOx species at the

interface between the cathode catalyst layer and the gas diffusion layer95. Temperature

effects on Pt/C catalysts have also been studied. The degradation observed revealed

straightforwardly that increased temperatures lead to decreased cell lifetimes. As with

previous studies, the degradation was manifest as loss of Pt ECSA and the deposition of

Pt in the membranes; the cathode too was affected, as indicated by the increased particle

size of the Pt layer in the cathode.96

Despite the possibility of exploiting the higher catalytic activity at elevated

temperatures, one problem is effectively traded for another as the effects of degradation

are accelerated as well. Thus, caution must be exercised before proposing apparently

simplistic solutions such as operation at higher temperatures in order to derive better

performance from fuel cells. Further, adequate research into membranes which can

perform well at high temperatures would be necessary before such a move can be used to

tackle the problem of catalyst degradation in PEMFCs.

38

1.3.1.3 Effects of metal alloying on degradation.

A number of alloyed electrocatalysts for both, anodes and cathodes, have

exhibited superior performance when compared to Pt/C 97-112 and efforts to enhance the

performance of oxidation (hydrogen, methanol and ethanol) and the ORR continues to be

an active research area. This enhancement in activity has been attributed to two principal

effects: a structural effect and an electronic or ligand effect. Alloying changes the Pt-Pt

bond distance and is known to affect both, the bond distances in the metals (and thus,

metal-adsorbate bonding geometry) 113-115 and naturally, the electronic properties of the

surface. While performance is no doubt important, note that most of the alloying

elements tend to be more oxophilic in nature (for instance, compared to platinum) and

thus, would be likely to undergo more severe dissolution if any of these atoms should be

near the surface of the catalysts. Yet, some studies have shown that, for instance, PtCo

catalysts are more stable than unalloyed platinum catalysts 116 while others, including

work described in this dissertation (chapter 4) have found evidence of significant

degradation in many alloy catalyst systems.117-121 Clearly, synthesis methods, size,

composition, internal-structure, surface morphology and operating conditions are all

expected to influence the performance and aging characteristics of alloy catalysts.

Understanding such aging phenomena will definitely prove more complicated than

studying single-component systems.

A review of the stability of alloy electrocatalysts has recently been published by

Antolini et al.122 They report that, in general, PtCr and PtCo alloys tend to be more stable

than PtV, PtNi or PtFe alloys but all the various factors just mentioned make a systematic

and rigorous comparison across studies by different research groups nearly impossible.

39

Finally, in situ studies, especially using element-specific techniques such as XAS, will

prove to be invaluable in studying aging phenomena in alloy catalysts. It is hoped that

detailed theoretical and in situ experimental investigations of catalyst aging processes on

various alloy catalyst systems of comparable size and synthesized using the same

procedure, will eventually lead to a more sound understanding of electrocatalyst

degradation.

1.3.2 Degradation of support

1.3.2.1 Dissolution of carbon.

Apart from the metal catalyst itself undergoing dissolution, the high surface area

carbon supports widely used in all PEMCs and DMFCs are also known to undergo

significant degradation. The oxidation of carbon proceeds according to the reaction - 123

C + 2H2O CO2 + 4H+ + 4e- E0 = 0.207 V (vs. RHE)

While the oxidation potential is low enough to be in the operating cell potential

window of many types of fuel cells, it is kinetically not favored, and therefore occurs

very slowly, especially at lower temperatures encountered in PEMFCs and DMFCs.

However, even minute amounts of support corrosion are sufficient to cause long term

degradation through eventual erosion of the carbon support, isolating the Pt islands

electronically from the electrode. This can lead to loss in ECSA either through loss of

contact with the surface or agglomeration as the small portions of the catalyst are no

longer highly dispersed on the support. An excellent survey of carbon corrosion in

PEMFCs as well as other types of fuel cells has been given by Borup et al.82 However, a

brief account of papers published since their review article (2007) was published is given

here.

40

Effects of carbon support and humidification of platinum electrocatalysts have

revealed that the rate of carbon loss (oxidative) increases rapidly with humidification.

Further, it has been found that graphitized carbon black is much more stable than un-

graphitized supports suggesting that carbon surface chemistry is also a critical factor in

the long-term stability of supported metal catalysts.124 Degradation of carbon supports in

PEMFC cathodes as studied with in situ scanning tunneling microscopy (STM) revealed

surface oxide formation that started on steps/edge sites and progressed onto terrace sites.

The corrosion was aggravated in the presence of the Pt catalyst as well as oxygen in the

environment. 125 A study by Virkar et al. (see Figure 1.5) showed unequivocally that

carbon supports play an instrumental role in their own degradation by serving as a

channel for electronic transport between the dissolving Pt nanoparticles, effectively

creating local electrochemical cells on the anode itself (much like concentration cells

found in many cases of corrosion). A likely mechanism which accounts for both,

corrosion of carbon support as well as the Ostwald ripening of supported Pt particles was

given. The study also showed that a small concentration of Ptn+ (n = +2 or +4) ions may

have a significant effect on the roughening of the electrode surface by promoting the

dissolution/re-deposition of Pt by Ostwald ripening. They conclude by suggesting that

operation at elevated temperatures (above 100 ºC) could mitigate the problem. 61 That Pt

and carbon when present together and in close contact cause more dissolution as well as

carbon corrosion has also been observed by other groups.126-129

Carbon support oxidation on the cathode is likely to be more severe than on the anode

due to the increased availability of molecular O2 in the oxidant. This was confirmed in a

study on the structural degradation of PEMFC cathodes by Young et al.130 The catalyst

41

layer was subjected to 30 hour stress tests wherein the potential was cycled from 0.1 to

1.5 V (vs. RHE). The results clearly showed that the carbon sub-layer in the GDL and

cathode catalyst layer were thinned-down due to oxidative corrosion, effectively reducing

the Pt surface area on the cathode catalysts. This in-turn affected the ohmic resistance,

oxygen mass transport and the water-management properties of the cathode. Aging

studies of Pt/glassy carbon electrodes investigated through potential cycling also showed

a loss in the ECSA, as evidenced by a shift in the CO oxidation peak potential. They also

provided evidence through AFM images for Pt dissolution, coalescence and migration

and also suggested that carbon oxidation, apart from the Pt dissolution and redeposition,

could be contributing to the roughness of the surface during aging. 131

1.3.1.2 Alternatives to carbon support.

It is quite unlikely that more noble and electrocatalytically active elements for fuel

cell technology will be discovered as gold and platinum are arguably the most noble of

all known metals. Fairly large reserves of the platinum group metals exist when

compared to say, the rare earths, which, should they turn out to possess novel catalytic

properties, would still be a futile developmental effort. However, alternate supports

which are more stable than carbon offer some hope for addressing the degradation

problem. In the hope of extending the operating lifetimes of catalysts in fuel cells, various

forms of carbon, treated supports and even modified carbon supports have been

researched in order to reduce the extent of carbon corrosion just described. More novel

alternatives are also being pursued in order to avoid using carbon altogether. Some

materials explored to date include nanotubes of all kinds 132-134 (for two recent reviews on

use of nanotubes in catalysis and as fuel cell supports, see references 135 and 136), oxide

42

Figure 1.5 Mechanism of degradation in a carbon-supported platinum catalyst showing both, Ostwald ripening and role of carbon in serving as a channel for electronic transport. Figure adapted from a study by Virkar et al.61

43

materials,137 silicon,138 conducting polymers,139-141 nonconductive whisker-like

materials,142, 143 and even conductive diamond.144, 145 This area of research was briefly

mentioned here as it is directly relevant to the overall theme of the dissertation, viz. the

understanding of fundamental aspects of aging and degradation in fuel cells, especially

low-temperature fuel cells such as PEMFCs and DMFCs because very often, the catalysts

are supported in order to obtain better dispersion leading to higher surface areas.

However, a detailed treatment of the various avenues being explored to mitigate

degradation in fuel cells is out of scope of this dissertation. The interested reader is

referred to an excellent, comprehensive review article which covers many of the topics on

degradation in fuel cells discussed in this chapter.82

1.4 Organization of the Dissertation

In chapter 2 of this thesis we present a summary of the x-ray absorption technique,

the primary method utilized in this work to provide new insights into the aging and

poisoning of electrocatalysts. The principles of x-ray absorption spectroscopy are

presented along with details of EXAFS analysis as well as analysis of XANES data using

the ∆µ-XANES method. The experimental details for the in situ XAS experiments and

aspects of design and development of in situ spectroelectrochemical cells used in all of

our studies are discussed at length. Following these sections, a brief literature survey of

the development of XAS as a mature technique and a powerful in situ spectroscopy is

also presented.

Chapters 3, 4 and 5 describe studies on the two principal forms of electrocatalyst

degradation viz. loss of ECSA by a) poisoning of active catalytic sites and b) through

44

morphological changes occurring in the electrocatalysts. An in situ XAS study on the

poisoning of Pt/C catalysts by chloride ions was studied and reported in chapter 3. RDE

experiments show unequivocally that adsorbed chloride drastically affects the catalysis

by blocking active surface sites and increases the overpotential for the oxygen reduction

reaction (ORR) by approximately 85 mV for every 10-fold increase in chloride

concentration. Through the use of the Δμ-XANES method, direct spectroscopic evidence

was obtained for the site-specific adsorption of Cl- ions on 3-fold sites on the (111)

planes of Pt nanoparticles. A re-interpretation of previous studies in the literature

alongside our findings reveal interesting details on the complex interplay among

adsorption of commonly encountered ions such as OH-, Cl-, O2- and HSO4

-.

The aging properties of two commercial PtRu catalysts as observed through

voltammetric cycling and chronoamperometry is described in Chapter 4. From both, the

∆μ-XANES method and EXAFS analysis of the catalysts at both the Pt L3 and Ru K

edges, we found that the morphology of the alloy catalysts viz. the difference in the

RuOxHy island size on the two catalysts, could account for the difference in activity and

aging behavior of the catalysts. Characteristic of an Ostwald ripening process, smaller Ru

islands on the surface, which were largely metallic in nature were found to undergo

significant dissolution and redeposition to form larger Ru islands. Larger Ru islands (on

the other commercial catalyst) were heavily oxidized to begin with and were much more

stable. In summary, a mechanistic model for the likely aging mechanisms for the two

catalysts is also proposed. The model accounts for the changes in voltammetric features

of the two commercial catalysts that are observed on continual cycling for up to 500

45

cycles as well as explains the CO stripping data collected on the two catalysts before and

after chronoamperometric aging at 0.5 V for 8 hours.

Chapter 5 builds on the previous chapter and contains studies on the adverse effects

on the ORR due to poisoning of the cathode by Run+ ions that are formed as a result of

metal dissolution of PtRu anode catalysts (such as studied in chapter 4). The ∆μ-XANES

analysis for Ru deposition on Pt/C electrocatalysts involving comparison between

experimental results and theoretical full-multiple scattering calculations using the FEFF

8.0 code revealed that the poisoning was primarily of the site-blocking type wherein the

ions adsorb on the platinum surface in 3-fold hollow sites, blocking active sites for the

catalytic activity of the supported Pt catalysts. The electrochemistry, electron-spin

resonance (ESR) and EXAFS results in our studies collectively indicate that these species

travel through the polymer membrane and deposit onto Pt/C cathodes, affecting the

membrane as well as decreasing the ORR activity. ESR results show that Ru ions, when

present in the membrane, alter the hydration levels and transport properties of the

membrane. Interestingly, the deposition of Ru on Pt was found to be most severe at open

circuit potential (ca. 0.95 V vs. RHE). These and other findings are discussed in detail in

this chapter.

Chapter 6 describes a fundamental investigation into the interaction between a widely

used stabilizing agent and supported platinum electrocatalysts. Nanoparticles are

stabilized by the addition of certain functional polymers that prevent their agglomeration

or coalescence, effectively increasing the lifetime of the nanoparticle catalysts by

preserving their particle size distribution. Polyvinyl Pyrrolidone (PVP) is one such

polymer widely used in this capacity as a capping agent to not only prevents the

46

agglomeration of the metal nanoparticles, but to also aid in the synthesis of specifically-

shaped nanoparticles. In this chapter, the interaction between PVP and platinum is

directly probed in 0.1M HClO4 using the ∆μ-XANES method. In agreement with

previous studies in the literature, we found that the PVP binds to the platinum chiefly via

the free carbonyl groups of the pyrrolidone rings of the polymer. The binding site-

specificity of the method also allowed us to determine that the PVP binds to the platinum

surface at the atop sites. Finally, results on the effect of residual PVP on the catalytic

activity of platinum towards the oxidation of methanol and formic acid are presented and

discussed.

47

1.5 References

1. Steam Engine, History of. The World Book Encyclopedia 18 (So-Sz), 883 (1989).

2. History of the Battery. The World Book Encyclopedia 2 (B) (1989).

3. Grove, W. R. On voltaic series and the combination of gases by platinum. Phil. Mag. Ser. 2, 127-130 (1839).

4. Development of the Gasoline Engine. The World Book Encyclopedia 8 (G) (1989).

5. Important dates in automobile history. The World Book Encyclopedia 1 (A) (1989).

6. Grosser, M. Diesel: The Man and the Engine. Atheneum Press (1978).

7. Ludwig Mond and C. Langer, A new form of gas battery, Proceedings of the Royal Society of London 46 (1889).

8. E. Baur, J. T. Z. Elektrochem. Angew. Phys. Chem. 39 (1933).

9. F.T. Bacon, T. M. F. Review Lecture: The Development and Practical Application of Fuel Cells. Proceedings of the Royal Society of London, Series A, Mathematical and Physical Sciences 334, 431 (1973).

10. W.T. Grubb, L. W. N., Batteries with Solid Ion-Exchange Membrane Electrolytes, Journal of the Electrochemical Society 107, 131 (1960).

11. Grubb, W. T. U.S. Pat. 2, 913, 511 (1959).

12. Kamaruzzaman Sopian, W. R. W. D. Challenges and future developments in proton exchange membrane fuel cells. Renewable Energy 31, 719-727 (2006).

13. George Apanel, E. J. Direct methanol fuel cells - ready to go commercial? Fuel Cells Bulletin, 12-17 (2004).

14. Buchmann, I. The Fuel Cell: is it ready? Batteries in a Portable World (2001).

15. Perry, M. L. & Fuller, T. F. A Historical Perspective of Fuel Cell Technology in the 20th Century. Journal of The Electrochemical Society 149, S59-S67 (2002).

16. M.P. Hogarth, T. R. R. Catalysis for Low Temperature Fuel Cells. Platinum Metals Review 46, 146-164 (2002).

48

17. Zaidi, S. M. J. Technology of Direct Methanol Fuel Cell - Progress and Future Prospects. 6th Saudi Engineering Conference Proceedings, KFUPM, Dhahran 2, 215-230 (2002).

18. R. Parson, V. d. N. Methanol oxidation on platinum electrodes. Journal of Electroanalytical Chemistry 257, 9-13 (1988).

19. A. K. Shukla, R. K. R, Methanol-Resistant Oxygen-Reduction Catalysts for Direct Methanol Fuel Cells. Annual Reviews in Materials Research 33, 155 (2003).

20. A. Anderson, E. G., S. Seong Systematic Theoretical Study of Alloys of Platinum for Enhanced Methanol Fuel Cell Performance. Journal of The Electrochemical Society 143, 2075-2082 (1996).

21. J. O'M. Bockris, H. W., Electrocatalysis, Journal of Electroanalytical Chemistry 7 (6), 428 (1964).

22. M. Watanabe, S. M., Electrocatalysis by ad-atoms : Part III. Enhancement of the oxidation of carbon monoxide on platinum by ruthenium ad-atoms, Journal of Electroanalytical Chemistry 60 (3), 275 (1975).

23. S. Motoo, M. W., Electrocatalysis by Sn and Ge ad-atoms. Journal of Electroanalytical Chemistry 69, 429 (1976).

24. M.M.P. Janssen, P. M., Platinum—tin catalysts for methanol fuel cells prepared by a novel immersion technique, by electrocodeposition and by alloying. Electrochimica Acta 21, 861-869 (1976).

25. S. Gottesfeld, P. Z. Advances in Materials for DMFC systems. 122-124 (1999).

26. John Cruikshank, K. S. The degree and effect of methanol crossover in the direct methanol fuel cell. Journal of Power Sources 70, 40-42 (1998).

27. T.H. Kin, W. Y. S., C.C. Yang, George Yu. Estimating the methanol crossover rate of PEM and the efficiency of DMFC via current transient analysis. Journal of Power Sources 161, 1183-1186 (2006).

28. Keith Scott, A. K. S. Direct Methanol Fuel Cells: Fundamentals, Problems and Perspectives. Modern Aspects of Electrochemistry 40, Chapter 4, 127 (2005).

29. Roberts, M. W. Birth of the catalytic concept (1800-1900). Catalysis Letters 67, 1-4 (2000).

30. Jr., R. L. B. Heterogeneous catalysis before 1934. Heterogeneous Catalysis (ACS Symposium Series) 1, 3-12 (1983).

49

31. Sleight, A. W. Heterogeneous Catalysts. Science 208, 895-900 (1980).

32. Editors: Avelino Corma, F. V. M., S. Mendioroz, J.L.G. Fierro. Proceedings of the 12th International Congress on Catalysis Studies in Surface Science and Catalysis, 83 (2000).

33. Robinson, A. L. Heterogeneous Catalysis: Can Surface Science Contribute? Science 185, 772-774 (1974).

34. White, J. M. Surface Science of Heterogeneous Reactions. Science 218, 429-433 (1982).

35. Yin, Y. et al. Formation of Hollow Nanocrystals Through the Nanoscale Kirkendall Effect. Science 304, 711-714 (2004).

36. Maugh, T. H. Catalysis: "No Longer a Black Art". Science 219, 474-477 (1983).

37. Scientific Background on the Nobel Prize in Chemistry 2007. Kungl. Ventenskapsakademien, The Royal Swedish Academy of Sciences (2007).

38. H. Conrad, G. E., E.E. Latta. Adsorption of hydrogen on palladium single crystal surfaces. Surface Science 41, 435-446 (1974).

39. K. Christmann, G. E., T. Pignet. Adsorption of hydrogen on a Pt(111) surface Surface Science 54, 365-392 (1976).

40. Christmann, K., Schober, O., Ertl, G. & Neumann, M. Adsorption of hydrogen on nickel single crystal surfaces. The Journal of Chemical Physics 60, 4528-4540 (1974).

41. Christmann, K., Behm, R. J., Ertl, G., Hove, M. A. V. & Weinberg, W. H. Chemisorption geometry of hydrogen on Ni(111): Order and disorder. The Journal of Chemical Physics 70, 4168-4184 (1979).

42. S. Hagstrom, H. B. L., G.A. Somorjai. Surface Structures on the Clean Pt(100) Surface. Physical Review Letters 15, 491-493 (1965).

43. D.R. Kahn, E. E. P., G.A. Somorjai The Hydrogenolysis of Cyclopropane on a Platinum Stepped Single Crystal at Atmospheric Pressure. Journal of Catalysis 34, 294-306 (1974).

44. P.S. Cremer, X. S., Y.R. Shen, G.A. Somorjai. The First Measurement of an Absolute Surface Concentration of Reaction Intermediates in Ethylene. Catalysis Letters 40, 143-145 (1996).

50

45. D.W. Blakely, E. I. K., B.A. Sexton, G.A. Somorjai. New Instrumentation and Techniques to Monitor Chemical Surface Reactions on Single Crystals over a Wide Pressure Range (10-8-105 Torr) in the Same Apparatus. Journal of Vacuum Science and Technology 13, 1091 (1976).

46. D.E. Gardin, G. A. S. An Ultrahigh Vacuum Chamber Equipped with a Liquid Phase Reaction Cell. Reveiw of Scientific Instruments 64, 1304 (1993).

47. J.A. Jensen, K. B. R., Y. Chen, M. Salmeron, G.A. Somorjai. High Pressure, High Temperature Scanning Tunneling Microscopy. Journal of Vacuum Science 17, 1080 (1999).

48. L.A. West, E. I. K., G.A. Somorjai. Molecular Beam Scattering from Single Crystal Surfaces under Ultrahigh Vacuum Conditions. Journal of Vacuum Science and Technology 8, 430-436 (1971).

49. S.T. Ceyer, W. J. S., G.A. Somorjai. Design of a Molecular Beam Surface Scattering Apparatus for Velocity and Angular Distribution Measurements, Journal of Vacuum Science and Technology 19, 726 (1981).

50. P.S. Cremer, B. J. M., M. Salmeron, Y.R. Shen, G.A. Somorjai. Monitoring Surfaces on the Molecular Level during Catalytic Reactions at High Pressure by Sum Frequency Generation Vibrational Spectroscopy and Scanning Tunneling Microscopy. Catalysis Letters 34, 11-18 (1995).

51. Mariella Moldovan, S. R., Gregory M. Morrison, Milagros Gomez, M. Antonia Palachios. Impact of ageing on the distribution of platinum group elements and catalyst poisoning elements in automobile catalysts. Surface and Interface Analysis 35, 354-359 (2003).

52. C. G. Myers, W. H. L., P. B. Weisz. Aging of Platinum Reforming Catalysts. Industrial & Engineering Chemistry Research 53, 299-302 (1961).

53. J.H. Vleeming, B. F. M. K., G.B. Marin. Effect of platinum particle size and catalyst support on the platinum catalyzed selective oxidation of carbohydrates. Catalysis Letters 46, 187-194 (1997).

54. Jonas Andersson, M. A., Lisa Eurenius, Eva Olsson, Magnus Skoglundh. Deactivation of diesel oxidation catalysts: Vehicle- and synthetic aging correlations. A report by the Volvo Corporation and Chalmers University of Technology, Sweden (2006).

55. J.H. Vleeming; G.B. Marin, F. O., P. Courtine, B.F.M. Kuster. Graphite-supported platinum catalysts - Effects of gas and aqueous phase treatments. Journal of Catalysis 166, 148-159 (1997).

51

56. Ferreira, P. J. et al. Instability of Pt/C Electrocatalysts in Proton Exchange Membrane Fuel Cells. Journal of The Electrochemical Society 152, A2256-A2271 (2005).

57. Merzougui, B. & Swathirajan, S. Rotating Disk Electrode Investigations of Fuel Cell Catalyst Degradation Due to Potential Cycling in Acid Electrolyte. Journal of The Electrochemical Society 153, A2220-A2226 (2006).

58. M. Fowler, R. F. M., J.C. Amphlett, B.A. Peppley, P.R. Roberge (Eds: W. Vielstich, H.A. Gasteiger, A. Lamm). Handbook of Fuel Cells: Fundamentals, Technology and Applications. 33, 663 (2003).

59. Antolini, E., Formation, microstructural characteristics and stability of carbon supported platinum catalysts for low temperature fuel cells. Journal of Materials Science 38, 2995 (2003).

60. Darling, R. M. & Meyers, J. P. Mathematical Model of Platinum Movement in PEM Fuel Cells. Journal of The Electrochemical Society 152, A242-A247 (2005).

61. Virkar, A. V. & Zhou, Y. Mechanism of Catalyst Degradation in Proton Exchange Membrane Fuel Cells. Journal of The Electrochemical Society 154, B540-B547 (2007).

62. Y. Shao-Horn, W. C. S., S. Chen, P.J. Ferreira, E.F. Holby, D. Morgan. Instability of Supported Platinum Nanoparticles in Low-Temperature Fuel Cells. Topics in Catalysis 46, 285-305 (2007).

63. Ferreira, P. J. & Shao-Horn, Y. Formation Mechanism of Pt Single-Crystal Nanoparticles in Proton Exchange Membrane Fuel Cells. Electrochemical and Solid-State Letters 10, B60-B63 (2007).

64. C.M. Rangel, R. A. S., T.I. Paiva. Materials aging mechanisms and performance impact in PEM fuel cells. Journal of New Materials for Electrochemical Systems 12, 119-122 (2009).

65. Pourbaix, M. Atlas of Electrochemical Equilibria in Aqueous Solutions (Pergamon Press, London, 1966).

66. M.J.N. Pourbaix, J. V. M., N. de Zoubov. Electrochemical Properties of the Platinum Metals. Platinum Metals Review 47-53 (1959).

67. Komanicky, V. et al. Stability and Dissolution of Platinum Surfaces in Perchloric Acid. Journal of the Electrochemical Society 153, B446-B451 (2006).

52

68. M. Azaroul, B. R., P. Freyssinet, J.R. Disnar. Solubility of platinum in aqueous solutions at 25°C and pHs 4 to 10 under oxidizing conditions. Geochimica et Cosmochimica Acta 65, 4453 (2001).

69. F.W. Bowles, A. P. G., D.J. Vaughan, S.J. Norris. Chronique de la Recherche Miniere 529, 65 (1995).

70. Bindra, P., Clouser, S. J. & Yeager, E. Platinum Dissolution in Concentrated Phosphoric Acid. Journal of The Electrochemical Society 126, 1631-1632 (1979).

71. J.H. Kim, A. I., S. Mitsushima, N. Kamiya, K. Ota. Catalytic activity of titanium oxide for oxygen reduction reaction as a non-platinum catalyst for PEFC. Electrochimica Acta 52, 2492 (2007).

72. Wood, S. A. Experimental determination of the hydrolysis constants of Pt2+ and Pd2+ at 25°C from the solubility of Pt and Pd in aqueous hydroxide solutions. Geochimica et Cosmochimica Acta 55, 1759 (1991).

73. M. Azaroul, B. R., P. Freyssinet, J.R. Disnar. Response to the comment by R. H. Byrne on “Solubility of platinum in aqueous solutions at 25°c and pHs 4 to 10 under oxidizing conditions” (2001) Geochim. Cosmochim. Acta 65, 4453–4466. Geochimica et Cosmochimica Acta 67, 2511 (2003).

74. G. Benke, W. G. The Electrochemical Dissolution of Platinum. Hydrometallurgy 64, 205 (2002).

75. D.C. Johnson, D. T. N., S. A ring-disk electrode study of the current/potential behaviour of platinum in 1.0 M sulphuric and 0.1 M perchloric acids Bruckenstein. Electrochimica Acta 15, 1493 (1970).

76. K. Kinoshita, J. T. L., P. Stonehart, Potential Cycling Effects on Platinum Electrocatalyst Surfaces. Journal of Electroanalytical Chemistry 48, 157 (1973).

77. D.A.J. Rand, R. W. A study of the dissolution of platinum, palladium, rhodium and gold electrodes in 1 M sulphuric acid by cyclic voltammetry. Journal of Electroanalytical Chemistry 35, 209 (1972).

78. K. -I. Ota, S. N., N. Kamiya. Dissolution of Platinum Anodes in Sulfuric Acid Solution. Journal of Electroanalytical Chemistry 257, 205 (1988).

79. A.J. Arvia, J. C. C., E. Custidiano, C.L. Perdriel, W.E. Triaca. Electrochemical Faceting of Metal Electrodes. Electrochimica Acta 31, 1359 (1986).

80. J. C. Canullo, W. E. T., A.J. Arvia, Electrochemical faceting of single crystal platinum electrodes. Journal of Electroanalytical Chemistry 200, 397 (1986).

53

81. W.A. Egli, A. V., W.E. Triaca, A.J. Arvia. The development of facetting and roughening at platinum polyfacetted single-crystal electrodes in a chloroplatinic acid solution. Applied Surface Science 68, 583 (1993).

82. Rod Borup, J. M., Bryan Pivovar, Yu Seung Kim, Rangachary Mukundan, Nancy Garland, Deborah Meyers, Mahlon Wilson, Fernando Garzon, David Wood, Piotr Zelenay, Karren More, Ken Stroh, Tom Zawodzinski, James Boncella, James E. McGrath, Minoru Inaba, Kenji Miyatake, Michio Hori, Knichiro Ota, Zempachi Ogumi, Seizo Miyata, Atshushi Nishikata, Zyun Siroma, Yoshiharu Uchimoto, Kazuaki Yasuda, Ken-ichi Kimijima, Norio Iwashita. Scientific Aspects of Polymer Electrolyte Fuel Cell Durability and Degradation. Chemical Reviews 107, 3904-3951 (2007).

83. Wendt, H., Brenscheidt, T., Fischer, A. & Kendall, K. Optimization and Modelling of Fuel Cell Electrodes with Emphasis upon Catalyst Utilization Ageing Phenomena and Ageing Prevention [and Discussion]. Philosophical Transactions of the Royal Society of London. Series A: Mathematical, Physical and Engineering Sciences 354, 1627-1641 (1996).

84. B.J. Hwang, Y. W. S., J.F. Lee, P. Borthen. H.H. Strehblow. In situ EXAFS investigation of carbon-supported Pt clusters under potential control. J. Synchrotron Rad. 8, 484-486 (2001).

85. Shanna D. Knights, K. M. C., Jean St-Pierre, David P. Wilkinson. Aging Mechanisms and Lifetime of PEMFC and DMFC. Journal of Power Sources 127, 127-134 (2004).

86. David P. Wilkinson, J. S.-P. Durability. Handbook of Fuel Cell Technology and Applications 3 Ch. 47, 611-626 (2003).

87. Piela, P., Eickes, C., Brosha, E., Garzon, F. & Zelenay, P. Ruthenium Crossover in Direct Methanol Fuel Cell with Pt-Ru Black Anode. Journal of The Electrochemical Society 151, A2053-A2059 (2004).

88. Kazuaki Yasuda, A. T., Tomoki Akita, Tsutomu Ioroi, Zyun Siroma. Platinum dissolution and depostion in the polymer electrolyte membrane of a PEM fuel cell as studied by potential cycling. Physical Chemistry Chemical Physics 8, 746-752 (2006).

89. T. Akita, A. T., J. Maekawa, Z. Siroma, K. Tanaka, M. Kohyama, K. Yasuda. Analytical TEM study of Pt particle deposition in the proton-exchange membrane of a membrane-electrode-assembly. Journal of Power Sources 159, 461 (2006).

90. Yasuda, K., Taniguchi, A., Akita, T., Ioroi, T. & Siroma, Z. Characteristics of a Platinum Black Catalyst Layer with Regard to Platinum Dissolution Phenomena

54

in a Membrane Electrode Assembly. Journal of The Electrochemical Society 153, A1599-A1603 (2006).

91. Yamada, H., Shimoda, D., Matsuzawa, K., Tasaka, A. & Inaba, M. Stability of Pt-Ru/C Catalysts: Effects of Ru Content. ECS Transactions 11, 325-334 (2007).

92. Arruda, T. M., Shyam, B., Ziegelbauer, J. M., Ramaker, D. E. & Mukerjee, S. In situ XAS Investigation of Electrocatalyst Surface Poisoning by Halides. ECS Transactions 11, 903 (2007).

93. Zhou, X. Degradation of Pt Catalysts in PEFCs: A New Perspective from Molecular Dynamic Modeling. Electrochemical and Solid-State Letters 11, B59-B62 (2008).

94. Soowhan Kim, B. K. A., M.M. Mench. Physical degradation of membrane electrode assemblies undergoing freeze/thaw cycling: Diffusion media effects. Journal of Power Sources 179, 140-146 (2008).

95. Wang, C. et al. Three-Dimensional Superlattices Built from (M4In16S33)10- (M = Mn, Co, Zn, Cd) Supertetrahedral Clusters. J. Amer. Chem. Soc. 123, 11506-11507 (2001).

96. Bi, W. & Fuller, T. F. Temperature Effects on PEM Fuel Cells Pt/C Catalyst Degradation. Journal of The Electrochemical Society 155, B215-B221 (2008).

97. Fernandez, J. L., Walsh, D. A. & Bard, A. J. Thermodynamic Guidelines for the Design of Bimetallic Catalysts for Oxygen Electroreduction and Rapid Screening by Scanning Electrochemical Microscopy. M−Co (M: Pd, Ag, Au). Journal of the American Chemical Society 127, 357-365 (2004).

98. Geng, D. & Lu, G. Dependence of Onset Potential for Methanol Electrocatalytic Oxidation on Steric Location of Active Center in Multicomponent Electrocatalysts. The Journal of Physical Chemistry C 111, 11897-11902 (2007).

99. Stamenkovic, V., Schmidt, T. J., Ross, P. N. & Markovic, N. M. Surface Composition Effects in Electrocatalysis: Kinetics of Oxygen Reduction on Well-Defined Pt3Ni and Pt3Co Alloy Surfaces. The Journal of Physical Chemistry B 106, 11970-11979 (2002).

100. Ohtani, T., Takeuchi, H., Koh, K. & Kaneko, T. Electrical properties and phase transitions of TlCu3S2 and BaCu2S2. J. Alloys Comp. 317-318, 201-207 (2001).

101. Stamenkovic, V. R., Mun, B. S., Mayrhofer, K. J. J., Ross, P. N. & Markovic, N. M. Effect of Surface Composition on Electronic Structure, Stability, and Electrocatalytic Properties of Pt-Transition Metal Alloys: Pt-Skin versus Pt-

55

Skeleton Surfaces. Journal of the American Chemical Society 128, 8813-8819 (2006).

102. Zhang, J., Vukmirovic, M. B., Xu, Y., Mavrikakis, M. & Adžić, R. R. Controlling the Catalytic Activity of Platinum Monolayer Electrocatalyst from Oxygen Reduction with Different Substrates. Angew. Chem. Int. Ed. 44, 2132-2135 (2005).

103. Liu, Z. et al. PtMo Alloy and MoOx@Pt Core−Shell Nanoparticles as Highly CO-Tolerant Electrocatalysts. Journal of the American Chemical Society 131, 6924-6925 (2009).

104. Park, J. & Cheon, J. Synthesis of “Solid Solution” and “Core-Shell” Type Cobalt-Platinum Magnetic Nanoparticles via Transmetalation Reactions. J. Am. Chem. Soc. 123, 5743-5746 (2001).

105. Yeh, Y.-C. et al. Pd−C−Fe Nanoparticles Investigated by X-ray Absorption Spectroscopy as Electrocatalysts for Oxygen Reduction. Chemistry of Materials 21, 4030-4036 (2009).

106. Shao, M.-H., Sasaki, K. & Adzic, R. R. Pd−Fe Nanoparticles as Electrocatalysts for Oxygen Reduction. Journal of the American Chemical Society 128, 3526-3527 (2006).

107. Strasser, P. Combinatorial Optimization of Ternary Pt Alloy Catalysts for the Electrooxidation of Methanol. Journal of Combinatorial Chemistry 10, 216-224 (2008).

108. Jeon, T.-Y. et al. Influence of Oxide on the Oxygen Reduction Reaction of Carbon-Supported Pt−Ni Alloy Nanoparticles. The Journal of Physical Chemistry C 113, 19732-19739 (2009).

109. Murthi, V. S., Urian, R. C. & Mukerjee, S. Oxygen Reduction Kinetics in Low and Medium Temperature Acid Environment: Correlation of Water Activation and Surface Properties in Supported Pt and Pt Alloy Electrocatalysts. J. Phys. Chem. B 108, 11011-11023 (2004).

110. Yang, H., Vogel, W., Lamy, C. & Alonso-Vante, N. Structure and Electrocatalytic Activity of Carbon-Supported Pt−Ni Alloy Nanoparticles Toward the Oxygen Reduction Reaction. The Journal of Physical Chemistry B 108, 11024-11034 (2004).

111. Fernandez, J. L., Raghuveer, V., Manthiram, A. & Bard, A. J. Pd−Ti and Pd−Co−Au Electrocatalysts as a Replacement for Platinum for Oxygen Reduction in Proton Exchange Membrane Fuel Cells. Journal of the American Chemical Society 127, 13100-13101 (2005).

56

112. Yang, H., Alonso-Vante, N., Leger, J.-M. & Lamy, C. Tailoring, Structure, and Activity of Carbon-Supported Nanosized Pt−Cr Alloy Electrocatalysts for Oxygen Reduction in Pure and Methanol-Containing Electrolytes. The Journal of Physical Chemistry B 108, 1938-1947 (2004).

113. Jalan, V. & Taylor, E. J. Importance of Interatomic Spacing in Catalytic Reduction of Oxygen in Phosphoric Acid. Journal of The Electrochemical Society 130, 2299-2302 (1983).

114. Mavrikakis, M., Hammer, B. & Nørskov, J. K. Effect of Strain on the Reactivity of Metal Surfaces. Physical Review Letters 81, 2819 (1998).

115. F.A. Pedersen, J. G., J. K. Norskov. Understanding the Effect of Steps, Strain, Poisons, and Alloying: Methane Activation on Ni Surfaces. Catalysis Letters 105, 9-13 (2005).

116. P. Yu, M. P., P. Plasse. Journal of Power Sources 144, 11 (2005).

117. Holstein, W. L. & Rosenfeld, H. D. In-Situ X-ray Absorption Spectroscopy Study of Pt and Ru Chemistry during Methanol Electrooxidation†. The Journal of Physical Chemistry B 109, 2176-2186 (2004).

118. Ziegelbauer, J. M., Gatewood, D., Gullá, A. F., Ramaker, D. E. & Mukerjee, S. X-ray Absorption Spectroscopy Studies of Water Activation on an RhxSy Electrocatalyst for ORR Applications. Electrochem. Solid-State Lett. 9, A430-A434 (2006).

119. Mayrhofer, K. J. J., Hartl, K., Juhart, V. & Arenz, M. Degradation of Carbon-Supported Pt Bimetallic Nanoparticles by Surface Segregation. Journal of the American Chemical Society 131, 16348-16349 (2009).

120. H.R. Colon-Mercado, B. N. P. Stability of platinum based alloy cathode catalysts in PEM fuel cells. Journal of Power Sources 155, 253 (2006).

121. Ma, Y. & Balbuena, P. B. Surface Properties and Dissolution Trends of Pt3M Alloys in the Presence of Adsorbates. The Journal of Physical Chemistry C 112, 14520-14528 (2008).

122. E. Antolini, J. R. C. S., E.R. Gonzalez. The stability of Pt–M (M = first row transition metal) alloy catalysts and its effect on the activity in low temperature fuel cells: A literature review and tests on a Pt–Co catalyst. Journal of Power Sources 160, 957 (2006).

123. Kinoshita, K. Particle Size Effects for Oxygen Reduction on Highly Dispersed Platinum in Acid Electrolytes. J. Electrochem. Soc. 137, 845-848 (1990).

57

124. Stevens, D. A., Hicks, M. T., Haugen, G. M. & Dahn, J. R. Ex Situ and In Situ Stability Studies of PEMFC Catalysts. Journal of The Electrochemical Society 152, A2309-A2315 (2005).

125. Matsumoto, M., Manako, T. & Imai, H. Degradation of Carbon Supports in PEFC Cathode Electrode Investigated by Electrochemical STM: Effects of Platinum and Oxygen on Enhanced Carbon Corrosion. ECS Transactions 16, 751-760 (2008).

126. Roen, L. M., Paik, C. H. & Jarvi, T. D. Electrocatalytic Corrosion of Carbon Support in PEMFC Cathodes. Electrochemical and Solid-State Letters 7, A19-A22 (2004).

127. Paik, C. H., Jarvi, T. D. & O'Grady, W. E. Extent of PEMFC Cathode Surface Oxidation by Oxygen and Water Measured by CV. Electrochemical and Solid-State Letters 7, A82-A84 (2004).

128. K. Kinoshita, J. B. Determination of carbon surface oxides on platinum-catalyzed carbon. Carbon 12 (5), 525 (1974).

129. Yu, P. T., Gu, W., Makharia, R., Wagner, F. T. & Gasteiger, H. A. The Impact of Carbon Stability on PEM Fuel Cell Startup and Shutdown Voltage Degradation. ECS Transactions 3, 797-809 (2006).

130. Young, A. P., Stumper, J. & Gyenge, E. Characterizing the Structural Degradation in a PEMFC Cathode Catalyst Layer: Carbon Corrosion. Journal of The Electrochemical Society 156, B913-B922 (2009).

131. Gu, Y. et al. Aging Studies of Pt/Glassy Carbon Model Electrocatalysts. Journal of The Electrochemical Society 156, B485-B492 (2009).

132. Bessel, C. A., Laubernds, K., Rodriguez, N. M. & Baker, R. T. K. Graphite Nanofibers as an Electrode for Fuel Cell Applications. The Journal of Physical Chemistry B 105, 1115-1118 (2001).

133. T. Maiyalagan, B. V., U. Varadaraju. Nitrogen containing carbon nanotubes as supports for Pt – Alternate anodes for fuel cell applications. Electrochemistry Communications 7, 905 (2005).

134. B. Rajesh, V. K., S. Karthikeyan, K.R. Thampi, J.M. Bonard, B. Viswnathan. Pt–WO3 supported on carbon nanotubes as possible anodes for direct methanol fuel cells. Fuel 81, 2177 (2002).

135. P. Serp, M. C., P. Kalck. Carbon nanotubes and nanofibers in catalysis. Applied Catalysis A 253 (2), 337 (2003).

58

136. K. Lee, J. J. Z., H.J. Wang, D.P. Wilkinson. Progress in the synthesis of carbon nanotube- and nanofiber-supported Pt electrocatalysts for PEM fuel cell catalysis. Journal of Applied Electrochemistry 36, 507-522 (2006).

137. Huang, S.-Y., Ganesan, P., Park, S. & Popov, B. N. Development of a Titanium Dioxide-Supported Platinum Catalyst with Ultrahigh Stability for Polymer Electrolyte Membrane Fuel Cell Applications. Journal of the American Chemical Society 131, 13898-13899 (2009).

138. Hayase, M., Kawase, T. & Hatsuzawa, T. Miniature 250 mu m Thick Fuel Cell with Monolithically Fabricated Silicon Electrodes. Electrochemical and Solid-State Letters 7, A231-A234 (2004).

139. G. Wu, L. L., J.H. Li, B. Q. Xu. Polyaniline-carbon composite films as supports of Pt and PtRu particles for methanol electrooxidation Carbon 43, 2579 (2005).

140. Lefebvre, M. C., Qi, Z. & Pickup, P. G. Electronically Conducting Proton Exchange Polymers as Catalyst Supports for Proton Exchange Membrane Fuel Cells. Electrocatalysis of Oxygen Reduction, Hydrogen Oxidation, and Methanol Oxidation. Journal of The Electrochemical Society 146, 2054-2058 (1999).

141. Swathirajan, S. & Mikhail, Y. M. Methanol Oxidation on Platinum-Tin Catalysts Dispersed on Poly(3-methyl)thiophene Conducting Polymer. Journal of The Electrochemical Society 139, 2105-2110 (1992).

142. Bonakdarpour, A. et al. Studies of Transition Metal Dissolution from Combinatorially Sputtered, Nanostructured Pt1-x Mx (M = Fe, Ni; 0 < x < 1) Electrocatalysts for PEM Fuel Cells. J. Electrochem. Soc. 152, A61-A72 (2005).

143. M.K. Debe (Eds: W. Vielstich, A. L., H.A. Gasteiger) Handbook of Fuel Cell Technology and Applications, John Wiley and Sons, N.Y. (2003).

144. Wang, J. & Swain, G. M. Fabrication and Evaluation of Platinum/Diamond Composite Electrodes for Electrocatalysis. J. Electrochem. Soc. 150, E24-E32 (2003).

145. Lee, Y. J., Wang, F., Mukerjee, S., McBreen, J. & Grey, C. P. 6Li and 7Li magic-angle spinning nuclear magnetic resonance and in situ X-ray diffraction studies of the charging and discharging of LixMn2O4 at 4 V. J. Electrochem. Soc. 147, 803-812 (2000).

59

Chapter 2

In situ X-ray Absorption Spectroscopy: Experiment, Theory and

Analysis

2.1 XAS – An overview

It was mentioned in chapter 1 that x-ray absorption spectroscopy (XAS) is uniquely

positioned to study catalysts as they function, under realistic operating conditions. XAS

has a singular advantage over most other spectroscopic techniques in that one uses x-rays

to probe the system under study. X-rays belong to that part of the electromagnetic

spectrum, more poetically called ‘Maxwell’s rainbow’, having energies of the order of 1-

100 keV. Further, the core-level binding energies of most elements are well-separated in

energy and lie in this very range, making it possible to probe virtually all elements and

their compounds using XAS. By choosing a particular energy range, one can selectively

probe only the element of interest, making it an element-specific technique. X-rays

possessing energies beyond for e.g. the 2-3 keV energy range, are less susceptible to

absorption by many compounds of lighter elements, including for instance, water (H2O).

This a particularly important advantage as many important catalytic processes in nature

60

and those employed in various industries occur in solution. Another interesting aspect of

XAS is that the mean free path of a photoelectron (electron ejected by a photon-electron

interaction) is of the order of 8-10 Å. Thus, one is able to effectively probe the immediate

environment of any material without placing any constraints on its structural properties.

This allows one to study all kinds of compounds under various conditions including

polymers, molten salts, nanoparticles, minerals and enzymes; it has also been used in

unexpected, fascinating areas such as identifying historical inks in pigments found on

archeological remains and in famous paintings etc.

In recent years, XAS has been used to attack some very significant problems and in

the future this will surely increase. Examples include understanding the structure-

function relationship of one of most highly conserved of all proteins, Cytochrome C,1, 2 to

obtaining the most-likely structure of the photosynthetic Mn4Ca cluster,3-8 to even

revealing the local structure of water.9-11 It is also interesting to note that the timeframe of

the entire absorption event (excitation of the core-level electron and filling of the core-

hole by another electron) is of the order of a femtosecond.12,13 An x-ray absorption

spectrum is therefore very much like a snapshot of the molecule and its environment

(discussed in greater detail in ‘Principles of XAS’ section). In the near future, one will be

able to collect ultra-fast, time-resolved XAS data, which will revolutionize the study of

catalysis, as one can then observe both, structural changes and changes in oxidation state

of the catalyst during catalytic activity. Some very recent breakthroughs 14-17 and review

articles, most notably by Chergui and co-workers, highlight the progress and potential for

this field of spectroscopy.18-21 Considering all the advantages mentioned above, XAS

offers direct promise for the in situ studies of catalytic reactions 12, 22, 23 and indeed, has

61

already proven its worth as evidenced by the large number of studies using XAS that are

published every year.

In our research group, as well as most of the work described in this thesis, the many

advantages of XAS are suitably brought to bear on the study of electrocatalysis.

Electrocatalysts as used in fuel cells are typically nanoparticles dispersed on a high-

surface area support, commonly carbon, and are present in an electrolyte of some kind.

The electrolyte may be in the form of a polymer electrolyte membrane (PEM), such as

Nafion™ 117, or an acid such as phosphoric acid (in PAFCs). As nanoparticle catalysts

are typically between 2-5 nm, they do not possess long-range order and thus, cannot be

adequately probed by x-ray diffraction (XRD), which might otherwise be the technique

of choice to obtain structural information. Transmission electron microscopy (TEM)

provides great detail on nanoparticle morphology but is, as of today, chiefly useful for ex

situ studies of such catalysts. However, over the last decade, several groups have been

making progress in studying the solid-liquid interface using in-situ TEM.24-28 The

presence of electrolyte however greatly enhances the inelastic scattering of the electron

beam and does not allow one to obtain images with atomic-level resolution as is possible

in air. As was briefly mentioned earlier, the presence of a liquid layer around the catalyst

rules out a good many spectroscopic techniques as well, including all those that require

ultra-high vacuum or low pressures (XPS, Auger, HREELS) and low-energy

spectroscopic techniques (Raman, IR, FTIR, UV-Vis spectroscopy). Several modified

vibrational spectroscopic techniques for in situ electrochemical measurements such as

second harmonic generation (SHG), sum-frequency generation (SFG), subtractively-

normalized FTIR spectroscopy (SNIFTIRS), attenuated total reflectance (ATR) and

62

surface-enhanced infra-red absorption spectroscopy (SEIRAS), to name a few, exist, but

are severely limited in their applicability and the information they provide. In most of

these techniques, the signal-noise ratio can be quite low and observations of three-fold

adsorption sites are not possible due to dipole selection rules imposed on the

measurement by the processes.29 Further, many of these methods cannot be used in a

working fuel cell containing a fair number of additional components around the catalyst

layer including high surface area carbon support, electrolyte layer, graphite current

collector plates, a plastic body etc. Finally, while some of these techniques may be well-

suited to study CO adsorption on nanoparticle catalysts,29 to the best of our knowledge,

none of them can actually follow changes in the adsorption of H, OH or O, all very

important adsorbates in the study of electrocatalytic processes. These adsorbates however

can be followed with XAS, specifically using the ∆μ-XANES technique developed by

Ramaker and Koningsberger.30-33 Apart from getting surface-specific information about

adsorbates, valuable structural information from the fine structure (EXAFS region) is also

obtained which complements very well, the information contained in the XANES region.

2.2 Synchrotron Radiation and Experimental methods

The majority of x-ray absorption experiments around the world are carried out at a

synchrotron facility (see Figure 2.1). A synchrotron provides researchers with

exceptionally high-intensity photons for their experiments. Radiation from all parts of the

electromagnetic spectrum are obtained through synchrotron radiation, though it is

primarily used for producing x-rays; in many synchrotrons, super bright ultraviolet (UV),

visible (Vis) and infra-red (IR) radiation are also produced for spectroscopic studies. One

63

of the added advantages at such a facility is the ability to obtain either linearly or

circularly polarized light for dichroism studies, making a host of magnetic and surface

studies possible. A cone of synchrotron radiation is produced when electrons are

accelerated to very high velocities (v ≈ c) and is a consequence of the laws of

electromagnetism. Within the framework of the classical theory of electromagnetism as

described by Maxwell’s equations, the principle of conservation of energy requires that

an accelerating charge should emit radiation. Synchrotrons are almost always in the shape

of a ‘ring’ so that a current of electrons can be kept circulating such that they undergo

acceleration at all points of their trajectory. As electrons are injected into the ring at high

energy, they are continuously accelerated towards the center of the ring, and therefore

emit radiation tangentially to the path of the electrons. Strong, carefully designed

magnetic fields are used to keep the electrons going around the storage ring in circular

motion. Third generation synchrotron facilities also make use of specialized injection

devices such as undulators and wigglers to obtain radiation of very high intensity.

X-rays of very high energy (ca. 100 keV or greater) may also be produced in such

facilities, albeit at the expense of some intensity of the photon beam. All experimental

stations are located at the periphery of the ring to collect the photons and channel them

into the sample under study. Continuous x-rays over a large energy range are produced

by the synchrotron and sophisticated beamline optics allows one to manipulate the beam

for various experimental techniques. A key feature of every beamline is the ability to

obtain monochromatic x-rays. This optical arrangement called a monochromator

typically consists of two single crystals of very high purity that are arranged in parallel so

as to select particular energies out of the x-ray beam, and is shown in Figure 2.2.

64

Figure 2.1 The National Synchrotron Light Source located at Brookhaven National

Lab, Long Island, N.Y. Picture credit: Courtesy of NSLS, Brookhaven National Laboratory.

65

\

Figure 2.2 A schematic of a double-crystal monochromator commonly used to tune the energy of the photon beam

66

The energy selection is effected by varying the angle (θ) between the x-rays and the

crystal as only x-rays satisfying Bragg’s law at that angle continue on to constitute the x-

ray beam used in the experiment –

nλ = 2dhkl Sin θ (Eq. 2.1)

where dhkl is the lattice spacing of the crystal and the other terms have their usual

meanings. It is due to the monchromator that experimenters can ‘scan’ a sample over a

wide range of energies to obtain a full EXAFS spectrum. The angles of the

monochromator are also used to control the higher harmonics that are usually present in

the beam. These harmonics can adversely affect any absorption spectrum and have to be

eliminated or ‘rejected’ as far as possible before any photons impinge on the sample.

Depending on the nature of the sample being studied, x-ray absorption data are

collected in either transmission mode or fluorescence mode (Figure 2.3). In transmission

mode, all the three detectors viz. one for the incident beam (I0), one for the transmitted

beam (It) and finally, the detector for the collecting data on the reference foil. For

concentrated samples, data are collected in transmission mode while fluorescence data

are collected on low-concentration or dilute samples. Experiments are carried out in

transmission mode when an absorption height of ~1 can be obtained (ideally) on the

sample of interest. However, if the maximum absorption height obtained is ~ 0.2 or

lower, owing to a much lower signal-noise ratio, it is advisable to collect data in

fluorescence mode instead. In order to determine if you have enough material for an XAS

measurement, the theoretical absorption step height for the sample has to be calculated.

Tables containing photoabsorption data and effective cross-sections for the various

elements over a wide range of energies can be used to calculate a theoretical absorption

67

step height for any compound and are available in the following references.34-40

Alternatively, it is also quite convenient to use the HEPHAESTUS program to calculate

the absorption step heights for your samples.41 It is available as part of the IFEFFIT

suite,42 a set of programs developed specially for the XAS community. This suite and

other useful programs for analysis of XAS data is available free of charge at their

homepage: http://cars9.uchicago.edu/~ravel/software/Welcome.html.

In transmission mode, a three-detector setup is used to record the initial and

transmitted intensities, respectively. The detectors in this case are gas-ionization detectors

which produce a current whenever the gas in the detector is ionized by the photon beam.

This weak current is then amplified and correlated with the intensity of the x-ray beam

using an appropriate calibration procedure and finally converted to digital form through a

voltage-frequency converter. Typical gains used in the amplifiers are of the order of 108

V/A. The most commonly used gases in detectors include helium, nitrogen and argon.

Mixtures of helium and nitrogen are used for low energy x-rays whereas argon is used for

higher energy x-rays. Shown in Figure 2.4 is a typical in situ XAS experimental setup,

like the kind that was used for work described in this thesis. This setup is located at

beamline X-3B at the National Synchrotron Light Source (NSLS), Brookhaven National

Lab in Upton, N.Y. where all XAS data for our experiments are collected.

68

Figure2.3 Schematic of experimental setup at the beamline showing the two principal methods of collecting XAS data: transmission and fluorescence.

I0 It Iref

If

sample foil

69

Figure 2.4 A typical in situ XAS experiment setup. Shown here is a flow-through in situ XAS cell setup (center) at beamline X-3B at the NSLS. The gas ionization detectors are visible at the bottom of the picture and the cryostat-cooled, solid-state fluorescence detector is seen on the left.

70

In case of fluorescence measurements, the expression for the dependence of the

absorption coefficient on incident and transmitted intensities is fairly complex. Sufficient

care must also be exercised over the condition of the sample if data are collected in this

mode. The equation relating the intensities and absorption coefficient in case of data

collected in fluorescence mode is shown below –

(Eq. 2.2)

Where If and I0 are the emitted fluorescence intensity and intensity of incident rays

respectively, Є is the fluorescence efficiency, t is the sample thickness, ∆Ω is the solid

angle of rays intercepted by the detector, μχ(E) is the absorption coefficient of the

element, μt(E) is the total absorption, θ and φ are the incident and reflected angles of the

x-ray beam, and Ef is the energy of fluorescent x-rays. The equation can be simplified

with some minor assumptions that hold in many cases. For a thin sample, μt << 1, and

reduces the exponential term and the associated term in the denominator simply to ‘t’.

The equation now becomes –

(Eq. 2.3)

For a thick, dilute sample, μχ << μt and the energy dependence of μt can be ignored. In

this case, the equation reduces to –

μt ≈ If / I0 (Eq. 2.4)

This is the commonly used approximation for most fluorescence data collection.

If = I0ε∆Ω4π

μχ (E) tIf = I0ε∆Ω4π

μχ (E) t

If = I0ε∆Ω

4π

μχ (E) 1 – e – [μt(E)/sin θ + μt(Ef)/sin φ] t

μt(E)/sin θ + μt(Ef)/sin φIf = I0

ε∆Ω

4π

μχ (E) 1 – e – [μt(E)/sin θ + μt(Ef)/sin φ] t

μt(E)/sin θ + μt(Ef)/sin φ

71

Solid-state semiconductor detectors made of Si-Li and Ge are widely used for

collecting fluorescence data. Shown on the far left of figure 2.4 is a 13-element, state-of-

the-art germanium detector that is cooled by a liquid-nitrogen cryostat to reduce thermal

noise in the detector. Other kinds of detectors used in many fluorescence measurements

include the passivated implanted planar silicon detector (PIPS ™) 43 and the Lytle

detectors.44

2.2.1 In situ spectroelectrochemical cell for XAS experiments: aspects of design

and development

Over the course of the work described in this thesis, many different in situ

spectroelectrochemical cells were designed, fabricated and tested. At least three or four

different versions were actually used and tested at the synchrotron. In the latest version of

our spectroelectrochemical cell, one can purge the electrolyte with any gas (O2, N2 or Ar)

and introduce poisons of interest (chloride, sulfide etc.) at specific concentrations while

under accurate electrode potential control. Some of the various design constraints and

experimental difficulties that had to be overcome in order to successfully collect data for

our experiments are described here.

a. A multi-purpose cell: XAS data can be collected either in transmission mode

or fluorescence mode. Depending on the beamline at which the experiment is

conducted, the fluorescence detectors may be located either to the right or left of

the incoming photon beam. This consideration is of little consequence if a Lytle

detector is to be used as it is small enough to be placed on any side of the cell.

The issue however becomes important if one wishes to collect very high-quality

72

data from samples at very low concentrations or samples from which significant

attenuation can be expected – as is the case in many in situ XAS experiments.

Most of these experiements require the use of sophisticated helium-cooled, multi-

element solid-state detectors that are available at some facilities. Due to the

experimental and spatial constraints of these larger detectors, the location of the

detector may not be conveniently changeable at many of these beamlines. The

earliest versions of the in situ cells used for our studies were only designed to

collect data in transmission mode. These cells contained an open electrolyte

reservoir on the top of the cell and could only accommodate the photon beam if

placed in a specific orientation. Thus, these cells could not be used in an

ambidextrous fashion and necessitated the use of two different handed cells that

had to be fabricated for use at the beamlines. The latest flow-through cell design

contains the ports of the electrolyte reservoir on the vertical face of the cell (as

opposed to the top of the cell) resulting in a sealed and ambidextrous unit which

can be employed at different beamlines with no modifications. Finally, although

the cell is designed for ‘flow-through’ operation, this feature is optional and can

still be used in the majority of in situ XAS experiments where flow of electrolyte

is not required.

b. Leakage of electrolyte: The interior electrolyte reservoir around the electrodes

is designed to contain only 3-5 ml of electrolyte, which is supplied by a peristaltic

pump. Tiny leaks through the joints of the cell as well as through any open

channels, such as those meant for electrical contact with the electrodes, were

sometimes observed. Furthermore, a leak near the current collection area

73

sometimes occurred, and such leaks could be potentially damaging to the

experiment if it shorted the half-cell under study. The latter was overcome by the

strategic placement of the electrode current collectors and more careful design of

the cell. The use of high-quality gaskets also prevented most leaks. Our most

recent cell also has a specially designed secondary containment to contain any

leakage as many electrolytes are potentially corrosive (strong acids or bases) and

cannot be allowed to leak onto the x-ray setup inside the hutch. The secondary

containment also contains built in pegs on which the cell rests while in use. This

allows the cells to be changed conveniently without having to reposition the

experimental table or realign the new cell. This is particularly helpful as the

alignment of a cell to maximize the flux of the photon beam may take up to a few

hours, thereby allowing more efficient use of the allocated beamtime.

c. Gas bubbles in the x-ray window: This is still a concern during our

experimental runs but has been overcome to a large extent. Gas bubbles such as

dihydrogen or dioxygen at very low and certain higher potentials, respectively,

often arise from redox reactions. This phenomenon has been particularly

troublesome for some of our fundamental studies on electrode materials used in

the chlor-alkali industry, where oxygen or chlorine evolution has severely

impeded the collection of high-quality XAS data.

Another key issue that arises occasionally with use of the flow-through

spectroelectrochemical cell is the occasional air bubble in the electrolyte which

makes it way to line of sight of the x-ray window. This issue is particularly severe

when the height of the cell is positioned differently from the height of the

74

peristaltic pump and the external electrolyte reservoir. In order to avoid this

problem, it is imperative that both the in situ XAS cell and the pump/reservoir be

maintained at the same level, which can be conveniently done using common

table jacks. It is believed that the source of the air bubbles in the latest cell design

arises from a vacant electrolyte port in the main bore of the cell. The cell contains

four bores but only employs three (electrolyte inlet, outlet and reference electrode

salt bridge) during any experiment. A slight drop in electrolyte pressure in this

vacant port may be enough to allow an air bubble or two to enter the electrolyte,

with eventual movement into the line of sight of the x-ray window. Forthcoming

cell designs will have to take this into account so as to avoid this problem

altogether.

d. Incomplete polarization of the electrode: Incomplete or non-uniform

electrode polarization can occur due to poor current distribution over the

electrode. This often arises from geometric incongruence between the working

electrode and counter electrode or reference electrode. This leads to an altered

surface chemistry in certain ‘dead zones’ on the electrode, and can make the

experimental XAS data un-representative of the general electrochemical processes

occurring at the desired potential. In general the flow-through

spectroelectrochemical cell employs a carbon-based counter electrode placed

directly across from the working electrode. This configuration minimizes the

asymmetry in the current distribution as the two electrodes are then separated

uniformly by only ca. 4 mm.

75

e. Poisoning of the reference electrode: As with all electrochemical

measurements, care must be taken to employ a stable reference electrode.

Changes to the reference electrode over the course of an experiment will result in

an inaccurate recorded potential of the working electrode. Further, the reference

electrode used for these studies must not release any ions into the system being

studied. This is often the case with the Ag/AgCl reference electrode resulting in

poisoning of the catalyst surface with unwanted ions such as Cl- etc. It is for this

reason that the standard hydrogen electrode (SHE) was used in all of our in situ

XAS studies.

f. X-ray window tape: Currently, a PTFE tape (3M) is employed along with a

silicone adhesive to create a tightly-sealed, transparent x-ray window for the in

situ cells. This tape exhibits excellent stability in high beam energies, for instance

at the Pt L3 edge at 11564 eV; however, at lower beam energies the tape is found

to degrade more rapidly. For example, at the Fe K edge (7112 eV) the tape is only

stable for up to 8 hours under irradiation. Leakage due to tape damage in the x-ray

window has occurred during some of our longer experimental runs at lower edge

energies.

76

2.3 Principles of x-ray absorption spectroscopy

X-ray absorption spectroscopy is a core-level spectroscopy that involves the

excitation of an electron from an inner orbital to a valence orbital, and at higher energies

directly into the continuum, creating a core-hole which is promptly occupied by an

electron from one of the valence levels to stabilize the momentarily ionized atom. This

relaxation may result in one of the following secondary processes: emission of x-ray

fluorescence radiation, emission of an Auger electron or secondary electron or so called

photo-production.45 The source of excitation in XAS is a photon beam of sufficiently

high energy directed at the material under study. In transmission mode, the intensity of a

beam of photons (I0) decreases proportional to its incident intensity on passing through a

medium of thickness (x). The transmitted intensity (I) is obtained through the well-known

Beer-Lambert law and is derived as follows –

dI α I.dx (Eq. 2.5)

Introducing a constant of proportionality, the inequality becomes

dI = I.μ(E).dx (Eq. 2.6)

Here, the constant of proportionality, the absorption coefficient μ, is an energy-dependent

term 46: μ(E) ~ ρZ4/AE3 (Eq. 2.7)

On integrating and rearranging terms in Equation 2.6, one obtains the Beer-Lambert law:

I = I0.e-μ(E).x (Eq. 2.8)

Thus, the absorption coefficient has a 1/E3 dependence which results in the slowly

decreasing absorption intensity as a function of the x-ray energy (Figure 2.5). However,

at certain energies, i.e. when the energy of the incident photons is exactly equal to the

binding energy of a core-level, a sharp increase in the absorption occurs due to an atomic

77

Figure 2.5 A schematic of an X-ray absorption spectrum over a large energy range showing the K, LI and LII edges. Note that the assignment of edge energies starts from the highest-energy transition.

78

absorption event followed by the slow fall-off in the absorption as mentioned earlier (Eq.

2.3). Superimposed on this decrease in overall absorption as a function of energy, are a

number of rapidly varying oscillations and is a characteristic feature of XAS spectra of an

element in the condensed phase. The edges are named K, L, M etc. to indicate the

subshell from which the photoelectron is ejected. For instance, when the ionization

occurs due a loss of an electron from the 1s orbital, it is termed a K-edge transition; if it is

from the 2s orbital, it is termed an LI edge transition and so on. It follows that the

energies of the different edges decrease in the order K > L > M etc. The edge notations

that are most commonly encountered are shown in Figure 2.6 (a) along with their

respective assignments.40 A typical XAS spectrum is broadly distinguished into two

regions: the region containing the sharply rising feature due to intense atomic absorption

at the edge energy, commonly referred to as the ‘white line’, is called the x-ray near edge

spectrum, or XANES region. The rest of the spectrum at higher energies, typically

between 150 - 1000 eV, contains the oscillatory component of the absorption spectrum.

This fine structure in the spectrum is called the extended x-ray absorption fine structure,

or EXAFS. The energy ranges assigned to the two regions are not to be taken as absolute

and may vary depending on the system being studied. However, for the most part, it is

agreed that the 50 – 100 eV region is called the XANES region while the rest of it above

100 or 150 eV is the EXAFS region. The XANES region is sensitive to the oxidation

state of the atom, the local symmetry around it and even the nature of ligands coordinated

to the atom, while the EXAFS is affected chiefly by the geometry of the structure of the

material. It is important to note that there is no fundamental distinction in the phenomena

giving rise to the XANES and the EXAFS regions.

79

Figure 2.6 Fundamental processes occurring during an x-ray absorption event. a. Excitation of a core-level electron, and b. backscattering of the ejected photoelectron due to neighboring atoms surrounding the absorber atom.

Ebinding

d

s p

M

L

K

EFermi

a

Photoelectron

hνAbsorber

Neighbor

b

Constructive Destructive

80

When a photon is absorbed by an atom, a photoelectron is ejected from it and propagates

radially outward as a wave. In an isolated atom, as can be approximated in case of a gas,

this wave just propagates into space without encountering any change in the energy

landscape around it. Thus, after the absorption event, there is a monotonic decrease in

absorption (the 1/E3 dependence) until another absorption edge is reached when a similar

process occurs. However, if the atom is not isolated but instead, is in a condensed phase

such as a liquid or a solid, the outgoing wave encounters the potential wells of the various

atoms surrounding the absorbing atom and thus, undergoes some backscattering as well

(see Figure 2.6b)

This backscattered wave (or reflected wave) can interfere constructively or

destructively with the outgoing waves from all the neighboring atoms and depending on

the phase of the waves, gives rise to an interference pattern. The intensity of the reflected

wave will depend primarily on two factors viz. the number of nearest neighbors to the

absorber atom and the distance at which these neighbors are located. Since the

interference pattern will depend significantly on these two parameters, structural

information is inherent in the EXAFS and thus, one can obtain coordination number (N)

and bond distances (R) through an analysis of this region of the spectrum. Two types of

scattering occur in an x-ray absorption spectrum viz. single-scattering (SS) and multiple-

scattering (MS), as shown in Figure 2.7. A single-scattering process only involves the

absorber atom and one scattering atom (process A). This scattering atom may or may not

be in the first coordination sphere of the absorbing atom. In a multiple-scattering process,

the outgoing photoelectron may interact with a number of different neighboring atoms in

any fashion as it undergoes a constructive or destructive interference (process B).

81

Figure 2.7 The various types of backscattering that occur during an absorption event

A

C

B

82

backscattering from atoms that are collinear to the absorber can be particularly strong due

to a ‘focussing effect’, an example of which is seen in process C.

It was mentioned in passing earlier that there is no fundamental difference between

the XANES and the EXAFS region. While this is believed to be true,13 MS processes are

much more frequent and take precedence over SS events in the low-energy scattering

range (XANES) while at higher energies above the edge energy, SS events dominate

(EXAFS) and MS events are less-likely (see Figure 2.7). It is for this reason that the

XANES region is extremely sensitive to the symmetry around the absorbing atom and

can be quite complex to describe in simple mathematical expressions. Thus analytical

expressions describing the EXAFS region were derived long back whereas no such

expression is available for the XANES region. However, recent advances in ab initio

quantum mechanical calculations such as the codes developed by members of the FEFF

project (http://leonardo.phys.washington.edu/feff/welcome.html) enables one to

accurately calculate a theoretical XANES spectrum for a given model cluster (see section

on FEFF 8.0).

2.3.1 Historical note on the development of the theory of the x-ray absorption

spectrum

Although x-rays were first discovered by Wilhelm Roentgen in 1895,47 the first

observations of the phenomenon of x-ray absorption were made in 1913 by Maurice de

Broglie, older brother to the more famous Louis de Broglie, when he noticed intense

absorption bands in the photographic plate of an x-ray diffraction spectrum of his

crystals. These first edges, as such absorption features came to be called, were the K

edges of Ag and Br. Observations of the associated oscillations beyond the edge, known

83

as the fine structure, were first made by Fricke in 1920.48 After a series of investigations

on various absorption spectra by Fricke and Hertz, Walter Kossel put forth the first

theoretical interpretation of the XANES region.49 He correctly interpreted the absorption

edge to be due to electronic transitions to unfilled orbitals but his theory nevertheless

failed to explain the extended fine structure, which by then was clearly established as a

very real spectroscopic feature of x-ray absorption spectra. We owe the first serious

attempts to explain the fine structure to Kronig, 50, 51 who based them on a theory of long-

range order (LRO), and were soon found to be incorrect. However, in a subsequent paper,

he modified his interpretation which eventually contained some ideas that remain valid

within the modern theory of EXAFS. 52 Following several incremental, albeit important

developments over the course of the next 40 years by a number of researchers in Japan,

Russia and the United States, 53-63 a nearly complete foundation for the theory of x-ray

absorption developed. It is interesting to note that the long-standing debate as to whether

EXAFS was adequately described by a theory that was long-range versus one that was

short-range, was still unresolved even as late as 1970. 64, 65

It was in the early 1970’s that XAS was finally correctly interpreted and theoretically

explained satisfactorily, an advance that ensured its potential as a spectroscopic tool.

84

Figure 2.8 A reproduction of an early x-ray absorption spectrum showing assignments of characteristic features that are seen in the EXAFS region of the spectrum.

85

This advance can be attributed to the following –

1. A concise mathematical description of the fine structure was put forth by Sayers,

Stern and Lytle in 1971 66, 67 and the entire treatment of what today constitutes

‘modern EXAFS theory’, were published in three seminal papers.68-70 Taking

advantage of the fact that single-scattering events dominate at energies much

higher than the edge-energy (E0), they derived a single equation which could

adequately model the oscillatory component of the spectrum. They also realized

that a discrete Fourier-transform of the oscillations would give rise to a radial

structure function containing peaks which could be assigned to the various

‘shells’ of atoms surrounding the absorber atom. This was then used to prove that

the technique is chiefly sensitive to short-range order, settling the debate about

this topic. These advances greatly facilitated the interpretation of EXAFS data and

made the technique seem less esoteric and more accessible to researchers.

2. The increasing availability and advances in computing power to perform complex

and lengthy calculations which were virtually impossible to carry out by hand

3. The advent of synchrotron sources where one had access to x-rays of

exceptionally high intensities as well as an easily ‘tunable’ x-ray source. The high

photon flux straightaway enabled XAS to be used for identifying elements,

studying minerals, important non-crystalline materials like polymers and even

ions in solution, turning it into an invaluable spectroscopic technique.

86

In reviewing the history and development of this fascinating field of spectroscopy,

wonderfully personal, biographical accounts have been written by Stern 71 and Lytle.72

2.3.2 A mathematical description of the EXAFS region

The kinetic energy of the photoelectron ejected from the absorber atom can be

expressed as-

Ek = Eincident – φ – B.E. (Eq. 2.9)

Here Eincident is the incident photon energy (hν), φ is the work-function of the material,

and B.E. is the binding energy of the core-shell electron relative to the Fermi-level of the

atom (Ef). From the fundamental laws of conservation of energy and using an alternate

expression for the kinetic energy-

½ mv2 = p2/2m (Eq. 2.10)

k = 2π/λ (Eq.2.11)

where m is the mass of the electron and p is the momentum of the electron, λ is the

wavelength of radiation and k is the wave number of the photoelectron, we obtain the

relationship between k and the energy above the edge (∆E = E – E0) as

(Eq. 2.12)

The absorption event itself is adequately described using time-dependent perturbation

theory. Within this framework, the absorption probability μ(E) is function of the square

of the transition matrix element –

(Eq. 2.13)

| < ψf | ε. r e i k . r | ψi > | 2

√ 2m (E – E0) / h2k = √ 2m (E – E0) / h2k =

87

Where ا ψi > and ا ψf > are the initial and final state wavefunctions respectively, while ε

and k are the electric polarization and wave vectors respectively.13, 73 If the polarization

dependence is taken out of the equation, a simpler form results, which is the well-known

‘Fermi’s Golden Rule’ or the ‘Dipole approximation’; if the equation is left as is, one

obtains the form that can describe even quadrupole transitions. The photoabsorption cross

section σ (ω) described in terms of the transition probability for each of the two cases are

as follows –

σ(ω) = 4πα2 ħω1

4 ∑‹ f |εr| i ›2 δ(Ef – Ei – ħω)σ(ω) = 4πα2 ħω14 ∑‹ f |εr| i ›2 δ(Ef – Ei – ħω)

(Eq. 2.14)

and

(Eq. 2.15)

where α is the fine structure or coupling constant (e2/ħc ~ 1/137), ω = 2πν, Ef is the

energy of the final state, Ei is the energy of the initial state and ħω is the energy of the

incident radiation. From equations 2.14 and 2.15, we obtain the following - 74

1. the dipole selection rule(s) for absorption viz. ∆ l = ± 1 and ∆ J = ± 1.

2. Effect of the core-hole (final state) on the total absorption probability

3. Energy dependence of the absorption and selectivity due to the ħω term

4. Dichroism, or a polarization dependence, that is completely determined by the

vector operator kr

Quadrupole transitions are generally very weak and only around 5-10 % of the edge

height of the corresponding dipole transition (white line intensity). However, in certain

geometries where significant mixing of orbitals occur, for e.g. in p-d mixing seen in

σ(ω) = 4πα2 ħω14 ∑‹ f | (εr)(kr) | i ›2 δ(Ef – Ei – ħω)σ(ω) = 4πα2 ħω14 ∑‹ f | (εr)(kr) | i ›2 δ(Ef – Ei – ħω)

88

molecules containing tetrahedral ions like K2CrO4 or BaTiO4, the quadrupole transition

suddenly assumes a ‘dipole-like intensity’ and the so-called pre-edge peak may be as

intense as the main edge height itself. Again, this is due to the sensitivity of the XANES

region to the local symmetry around the absorbing atom and in certain cases, can reveal

subtle changes in oxidation state indirectly even if they are not directly observable

through edge-shifts.

The time-frame of an absorption event is governed by Heisenberg’s uncertainty

relation in energy and time (∆E.∆t ≥ ħ). The natural line-widths for an XAS spectrum are

typically a few electron volts. Inserting this into the uncertainty relation, one sees that the

entire process occurs very rapidly and is of the order of a femtosecond. This time-frame

is at least at 100-1000 times faster than any vibrational relaxation in molecules75 and an

XAS spectrum can be considered an instantaneous picture of the atom and its immediate

environment. The absorption coefficient μ(E) is often expressed as

μ(E) = μ0 [1+χ(E)] (Eq. 2.16)

where μ0 is the atomic absorption coefficient and χ(E) is the oscillatory component of the

EXAFS which can be considered ‘superimposed’ onto the variation in the atomic

absorption coefficient (μ0). Noting that single-scattering events dominate at higher

energies, Sayers, Stern and Lytle derived the following expression for the χ(E) (or

equivalently, χ(k)) after making several assumptions (see below).68 It is also known as

‘The EXAFS equation’ -

(Eq. 2.17)

The parameters in the equation are defined as follows: Nj is the coordination number, S02

χ (k) = Σj

Nj S02 fj (k) e-2Rj/λ(k) e- 2k σj

2 2

k Rj2

Sin [2kRj + δj (k)]χ (k) = Σj


2 2

k Rj2

Sin [2kRj + δj (k)]

89

is the many body amplitude reduction factor, Rj is the inter-atomic bond distance, λ (k) is

the mean free path of the photoelectron, σj2 is the mean-square radial disorder term (also

known as the Debye-Waller factor) and is the sum of both, thermal and entropic disorder;

fj(k) is the scattering factor and δj (k) is the phase term in the equation. The technique

does not require long-range order and only samples the immediate environment around

the absorber atom – this feature of EXAFS is what gives the technique its tremendous

scope of applicability over for e.g., XRD, which can only give structural information for

systems possessing long-range order, such as crystals. This is a natural consequence of

the dependence on two aspects of the equation described here: a) the inverse R2

dependence in the equation and b) the exponential fall-off due to the mean free path of

the photoelectron i.e. the e-2Rj/λ(k) term. Both these terms suggest that the photoelectron

loses a significant portion of its energy over very short distances (under 6-8 Ǻ). Before

describing the procedures used in an analysis of EXAFS data, it is worthwhile to take a

brief look at some of the assumptions made in deriving the original equation. They are as

follows:

a. Variations in atomic potentials within a given ‘shell’ are small or negligible

b. ‘Muffin tin’ potentials at fixed distances are used to model the energy landscape

surrounding the absorber (see Figure 2.9)

c. Only a single photoelectron is ejected during a photon-electron interaction

d. The photoelectron propagates as a plane wave

e. Only single-scattering events are considered and multiple-scattering processes are

neglected.

90

Figure 2.9 Example of a muffin-tin potential (solid black line) that is frequently used to compute a theoretical EXAFS spectrum

0

distance

Pt Ru

potential

91

Over the years, several inadequacies in the equation have been corrected, especially

through the contributions of John Rehr and co-workers at the University of Washington

at Seattle.76-80 They have made tremendous progress in making improvements to the

original equation, developing faster algorithms and computational methods to a point

where the EXAFS equation is routinely used to model experimental spectra of a whole

range of compounds to a high level of accuracy. One of the main corrections to the

equation includes replacing the plane-wave formalism by spherical waves to more

accurately reproduce scattering amplitudes and phase-shifts. In order to retain the basic

structure of the equation, an effective scattering factor, feff (k,r) was introduced into the

equation (hence the name ‘FEFF’ for the project). Secondly, numerous expressions were

derived by many people in the field for the full multiple scattering description of χ(k).

However, almost all of them were severely demanding and inefficient in the use of

computing power applied for the calculations. Rehr and Albers developed an algorithm

that is less computationally intensive but sufficiently accurate to generate a theoretical

full multiple scattering χ(k) function.81

2.4 EXAFS analysis

Given that the EXAFS region of the absorption spectrum contains structural

information, one still has to extract this information from the spectrum. The procedure

followed to obtain parameters such as coordination numbers (N) and bond lengths (R)

through EXAFS analysis is described in this section.

All the data were analyzed using the IFEFFIT © suite,41 a set of programs specifically

developed for the XAS community and is available free of charge. Most of the programs

92

have an excellent graphical user interface that makes the entire package very user-

friendly and accessible. The two most important programs in the software include –

1. ATHENA – this program is used for all kinds of data processing, especially all

the initial processing of XAS data including removing glitches (deglitching),

alignment and calibration of scans, background removal, normalization etc. and,

2. ARTEMIS – this program is used to fit the EXAFS data using a fitting routine.

The experimental χ(k) function (EXAFS) is extracted from ATHENA and

imported into ARTEMIS for analysis. Once a theoretical model has been defined

and the parameters to be used for the fits are set, the fits are carried out using non-

linear regression analysis. A complete statistical analysis is outputted in a ‘palette’

at the end of each fit.

Both programs allow the user to visualize the data in energy, k or R-space by

conveniently clicking on the required option. A number of options for appropriately

transforming the data using fourier-transforms; commonly used ‘windows’ such as the

‘Kaiser-Bessel’, ‘Hanning’, ‘Sine’ window etc. are available for use in order to minimize

the ‘ringing’ that occurs when the data to be transformed is terminated too quickly.

Let’s briefly revisit the standard EXAFS equation (Eq. 2.17) –

Here, S02 and δj (k) depend only on the atomic number (Z) of the scattering atom and can

be generated theoretically by the FEFF code (this is already included as a routine and is

carried out automatically within FEFF 6.0 which ARTEMIS uses to fit EXAFS data).

The equation has to be summed for all possible paths before a fit can be made to the data.

χ (k) = Σj


2 2

k Rj2

Sin [2kRj + δj (k)]χ (k) = Σj


2 2

k Rj2

Sin [2kRj + δj (k)]

93

The parameters that are input as variables for the fit include coordination number (N),

bond-distance (R), the disorder term (σ2) and a fourth term, ∆E which is related to the

inner potential (E0) of the atom. As mentioned earlier, the fits are then carried out by

ARTEMIS using non-linear regression analysis. Once this is done, both the experimental

data and fit are displayed on screen along with statistical information which is outputted

in a separate window. Details regarding each of the steps taken in order to obtain these

parameters from raw XAS data are as follows –

a. Pre-edge background removal – The background is removed by fitting a

quadratic (or linear) polynomial to data within 50 eV from the edge in the pre-

edge region and is extrapolated well beyond the edge. Often, this region contains

some background noise, glitches, spikes or even higher harmonics from previous

edges which have to be taken care of before removing the background. This

however, is more critical in a XANES analysis than it is for an EXAFS analysis.

b. Normalization – The data is normalized through an estimation of the edge height,

or step height μ(E0), and normalizing it to a value of 1 over a large energy range.

Here, E0 is the edge energy typically obtained by locating the inflection point on

the derivative spectrum i.e. ∂μ/∂E. The normalization is done such that all the

oscillations in the EXAFS region now oscillate about the horizontal line μ = 1.

c. Extraction of the χ(k) function – To extract the most important part of the

spectrum which contains all the structural information, viz. χ(k), one must first

determine and effectively subtract out the slowly varying atomic component (or

background) of the function, μ0(E) (recall that μ(E) = μ0(E)[1 + χ(E)]). This

94

procedure is also known as ‘post-edge background removal’. It is the slowly

varying (low-frequency) component of the post-edge region extending all the way

out to around 1000 eV from the edge. A note of caution here: this atomic

background can easily be removed by fitting a cubic spline through a regular

least-squares fitting procedure. Ineffective removal of this component will result

in unexpected peaks at very small values of R in the Fourier-transformed data; if

too much is removed, information about the structure from the EXAFS region

will be lost leading to erroneous conclusions from fitted data. Several

improvements in standard background-removal algorithms available in most

standard XAS processing software have led to making this step routine. However,

care must still be taken in specifying the extent and range for the background

removal.

d. Fourier-transforms and the χ(k) – Once the χ(k) function is obtained, the data

may be Fourier-transformed to visualize the data in ‘r’ space to produce a radial

structure function. This is carried out by the following operation –

(Eq. 2.18)

Notice how the data is transformed from ‘k’ space to ‘r’ space. Further, given that

low-Z elements scatter more effectively at low k values and high-Z elements, at

higher k values, a kn term in the equation allows one to preferentially ‘weight’ the

χ(k) function to emphasize the scattering over a desired k-range. The transformed

χ (R) = 1

√2π∫kmin

kmax

kn χ(k) e i2kR dkχ (R) = 1

√2π∫kmin

kmax

kn χ(k) e i2kR dk

95

data gives rise to anumber of peaks located at different distances from the absorber

atom. The peaks are the various frequency components that constitute the original

χ(k) function. The location of each of the peaks however, are slightly shifted to lower

values of r from the true values of the bond-distance (R) and is due to the phase term

that exists in the EXAFS equation.

e. Building and choosing a model structure – In order to attempt to carry out a fit

to any EXAFS data, a model structure described either by its crystallographic

parameters (for e.g. a .cif file) or by defining the coordinates of the various atoms

with respect to the absorbing atom is required. The IFEFFIT suite already has

within it, both, a database of commonly encountered structures of metals and their

compounds (folder ‘atomsdb’), as well as a program (ATOMS) that can generate

a set of coordinates if the crystallographic parameters are inputted into the

program.82 These coordinates are then input into the FEFF calculation routine

within ARTEMIS, which on execution, generates a list of paths and associated

amplitudes. One then chooses the required paths (in order of importance and

relevance with respect to the compound being investigated) to carry out a fit to the

data.

f. Fitting the data – The values of the four parameters commonly fit to EXAFS

data are obtained through a non-linear least-square regression analysis procedure

incorporated in the ARTEMIS program; other parameters required for a fit such

as scattering factors f(k), the many-body amplitude reduction factors S02 and the

phase term δ(k) are theoretically generated by FEFF. Alternatively, they may also

be obtained by fitting standards or reference compounds of the same element

96

which have a well-characterized structure and of known oxidation state. These

parameters can then be used in fitting the data.

Thus from the above procedure, values for N, R and σ2 are obtained. While the

uncertainties in the values of the parameters are largely dictated by the quality of data

and quality of fit, it is not uncommon to obtain bond distances to an accuracy of 0.02 Ǻ

and coordination numbers to within 10-15% of the correct or known values. Indeed,

given that XAS is a very sensitive probe over short distances, R values derived from

such an analysis may be more accurate than values determined from XRD.83-87 The

Debye-Waller factor (σ2) is a measure of change in the absorber-scatterer distance and is

actually a sum of two forms of disorder: static or entropic disorder (σst2), and dynamic or

thermal disorder (σth2).88 Thus, compared to values for ordered, homogeneous materials

at normal conditions of temperature and pressure, the σ2 values are higher for disordered

materials such as alloys, ions in solution etc. as well as for data at higher temperatures.

Most of our work involves analyzing in situ XAS data for electrocatalytic systems

and as such, we are generally interested in following changes in values of N as a function

of the applied electrode potential, V. Given that in EXAFS fitting, the obtained values of

N are significantly correlated with σ2, one cannot meaningfully compare values of N

obtained that are associated with different values of σ2 and sufficient care must be

exercised in such analysis. In order to make such N vs. V graphs more meaningful, most

of our data sets are fit twice over. The first time, all four parameters are allowed to vary

independently, as is usual practice in EXAFS fitting. From this set of fits, the average

value < σ2 > is determined. In a second round of fits to the data, this term is kept fixed at

the average value for data at all potentials and the fits are carried out allowing only the

97

remaining three parameters to vary. It is this set of parameters that are used to make an N

vs. V graph. No doubt the fits may not be as good as those obtained when all four

parameters are allowed to vary. Nevertheless, the variation in N as a function of V is

much more real and acceptable after such a modified fitting procedure.

In case of a bimetallic alloy (PtM), data at both edges (for e.g. Pt LIII and Ru K edges)

were fit simultaneously setting RPt-M = RM-Pt and NPt-M = NM-Pt (for a uniform mixture).

Details about such procedures are discussed in the relevant chapters.

2.5 XANES analysis

The remarkable sensitivity of the XANES region to the oxidation state, local

geometry and even presence of ligands around the absorber atom is well-known. This is

largely due to the fact that at low-energies, multiple-scattering dominates over single-

scattering paths and makes the overall shape of the XANES region particularly sensitive

to the immediate scattering environment of the absorber atom. For instance, the white-

line intensity for the Pt L3 edge (2p3/2 5d) is directly proportional to the unfilled-

density of states in the d-band (or d-band vacancy), and changes in the white-line can be

used to infer changes in the bonding of Pt to its neighbors as ligands coordinated to it

will either donate or withdraw charge from the Pt d-band. It is for this very reason that

traditionally, XANES analysis is labeled as a ‘fingerprinting technique’ as most

compounds have rather unique XANES spectra. Yet, when one is looking at small

perturbations in a catalytic system, tiny changes can easily be overlooked especially

since such changes are only 3-5% of the total intensity and are masked by the huge

atomic absorption contribution of any XAS spectrum.

98

In analyzing XAS data on catalytic processes, it is then quite convenient to turn the

method into a difference technique wherein only changes in the XANES region are

isolated, magnified, observed and finally, modeled with theory in order to successfully

interpret the changes. This idea forms the basis of the ∆μ-XANES method developed by

Ramaker and Koningsberger 30, 31, 89and is described in greater detail below.

2.5.1 The ∆μ-XANES method

The ∆μ-XANES method is ideally suited to study changes in a system where one has

a significant control over the environment around the absorber atom, such as in

electrocatalysis or even gas-phase catalysis. In electrocatalysis, for instance, different

species/adsorbates are in equilibrium with the electrode surface depending on the applied

potential (e.g., H, OH, O) while in gas-phase catalysis, one can introduce gases such as

H2, O2, CO etc. at will, at a desired temperature and pressure. The scattering due to the

adsorbate is isolated by taking out the bulk-scattering component from the spectrum by

subtracting out an XAS spectrum of a clean surface. As was mentioned previously,

several investigations in both areas of catalysis have been successfully carried out using

this method. The adsorption of anionic and cationic poisons (e.g. Cl- and Run+) on Pt/C

nanoparticle catalysts widely used in fuel cells, have been studied for the first time using

this technique and are reported in this thesis. The nature of poisoning, possible adsorption

site and their detrimental effects on the activity of the electrocatalysts are discussed in

Chapters 3 and 5 respectively.

2.5.2 Data analysis

The raw XAS data are first imported into ATHENA and a careful normalization in

the 25-150 eV range is carried out. All the reference foil spectra are aligned to one

99

particular spectrum, e.g., the reference foil spectra at 0.54 V whose edge energy has been

calibrated to the known standard value (Pt L3 edge at 11564 eV). ATHENA then

automatically applies all shifts on the foil spectra to the corresponding sample spectra.

This alignment is done to correct for experimental errors. At the synchrotron, the photon

beam drifts slightly in energy and thus, the energy calibration of the beam may not be

valid over a long time period. To illustrate the point, let us assume that data were being

collected at the Ru K edge for a ruthenium catalyst over a period of 8 hours. However,

after five or six scans, the 20,012.0 eV of the K edge may appear to shift to 20,012.3 eV.

This minute difference in the calibration is likely to affect the final shape of the

difference spectra that are calculated. Thus, this alignment and calibration procedure

ensures that any energy drift in the collected spectra are accounted for and all changes in

the sample data accurately reflect only changes in the sample due to the applied electrode

potential. Next, a XANES spectrum at a potential where the surface is relatively free

from any adsorbates (such as in the double-layer region) is chosen as a ‘reference’

spectrum. This ‘reference’ spectrum is not to be confused with the ‘reference foil’

spectrum mentioned earlier. A set of experimental difference spectra are then generated

by subtracting out this ‘clean’ reference spectrum. It is more clearly expressed in an

equation –

∆μexp = μ(ads/M)v – μ(clean M)Vref (Eq. 2.19)

where ‘ads’ refers to an adsorbate, V is a potential of interest, M is the catalyst being

studied (the XAS data would be collected at an edge that is characteristic of element M)

and Vref is the reference potential at which the catalyst surface is free from any

100

adsorbates. Such a calculation would ideally cancel out all of the bulk absorption and

reflect only changes due to the effect of adsorbate on the catalyst.

The experimental difference spectra by themselves do not reveal much information.

In order to understand and correctly interpret these experimental ∆μ-XANES curves,

corresponding theoretical difference spectra of adsorbates in various positions on the

catalyst have to be calculated using the FEFF 8.0 code. Unlike earlier versions of FEFF,

version 8.0 was the first code to carry out real space, full multiple scattering calculations

using muffin-tin potentials, perform a Hedin-Lundqvist exchange correlation

approximation and determine the Fermi-level and charge-transfer effects using self-

consistent field theory. A core-hole may be included on the absorber atom to simulate the

final-state of the absorber atom during the x-ray absorption event, but this is not always

desirable or necessary. When the valence band is highly occupied, these electrons may

screen that core hole (eliminating its effects), so then better agreement with experiment is

obtained without the core hole, as is the case for Pt.1 (see footnote)

A bare Pt6 cluster is used to calculate a ‘reference’ spectrum which results in a

XANES spectrum for a ‘clean’ cluster. Following this, adsorbates of interest (H, OH, CO

etc.) are placed on the cluster in various adsorption geometries. The most common

adsorption geometries adopted tend to be adsorbates in atop, bridge-bonded and three-

fold sites, as shown in Figure 2.10. A second calculation is then made for this cluster

containing adsorbate which results in another XANES spectrum for the cluster at

potential V. A theoretical difference spectrum (∆μtheoretical) is calculated as -

1 All the data analyzed in this dissertation were collected only on two edges: the Pt L3 edge and the Ru K edge and as such, most of the discussion will pertain to these two metals but can be generalized to data collected on any element, at any edge.

101

∆μtheoretical = μ(ads/Pt6) – μ(Pt6) (Eq. 2.20)

The two sets of ∆μ curves as described in Equations 2.19 and 2.20 are then compared

to correctly interpret the experimental data along with supporting evidence from the

electrochemistry as well as other spectroscopic techniques.

Many adsorbates seem to have Δμ-XANES spectra that are characteristic of a

particular binding site. This site-dependence of the ∆μ-XANES spectra can be understood

by expanding the ∆μ in terms of its’ components. From Equation 2.16, we have μ(E) =

μ0(E)[1 + χ(E)] and deriving an expression for ∆μ(E) leads to the expression –

∆μ = ∆μ0 + ∆(μ0 χ Pt-Pt) + μ0, ad/Pt χ Pt-ad (Eq. 2.21)

where ∆μ0 is the ‘atomic’ XAFS due to adsorbate coverage, ∆(μ0 χ Pt-Pt) is the change

in Pt-Pt scattering induced due to adsorption of a species or adsorbate, μ0, ad/Pt is the free

atom absorption in the presence of an adsorbate, and finally, χ Pt-ad is the additional

scattering between the adsorbate and platinum. Theoretical calculations using FEFF 8.0

reveal that the second term, i.e. the change in Pt-Pt scattering dominates over the

contributions of the other terms to the right, and is responsible for the binding-site

sensitivity of the ∆μ curves.

To obtain an intuitive picture of this dependence, note what happens to the underlying

Pt-Pt scattering due to adsorption of a species in any site on the surface (Figure 2.11).

102

Figure 2.10 Common adsorption site geometries seen for many small molecule adsorbates

Plain Pt cluster

1-fold, atop site

2-fold, bridge-bonded site3-fold site

103

Figure 2.11 Adsorbate-induced redistribution of electronic charge on substrate metal atoms closest to the adsorbate. Note that the adsorption reduces the electron density between the surface Pt atoms.

Plain Pt Pt cluster with adsorbate

δ -

δ +

104

The adsorption process is bound to redistribute the electronic charge around the substrate

atom, thus effectively reducing the electronic density between the substrate metal atom

and its’ neighboring atoms. This in turn affects their photoabsorption cross sections,

which lead to changed backscattering properties. Further, due to increased electron-

sharing between the adsorbate and the substrate atom, bridge-bonded and 3-fold bonded

adsorption is generally slightly stronger than the bond in an atop site on a (111) surface.

We now see how the binding site affects the scattering between the substrate

metal atom and its’ neighbors and thus, arrive at the site-dependence of the ∆μ-XANES

spectra. Conversely, it then follows that the binding site of adsorbates (from energy of ∆μ

features) and the relative coverage of adsorbates (from the ∆μ magnitudes) on the surface

may be obtained from a careful analysis of the ∆μ-XANES spectra. If the data quality is

very good, even quantitative estimates may be made from the data as was done in the

case of Run+/Pt adsorption, wherein the coverage estimates were in good agreement with

other reports in the literature (chapter 5).

2.6 In situ vs. in operando XAS on electrodes and electrocatalysts – a literature

review

It must be clarified at the outset that the prefix ‘in situ’ is typically used to indicate any

study under the environment or conditions existing for the system being studied. The

term in operando is usually employed to indicate actual studies of reactions, the latter to

distinguish it from the term in situ. Thus in the in situ case, such as existing in an

electrochemical cell, a current might not be flowing because the system after some initial

105

time has reached steady-state. In a fuel cell, current can continuously flow as reactants

are continuously provided, so this is referred to as in operando. However, one must also

recognize that the Δμ-XANES technique does not see temporal reactants (i.e., those

coming on and leaving in a relatively short “turn over” time) on the surface, only more

long term adsorbates that might be referred to as poisons or excess reactants or products.

Thus the differences between in situ and in operando results are rather subtle, and

possibly may even be hard to predict or see. The in situ results are controlled more by the

thermodynamics, while the in operando results will be determined by the dynamic

kinetics of the reaction under study. Careful comparisons of in situ and in operando

results for O2 reduction at the anode in a fuel cell (in operando), vs. in the absence of

current in O2 saturated electrolyte (in situ) have shown significant differences in OH

coverage due to these differences.29

In the field of heterogeneous catalysis, there has been a number of in situ XAS

studies of gases adsorbed onto catalysts and effects of various reactions conditions such

as temperature, pressure, gas composition etc. In liquid medium, in situ studies have been

carried out by varying solvent, pH, concentration etc. to study homogeneous catalysts by

measuring absorption intensities, metal-ligand distances, coordination numbers and the

like. In case of a solid-liquid interface, extensive studies on the sorption of ions onto

minerals and other natural surfaces, again under various conditions of pH, temperature,

ion concentration, etc. have been undertaken. There have also been several studies using

in situ XAS to study all forms of electrochemical processes including the behavior of ion-

exchange materials and corrosion in common metals and alloys. For sake of relevance,

this brief review of the literature will be restricted to in situ XAS studies of batteries,

106

half-cells and fuel cells, and will always imply the existence of a solid-liquid interface

and occurring under realistic operating conditions. Given that most of the research

discussed here only pertains to in situ XAS studies, the use of the term ‘in situ XAS’ will

be avoided whenever possible to avoid redundancy in the language, hopefully enabling

an easier reading of the section. Finally, the literature survey in this section is intended to

be primarily illustrative and not exhaustive and is presented in loose chronological order.

The application of XAS to study electrode materials under realistic, operating

conditions occurred first in the early 1990s. This is due to the fact that much research was

being carried out on fuel cells and advanced batteries as alternate energy sources, and

more importantly, synchrotron sources were becoming increasingly accessible to

researchers. It is interesting to note that XAS has been applied to study species under

potential control since the mid-1980s, the earliest studies of which were carried out by

Heineman et al. to study a particular species that would otherwise be subject to reduction

by prolonged exposure to the photon beam; 90 they also used in situ XAS to study

transition metal ions in solution using a specially designed spectroelectrochemical cell.91

In an article titled “Is there any beam yet? Use of synchrotron radiation in the in situ

study of electrochemical interfaces”, Abruna and co-workers at Cornell foresaw that in

situ XAS could be expected to play an important role in understanding the solid-liquid

electrolyte interface, which was just beginning to be explored using various spectroscopic

techniques available at synchrotron facilities. Abruna, Samant and co-workers carried out

the first series of studies of adsorbed ions such as bromide and chloride on single crystals,

and adsorbed upd layers of Cu, Ag and Pb on gold, silver and platinum single crystals

using XAS, grazing-incidence XAFS (GI-XAFS) and x-ray standing wave (XRSW)

107

spectroscopy.92-99 Following this, various studies that used simple in situ

spectroelectrochemical cell designs were also reported.100-102 Enough momentum had

picked up in the field by 1990 for a review article entitled ‘EXAFS

spectroelectrochemistry’ to appear.103 It was around this time that several groups began to

use XAS to study electrode materials for use in batteries and fuel cells (ethanol oxidation

in alkaline media) chiefly containing nickel and cobalt.104-107 Yoshitake et al. carried out

one of the earliest in situ XAS studies on Pt/C nanoparticles and reported systematic

changes with the potential in the XANES region.108 Dan Scherson’s group carried out

studies on nickel-based electrodes in strong alkaline electrolytes to determine the effect

of charging and discharging on the d-band vacancy of materials,109 and carried out a

similar study in an operating alkaline battery.110 Their group also reported observing x-

ray induced photocurrents, but we do not have any evidence that it is significant enough

to disturb any in situ spectroelectrochemical experiments.111 The first sustained efforts

towards using in situ XAS routinely to study battery electrodes and fuel cell catalysts

under realistic conditions similar to actual operating environments were carried out most

notably by O’Grady, McBreen, Mukerjee, Pandya, Mansour Ramaker and others.112-124

Through these studies, it was established beyond doubt that minute changes in d-band

occupancy of the investigated metals, either due to alloying, effects of potential, or

adsorbed species, were reflected in the shapes and intensity of the white-lines observed in

the XANES regions of the XAS spectra. Further, the EXAFS region provided structural

information about electrode materials under operating conditions, information that cannot

be obtained by most other techniques. O’Grady and Ramaker soon found systematic

chemical effects in the ‘Atomic EXAFS’ region of the in situ absorption spectra of PtRu

108

alloys. These changes were attributed to small electrode potential-induced changes in the

potential-well of the absorbing atom itself, thereby causing a variation in the atomic

EXAFS.125 The presence of adsorbed CO on Pt/C electrocatalysts was indicated using the

‘difference file method’ on in situ data by Russell and co-workers. This study was one of

the first reports of the direct observation of small-molecule adsorbates in presence of a

liquid electrolyte using XAS.126

Ramaker and Mukerjee have since investigated several electrocatalysts used in fuel

cells using in situ XAS. The first application of the ∆μ-XANES method developed by

Ramaker and Koningsberger to study adsorbates in a half-cell i.e. in the presence of an

electrolyte (0.1 M HClO4) was in 2004. Around this time, another review paper titled ‘X-

ray Absorption Spectroscopy of Low Temperature Fuel Cell Catalysts’ was published,127

making apparent the utility of XAS as a technique to study electrocatalysts regardless of

it being used ex situ or in situ.

Using the ∆μ-XANES method on Pt/C electrocatalysts in 0.1 M HClO4, Teliska et al.

were able to identify adsorbed hydrogen and showed that at various potentials, the

hydrogen may occupy either atop (weakly-bonded, upd hydrogen) or n-fold adsorbed

sites (at very low potentials). The adsorption lineshapes found for hydrogen on Pt were

also found to be similar to those found in the gas phase.89 The similarity of the ∆μ-

XANES signatures in both, gas and liquid media greatly enhanced the credibility of the

technique as a truly surface sensitive technique. This was followed with a study on the O

and OH adsorption under similar conditions.31

In situ XAS experiments on PtRu anode catalysts in an operating fuel cell have also

been carried out by Smotkin, Segre and co-workers. Their studies on a reformate-air fuel

109

cell under various operating conditions revealed that the PtRu phase was metallic under

normal reducing conditions of an operating fuel cell.128 In another study on similar

catalysts in an operating DMFC, no evidence for phase segregation was found during

operation and an EXAFS analysis revealed a very short Ru-O bond length that was

significantly shorter than that typically seen in oxides of ruthenium, suggesting the

presence of an oxide phase of a different kind.129

The first operando studies in a specially-designed PEM fuel cell was reported by

Roth et al. in which the OH and CO coverage on the electrodes were monitored using the

∆μ-XANES method as the fuel cell was discharged.130

Hwang et al., in a study on Pt/C nanoparticle catalysts in 0.1 M HClO4, reported

changes in coordination number as a function of electrode potential, suggesting that the

nanoparticles suffered some form of aggregation during cycling. Our own studies have

repeatedly confirmed this finding; it has also been suggested that these changes may also

be due to shape changes that are induced due to strong chemisorption and desorption

processes occurring on nanoparticles at various electrode potentials.131-133 Such dynamic

changes in nanoparticle shapes, observed through in situ spectroscopic and microscopic

techniques, have been receiving a lot of attention lately as they are no doubt, expected to

have significant effects upon the surface morphology and electronic state,134-141 and

therefore have important consequences for catalysis. Further, repeated changes in shape

during a catalytic reaction may lead to increased rates of degradation, a very real concern

for any catalyst that is to be widely employed in low-temperature fuel cells.

The ∆μ-XANES method has been used to study several electrocatalytic systems

including novel electrocatalysts such as RhxSy chalcogenide materials,142-144 observing

110

CO coverage on Pt/C electrocatalysts,32, 145 probing the nature of electrochemical activity

of novel supports used in catalysts such as Au/SnOx,132 understanding water activation on

ligand-stabilized Pt nanoparticles,133 the poisoning of Pt/C electrocatalysts by halide

adsorption, specifically chlorides,146 (chapter 3) studying the aging of PtRu black alloy

catalysts under effects of potential cycling,131 (chapter 4) and much of the work that

constitutes the rest of the thesis. The work described in chapter 4 also builds on a study

by Holstein and Rosenfeld who studied changes occurring on Pt and Ru using in situ

XAS.147 We noticed that they had cycled the catalysts to 1.36 V (vs RHE) and at which

significant oxidative loss is expected to occur. Further, they found that Pt loses its activity

at higher potentials chiefly due to oxidation. They further found that not only do surface

Ru atoms supply OH species for the oxidation of methanol on Pt sites, they also aid in

preventing Pt from excessive oxidation and thus, enhance the overall activity of the alloy

catalyst for methanol oxidation. We were curious to see if any signs of degradation could

be found due to potential cycling even under milder conditions i.e. by cycling the

catalysts only up to around 0.84 V (vs. RHE). Interestingly enough, an EXAFS analysis

revealed statistically significant changes even after 20 and 40 cycles. Using the ∆μ-

XANES method, it was shown that the aging processes on the two different commercial

catalysts occur via slightly different processes.

Several studies on anode and cathode catalysts using in situ and in operando XAS as

applied to fuel cells have continued to appear, many of them having been very recently

published and will not be reviewed here.148-153 It is quite safe to assume that research in

this field of study will be crucial towards the development of fuel cells and indeed, in any

field that employs catalysts.

111

2.7 References

1. Korszun, Z. R., Moffat, K., Frank, K. & Cusanovich, M. A. Extended x-ray absorption fine structure (EXAFS) studies of cytochromes c: structural aspects of oxidation-reduction. Biochemistry 21, 2253-2258 (1982).

2. Scott, R. A. X-Ray Absorption Spectroscopic Investigations of Cytochrome c Oxidase Structure and Function. Annual Review of Biophysics and Biophysical Chemistry 18, 137-158 (1989).

3. George, G., Prince, R. & Cramer, S. The manganese site of the photosynthetic water-splitting enzyme. Science 243, 789-791 (1989).

4. Roelofs, T. A. et al. Oxidation states of the manganese cluster during the flash-induced S-state cycle of the photosynthetic oxygen-evolving complex. Proceedings of the National Academy of Sciences of the United States of America 93, 3335-3340 (1996).

5. Yano, J. et al. Where Water Is Oxidized to Dioxygen: Structure of the Photosynthetic Mn4Ca Cluster. Science 314, 821-825 (2006).

6. Pushkar, Y., Yano, J., Sauer, K., Boussac, A. & Yachandra, V. K. Structural changes in the Mn4Ca cluster and the mechanism of photosynthetic water splitting. Proceedings of the National Academy of Sciences 105, 1879-1884 (2008).

7. Cinco, R. M. et al. Comparison of the Manganese Cluster in Oxygen-Evolving Photosystem II with Distorted Cubane Manganese Compounds through X-ray Absorption Spectroscopy. Inorganic Chemistry 38, 5988-5998 (1999).

8. Yano, J. & Yachandra, V. K. Where Water Is Oxidized to Dioxygen: Structure of the Photosynthetic Mn4Ca Cluster from X-ray Spectroscopy. Inorganic Chemistry 47, 1711-1726 (2008).

9. Smith, J. D. et al. Energetics of Hydrogen Bond Network Rearrangements in Liquid Water. Science 306, 851-853 (2004).

10. Head-Gordon, T. & Johnson, M. E. Tetrahedral structure or chains for liquid water. 103, 7973-7977 (2006).

11. Wernet, P. et al. The Structure of the First Coordination Shell in Liquid Water. Science 304, 995-999 (2004).

12. Evans, J. Brilliant opportunities across the spectrum. Physical Chemistry Chemical Physics 8, 3045-3058 (2006).

13. Bunker, G. Tutorials on XAFS. http://gbxafs.iit.edu/training/tutorials.html.

112

14. van der Veen, R. M. et al. Structural determination of photochemically active diplatinum molecule by time-resolved EXAFS spectroscopy. Angewandte Chemie International Edition 48, 2711-2714 (2009).

15. Bressler, C. et al. Femtosecond XANES Study of the Light-Induced Spin Crossover Dynamics in an Iron(II) Complex. Science 323, 489-492 (2009).

16. Gawelda, W. et al. Structural analysis of ultrafast extended x-ray absorption fine structure with subpicometer spatial resolution: Application to spin crossover complexes. The Journal of Chemical Physics 130, 124520 (2009).

17. Pettifer, R. F., Mathon, O., Pascarelli, S., Cooke, M. D. & Gibbs, M. R. J. Measurement of femtometre-scale atomic displacements by X-ray absorption spectroscopy. Nature 435, 78-81 (2005).

18. Bressler, C. & Chergui, M. Molecular Structural Dynamics Probed by Ultrafast X-Ray Absorption Spectroscopy. Annual Review of Physical Chemistry 61 (2010).

19. Chergui, M. & Zewail, A. H. Electron and X-Ray Methods of Ultrafast Structural Dynamics: Advances and Applications. ChemPhysChem 10, 28-43 (2009).

20. Christian Bressler, R. A., Majed Chergui. Exploiting EXAFS and XANES for time-resolved molecular structures in liquids. Photocrystallography 223, 307-321 (2008).

21. Bressler, C. & Chergui, M. Ultrafast X-ray Absorption Spectroscopy. Chemical Reviews 104, 1781-1812 (2004).

22. Thorsten Ressler, J. W., Rolf E. Jentoft, Thomas Neisius, Marco M. Gunter. Kinetics of solid-state reactions in heterogeneous catalysis from time-resolved X-ray absorption spectroscopy. Topics in Catalysis 18, 45-52 (2002).

23. Mark A. Newton, A. J. D., John Evans. Bringing time resolution to EXAFS: recent developments and application to chemical systems. Chemical Society Reviews 31, 83-95 (2002).

24. Eswaramoorthy, S. K., Howe, J. M. & Muralidharan, G. In Situ Determination of the Nanoscale Chemistry and Behavior of Solid-Liquid Systems. Science 318, 1437-1440 (2007).

25. Hanson, K. J., Gibson, J. M. & McDonald, M. L. A Solid-State Thin Film Cell for In Situ Transmission Electron Microscopy of Electrodeposited Silver on Gold. Journal of The Electrochemical Society 136, 2214-2218 (1989).

26. M. Dollé, S. Grugeon, B. Beaudoin, L. Dupont, J-M. Tarascon. In situ TEM study of the interface carbon/electrolyte. Journal of Power Sources 97-98, 104-106 (2001).

113

27. Arai, S., Tsukimoto, S. & Saka, H. In Situ Transmission Electron Microscope Observation of Melting of Aluminum Particles. Microscopy and Microanalysis 4, 264-268 (1998).

28. J. M. Howe, S. K. E., G. Muralidharan. Determining the nanoscale chemistry and behavior of interfaces and phases in Al-Si(-Cu-Mg) nanoparticles using in-situ TEM. EMC 2008 14th European Microscopy Congress 1–5 September 2008, Aachen, Germany M3, 633-634 (2008).

29. Christina Roth, D. R. XAS Investigations of PEM Fuel Cells. Modern Aspects of Electrochemistry (Ed. C. Vayenas) (2008).

30. Ramaker, D. E. & Koningsberger, D. C. Comment on "Effect of Hydrogen Adsorption on the X-Ray Absorption Spectra of Small Pt Clusters". Physical Review Letters 89, 139701 (2002).

31. Teliska, M., O'Grady, W. E. & Ramaker, D. E. Determination of O and OH Adsorption Sites and Coverage in Situ on Pt Electrodes from Pt L23 X-ray Absorption Spectroscopy. The Journal of Physical Chemistry B 109, 8076-8084 (2005).

32. Scott, F. J., Roth, C. & Ramaker, D. E. Kinetics of CO Poisoning in Simulated Reformate and Effect of Ru Island Morphology on PtRu Fuel Cell Catalysts As Determined by Operando X-ray Absorption Near Edge Spectroscopy. The Journal of Physical Chemistry C 111, 11403-11413 (2007).

33. M. Teliska, W. E. O. G., D.E. Ramaker. Determination of H Adsorption Sites on Pt/C Electrodes in HClO4 from Pt L23 X-ray Absorption Spectroscopy. The Journal of Physical Chemistry B 108, 2333 (2004).

34. Scofield, J. H. Theoretical Photoionization Cross Sections from 1 to 1500 keV. Lawrence Livermore National Laboratory Report UCRL - 51326 (1973).

35. Hubbell, J. R. Photon Cross Sections, Attenuation Coefficients, and Energy Absorption Coefficients from 10 keV to 100 GeV. Report NSRDS-NBS 29 (National Bureau of Standards, Washington DC.) (1969).

36. W.H. McMaster, N. K. D. G., J.H. Mallet, J.H. Hubbell. Lawrence Livermore National Laboratory Report UCRL - 50174 Section II Revision I (1969).

37. B.L. Henke, E. M. G., J.C. Davis Atomic Data and Nuclear Data Tables. 54, 181 (1993).

38. W.T. Elam, B. R., J.R. Sieber. A New Atomic Database for X-ray Spectroscopic Calculations. Radiation Physics and Chemistry 63, 121 (2002).

39. A. Shaltout, H. E., R. Svagera. Update of photoelectric absorption coefficients in the tables of McMaster. X-ray Spectrometry 35, 52-56 (2005).

114

40. http://www4.nau.edu/microanalysis/microprobe/BulkAbsorp.html.

41. Newville, M. IFEFFIT : interactive XAFS analysis and FEFF fitting. Journal of Synchrotron Radiation 8, 322-324 (2001).

42. Ravel, B. & Newville, M. ATHENA, ARTEMIS, HEPHAESTUS: data analysis for X-ray absorption spectroscopy using IFEFFIT. Journal of Synchrotron Radiation 12, 537-541 (2005).

43. PIPS is a registered trademark of Canberra Industries, Inc.

44. F.W. Lytle, R. B. G., D.R. Sandstrom, D.C. Marques, J. Wong, C.L. Spiro, G.P. Huffman, F.E. Huggins. Nuclear Instruments and Methods in Physics Research, Section A 226, 542 (1984).

45. Teo, B. K. EXAFS Spectroscopy: Basic principles and data analysis. Springer-Verlag, New York (1986).

46. D.C. Koningsberger, R. P. X-ray Absorption, Principles, techniques of EXAFS, SEXAFS and XANES. John Wiley & Sons, New York, ISBN 0-471-87547-3 (1988).

47. Roentgen, W. C. On a new kind of rays. Sitzungsberichte der Physikalisch-medizinischen Gesellschaft zu Wurzburg (1895).

48. Fricke, H. The K-Characteristic Absorption Frequencies for the Chemical Elements Magnesium to Chromium. Physical Review 16, 202-215 (1920).

49. Kossel, W. Zum Bau der Röntgenspektren. Zeit. Phys. 1, 119-134 (1920).

50. Kronig, R. d. L. Zur Theorie der Feinstruktur in den Röntgenabsorptionsspektren. Zeit. Phys. 70, 317-323 (1931).

51. Kronig, R. d. L. Zur Theorie der Feinstruktur in den Röntgenabsorptionsspektren II Zeit. Phys. 75, 191-210 (1932).

52. Kronig, R. d. L. Zur Theorie der Feinstruktur in den Röntgenabsorptionsspektren. III Zeit. Phys. 75, 468-475 (1932).

53. Prins, J. Transitions to Optical Levels in the Argon L X-Ray Absorption Spectrum. Nature 133, 795-796 (1934).

54. Hanawalt, J. D. The Dependence of X-ray Absorption Spectra upon Chemical and Physical State. Physical Review 37, 715-726 (1931).

55. Hayasi, T. Sci. Rep. Tohoku Univ. Ser. 1, 123-132 (1949).

56. Kostarev, I. Zh. Eksperim. Teor. Fiz. 11, 60-73 (1941).

115

57. Kozlenkov, A. Bull. Acad. Sci. USSR, Phys. Ser. 25, 968-987 (1961).

58. M. Sawada, K. T., T. Shiraiwa, M. Obashi. On the Fine Structures of X-ray Absorption Spectra of Amorphous Substances The Amorphous State of the Binary System of Nickel-Sulfur. II. J. Phys. Soc. Jpn. 10, 464-468 (1955).

59. M. Sawada, K. T., T. Shiraiwa, M. Obashi. Annual Rep. Sci. Works Osaka University 7, 1-87 (1959).

60. L. Parratt, L. J. X-Ray Spectroscopy of the Solid State: Potassium Chloride. Physical Review 97, 916-926 (1955).

61. Parratt, L. Electronic Band Structure of Solids by X-Ray Spectroscopy. Rev. Mod. Phys. 31, 616-645 (1959).

62. Nordstrand, R. V. The Use of X-Ray K-Absorption Edges in the Study of Catalytically Active Solids. Advances in Catalysis 12, 149-187 (1960).

63. Nordstrand, R. V. Handbook of X-rays (Edited by E. Kaelble) McGraw-Hill 43, 1-18 (1967).

64. J. Perel, R. D. D. Extended Fine Structure in X-Ray Absorption Spectra of Certain Perovskites. Physical Review B 2, 1317-1323 (1970).

65. Stern, E. A. Dale Sayers Festschrift. Proceedings of the X-Ray Absorption Fine Structure 13th International Conference (XAFS-13), 9-14 July (2006).

66. D. Sayers, F. L., E. Stern. Bulletin of the American Physical Society 16, 302 (1971).

67. Sayers, D. A New Technique to Determine Amorphous Structure Using Extended X-ray Absorption Fine Structure. BSRL Doc. D180-14436-1, Submitted to the University of Washington in partial fulfillment of the degree of Doctor of Philosophy (1971).

68. D. Sayers, F. L., E. Stern. New Technique for Investigating Noncrystalline Structures: Fourier Analysis of the Extended X-Ray Absorption Fine Structure. Phys. Rev. Lett. 27, 1204-1207 (1971).

69. E. Stern, D. S., F. Lytle. Extended x-ray absorption fine structure technique II. Experimental practice and selected results. Phys. Rev. B 11, 4825-4835 (1975).

70. E. Stern, D. S., F. Lytle. Extended x-ray absorption fine structure technique. III. Determination of physical parameters Phys. Rev. B 11, 4836-4846 (1975).

71. Stern, E. Musings about the development of XAFS. Journal of Synchrotron Radiation 8, 49-54 (2001).

116

72. Lytle, F. The EXAFS family tree: a personal history of the development of extended X-ray absorption fine structure. Journal of Synchrotron Radiation 6, 123-134 (1999).

73. Dirac, P. A. M. The Principles of Quantum Mechanics. Oxford University Press (1982).

74. V.L. Aksenov, M. V. K. c., A.Yu. Kuz'min, Yu. Purans, S.I. Tyutyunnikov. Development of Methods of EXAFS Spectroscopy on Synchrotron Radiation Beams: Review. Crystallography Reports 51, 908-935 (2006).

75. Fonda, L. Multiple-scattering theory of X-ray absorption: a review. Journal of Physics: Condensed Matter, 8269 (1992).

76. Stern, E. A. & Rehr, J. J. Many-body aspects of the near-edge structure in x-ray absorption. Physical Review B 27, 3351 (1983).

77. Mustre, J., Yacoby, Y., Stern, E. A. & Rehr, J. J. Analysis of experimental extended x-ray-absorption fine-structure (EXAFS) data using calculated curved-wave, multiple-scattering EXAFS spectra. Physical Review B 42, 10843 (1990).

78. Rehr, J. J., Mustre de Leon, J., Zabinsky, S. I. & Albers, R. C. Theoretical x-ray absorption fine structure standards. Journal of the American Chemical Society 113, 5135-5140 (1991).

79. Zabinsky, S. I., Rehr, J. J., Ankudinov, A., Albers, R. C. & Eller, M. J. Multiple-scattering calculations of x-ray-absorption spectra. Physical Review B 52, 2995 (1995).

80. Mustre de Leon, J., Rehr, J. J., Zabinsky, S. I. & Albers, R. C. Ab initio curved-wave x-ray-absorption fine structure. Physical Review B 44, 4146 (1991).

81. Rehr, J. J. & Albers, R. C. Scattering-matrix formulation of curved-wave multiple-scattering theory: Application to x-ray-absorption fine structure. Physical Review B 41, 8139 (1990).

82. Ravel, B. ATOMS: crystallography for the X-ray absorption spectroscopist. Journal of Synchrotron Radiation 8, 314-316 (2001).

83. Hennig, C. et al. EXAFS as a tool for bond-length determination in the environment of heavy atoms. Journal of Synchrotron Radiation 8, 695-697 (2001).

84. Maeda, H. Accurate Bond Length Determination by EXAFS Method. Journal of the Physical Society of Japan 56, 2777.

85. Arcovito, A. et al. X-ray structure analysis of a metalloprotein with enhanced active-site resolution using in situ x-ray absorption near edge structure

117

spectroscopy. Proceedings of the National Academy of Sciences 104, 6211-6216 (2007).

86. Randaccio, L. & Geremia, S. Comparison of the Coordination Distances Derived by Extended X-ray Absorption Fine Structure and X-ray Crystallography in a Vitamin B12 Model. Organometallics 16, 4951-4953 (1997).

87. A. Levina, R. S. A., P.A. Lay. Three-dimensional structure determination using multiple-scattering analysis of XAFS: applications to metalloproteins and coordination chemistry. Coordination Chemistry Reviews 249, 141-160 (2005).

88. Dalba, G. & Fornasini, P. EXAFS Debye-Waller Factor and Thermal Vibrations of Crystals. Journal of Synchrotron Radiation 4, 243-255 (1997).

89. Teliska, M., O'Grady, W. E. & Ramaker, D. E. Determination of H Adsorption Sites on Pt/C Electrodes in HClO4 from Pt L23 X-ray Absorption Spectroscopy. The Journal of Physical Chemistry B 108, 2333-2344 (2004).

90. Smith, D. A., Elder, R. C. & Heineman, W. R. Extended x-ray absorption fine structure thin-layer spectroelectrochemistry. Analytical Chemistry 57, 2361-2365 (1985).

91. Dewald, H. D., Watkins, J. W., Elder, R. C. & Heineman, W. R. Development of extended x-ray absorption fine structure spectroelectrochemistry and its application to structural studies of transition-metal ions in aqueous solution. Analytical Chemistry 58, 2968-2975 (1986).

92. Abruna, H. D. et al. Is there any beam yet? Uses of synchrotron radiation in the in situ study of electrochemical interfaces. The Journal of Physical Chemistry 92, 7045-7052 (1988).

93. Bommarito, G. M., White, J. H. & Abruna, H. D. Electrosorption of iodide on platinum: packing density and potential-dependent distributional changes observed in situ with x-ray standing waves. The Journal of Physical Chemistry 94, 8280-8288 (1990).

94. Yee, H. S. & Abruna, H. D. In-situ x-ray studies of the underpotential deposition of copper on platinum(111). The Journal of Physical Chemistry 97, 6278-6288 (1993).

95. Yee, H. S. & Abruna, H. D. In situ x-ray absorption spectroscopy studies of copper underpotentially deposited in the absence and presence of chloride on platinum (111). Langmuir 9, 2460-2469 (1993).

96. L. Blum, H. D. A., J.H. White, M.J. Albarelli, J.G. Gordon, G. Borges, M. Samant, O.R. Melroy. Journal of Chemical Physics 85, 6732 (1986).

118

97. O.R. Melroy, M. G. S., G.C. Borges, J.G. Gordon, L. Blum, J.H. White, M.J. Albarelli, M. McMillan, H.D. Abruna. Langmuir 4, 728 (1988).

98. White, J. H. et al. Surface extended x-ray absorption fine structure of underpotentially deposited silver on gold(111) electrodes. The Journal of Physical Chemistry 92, 4432-4436 (1988).

99. Samant, M. G., Borges, G. L., Gordon, J. G., Melroy, O. R. & Blum, L. In situ surface extended x-ray absorption fine structure spectroscopy of a lead monolayer at a silver(111) electrode/electrolyte interface. Journal of the American Chemical Society 109, 5970-5974 (1987).

100. Villain, F., Briois, V., Castro, I., Helary, C. & Verdaguer, M. Multipurpose x-ray absorption cell. Analytical Chemistry 65, 2545-2548 (1993).

101. Nicola R.S. Farley, S. J. G., A. Robert Hillman. Simple cell for in situ X-ray absorption spectroelectrochemistry. Electrochemistry Communications 1, 449-452 (1999).

102. Igo, D. H., Elder, R. C., Heineman, W. R. & Dewald, H. D. Bulk-electrolysis flow-cell system for UV-visible and x-ray absorption spectroelectrochemical analysis. Analytical Chemistry 63, 2535-2539 (1991).

103. Sharpe, L. R., Heineman, W. R. & Elder, R. C. EXAFS spectroelectrochemistry. Chemical Reviews 90, 705-722 (1990).

104. N.R.S. Farley, S. J. G., A.R. Hillman. Dynamic EXAFS study of discharging nickel hydroxide electrode with non-integer Ni valency. Electrochimica Acta 46, 3119-3127 (2001).

105. A.J. Kropf, H. T., C.S. Johnson, J.T. Vaughey, M.M. Thackeray. An in situ X-ray absorption spectroscopy study of InSb electrodes in lithium batteries. Electrochemistry Communications 3, 244-251 (2001).

106. Kim, J.-W. & Park, S.-M. In Situ XANES Studies of Electrodeposited Nickel Oxide Films with Metal Additives for the Electro-Oxidation of Ethanol. Journal of The Electrochemical Society 150, E560-E566 (2003).

107. Pandya, K. I., Hoffman, R. W., McBreen, J. & O'Grady, W. E. In Situ X-Ray Absorption Spectroscopic Studies of Nickel Oxide Electrodes. Journal of The Electrochemical Society 137, 383-388 (1990).

108. Yoshitake, H., Yamazaki, O. & Ota, K.-i. In Situ X-Ray Absorption Fine Structure Study on Structure Transformation and Electronic State of Various Pt Particles on Carbon Electrode. Journal of The Electrochemical Society 141, 2516-2521 (1994).

119

109. Tryk, D. A. et al. In Situ X-Ray Absorption Fine Structure Measurements of LaNi[sub 5] Electrodes in Alkaline Electrolytes. Journal of The Electrochemical Society 142, 824-828 (1995).

110. Tryk, D. A. et al. In Situ La, Ce, and Nd L-Edge X-Ray Absorption Fine Structure Study of an Intermetallic Metal Hydride Electrode in an Operating Alkaline Battery. Journal of The Electrochemical Society 142, L76-L78 (1995).

111. Tolmachev, Y. V., Bae, I. T. & Scherson, D. A. X-ray Induced Photocurrents at the Metal/Solution Interface. The Journal of Physical Chemistry B 104, 7663-7667 (2000).

112. McBreen, J. et al. In situ time-resolved x-ray absorption near edge structure study of the nickel oxide electrode. The Journal of Physical Chemistry 93, 6308-6311 (1989).

113. Mukerjee, S., Srinivasan, S., Soriaga, M. P. & McBreen, J. Role of Structural and Electronic Properties of Pt and Pt Alloys on Electrocatalysis of Oxygen Reduction. Journal of The Electrochemical Society 142, 1409-1422 (1995).

114. Mukerjee, S. & McBreen, J. Hydrogen Electrocatalysis by Carbon Supported Pt and Pt Alloys. Journal of The Electrochemical Society 143, 2285-2294 (1996).

115. Yoon, W.-S. et al. Investigation of the Local Structure of the LiNi[sub 0.5]Mn[sub 0.5]O[sub 2] Cathode Material during Electrochemical Cycling by X-Ray Absorption and NMR Spectroscopy. Electrochemical and Solid-State Letters 5, A263-A266 (2002).

116. Mansour, A. N., Smith, P. H., Baker, W. M., Balasubramanian, M. & McBreen, J. A Comparative In Situ X-Ray Absorption Spectroscopy Study of Nanophase V[sub 2]O[sub 5] Aerogel and Ambigel Cathodes. Journal of The Electrochemical Society 150, A403-A413 (2003).

117. Mansour, A. N., Smith, P. H., Baker, W. M., Balasubramanian, M. & McBreen, J. A Comparative In Situ X-Ray Absorption Spectroscopy Study of Nanophase [bold V][sub 2][bold O][sub 5] Aerogel and Ambigel Cathodes [Journal of The Electrochemical Society, [bold 150], A403 (2003)]. Journal of The Electrochemical Society 150, L13 (2003).

118. Yoon, W.-S. et al. Combined NMR and XAS Study on Local Environments and Electronic Structures of Electrochemically Li-Ion Deintercalated Li[sub 1 - x]Co[sub 1/3]Ni[sub 1/3]Mn[sub 1/3]O[sub 2] Electrode System. Electrochemical and Solid-State Letters 7, A53-A55 (2004).

119. Mansour, A. N., Smith, P. H., Balasubramanian, M. & McBreen, J. In Situ X-Ray Absorption Study of Cycled Ambigel V[sub 2]O[sub 5] [center-dot] nH[sub 2]O(n [approximate] 0.5) Composite Cathodes. Journal of The Electrochemical Society 152, A1312-A1319 (2005).

120

120. Mukerjee, S., Srinivasan, S., Soriaga, M. P. & McBreen, J. Effect of Preparation Conditions of Pt Alloys on Their Electronic, Structural, and Electrocatalytic Activities for Oxygen Reduction - XRD, XAS, and Electrochemical Studies. The Journal of Physical Chemistry 99, 4577-4589 (1995).

121. O'Grady, W. E., Hagans, P. L., Pandya, K. I. & Maricle, D. L. Structure of Pt/Ru Catalysts Using X-ray Absorption Near Edge Structure Studies. Langmuir 17, 3047-3050 (2001).

122. Pandya, K. I., O'Grady, W. E., Corrigan, D. A., McBreen, J. & Hoffman, R. W. Extended x-ray absorption fine structure investigations of nickel hydroxides. The Journal of Physical Chemistry 94, 21-26 (1990).

123. Qian, X., Sambe, H., Ramaker, D. E., Pandya, K. I. & O'Grady, W. E. Quantitative Interpretation of K-Edge NEXAFS Data for Various Nickel Hydroxides and the Charged Nickel Electrode. The Journal of Physical Chemistry B 101, 9441-9446 (1997).

124. McBreen, J., O'Grady, W. E., Pandya, K. I., Hoffman, R. W. & Sayers, D. E. EXAFS study of the nickel oxide electrode. Langmuir 3, 428-433 (1987).

125. O'Grady, W. E., Qian, X. & Ramaker, D. E. Systematic Chemical Effects Observed in “Atomic” X-ray Absorption Fine Structure. The Journal of Physical Chemistry B 101, 5624-5626 (1997).

126. Maniguet, S., Mathew, R. J. & Russell, A. E. EXAFS of Carbon Monoxide Oxidation on Supported Pt Fuel Cell Electrocatalysts. The Journal of Physical Chemistry B 104, 1998-2004 (2000).

127. Russell, A. E. & Rose, A. X-ray Absorption Spectroscopy of Low Temperature Fuel Cell Catalysts. Chemical Reviews 104, 4613-4636 (2004).

128. Viswanathan, R. et al. In-Situ XANES of Carbon-Supported Pt−Ru Anode Electrocatalyst for Reformate-Air Polymer Electrolyte Fuel Cells. The Journal of Physical Chemistry B 106, 3458-3465 (2002).

129. Stoupin, S., Chung, E.-H., Chattopadhyay, S., Segre, C. U. & Smotkin, E. S. Pt and Ru X-ray Absorption Spectroscopy of PtRu Anode Catalysts in Operating Direct Methanol Fuel Cells. The Journal of Physical Chemistry B 110, 9932-9938 (2006).

130. Roth, C. et al. Determination of O[H] and CO Coverage and Adsorption Sites on PtRu Electrodes in an Operating PEM Fuel Cell. Journal of the American Chemical Society 127, 14607-14615 (2005).

131. Shyam, B., Arruda, T. M., Mukerjee, S. & Ramaker, D. E. Effect of RuOxHy Island Size on PtRu Particle Aging in Methanol. The Journal of Physical Chemistry C 113, 19713-19721 (2009).

121

132. Gatewood, D., Ramaker, D., Sasaki, K. & Swider-Lyons, K. Establishing the Mechanism for Oxygen Reduction on Au/SnOx Using In Situ X-ray Absorption Spectroscopy. ECS Transactions 11, 271-276 (2007).

133. Gatewood, D. S. et al. Characterization of Ligand Effects on Water Activation in Triarylphosphine-Stabilized Pt Nanoparticle Catalysts by X-ray Absorption Spectroscopy. The Journal of Physical Chemistry C 112, 4961-4970 (2008).

134. Yanase Akihisa, K. H. In situ optical observation of oxygen-adsorption-induced reversible change in the shape of small supported silver particles. Surface Science 264, 147-156 (1992).

135. Zhdanov, V. P. & Norton, P. R. Surface Reconstruction and Rate Processes in Adsorbed Overlayers†. Langmuir 12, 101-108 (1996).

136. Nolte, P. et al. Shape Changes of Supported Rh Nanoparticles During Oxidation and Reduction Cycles. Science 321, 1654-1658 (2008).

137. Soon, A., Wong, L., Delley, B. & Stampfl, C. Morphology of copper nanoparticles in a nitrogen atmosphere: A first-principles investigation. Physical Review B 77, 125423 (2008).

138. N. Inoglu, J. R. K. Atomistic thermodynamics study of the adsorption and the effects of water–gas shift reactants on Cu catalysts under reaction conditions. Journal of Catalysis 261, 188-194 (2009).

139. Mark a. Newton, C. B.-C., Arturo Martinez-Arias, Marcos Fernandez-Garcia. Dynamic in situ observation of rapid size and shape change of supported Pd nanoparticles during CO/NO cycling. Nature Materials 6, 528-532 (2007).

140. Rentao Mu, Q. F., Hongyang Liu, Dali Tan, Runsheng Zhai, Xinhe Bao. Reversible surface structural changes in Pt-based bimetallic nanoparticles during oxidation and reduction cycles. Applied Surface Science 255, 7296-7301 (2009).

141. Tao, F. et al. Reaction-Driven Restructuring of Rh-Pd and Pt-Pd Core-Shell Nanoparticles. Science 322, 932-934 (2008).

142. Mukerjee, S., Ramaker, D., Gatewood, D. & Ziegelbauer, J. M. In Situ X-Ray Absorption Spectroscopy Studies of Water Activation on Novel Electrocatalysts for Oxygen Reduction Reaction in Acid Electrolyte. ECS Transactions 1, 119-128 (2006).

143. Ziegelbauer, J. M., Gatewood, D., Gulla, A. F., Ramaker, D. E. & Mukerjee, S. X-Ray Absorption Spectroscopy Studies of Water Activation on an Rh[sub x]S[sub y] Electrocatalyst for Oxygen Reduction Reaction Applications. Electrochemical and Solid-State Letters 9, A430-A434 (2006).

122

144. Ziegelbauer, J. M. et al. Fundamental Investigation of Oxygen Reduction Reaction on Rhodium Sulfide-Based Chalcogenides. The Journal of Physical Chemistry C 113, 6955-6968 (2009).

145. Scott, F. J., Mukerjee, S. & Ramaker, D. E. CO Coverage/Oxidation Correlated with PtRu Electrocatalyst Particle Morphology in 0.3 M Methanol by In Situ XAS. Journal of The Electrochemical Society 154, A396-A406 (2007).

146. Arruda, T. M., Shyam, B., Ziegelbauer, J. M., Mukerjee, S. & Ramaker, D. E. Investigation into the Competitive and Site-Specific Nature of Anion Adsorption on Pt Using In Situ X-ray Absorption Spectroscopy. The Journal of Physical Chemistry C 112, 18087-18097 (2008).


148. Wiltshire, R. J. K. et al. Channel-Flow Cell for X-ray Absorption Spectroelectrochemistry. The Journal of Physical Chemistry C 113, 308-315 (2008).

149. Hudson, S. L. et al. Probing the Structure of Operating Fuel Cell Cathode Catalysts using XAS. ECS Transactions 16, 1395-1404 (2008).

150. Richard J.K. Wiltshire, C. R. K., Abigail Rose, Peter P. Wells, Hazel Davies, Martin P. Hogarth, David Thompsett, Brian Theobald, Fredrick W. Mosselmans, Mark Roberts, Andrea Russell. Effects of composition on structure and activity of PtRu/C catalysts. Physical Chemistry Chemical Physics 11, 2305-2313 (2009).

151. Stanislav Stoupin, H. R., Zhengrong Li, Carlo U. Segre, Carol Korzeniewski, Dominick J. Casadonte Jr., Hisashi Inoue, Eugene S. Smotkin. Structural analysis of sonochemically prepared PtRu versus Johnson Matthey PtRu in operating direct methanol fuel cells. Physical Chemistry Chemical Physics 10, 6430-6437 (2008).

152. Imai, H. et al. In Situ and Real-Time Monitoring of Oxide Growth in a Few Monolayers at Surfaces of Platinum Nanoparticles in Aqueous Media. Journal of the American Chemical Society 131, 6293-6300 (2009).

153. Emiliano Principi, A. W., Sonia Dsoke, Roberto Marassi, Andrea Di Ciccio. An XAS experimental approach to study low Pt content electrocatalysts operating in PEM fuel cells. Physical Chemistry Chemical Physics 11, 9987-9995 (2009).

123

Chapter 3

An Investigation into the Competitive and Site-Specific Nature of Anion

Adsorption on Pt Using In Situ X-ray Absorption Spectroscopy∗

3.1 Introduction

Highly dispersed polycrystalline platinum nanoparticles are still considered to be the

premier electrocatalysts for use in low and medium temperature fuel cells such as

polymer electrolyte membrane fuel cells (low temperature, PEMFC) and the acid-

imbibed analogs (medium temperature PBI-Phosphoric acid based). They are also

candidates for potential applications in new hybrid concepts as reversible electrodes for

large scale energy storage using modified flow-through battery configurations. To date

they still exhibit the highest activity toward the oxygen reduction reaction (ORR) and are

relatively resistant to corrosion under standard fuel cell operating conditions. When used

as an anode, platinum demonstrates near perfect electrokinetics as the overpotential of

∗ Published in the Journal of Physical Chemistry C, 2008, 112 (46), 18087-18097 Authors: Thomas Arruda, Badri Shyam, Joseph Ziegelbauer, Sanjeev Mukerjee, and David E.Ramaker Electrochemistry data collection and analysis was carried out by authors affiliated with Northeastern University. XAS data collection and analysis were carried out by authors from The George Washington University and Northeastern University.

124

hydrogen oxidation/evolution is negligible in pure, hydrated H2 streams.1 The high

activity exhibited by Pt is also the reason for its higher susceptibility to poisoning by

species which have higher chemisorption ability. One of the significant classes of such

surface poisons are halides.

Due to the aforementioned importance of platinum and its susceptibility to poisoning,

a myriad of studies on Pt poisoning have been published over the years. Many of these

endeavors have investigated the adsorption of anions such as bisulfate and halides

revealing adverse electrochemical effects.2-9 In the presence of strongly adsorbing anions,

ORR on platinum suffers further overpotential losses of several hundred millivolts

beyond that due to the sluggish ORR kinetics in clean electrolytes.10, 11 In addition, H2O2

byproduct formation has been shown to increase in the presence of adsorbed anions.10

The presence of H2O2 in a fuel cell electrode-electrolyte interface is known to lead to

formation of free radicals such as hydroxyl and peroxy-hydroxyl which ultimately attack

the polymer electrolyte membrane, causing durability issues.12 It is generally accepted

that the adsorption strength of adsorbed halides on platinum increases in the order Cl- <

Br- < I-.5 The reversibility of halide adsorption, however, has been somewhat

controversial. Lane et al.13 suggested that halides adsorb strongly enough to withstand

rinsing with an inert electrolyte. Such irreversibility however has been contradicted by

subsequent investigations employing radiotracers to demonstrate dislodgement of

adsorbed Cl- by other halides or rinsing with clean electrolyte.3

Halide poisoning is commonly observed during cyclic voltammetry (CV); one such

example in acidic and alkaline media revealed significant alteration of the H

underpotential deposition (Hupd) peaks particularly with respect to amplitude and position

125

in platinum CVs.4 This suggested chloride anions were present on the platinum surface

even at low potentials. Other studies have demonstrated halide anion influences at

concentrations as low as μM levels.2, 3 In the context of O[H] adsorption from water

activation, adsorbed halide anions were shown to frustrate Pt-O[H] formation as indicated

by the anodic peak suppression/shift in the CV.10 Further Rotating Ring Disk Electrode

(RRDE) measurements also revealed an increased production of H2O2 on Pt(100)

facets.10 The competitive nature of O[H] and Cl- adsorption was probed by

chronocoulometric measurements.14 Thermodynamic analyses of the charge density was

used to determine the Gibbs energy of adsorption (142 kJ mol-1 Cl-, 136 kJ mol-1 O[H] on

Pt(111)), suggesting that the coexistence of Cl- and O[H] inhibited the development of an

ordered Cl- adlayer.

Since anion adsorption is a surface phenomenon, most of the available surface

sensitive techniques have been exploited to elucidate adsorbate coverage and structure.

Such techniques include Auger Electron Spectroscopy (AES), Low Energy Electron

Diffraction (LEED), second harmonic generation (SHG) and Surface X-Ray Scattering

(SXS), all of which revealed significant results. Through AES/LEED studies, it was

determined that Clads occurs on Pt(100) and Pt(111) surfaces in two separate

electrochemical windows; the hydrogen desorption region and the onset of O[H]

adsorption region.15 The Pt(100) surface was also shown to be affected by Cl- at a lower

potential than Pt(111) indicating Pt(100) was more susceptible to Cl- poisoning. SHG

experiments indicate a strong correlation between halide concentration and monolayer

coverage (0.1 ML to >0.9 ML for Cl- concentration of 10-6 – 10-4 M respectively).16

Similar findings were obtained for Br- anions adsorbed on platinum. Modern X-ray

126

scattering methods have also proved invaluable for surface studies. For example,

Marković et al.5, 17 established that Cl- adsorbs on platinum with a Pt-Cl interatomic

separation of 2.4Å with no ordered, in-plane structure. Analysis of Crystal Truncation

Rods (CTR) in the above studies yielded a monolayer coverage of 0.6 ML at 0.7 V vs.

Reversible Hydrogen Electrode (RHE) with a Pt-Cl interatomic distance of

approximately 2.4 Å. Interatomic Cl-Cl distances were also determined to be in the range

of 3.58 Å to 4.39 Å at 0.7 V and 0.25 V respectively.

As mentioned above, the spectroscopic methods used to study anion adsorption have

revealed pertinent fundamental information; however, most methods are either dependant

on ultra high vacuum conditions (UHV) or have been employed on single crystals. Such

results may not be comparable to in operando fuel cell catalyst surfaces, which are

polycrystalline in nature and strongly influenced by their surroundings, such as the

presence of electrolyte or gases. In order to obtain a more complete description of the

catalyst surface, in situ experiments should be employed on actual fuel cell materials.

Over the past two decades X-Ray Absorption Spectroscopy (XAS) has been

developed into a reliable method to study electrode processes in situ via Extended X-ray

Absorption Fine Structure (EXAFS) and X-ray Absorption Near Edge Structure

(XANES).18-20 Because of the high photon flux of modern day synchrotron sources, XAS

is available to probe materials in situ without significant x-ray beam attenuation. XAS

allows for the determination of bond distances, coordination numbers, Debye-Waller

factors and oxidation state without the necessity of long-range order. When acquired in

situ, the above parameters can be evaluated under operating fuel cell conditions.

Historically, the inherent bulk averaging nature of XAS has limited its utility for surface

127

studies. Recently however, the delta mu (Δμ) method of XANES analysis has been

developed as a surface sensitive method to determine surface/adsorbate interactions.21-24

The Δμ analysis isolates surface/adsorbate interactions by subtracting out the bulk metal-

metal interactions. Once obtained, the Δμ spectra are compared to theoretical Δμ curves

generated based on crystallographic models.

Water activation studies utilizing Pt L3 edge XANES data in acidic media have

revealed that OHads on platinum occurs in an atop configuration (one-fold) at low

potentials, followed by two and three fold symmetries with increasing electrode

polarization until finally place exchange occurs at 1.05V and higher.25-27 Prior to these

studies it was believed that place exchange occurred only at potentials greater than 1.2V.

The Δμ technique was also employed to investigate hydrogen adsorption and mobility on

platinum catalysts in the UPD region of the platinum CV.28, 29 At low coverage, H was

found to be delocalized on the platinum surface, while at higher coverage, H occupied fcc

sites as well as edge/step locations. Other platinum based in situ XAS systems that have

been explored utilizing the Δμ technique include bisulfate poisoning of platinum,30 and

COads on PtRu alloy electrocatalysts.31 The Δμ technique was also applied to a non-

metallic, heterogeneous (3-phase) RhxSy system to examine H2O activation.32 After a

complex analysis of RhxSy clusters, it was established that O adsorbs in one-fold sites at

low potentials and bridge-bonded sites at 0.8V and above. This XAS investigation also

identified the Rh3S4 moiety as the electroactive phase of the heterogeneous RhxSy

material when the previous inclination was that Rh17S15 was the active phase.

Previously, we determined the Cl- poisoning site on Pt in Cl- contaminated

electrolytes by using the Δμ method.33 In this work we extend our study to include a

128

detailed analysis of Cl- adsorption on all low index Pt faces, corner, and edge sites using

both electrochemical and spectroscopic methods. Through in situ XAS measurements

(Δμ technique and EXAFS) we are able to show that Cl- anions affect water activation

differently depending on binding site. There is also a Cl- anion concentration effect

which is apparent over the concentration range we study. Further, the Δμ technique

provides an atomic level view of the surface-adsorbate interactions, which allows us to

determine the adsorption site of Cl- on a polycrystalline Pt/C electrocatalyst.

3.2 Experimental

3.2.1 Electrochemical Characterization

RDE studies were performed as described in detail elsewhere.25 Briefly, inks of 30

wt% Pt/C (Vulcan XC72 from E-TEK) were prepared by combining 10 mg catalyst, 5

mL deionized water (18.2MΩ, Millipore MilliQ® system), 5 mL 2-propanol (HPLC

grade, Aldrich), and 40 μL 5 wt% Nafion® solution (Aldrich). The ink was then

sonicated for 15 minutes and stirred for 2 hours. The ink was then cast onto a polished

glassy carbon (GC) RDE (Pine Instrument Co.) via two 5 μL aliquots, and allowed to dry

in air after each application. The final metal loading on the GC was 14 μg cm-2 Pt. The

corresponding Nafion® to catalyst ratio was 1:50 by weight. Room temperature CVs

were obtained in 1M HClO4 (GFS Chemicals, doubly distilled), 1M HClO4 + 10-3 M KCl

(Aldrich, 99.999% pure) and 1M HClO4 + 10-2 M KCl under an Ar (Med-Tech Gasses)

atmosphere. For oxygen reduction experiments the above electrolytes were purged with

ultra high purity O2. All data were collected with an Autolab® potentiostat (model

PGSTAT30, Ecochemie-Brinkmann). For all of the experiments a sealed RHE, filled

with 1M HClO4 was utilized as the reference electrode and a high surface area Pt mesh

129

(Alfa Aesar) functioned as the counter electrode. All current densities reported in this

text refer to geometric surface area as the poisoning effects of chloride render Hupd

adsorption useless for surface area measurements. As reported earlier,34 average particle

size of these electrocatalysts were in the range of 30 Å ± 4 Å.

3.2.2 In Situ XAS Data Collection

All experiments were conducted at room temperature in an in situ electrochemical

XAS cell based on a previously reported design.35 The cells used consisted of a 30 wt%

Pt/VXC72 (E-TEK) working electrode (WE), an acid washed (0.5M H2SO4) Grafoil®

counter electrode (CE) and an RHE reference electrode. The WE and CE were separated

by a piece of Nafion® 112 (DuPont) polymer electrolyte membrane and the cell was

flooded with 1M HClO4 + xM KCl (x = 10-3 or 10-2 M). Grafoil® was chosen as a CE to

eliminate any interference at the Pt L3,2 edges with little x-ray beam attenuation and

HClO4 was utilized for its low anion adsorption effects on platinum. In all cases, Au wire

(99.999%, Alfa-Aesar) was utilized as a current collector and mechanically pressed

against the back side of the electrode in a fashion which did not expose the gold to the x-

ray beam. Silicone gaskets (Auburn Chemical Co.) were used to seal the cell. The

electrodes were prepared by hand painting catalyst suspensions of 30 wt% Pt/VXC72, 1:1

deionized H2O / 2-propanol mixture and 95:5 mixture of catalyst (wt) / 5 wt % Nafion

solution. The inks were painted onto commercially available carbon cloth (Zoltek®) with

a loading of ~5 mg cm-2 Pt to obtain an absorption cross section of ~ 1. The total

geometric area of the Pt WE used in the cell was 5 cm2. To ensure proper electrode

130

wetting, each electrode was vacuum impregnated in clean electrolyte prior to cell

assembly.

The platinum working electrodes were activated by potential cycling (0.05 V to 1.2 V

vs. RHE at 10 mV s-1) in clean 1M HClO4. Following the activation step, the clean

electrolyte was removed from the cell by syringe and replaced with 1M HClO4 + xM KCl

(x = 0, 10-3, 10-2). Full range Pt L3 extended x-ray absorption fine structure (EXAFS)

were taken (-250 eV to 18k) with the WE fixed at various static potentials along the

anodic sweep of the CV. Between EXAFS scans the potential was cycled around

completely to ‘clean’ the electrode surface. A full set of EXAFS were also obtained in

clean 1M HClO4 to provide clean reference scans and as H2O activation standards. The

measurements were performed at beam line X11-B (National Synchrotron Light Source,

Brookhaven National Laboratory, Upton, NY) with the Si(111) monochromator detuned

by 40% in order to reject the higher harmonics from the beam. Data were collected in

transmission mode using gas ionization detectors (I0, I1 or I2) with a nominal

Nitrogen/Argon gas mixture to allow ~ 10% photon absorption in I0 and 70 % in I1. The

sample was placed between the I0 and I1 detectors, while a Pt reference foil (4 μm, Alfa

Aesar) was positioned between I1 and I2.

3.2.3 EXAFS and Δμ analysis

The IFEFFIT suite36 (version 1.2.9, IFEFFIT Copyright 2005, Matthew Newville,

University of Chicago, http://cars9.uchicago.edu/ifeffit/) was utilized for background

subtraction (AUTOBK)37 and normalization. A k-range window of 1.988 – 14.072 Å-1

(Kaiser-Bessel) and a, R-window of 1.532 – 3.379 Å was used for all the EXAFS fits.

3.2.4 Alignment and normalization of XAS data

131

The Δμ analysis technique has been described in great detail elsewhere.22, 24, 38, 39

Briefly, XAS reference scans were carefully calibrated to the edge energy (11564 eV, Pt

L3) and aligned to one standard reference scan as the beam energy is known to “drift”

over the duration of the beam lifetime (12 hrs.). Any edge shift corrections applied to the

reference foils are automatically applied to their respective sample scans. A post-edge

normalization procedure was then applied to the aligned scans via a cubic spline function

(AUTOBK)37 which normalizes the oscillations over a specific energy range (for Pt Δμ

typically 25 – 150 eV with respect to E0, 150 – 1000 eV for EXAFS) to present the data

on a per-atom basis. These parameters often vary from scan to scan and are assessed on

an individual basis.39 Difference spectra were constructed using the equation

Δμ = μ(V) – μ(0.54 V) (Eq. 3.1)

where μ(V) is the sample at various potentials and μ(0.54) is the reference signal at 0.54

V, which is considered the cleanest region of platinum; i.e., relatively free of any

adsorbed H, OH or Ox species. The Δμ spectra are then compared to theoretical curves

(Δμt) constructed using the FEFF 8.0 code. This was accomplished using the relationship

Δμt = μ(Pt6X) – μ(Pt6) (Eq. 3.2)

where X is Cl- or O in a specified binding site with respect to the absorbing Pt atom and

Pt6 is a 6-Pt cluster with a Pt-Pt bond distance of 2.77 Å as described by Janin et al.40 It

should be noted that theoretical Δμ curves are generally shifted by 1-5 eV and scaled by a

multiplication factor for optimal comparison with experimental data.

3.3 Results and Discussion


132

As mentioned above, the presence of Cl- anions severely impedes the oxygen

reduction reaction (ORR) on Pt/C electrocatalysts. Figure 3.1a presents the rotating disk

electrode (RDE) curves for Pt/C in clean and Cl- contaminated electrolyte at several

rotation rates. In clean electrolyte a reasonable ORR activity is observed with an onset

potential close to 1.0 V vs. RHE. In addition, a well defined diffusion limiting current is

obtained as a function of rotation rate as described by the Levich equation:

ilim = 0.62neFD2/3ν-1/6Co (Eq. 3.3)

where ilim is the diffusion limiting current density, ne is the number of electrons, F is

Faraday’s constant, D is the diffusion coefficient of O2 in the electrolyte, ν is the

kinematic viscosity and Co is the concentration of O2. With each 10-fold addition of

chloride, the ORR overpotential is increased by approximately 85 mV with respect to the

curve in 1M HClO4. Though the Levich relationship still appears to exist, a clear

delineation of a diffusion limiting current is less evident in the presence of Cl-. This is

more manifest for ORR with 10-2 M chloride concentration (dotted line, inset). At this

concentration a significant change in the magnitude of the limiting current is observed.

The Tafel plots shown in Figure 3.1b were extracted from the ORR polarizations (anodic

scan) following a correction for mass transport by the equation

ik = ilim * i / (ilim – i) (Eq. 3.4)

where ik is the kinetic current density, ilim is the diffusion limiting current as described by

Equation 3.4 and i is the measured current during the ORR polarization. The anodic scan

was used since this represents the ORR activity initiated on a relatively clean Pt surface

133

log (ik, mA/cm2)-1.0 -0.5 0.0 0.5 1.0 1.5

E, V

vs.

RH

E

0.5

0.6

0.7

0.8

0.9

1.0E, V vs. RHE

0.4 0.6 0.8 1.0

i, m

A/c

m2

-4

-3

-2

-1

0

E, V vs. RHE0.4 0.6 0.8 1.0

i, m

A/c

m2

1M HClO4

10-3 M KCl400RPM

625RPM

900RPM

a

b

10-2 M KCl

1M HClO4

10-3 M KCl

10-2 M KCl

1 mA.cm-2

Figure 3.1 ORR polarization curves and Tafel plots. (a) ORR polarization curves (anodic sweep) for 30 wt% Pt/C on a glassy carbon disk in O2 saturated 1M HClO4 and 1M HClO4 + 10-3 M KCl at 20oC using a sweep rate of 20 mV s-1. The inset includes 1M HClO4 + 10-2 M KCl (dotted); (b) Mass transfer corrected Tafel plots taken at 900 RPM for 30 wt% Pt/C in 1M HClO4 (circles), 1M HClO4 + 10-3 M KCl (triangles), and 1M HClO4 + 10-2 M KCl (squares). All current densities utilize geometric surface area.

134

with the competing effects of Cl- and oxide growth as a function of positive potential

scan. The cathodic scan proceeds with a pre-existing oxide covered surface, thereby

representing a more complex kinetic analysis perspective. Though it was mentioned

above that Cl- contamination resulted in the loss of a well-defined diffusion limiting

current, we utilize the current density at 0.3V in each electrolyte as ilim to maintain

consistency. The overall shapes of the Tafel curves remain relatively unchanged with

chloride present, indicating no significant change in the rate limiting step of ORR. The

decrease in electrocatalytic activity reflects the increased overpotentials caused by Cl-,

and is consistent with a site blocking mechanism.10, 41 These observations agree with

work recently reported by Schmidt et al.10 Although their work was done at elevated

temperature (60oC) and with a slower sweep rate (5 mVs-1), they observed similar

overpotentials as a function of chloride concentration.

To further illustrate the effect of chloride adsorption on platinum, cyclic

voltammograms with and without chloride are presented in Figure 3.2. The solid line

representing polycrystalline platinum in clean HClO4 reveals all the signature platinum

features. Perhaps the most notably are Pt-O[H] formation (anodic sweep, > 0.70 V) and

reduction (cathodic sweep, ca. 0.80 V). From inspection of the chloride contaminated

voltammograms (dashed 10-3 M, dotted 10-2 M), it is evident that adsorbed Cl- hinders

O[H] adsorption at 0.70V. It is not until the electrode is polarized to approximately 1.0 V

that an appreciable increase in current density is attained, whence Pt-O formation occurs

despite the chloride presence. This delay in the onset potential for O[H] adsorption has

been confirmed by earlier studies on both single crystals5, 41-43 and polycrystalline Pt.10, 44,

45 Though it is not the objective of this work to investigate the effect of chloride on Hupd,

135

E, V vs RHE

0.0 0.2 0.4 0.6 0.8 1.0 1.2

i, m

A/c

m2

-0.4

-0.2

0.0

0.2

0.4

1M HClO4

1M HClO4 + 10-3M KCl1M HClO4 + 10-2M KCl

Figure 3.2 Cyclic Voltammograms of 30 wt% Pt/C (E-TEK) in Ar-saturated 1M HClO4 (solid), 1M HClO4 + 10-3 M KCl (dashed) and 1M HClO4 + 10-2 M KCl (dotted) with a scan rate of 50 mV s-1 on a 5 mm glassy carbon RDE tip at 0 RPM. The vertical lines indicate potentials at which EXAFS measurements were made

136

it is worth mentioning that changes in Hupd peaks are also observed. Among the halides,

I- adsorption is known to be the strongest,2, 10 and thus affects the Hupd region to a much

larger extent, and in a very different manner than either Cl- or Br-. The adsorption of Cl-

is known to take place on at least three distinct sites (2 for Br) with differing energies of

activation. This however does not change the total amount of adsorbed hydrogen. The

potential range over which Hads occurs is also reduced in the presence of halides and is

generally recognized to be due to the competitive nature of the adsorbed ions on the Pt

surface. There is also evidence in the literature for the partial desorption of Cl- and Br- in

this region. Such phenomena, however, have already been adequately explained2, 9 and

thus will not be discussed here in further detail.

3.3.2 EXAFS Results

Prior to the Δμ XANES analysis, traditional Fourier Transform (FT) EXAFS analysis

was performed to ensure no major changes in Pt-Pt bond length occur as a function of

potential. Such changes would yield unreliable results as Δμ relies on crystallographic

modeling with consistent bond lengths to generate theoretical Δμ signatures. To elucidate

quantitative structural information, the EXAFS data were fit using Artemis, a subroutine

of the IFEFFIT code.36 A representative fit is presented in Figure 3.3 for Pt/C in clean 1M

HClO4 at 0.54 V vs. RHE. In this region no adsorbed species (such as hydrides or

oxides) are expected and therefore it is generally representative of a clean metal surface.

The fit utilized a single shell Pt-Pt scattering path to isolate the effect that adsorbates

exhibit on local Pt structure. Table 3.1 contains a summary of the EXAFS fitting

parameters for the Pt L3 edge in each electrolyte as a function of electrode potential in the

137

R, Å1 2 3 4

| χ(R

)|, Å

- 3

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

No Cl-, 0.54VFit

Figure 3.3 Fourier Transformed EXAFS for 30 wt% Pt/C in 1M HClO4 at 0.54 V vs. RHE measured in situ at the Pt-L3 edge. Phase and amplitude parameters were fit using those generated with IFEFFIT 1.2.9 and sample data. Single shell (Pt-Pt) fit, (1.5 < k < 15.8 Å-1, k2 weighted), performed in R-space.

138

Table 3.1 Summary of EXAFS Resultsa

Potential (V, RHE)

NPt-Pt

ΔN = ±0.4b RPt-Pt (Å)

ΔR = ±0.035b E0 (eV)

ΔE0 ±1.50 σ2 (Å2)

(a) 1M HClO4 0.40 8.9 2.74 8.99 0.0065 0.54 9.1 2.74 8.99 0.0065 0.70 9.1 2.74 8.94 0.0065 0.84 9.2 2.74 8.78 0.0065 1.00 8.8 2.74 8.91 0.0065

(b) 1M HClO4 + 10-3 M KCl 0.40 8.7 2.73 9.08 0.0065 0.54 8.8 2.73 8.86 0.0065 0.70 8.8 2.74 8.83 0.0065 0.84 9.0 2.74 8.88 0.0065 1.00 8.4 2.74 8.59 0.0065

(c) 1M HClO4 + 10-2 M KCl 0.40 7.6 2.74 9.35 0.0065 0.54 7.9 2.74 9.23 0.0065 0.70 7.2 2.74 9.75 0.0065 0.84 7.5 2.74 9.77 0.0065 1.00 7.0 2.74 9.46 0.0065

a 20S fixed at 0.934 as calculated by FEFF8.0. bAlthough the absolute values of ΔN

and ΔR are larger than those indicated, the variation of N and R with potential is believed to be meaningful down to the values indicated.

139

range between 0.4 and 1.00 V vs. RHE. As evidenced by the R values, no major changes

in bond length occur although the coordination number (N) is altered as a function of

potential representing the effect of adsorbed species (both Cl- and oxides). In order to

include the Pt-Cl and Pt-O interactions, corresponding 3 shell fits were conducted (Pt-Pt,

Pt-Cl and Pt-O), however for simplicity only the Pt-Pt interactions are shown in Table

3.1. It should be noted that the EXAFS results obtained in 1M HClO4 were used to

determine the average bulk Pt-Pt coordination number (NPt-Pt) which was determined to

be in the range of 8 & 9. From these results, the Pt/C nanoparticles can be estimated to

be between 1 to 1.5 nm in diameter,46 with a dispersion factor of around 50-60%,47 and

thereby consisting of approximately 55 to 150 atoms.5 These were calculated using

algorithms developed earlier by Benfield.48 These algorithms provide an important link

between the geometric models of clusters described by Hardeveld and coworkers49, 50 and

the shell-by-shell approach used in EXAFS analysis. Accordingly, a cubo-octahedron

model was used in this analysis as previously by Kinoshita,47 and Stonehart.51 Results

from the cluster analysis are shown in Table 3.2 for various cluster sizes in the range of

10 to 60 Å. The most reactive sites are those which correspond to low surface

coordination numbers. These are depicted in Table 3.2 in terms of number of atoms at

particle edges (coordination no. 7) and at square faces or steps (coordination no. 8).

When compared to the number of sites on bulk crystallographic planes such as the

Pt(100), Pt(111) faces, the edges and steps predominate below 15 Å. However with the

particle size greater than 20 Å, the bulk crystallographic planes assume importance.

From our previous study on the effect of particle size on electrocatalysis, 52 it is clear that

140

Table 3.2 Distribution of surface sites in a cubo-octahedral Pt cluster as a function of particle size

Particle Size (Å)

Total # Atoms

Ns/Nt Nedge(CN7)/Nt Nstep(CN8)/Nt N111/Nt N100/Nt

11.75 55 0.76 0.29 0.57 0.0 0.142 16.28 147 0.63 0.13 0.52 0.22 0.26

20.86 309 0.52 0.074 0.25 0.37 0.33 30.05 561 0.39 0.033 0.034 0.55 0.41 40.00 1420 0.30 0.018 0.025 0.56 0.46 53.00 3871 0.2384 0.0099 0.019 0.56 0.47

The ratio of sites are not normalized and reflect relative changes only. This is a consequence of not including sites covering the entire range of coordination numbers.

141

the predominance of low coordination sites, such as the edge and steps, are detrimental to

the ORR activity. This was based on in situ XAS investigations of a series of Pt/C

electrocatalysts with varying particle sizes.52 The XAS data in this prior study showed

unequivocally that the imbalance of charge on the edge and step sites in these small

clusters below 20 Å results in the strong adsorption of both hydrides and oxides on the Pt

surface (depending on the potential region it is exposed to) thereby rendering the surface

largely inactive for electrocatalysis. A visual representation of the variation of the

population of various surface crystal faces as a ratio to the total number of surface atoms

is expressed as a plot with respect to overall particle size in Figure 3.4.

To visualize the effect of Cl- on Pt clusters, the Pt coordination numbers tabulated in

Table 3.1 are plotted in Figure 3.5. Without Cl- present, NPt-Pt is observed decreasing at

low and high potentials. The former has been explained29 by the desorption of H+ from

the surface while the latter is attributed to the formation of Pt-Ox species.27 In general,

adsorbed H+ (and other species in atop positions) tend to make the cluster more spherical,

which increases N by increasing the bulk to surface character of the clusters. Conversely,

adsorbed species in 3-fold positions tend to decrease N, similar to that indicated for Ox

and as we will show with Δμ for Cl-. In the presence of Cl- there is a noticeable decrease

in bulk NPt-Pt with the effect compounded as [Cl-] is increased. In 10-2M Cl- the general

downward trend in N Pt-Pt is observed, however, with some unexpected ‘bumps’ not

observed in the other electrolytes. Such phenomena are complex and will be explained in

detail later.

142

Figure 3.4 Variation of the ratio of population of various coordination sites on the

surface of clusters and the total number of surface sites as a function of the particle size of the cluster. Calculations were performed using the methodology developed by Benfield.48 Also shown for comparison is the evolution of the total coordination number and those of the individual sites. All calculations were made using a cubo-octahedron model cluster.

Particle Size, Å20 40 60 80 100

Ni /

Ns

0.00

0.25

0.50

0.75

1.00

Coo

rdin

atio

n N

umbe

r, cu

booc

tahe

dron

5

6

7

8

9

10

11

12

Nsurf/Ntot

Nvertex/Nsurf Nedge(cubo)/Nsurf

N100(cubo)/Nsurf

N111(cubo)/Nsurf

CN(cubo)

143

Figure 3.5 EXAFS results showing change in Pt-Pt coordination as a function of electrode potential for Pt/C electrocatalyst in (a) clean 1M HClO4, (b) 1M HClO4 + 10-3 M Cl and (c) 1M HClO4 + 10-2 M Cl.

0.0 0.2 0.4 0.6 0.8 1.0 1.2

E, V vs RHE

7.0

7.5

8.0

8.5

9.0

9.5

10.0

N Pt

-Pt

1M HClO410-3 M Cl10-2 M Cl

.

.a

b

c

0.0 0.2 0.4 0.6 0.8 1.0 1.2

E, V vs RHE

7.0

7.5

8.0

8.5

9.0

9.5

10.0

N Pt

-Pt

1M HClO410-3 M Cl10-2 M Cl

.

.a

b

c

144

It is also interesting to note that the increase in chloride concentration has little or no

effect on the rate of change (i.e. the negative slope) of NPt-Pt in Figure 3.5 above 0.95V.

Sub-surface oxygen is known to form on Pt above 1.00 V, reducing the Pt-Pt scattering

sharply. 27 Since the oxygen that goes sub-surface is actually the oxygen that was

previously adsorbed at lower potentials, adsorbed chloride is not expected to hinder the

movement of these oxygen atoms and therefore the rate of change in NPt-Pt is not

significantly affected.

3.3.3 Δμ-XANES Results

Figure 6 shows the Δμ spectra for the Pt/C electrodes fixed at the indicated potentials

(0.4 to 1.0 V vs. RHE). The spectra in Figure 6a indicate a normal H2O activation

pathway for platinum in HClO4. In the potential region where Pt-OH formation occurs,

typically 0.7 V, a small positive peak is obtained in the Δμ spectrum near the absorption

edge energy. Subsequently, as the potential is increased, the peak increases in both

magnitude and energy. This signature change is consistent with a previous report27 which

described it as a transition from atop OH (ca. 0.7 V) to n-fold O (n = 2 or 3, typically

indistinguishable). Teliska et al.27 ran Δμ simulations as described by Equation 3.2, using

the same Pt6 clusters employed in this work, with O situated in 1, 2, 3 and 4-fold

locations. As the Pt-O coordination increased from 1 to 4, the Δμ lines were observed

behaving as described above. These line shapes were confirmed as they correlated very

well with their in situ Δμ taken in 0.1M HClO4. Due to the weak scattering properties of

H, the Δμ technique cannot distinguish directly between OH and O adsorption; however,

DFT calculations40, 53, 54 have shown that OH prefers 1-fold coordination (atop) and O

145

prefers 3-fold and therefore the Δμ can indirectly distinguish the two because of their

adsorption site preference.

The Δμ in Figure 3.6b (10-3 M Cl-) exhibits several additional features not present in

the clean HClO4. A consistent negative contribution near 0 eV precedes the Pt-O[H]

lines at all potentials investigated. Also noticeable is an apparent change in the H2O

activation pathway as evidenced by the location of the Pt-O[H] maxima. As mentioned

above, a clean Pt surface should undergo Pt-O[H] formation by the transition of Δμ peaks

to higher eV values as in Figure 3.6a and Reference 27. This, however, is not observed

with 10-3 M Cl. Not only do all the Δμ peaks shift to a higher energy position, they also

‘stack’ on top of each other in Figure 3.6b, representing a deviation from the normal H2O

activation pathway. Also, the positive maxima in the Δμ spectra between 0.54 V to 0.70

V exhibit virtually no increase in magnitude (i.e. Pt-OH suppression in that potential

range). Conversely, there is significant growth in the Δμ from 0.7 V to 0.84 V, indicating

Pt-O[H] formation in agreement with observations from the CV in Figure 3.2. Between

0.54 V and 0.7 V, no appreciable increase in current density was observed, however; at

0.84 V there is a small but discernable Pt-O[H] current in the CV. It is also worth

mentioning that the Δμ magnitudes for 0 and 10-3 M Cl- at 0.84 V are quite close

indicating the O[H] coverage is approximately equal even though the adsorption process

may have occurred differently. The Δμ magnitudes at 1.0 V do not correlate as closely

for reasons which will be discussed in the FEFF 8.0 analysis.

In the presence of 10-2 M Cl- (Figure 3.6c) an overwhelming negative contribution

dominates the spectrum at all potentials. In the following 5-10 eV region where the O[H]

peak typically appears, only a very small peak is obtained at 1.0 V, indicating little, if any

146

Figure 3.6 Pt L3 edge Δμ = μ(V, xM Cl-) - μ(0.54 V clean) spectra for 30 wt% Pt/C in 1M HClO4 and the indicated KCl concentrations.

0.00

0.02

0.04

0.06

-0.01

0.00

0.01

0.02

0.03 0.40V0.54V0.70V0.84V1.00V

E, eV (rel. to Pt L3 edge)-10 0 10 20 30

-0.06

-0.04

-0.02

0.00

0.02

Δμ

a

b

c

No Cl

10-3 M KCl

10-2 M KCl

147

Figure 3.7 Theoretical Δμ = μ(Pt6-Cl) – μ(Pt6) signatures for atop (solid), bridged (dashed) and 3-fold fcc (dotted) chloride on Pt6 clusters

Erel, eV (Pt L3 edge)-10 0 10 20

Δμ

-0.04

-0.02

0.00

0.02

0.04

Erel, eV (Pt L3 edge)-10 0 10 20

Δμ

-0.04

-0.02

0.00

0.02

0.04

148

Figure 3.8 (a) Comparison of theoretical 3-fold O (solid) and 3-fold Cl (dotted) Δμ signatures. The dash-dot line shows the sum of the two curves. (b) Comparison of experimental Δμ in 10-2 M Cl- at 0.54 V (solid), 10-2 M Cl- at 1.00 V (dashed) and theoretical Δμ signature for 3-fold Cl-.


Δμ

-0.04

-0.02

0.00

0.0210-2 M KCl @ 0.54V10-2 M KCl @ 1.00VPt6-Cl fcc

-0.04

-0.02

0.00

0.02

0.04

0.06

0.08

Average of 3-fold lines

a

b


Δμ

-0.04

-0.02

0.00

0.0210-2 M KCl @ 0.54V10-2 M KCl @ 1.00VPt6-Cl fcc

-0.04

-0.02

0.00

0.02

0.04

0.06

0.08

Average of 3-fold lines

a

b

149

O[H] adsorption occurs. This again is confirmed by inspection of the CV in Figure 3.2.

The small bump on the anodic sweep is almost nonexistent in 10-2 M Cl- at 0.84 V, and at

1.0V only a miniscule current density is achieved. This dip in the Δμ spectra for 10-3 and

10-2 M Cl- seems to suggest the line shape for Cl- adsorption is a negative peak near the

edge; however, definitive adsorption site conclusions cannot be made until theoretical

simulations by FEFF 8.0 are discussed below.

In the case of 10-3 M chloride the adsorption geometry is not as obvious as it was in

10-2 M Cl-. The theoretical Δμ line shapes suggest bridge bonded adsorption; however,

we interpret the spectra to be a combination of H2O activation (hence presence of surface

oxides) and 3-fold Cl- adsorption occurring concurrently. There is direct evidence of both

chloride and O[H] adsorption in the Δμ. To illustrate this, Figure 8a shows the FEFF 8.0

Δμ simulations for 3-fold O and 3-fold Cl- with boxes to emphasize the major elements of

each adsorbing species. Both Pt6 clusters show 3-fold adsorption of chloride and O[H]

with vastly different Δμ signatures. The dash-dot line, showing the sum of 3-fold Δμ

theory exhibits an astonishing likeness to the experimental data in 10-3 M chloride. This

is, in our opinion, the best evidence indicating 3-fold chloride adsorption occurring in

parallel with 3-fold O[H] formation.

3.3.4 The 0.4 - 0.7 V region

Perhaps the most interesting potential region in Figure 3.5 lies between 0.40 and 0.84 V.

This region is scrutinized in greater detail because it is within the operating range of most

fuel cells and therefore is as significant as the region at higher potentials. This potential

region is where adsorption and rearrangements occur in the Cl- overlayer. Further, it is

the potential region where the more reactive Pt (100) planes plus the corners, steps and

150

edges catalyze water activation and become covered with OH. We discuss Cl- adsorption

and water activation in separate sections below.

3.3.5 Chloride adsorption and rearrangement

It is reasonable to estimate relative surface Cl- coverage changes based on the Δμ peak

amplitudes as was done previously,26, 27 even though absolute monolayer coverage values

are much more difficult to obtain. In Figure 3.9 we use the absolute value of the negative

going Δμ located near 1 eV as it has been shown to reflect 3-fold Cl-. The Δμ at

potentials less than 0.40 V were excluded here as the Pt-H and Pt-Cl Δμ signatures are

too similar, making them impossible to distinguish. Figure 3.9 also includes previously

published data55 estimating Br coverage on a single crystal Pt (111) face in 0.01 M Br,

the Pt-Cl Gibbs Free Energy (ΔGPt-Cl) reflecting the Pt- Cl bond strength as well as the

EXAFS NPt-Pt results from Figure 3.5. Although a similar plot of Cl- coverage would be

preferable, we are not aware of such data, although Lucas et al.17 have reported the

coverage of Cl- on Pt(111) at 0.25 V and 0.7 V as indicated in Figure 3.9. Note the

dramatic increase in Br coverage exactly in the region where the adsorbed H is known to

leave. This is entirely consistent with our reduced NPt-Pt values as noted in Figure 3.5, and

significant increase in Δμ magnitude already evident at 0.4 V as shown. Above 0.25 V,

the Br coverage continues to increase with potential but at a much slower rate, and this is

believed to occur because of a “continuous compression of the Br adlayer” producing a

partially disordered or incommensurate adlayer on the Pt(111) surface,5 occurring in the

region between 0.30-0.55 V. Such compression will move some of the halide ions into

bridged and atop sites. This is the probable cause for the reduction in apparent Cl-

coverage as suggested by the Δμ results between 0.3-0.5V. Note that in Figure 3.7, the

151

Figure 3.9 Plot of Br coverage55 and Δμ amplitudes representing Cl- coverage (this work) using left axis, and the Gibbs free energy for Cl- adsorption14 using the right axis. The Pt-Pt coordination numbers from Table 3.2 for the 10-2 M Cl- case are indicated with arbitrary units and the Δμ amplitude has been scaled so that it approximately represents coverage in ML. The small shaded lines indicate Cl- coverage at 0.25 and 0.70 V as estimated by Lucas et al.17 The vertical lines roughly separate the regions where Cl- adsorption, compression in the Cl- overlayer, more Cl- adsorption, and OH adsorption dominate as indicated. The symbols at the bottom indicate the dominant Δμ signatures from Figure 3.10 in each region.

0 200 400 600 800 1000

0.0

0.1

0.2

0.3

0.4

0.5

0.6

140

150

160

170

180

Potential, V vs. RHE

Br C

ov.,

ML

or Δ

μ

ΔGPt

-Cl, k

J mol

-1

Ads. Comp. Ads. OH Br CovΔμ AmpΔGPt-ClNPt-Pt

Cl3f Clat Cl3f O(H) 0 0.2 0.4 0.6 0.8 1.0 0 200 400 600 800 1000

0.0

0.1

0.2

0.3

0.4

0.5

0.6

140

150

160

170

180


Br C

ov.,

ML

or Δ

μ

ΔGPt

-Cl, k

J mol

-1

Ads. Comp. Ads. OH Br CovΔμ AmpΔGPt-ClNPt-Pt

Cl3f Clat Cl3f O(H) 0 0.2 0.4 0.6 0.8 1.0

152

FEFF 8.0 results show that atop/bridged Cl- does not have the large negative feature

around 1 eV, so the movement of 3-fold Cl into atop/bridged sites will decrease this

negative contribution even though additional Cl- might be adding to the surface. Above

0.55 V, apparently more Cl- is added again as suggested by the Δμ magnitudes, and the

results of Lucas et al.17 shown in Figure 3.9. The ‘hump’ in NPt-Pt, between 0.4-0.7 V is

now easily understood. Recall as mentioned above that 3-fold Cl reduces NPt-Pt and atop

Cl will increase it, just as we see for O[H] and O. Thus the increase in NPt-Pt falls right in

the ‘compression’ region, and the decrease again during the second Cl adsorption zone.

Thus, the Δμ and changes in NPt-Pt from EXAFS analysis correlate very well.

Figure 3.9 also shows that the Pt-Cl Gibbs free energy reaches a maximum right

where the Cl- coverage begins to increase again. The increase in ΔGPt-Cl apparently arises

because of the increasing charge on the Pt surface atoms with potential. The decrease in

ΔGPt-Cl above 0.5 V is believed to arise because of additional Cl- adsorption, which

increases the repulsive lateral interactions and hence reducing the net Pt-Cl bond

strength. Other studies5, 15 have also shown that Cl- adsorbs following H desorption and

concurrently or slightly before O[H] adsorption. The decrease in Cl- coverage as

indicated by the Δμ above 0.7 V, obviously arises because of O[H] adsorption.

To further investigate the Cl- rearrangement from 3-fold into atop sites during the Cl-

adlayer compression stage, a set of Δμ spectra are given in Figure 3.10, only this time

using 0.40 V in the same electrolyte as the reference spectrum (e.g., Δμ = μ( xV, 10-3 M

Cl-) - μ(0.4V, 10-3 M Cl-). If atop Cl- adsorption is indeed occurring, it should be

apparent by isolating such a Δμ signature in this potential region by eliminating the large

contribution from the 3-fold Cl- that has adsorbed below 0.4 V. In the case of 10-2 M Cl-

153

Figure 3.10 Plot of Δμ= μ(V) - μ(0.40 V) for the indicated Cl- concentrations and comparison with FEFF 8.0 results from Figure 6. Vertical line separates the energy where below the atop Cl- Δμ signature dominates and above the O[H] dominates in magnitude.

Atop Cl

Cl O(H)

E, eV (rel. Pt L3 edge)

0.015

0.000

-0.015

-0.030

0.015

0.000

-0.015

-0.030

10 0 10 20 30

10mM

1mM

Δμ

0.540.700.84

3f ClFEFF8

a

bAtop Cl

Cl O(H)

E, eV (rel. Pt L3 edge)

0.015

0.000

-0.015

-0.030

0.015

0.000

-0.015

-0.030

10 0 10 20 30

10mM

1mM

Δμ

0.540.700.84

3f ClFEFF8

a

b

154

(Figure 3.10b), the Δμ signature clearly indicates the addition of atop Cl- at 0.54 V and 3-

fold Cl- at 0.70 V. This is again consistent with our discussion of Figure 3.9. The line

shape at 0.84 V reflects the adsorption of O[H] along with a small amount of additional

Cl-.

Examination of Figure 3.10a for the 10-3 M Cl- is also revealing. The Δμ spectra at

0.54 and 0.70 V reflect primarily atop Cl- combined with some atop O[H]. The signature

for atop Cl- and O[H] are very similar, but that for atop Cl- is shifted to lower energy by

approximately 5 eV. This is confirmed by the comparison of the Δμ signature from

Figure 10b for 10-2M at 0.54 V, which is believed to reflect entirely atop Cl- without any

adsorbed O[H]. One can then detect a small amount of O[H] present at 0.70 V in 10-3 M

Cl- (shaded area in Figure 3.10a that is not present at 0.54 V. This also explains the large

amplitudes of the Δμ signatures in Figure 3.6b for the 10-3M Cl-; amplitudes larger than

those in Figure 3.6a without Cl-. Seemingly both atop Cl- and O[H] are adsorbing

together at all potentials above 0.54 V. Comparison of the CVs in Figure 3.2 and the NPt-

Pt curves in Figure 3.5 both show that even the 10-3M Cl- has a harmful effect on the

adsorption of OH below 0.8 V, so indeed we suspect that the Δμ curves in Figure 3.6a

reflect primarily atop Cl-, not O[H] (the one at 0.84 V may reflect about equal amounts of

each). Further, since the NPt-Pt changes well below 0.84 V reflects adsorption more at the

corners/edges than on the Pt(111), this indeed provides evidence that O[H] cannot

displace Cl- at the corners/edges until about 200 mV higher than normal, just as on the

Pt(111) planes.

3.3.6 Water activation on low index Pt planes, corners and edges

155

Figure 3.11 Cyclic voltammograms of 20 wt% Pt/C (E-TEK)in 0.5 M HClO4 and 0.5M HClO4 + 10-2M Cl- as reported by Schmidt et al.10 (50 mV s-1, 900 RPM, 7 μgPt cm-2). Also shown are fraction of H2O2 formed during ORR on these same samples (Ering = 1.2 V, 5 mV s-1, 1600 RPM) as reported by Schmidt et al.10 Finally the NPt-Pt data from Table 3.2 are plotted scaled and shifted as noted to fit on the right axis. Rectangle indicates region where O[H] from water activation occurs on the cluster corners/edges and on the Pt(100) planes.

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

-0.8

-0.6

- 0.4

- 0.2

0.0

0.2

0.4

0

10

20

30

40i,

(mA

cm-2

)

% H

2O2/H

2O

010

Cl(mM)

10 x (NPtPt- 6)

Corners/edges/(100)

Potential, V vs. RHE0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

-0.8

-0.6

- 0.4

- 0.2

0.0

0.2

0.4

0

10

20

30

40i,

(mA

cm-2

)

% H

2O2/H

2O

010

Cl(mM)

10 x (NPtPt- 6)

Corners/edges/(100)


156

Previous studies regarding Cl- adsorption on larger Pt clusters as summarized in Figure

3.11 shed light on the interplay between O[H] and Cl- on the different cluster planes. 10

In pure 1M HClO4, and in 1M HClO4 + 10-2 M Cl-, the CVs reveal a feature in the 0.4 -

0.7 V region. Note that the Pt/C electrocatalysts in Figure 3.11 have an average particle

size of 30 Å, which according to correlation with the cubo-octahedron model indicates

significant presence of Pt(100) sites (Table 3.2). Since the Pt-Cl interaction on lower

coordinated Pt atoms at the corners and edges, as well as on the more open Pt(100)

surface, is known to be stronger than on the Pt(111) faces5, these more reactive sites are

expected to be covered with Cl- well below 0.70 V. However, also note that the second

rise in the CV, beyond 0.70 V, doesn’t occur until higher potentials in the presence of 10-

3 M Cl-. It should be apparent that OH can displace Cl- on the Pt (100) faces and

corners/edges but not so easily on the Pt(111) faces.

Also shown in Figure 3.11, the production of H2O2, which occurs when the O2

dissociation is partially obstructed forcing it to adsorb end-on, clearly sharply increases in

the 0.4-0.7 V region when compared with the absence of Cl-. This indicates that OH

coming from water activation can displace Cl- on the Pt sites such as Pt (100) but O2

during ORR cannot.

Figure 3.11 also shows NPt-Pt for the 10-2 M Cl- concentration from Figure 3.5 to

emphasize the different behavior between NPt-Pt and the CV in this region where the

feature is much broader. Therefore, we conclude that O[H] adsorbing on the

corners/edges and Pt(100) sites in this region in 10-2 M Cl- is not believed to be the cause

of the ‘hump’ in NPt-Pt, but rather the compression of the Cl- adlayer on the Pt(111) faces

and corners/edges as discussed above.

157

3.3.7 Interplay of bisulfate and halide ions on Pt

The results above reveal the interaction between adsorption of O[H] and Cl- and its

dependence on the Pt adsorption site. A comparative study of the interplay between other

anions can provide further insight. Here, we will extend our findings with some recently

published results45 to understand the complex scenario of competing adsorption among

other commonly found anions. Of particular interest are some recent findings involving

Cl- and Br poisoning of Pt surfaces in both HClO4 and H2SO4. While studies performed in

H2SO4 are disadvantageous in the sense that the problem is complicated due to bisulfate

adsorption, they nevertheless furnish information on how bisulfate behaves in an

environment containing both Cl- and O[H]. Although these results are by no means new,

it is our intent to offer an alternative interpretation than that offered by Zolfaghari et al.45

as their paper made no mention of bisulfate adsorption throughout their discussion.

Figure 3.12 shows a re-creation of Electrochemical Quartz Crystal Nanobalance (EQCN)

results from their experiments involving Cl- and Br adsorption in H2SO4. The

representation of the data we present is different from that in their manuscript for the sake

of clarity. Two sets of curves are shown indicating the mass-frequency response in

H2SO4, used here for a standard/reference, with that in the presence of Cl- and Br under

various concentrations. Sulfuric acid furnishes bisulfate ions that adsorb weakly on the

surface and thus should be expected to increase the mass density of adsorbate. The onset

of oxide formation is indicated by a change of slope in the curve, and occurs around 0.80

V. In the presence of chloride, we observe that as the concentration of chloride in

solution increases, the amount of total mass on the surface decreases. This can only occur

as a result of bisulfate anion desorption.

158

Figure 3.12 Adsorbate mass change (from that at 0.0 V) with potential as estimated from EQCN data reported by Zolfaghari et al.45 in 0.5M H2SO4 and the indicated concentrations of Br- or Cl-. Arrows indicate anodic/cathodic potential direction. The data for Br have been shifted up by 20 g mol-1 Pt for clarity.


Δm, g

mol

-1Pt

Br (mM)1010

Cl (mM)0110

0.5 M H2SO4

0.0 0.4 0.8 1.2 1.6 2.00

10

20

30

40

50

60

70


Δm, g

mol

-1Pt

Br (mM)1010

Cl (mM)0110

0.5 M H2SO4

0.0 0.4 0.8 1.2 1.6 2.00

10

20

30

40

50

60

70

0.0 0.4 0.8 1.2 1.6 2.00

10

20

30

40

50

60

70

159

For the ensuing discussion, the following approximations are made with respect to

mass density. The mass of adsorbate/mole Pt-atom is approximately 18 for O2-/OH-, 50

for HSO4-, 36 for Cl- and 80 for Br-. These mass densities assume one Cl- and one Br

anion per surface Pt, ½ bisulfate per Pt atom, and 1.5 O per Pt. The results can then

easily be accounted for by assuming Cl-, which has a smaller mass density, adsorbs more

strongly than bisulfate. Thus, on displacing bisulfate, the overall mass decreases with

increasing Cl- concentration, as the heavier bisulfate leaves the surface for the lighter,

more strongly bound chloride ions. Since Br ion is heavier, the mass density increases

with Br concentration.

These curves clearly indicate that not all of the halide anions are displaced from the

surface by O even at 1.4 V, because if that were the case, the resulting curves at 1.4 V

should fall at the same place with whatever the O mass density was at that point. Some

disagreement on this point has occurred in the literature,45 however, Figure 3.12 makes

clear that the halides remain in part on the surface all the way up to 1.4 V.

Figure 3.12 reveals a very interesting difference between Br and Cl- adsorption

relative to the bisulfate; Cl- shows a hystersis effect in the region between 0.3-0.8 V,

while Br does not. In the case of the Cl-, during the anodic sweep the mass density is

independent of the Cl- concentration which indicates that bisulfate is on the surface.

However, during the cathodic sweep the mass density changes as expected with Cl-

concentration suggesting the presence of Cl- ion. The adsorption of a bisulfate/water

adlayer is known to give rise to the “butterfly” feature in CV curves Pt(111) single

crystals. Apparently, Cl- is not able to penetrate this adlayer until it is disrupted by O

adsorption at higher potentials. Then, during the cathodic sweep the Cl- ions re-adsorb

160

and bisulfate is not able to displace the adsorbed Cl-. This effect seemingly does not

occur in mixed Br/H2SO4 electrolytes as Br adsorption occurs so strongly that it can

displace the bisulfate either way.

This interpretation of EQCN data also calls into question the conclusions reached

more recently by Yadav et al.56 They report similar EQCN results for halide ions in

H2SO4. They also report analogous hysteresis effects in the case of Cl- and attributed it to

Pt dissolution. While we cannot rule out some Pt dissolution during potential cycling,

clearly the effects of bisulfate adsorption should not be ignored; otherwise the extent of

Pt dissolution can be grossly over-estimated.

3.4 Summary and Conclusions

The combination of in situ X-ray absorption spectroscopy, and electrochemical

measurements (CV and RDE), and previously published EQCN data has provided further

understanding of the nature of chloride poisoning on different faces/sites of platinum

catalysts in acidic medium (HClO4). The sensitivity to anion adsorption effects of the Δµ

XANES procedure is clear. In this work, Cl- ions produced a direct contribution to the

Δμ in contrast to that found previously for bisulfate which did not appear in the Δμ,

unless OH or other adsorbates were co-adsorbed on the surface to force the bisulfate

anions into site-specific binding.

To the best of our knowledge, this is the first study to provide conclusive evidence for

the site-specific 3-fold adsorption of chloride species on Pt(111) faces in a working fuel

cell environment, though the compression of the Cl- adlayer, which apparently forces

some Cl- into atop/bridge sites between 0.40-0.55V, also occurs. The adsorbed chloride

on the Pt(111) faces at the investigated concentrations (10-3 and 10-2 M) are believed to

161

be adsorbed in equilibrium; i.e. Cl- + Pt → Cl-/Pt at nearly all potentials consistent with

our Δμ results and Figure 3.9. At 10-2 M halide concentration, the coverage at 0.3V

appears to be around 0.35-0.4 ML for both Br and Cl-. This probably occurs because at

this low potential the halide ions are still highly negatively charged, so that the large

Coulomb lateral interactions keep the coverage small in both cases. As the magnitude of

the halide ion charge decreases with potential, the Coulombic lateral interactions

decrease, enabling some compression of the adlayers. In the case of the Cl-, the smaller

covalent radius then enables a great deal more 3-fold Cl- to adsorb between 0.5 and 0.7V;

in Br this apparently cannot occur because of the much larger covalent radius.

Somewhat surprisingly, NPt-Pt found from the EXAFS analysis is also quite dependent

upon Cl- adsorption. Chloride adsorption into the 3-fold sites decreases NPt-Pt, similar to

that found previously for 3-fold O adsorption,27 and clearly arises because of the

proximity of the Cl- anion partially in between the Pt atoms. In contrast, atop Cl-

adsorption increases NPt-Pt, and rearrangement of some Cl- atoms into atop sites between

0.40-0.55V is therefore also evident from NPt-Pt. The change of NPt-Pt with potential

therefore reveals the nature of the adsorption, atop vs. 3-fold Cl- and O[H].

The interplay of anionic (Cl-, Br-, OH-, and HSO4-) adsorption on the different

surfaces of Pt are indeed complex as the results summarized in Table 3.3 indicate. For

example, we find that O[H] can displace atop chloride on the Pt(100) faces, but not on the

Pt(111) faces as well as corners and edges until much higher potentials than without

chloride. Chloride also drastically alters the ORR causing an increase of the overpotential

by approximately 85 mV for every 10-fold increase in Cl- concentration with a total 150-

162

Table 3.3 Summary of results at different faces and corner/edge sites on Pt particles.

Adsorbate

Species Corners/edges Pt(100) faces Pt(111) faces

OH potential region 0.4-0.7 0.4-0.7 V > 0.7 V

OH from H2O activation

Threshold pot. moved up by 200 mV in Cl

Not affected by Cl

Threshold pot. moved up by 200 mV in Cl

ORR Site blocked by Cl Cl enhances 2e over 4e red. Site blocked by Cl

Bisulfate Expected to have little effect on Cl ad.

Appears to have little effect on Cl

Bisulfate adlayer blocks Cl adsorption anodically; no

effect cathodically Apparent ion

adsorption preference

Br > Cl > OH > HSO4 OH > Cl > HSO4 Br > HSO4 > Cl > OH

163

200 mV increase in the overpotential at large concentrations at the Pt(111) sites. In

contrast, Cl- appears to force the ORR to the undesirable 2-electron reduction (peroxide

production) at the corner/edge sites. It is well known that peroxide production occurs

when the nearby surface is crowded thus making it difficult for the O2- intermediate to

“tip over” and result in dissociation of the O2 bond. Although we did not directly study

bisulfate adsorption in this work, a reinterpretation of the EQCN results shed further light

on the interplay between Cl- and Br vs. bisulfate. It seems obvious that Cl- ions cannot

displace the bisulfate-water adlayer formed on Pt(111) planes after it is formed at lower

potentials, however, once the bisulfate is disturbed at higher potentials, it cannot displace

the adsorbed Cl-. On the other hand the Br ion is able to displace bisulfate at lower

potentials in the anodic direction.

The relative order of the Pt-X adsorption preferences indicated in Table 3.3 at the

different binding sites is suggested by the results and discussion above as well as

previous results summarized by Markovic et al.5 They should be taken as only very

qualitative and phenomenological, but they do give some insight into the complex

interplay of anion adsorption occurring at different sites in an electrochemical cell.

Markovic et al.5 concluded that on nearly all low-index surfaces the adsorbate-Pt

interactions increase in the order HSO4 < Cl < Br. They further indicated that for the

halides the Pt-X interaction was stronger on the Pt(100) faces than on the Pt(111) faces,

but that the bisulfate interactions went in the opposite direction (Pt(111) < Pt(100)). In

contrast, the EQCN results suggest that the Pt-bisulfate interaction is in fact stronger than

the Pt-Cl on the Pt(111) faces. Further, although both the Pt-halide and Pt-OH

interactions are stronger on the corners/edges and Pt(100) sites than on the Pt(111) sites,

164

the difference is much stronger for the Pt-OH interaction (70-80 kJ mole-1)5 so that of the

adsorbates studied, the Pt-OH interaction is apparently the strongest on the Pt(100) faces,

and the weakest on the Pt(111) faces while intermediate on the corners/edges.

These relative interaction strengths indicated here should help to explain the different

dependencies of the important ORR reaction on anion adsorption, and suggests that the

effect of Cl- poisoning might be quite dependent on the particle size, as the relative

number of corner/edge sites to Pt (111) face sites changes with particle size. Markovic et

al.5 have already given an extensive review of some these effects on low index single

crystal faces, but not on real nanoparticle-sized catalysts.

3.5 Acknowledgments

Financial support for this effort was provided by the Army Research Office via both a

single investigator grant and a Multi University Research Initiative (Case Western

Reserve University, P.I.). One of the authorsb would like to acknowledge financial

support by way of a Summer Research Fellowship award towards part of this study from

the Sigma Xi foundation. The authors are grateful for the use of X-11B at the National

Synchrotron Light Source, Brookhaven National Laboratory, Upton NY, which is

supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy

Sciences, under Contract No. DE-AC02-98CH10886.

165

3.6 References

1. Gottesfeld, S. & Zawodzinski, T. A. Polymer electrolyte fuel cells. Advances in Electrochemical Science and Engineering 5, 195-301 (1997).

2. Breiter, M. W. Voltammetric study of halide ion adsorption on platinum in perchloric acid solutions. Electrochim. Acta 12, 925-35 (1963).

3. Horanyi, G. & Rizmayer, E. M. The problem of the irreversibility of the adsorption of halides on platinum. J. Electroanal. Chem. Interfacial Electrochem. 83, 367-73 (1977).

4. Huang, J. C., O'Grady, W. E. & Yeager, E. The effects of cations and anions on hydrogen chemisorption at platinum. J. Electrochem. Soc. 124, 1732-7 (1977).

5. Marković, N. M. & Ross, P. N. Surface science studies of model fuel cell electrocatalysts. Surf. Sci. Rep. 45, 117-229 (2002).

6. Markovic, N., Gasteiger, H. & Ross, P. N. Kinetics of Oxygen Reduction on Pt(hkl) Electrodes: Implications for the Crystallite Size Effect with Supported Pt Electrocatalysts. J. Electrochem. Soc. 144, 1591-1597 (1997).

7. Slygin, A. & Frumkin, A. Acta Physicochim. U. R. S. S. 3, 791 (1935).

8. Llopis, J. Corrosion of platinum metals and chemisorption. Catal. Rev. 2, 161-220 (1968).

9. Bagotzky, V. S., Vassilyev, Y. B., Weber, J. & Pirtskhalava, J. N. Adsorption of anions on smooth platinum electrodes. J. Electroanal. Chem. 27, 31-46 (1970).

10. Schmidt, T. J., Paulus, U. A., Gasteiger, H. A. & Behm, R. J. The oxygen reduction reaction on a Pt/carbon fuel cell catalyst in the presence of chloride anions. J. Electroanal. Chem. 508, 41-47 (2001).

11. Wang, J. X., Markovic, N. M. & Adžić, R. R. Kinetic Analysis of Oxygen Reduction on Pt(111) in Acid Solutions: Intrinsic Kinetic Parameters and Anion Adsorption Effects. J. Phys. Chem. B 108, 4127 (2004).

12. Zhang, L. & Mukerjee, S. Investigation of durability issues of selected nonfluorinated proton exchange membranes for fuel cell application. J. Electrochem. Soc. 153, A1062-A1072 (2006).

13. Lane, R. F. & Hubbard, A. T. Electrochemistry of Chemisorbed Molecules. 111. Determination of the Oxidation State of Halides Chemisorbed on Platinum. Reactivity and Catalytic Properties of Adsorbed Species. J. Phys. Chem. 79, 808-815 (1975).

166

14. Li, N. & Lipkowski, J. Chronocoulometric studies of chloride adsorption at the Pt(111) electrode surface. J. Electroanal. Chem. 491, 95-102 (2000).

15. Markovic, N. & Ross, P. N. The Effect of Specific Adsorption of Ions and Underpotential Deposition of Copper on the Electro-Oxidation of Methanol on Platinum Single-Crystal Surfaces. J. Electroanal. Chem. 330, 499 (1992).

16. Campbell, D. J., Lynch, M. L. & Corn, R. M. Second harmonic generation studies of anionic chemisorption at polycrystalline platinum electrodes. Langmuir 6, 1656-64 (1990).

17. Lucas, C. A., Marković, N. M. & Ross, P. N. Adsorption of Halide Anions at the Pt(111)-solution Interface Studied by in situ Surface X-ray Scattering. Phys. Rev. B 55, 7964-7971 (1997).

18. Mukerjee, S. & McBreen, J. An in situ X-ray absorption spectroscopy investigation of the effect of Sn additions to carbon-supported Pt electrocatalysts. Part I. J. Electrochem. Soc. 146, 600-606 (1999).

19. Mukerjee, S. et al. In situ x-ray absorption spectroscopy studies of metal hydride electrodes. J. Electrochem. Soc. 142, 2278-86 (1995).

20. O'Grady, W. E. & Ramaker, D. E. ‘Atomic' X-ray absorption fine structure: a new tool for examining electrochemical interfaces. Electrochim. Acta 44, 1283-1287 (1998).

21. Koningsberger, D. C., Mojet, B. I., Van Dorssen, G. E. & Ramaker, D. E. XAFS spectroscopy; fundamental principles and data analysis. Top. Catal. 10, 143-155 (2000).

22. Koningsberger, D. C., Mojet, B. L., Miller, J. T. & Ramaker, D. E. XAFS spectroscopy in catalysis research: AXAFS and shape resonances. J. Synchrotron. Radiat. 6, 135-141 (1999).

23. Koningsberger, D. C., Mojet, B. L., van Dorssen, G. E. & Ramaker, D. E. XAFS Spectroscopy; Fundamental Principles and Data Analysis. Top. Catal. 10, 143-155 (2000).

24. Ramaker, D. E., Mojet, B. L., Koningsberger, D. C. & O'Grady, W. E. Understanding Atomic X-ray Absorption Fine Structure in X-ray Absorption Spectra. J. Phys.: Condens. Matter 10, 8753-8770 (1998).


167

26. Teliska, M., Murthi, V. S., Mukerjee, S. & Ramaker, D. E. Correlation of Water Activation, Surface Properties, and Oxygen Reduction Reactivity of Supported Pt-M/C Bimetallic Electrocatalysts Using XAS. J. Electrochem. Soc. 152, A1259-A2169 (2005).

27. Teliska, M., O'Grady, W. E. & Ramaker, D. E. Determination of O and OH Adsorption Sites and Coverage in Situ on Pt Electrodes from Pt L23 X-ray Absorption Spectroscopy. J. Phys. Chem. B 109, 8076-8084 (2005).

28. Teliska, M. Determination of H, O and OH Chemisorption Sites on Platinum and Platinum-based Alloy Electrodes in an Electrochemical Cell using in situ X-ray Absorption Spectroscopy, Ph.D. Thesis, The George Washington University, Washington, D.C., 2004.

29. Teliska, M., O'Grady, W. E. & Ramaker, D. E. Determination of H Adsorption Sites on Pt/C Electrodes in HClO4 from Pt L23 X-ray Absorption Spectroscopy. J. Phys. Chem. B 108, 2333-2344 (2004).

30. Teliska, M., Murthi, V. S., Mukerjee, S. & Ramaker, D. E. Site-Specific vs Specific Adsorption of Anions on Pt and Pt-Based Alloys. J. Phys. Chem. C 111, 9267-9274 (2007).

31. Scott, F. J., Mukerjee, S. & Ramaker, D. E. CO Coverage/Oxidation Correlated with PtRu Electrocatalyst Particle Morphology in 0.3 M Methanol by In Situ XAS. J. Electrochem. Soc. 154, A396 (2007).


33. Arruda, T. M., Shyam, B., Ziegelbauer, J. M., Ramaker, D. E. & Mukerjee, S. In situ XAS Investigation of Electrocatalyst Surface Poisoning by Halides. ECS Transactions 11, 903 (2007).

34. Ramaswamy, N., Hakim, N. & Mukerjee, S. Degradation mechanism study of perfluorinated proton exchange membrane under fuel cell operating conditions. Electrochmica Acta 53, 3279 (2008).

35. McBreen, J., O'Grady, W. E., Pandya, K. I., Hoffman, R. W. & Sayers, D. E. EXAFS study of the nickel oxide electrode. Langmuir 3, 428-433 (1987).

36. Newville, M. IFEFFIT: Interactive XAFS Analysis and FEFF Fitting. J. Synchrotron. Radiat. 8, 322-324 (2001).

37. Newville, M., Livins, P., Yacoby, Y., Stern, E. A. & Rehr, J. J. Near-edge x-ray-absorption fine structure of Pb: A comparison of theory and experiment. Phys. Rev. B 47, 14126 (1993).

168

38. Ramaker, D. E., Mojet, B. L., Miller, J. T. & Koningsberger, D. C. An atomic X-ray absorption fine structure study of the influence of hydrogen chemisorption and support on the electronic structure of supported Pt particles. Top. Catal. 10, 157-165 (2000).

39. van Dorssen, G. E., Koningsberger, D. C. & Ramaker, D. E. Separation of double-electron and atomic XAFS contributions in x-ray absorption spectra of Pt foil and Na2Pt(OH)6. J. Phys.: Condens. Mat. 14, 13529-13541 (2002).

40. Janin, E., von Schenck, H., Göthelid, M. & Karlsson, U. O. Bridge-bonded atomic oxygen on Pt(110). Phys. Rev. B 61, 13144-13149 (2000).

41. Stamenkovic, V., Marikovic, N. M. & P. N. Ross, J. Structure-relationships in Electrocatalysis: Oxygen Reduction and Hydrogen Oxidation Reactions on Pt(111) and Pt(100) in Solutions Containing Chloride Ions. J. Electroanal. Chem. 500, 44-51 (2001).

42. Li, N. & Lipkowski, J. Chronocoulometric Studies of Chloride Anion Adsorption at the Pt(111) Electrode Surface. J. Electroanal. Chem. 2000, 95-102 (2000).

43. Marković, N. M., Gasteiger, H. A., Grgur, B. N. & Ross, P. N. Oxygen ReductionRreaction on Pt(111): Effects of Bromide. J. Electroanal. Chem. 467, 157-163 (1999).

44. Jerkiewicz, G., Vatankhah, G., Lessard, J., Soriaga, M. P. & Park, Y.-S. Surface-oxide growth at platinum electrodes in aqueous H2SO4. Reexamination of its mechanism through combined cyclic-voltammetry, electrochemical quartz-crystal nanobalance, and Auger electron spectroscopy measurements. Electrochim. Acta 49, 1451-1459 (2004).

45. Zolfaghari, A., Conway, B. E. & Jerkiewicz, G. Elucidation of the effects of competitive adsorption of Cl−and Br− ions on the initial stages of Pt surface oxidation by means of electrochemical nanogravimetry. Electrochim. Acta 47, 1173-1187 (2002).

46. Ramaker, D. E., Oudenhuijzen, M. K. & Koningsberger, D. C. Strong Support Effects on the Insulator to Metal Transition in Supported Pt Clusters as Observed by X-ray Absorption Spectroscopy J. Phys. Chem. B 109, 5608-5617 (2005).

47. Kinoshita, K. Particle Size Effects for Oxygen Reduction on Highly Dispersed Platinum in Acid Electrolytes. J. Electrochem. Soc. 137, 845-848 (1990).

48. Benfield, R. E. Mean coordination numbers and the non-metal–metal transition in clusters. J. Chem. Soc. Faraday Trans. 88, 1107 (1992).

49. Hardeveld, R. v. & Hartog, F. The statistics of surface atoms and surface sites on metal crystals. Surf. Sci., 15, 189 (1969).

169

50. Hardeveld, R. v. & Monfoort, A. V. The influence of crystallite size on the adsorption of molecular nitrogen on nickel, palladium and platinum: An infrared and electron-microscopic study. Surf. Sci. 4, 396 (1969).

51. Stonehart, P. Bur. Bunsenges Phys. Chem., 94, 913 (1990).

52. Mukerjee, S. & McBreen, J. Effect of particle size on the electrocatalysis by carbon-supported Pt electrocatalysts: an in situ XAS investigation. J. Electroanal. Chem. 448, 163-171 (1998).

53. Balbuena, P. B. et al. Adsorption of O, OH, and H2O on Pt-Based Bimetallic Clusters Alloyed with Co, Cr, and Ni Adsorption of O, OH, and H2O on Pt-Based Bimetallic Clusters Alloyed with Co, Cr, and Ni. J. Phys. Chem. A 108, 6378 (2004).

54. Xu, Y., Ruban, A. V. & Mavrikakis, M. Adsorption and Dissociation of O2 on Pt−Co and Pt−Fe Alloys. J. Am. Chem. Soc. 126, 4717 (2004).

55. Garcia-Araez, N., Climent, V., Herrero, E., Feliu, J. & Lipkowski, J. Thermodynamic studies of chloride adsorption at the Pt(111) electrode surface from 0.1 M HClO4 solution. J. Electroanal. Chem. 576, 33-41 (2005).

56. Yadav, A. P., Nishikata, A. & Tsuru, T. Effect of halogen ions on platinum dissolution under potential cycling in 0.5 M H2SO4 solution. Electrochim. Acta 52, 7444 (2007).

170

Chapter 4

Effect of RuOxHy Island Size on PtRu Particle Aging in Methanol∗

4.1 Introduction

The cost, performance levels, vulnerability to poisoning, and durability of Pt-based

Proton Exchange Membrane (PEM) fuel cells have each in part kept them from large-

scale commercialization. Earlier efforts concentrated on cost and performance levels, but

as the particle size and loading of modern electrocatalysts have decreased to meet the first

two issues, and bimetallics have been introduced to meet the third issue, the emphasis has

turned to the durability of the bimetallic particles. The degradation with time of a PEM

fuel cell involves both the membrane (e.g. Nafion), and the Pt-M (M = Ru, Co, Ni, Cr,

Mo, etc.) electrocatalysts, but of these the most studied are the long term changes to the

electrocatalysts.1 The degradation of the catalyst, resulting in current decay over time (i.e.

in chronoamperometry experiments), can involve surface M atom rearrangements,

particle growth, and/or M and Pt atom dissolution at the anode and cathode.

∗ Published in the Journal of Physical Chemistry C, 2009, 113 (45), 19713-19721 Authors: Badri Shyam, Thomas Arruda, Sanjeev Mukerjee and David E. Ramaker Electrochemistry and XAS data collection were carried out by authors affiliated with Northeastern University. XAS data analysis was carried out by authors from The George Washington University.

171

Both Pt dissolution and agglomeration from pure Pt anodes and cathodes2-14 have

been extensively studied as well as the slow dissolution of Rh, Ru or Sn from supported

Pt-M bimetallic particles15-22 in acid media during voltammetric cycling or fuel cell

aging. While it is well-known that cathodic losses due to sluggish oxygen reduction

reaction (ORR) kinetics are much more significant than anodic overpotential losses, in

the case of the direct methanol fuel cell (DMFC) where methanol is oxidized on the PtRu

surface, recent findings23-25 reveal that the two processes are linked. The above

mentioned reports illustrated the degradation of the anode can eventually result in the

rapid deterioration of cathode performance, drastically reducing the overall efficiency of

the fuel cell. For example, in these PtRu catalysts, significant Ru dissolution from the

anode has been observed and the leached Ru ions were found not only in the polymer

electrolyte membrane, but also deposited on the surface of the Pt cathode. Cyclic

voltammograms of the anode and cathode in the above mentioned situation clearly

indicate that the DMFC cathode becomes more anode-like and quickly leads to a steep

loss in operating potential. Thus, it is of considerable interest to understand the processes

of catalyst aging and degradation, not only from a fundamental standpoint but also with

the aim of eliminating the various bottlenecks which keep such catalysts from widespread

use in fuel cells.

A broad array of electrochemical, microscopic, and spectroscopic techniques have

previously been used to study these aging effects. The mass loss from the anode/cathode

has been observed directly using an electrochemical quartz crystal microbalance

(EQCM)26, 27 as well as the presence of Pt or M atoms in the electrolyte using energy

172

dispersive x-ray analysis (EDX).18 Atom rearrangements or particle growth have been

observed with atomic force microscopy (AFM),7 x-ray diffraction (XRD),

2,4,18,19,21scanning electron microscopy (SEM), transmission electron microscopy (TEM) 2,

4, 7, 14, 19, 20 and x-ray absorption spectroscopy (XAS).17, 18, 28

The general mechanism for metal dissolution has been attributed to

oxidation/reduction of the surface atoms causing atom rearrangement and dissolution into

the electrolyte, 29, 30 and sometimes even (both Pt and Ru) crossover from the anode to the

cathode. 4,14,25,28 Detailed mechanistic information and kinetic models have been

presented to characterize the particle agglomeration and atom dissolution.29, 30 Despite

these numerous efforts and proposed models, many questions and issues remain

regarding the relative rates of agglomeration and dissolution at the anode and cathode,

and their dependence on potential, particle loading and morphology. A very important

question, for example, is how does the M atom dissolution depend on the RuOxHy island

size on the surface? This question has not been considered, as generally the island size is

unknown. Measuring the island size however is no trivial task. Very few methods are

available to do such measurements in situ where it must be done as ex situ samples are

almost always known to be fully oxidized.

As mentioned above, Ru dissolution from PtRu electrocatalysts during cycling in acid

media with methanol present has been reported previously.17 Holstein and Rosenfeld

observed Ru dissolution when the catalyst was cycled over a large window (0 - 1.3 V vs.

RHE) while little or no dissolution was observed over much narrower potential ranges.

Further, the dissolution was observed with catalysts loadings of ~ 1/10 of those typically

used in a DMFC, and subsequently at much lower current densities. Chen et al.

173

established that the anode potential under normal operation in a DMFC was benign for

the PtRu/C electrocatalyst, but in the case of deep discharge or short circuit, the anode

potential value could exceed 0.6 V which apparently was detrimental to the catalyst.16

Recent studies by Wang and coworkers using XPS, XRD and electrochemical methods

on PtRu catalysts have shown that while Pt is relatively stable, the Ru atoms undergo

substantial oxidation.31,32 This oxidative process suggests the existence of these

oxides/hydroxides of Ru may actually hinder the dissolution and agglomeration of Pt in

the catalysts, thus slowing down the aging/degradation process in these alloy catalysts.

We suspect this conclusion may be highly dependent on Ru island size. Others have

observed decay of current with time even under normal operation.25 Thus it has not yet

been fully established exactly which potential ranges are required to see significant Pt or

Ru dissolution/agglomeration, and what the effects of Ru island size might be on this

dissolution process. Recently it has been reported that covering the Pt particles with a

monolayer of Au significantly slows the Pt dissolution process at nearly all potentials.33

In this work, electrochemistry and XAS were used to study the aging of two

commercially available catalysts. In most XAS studies on alloys, the analysis chiefly

involves studying the extended x-ray absorption fine structure (EXAFS) region to obtain

changes in coordination numbers and bond lengths as a function of some electrochemical

treatment. Here, in addition to conventional XAS analysis, the recently developed Δµ-

XANES (x-ray absorption near-edge structure) technique is used to observe small

changes occurring on the surface of the catalysts.34 The primary scope of this work is to

examine Ru and Pt dissolution/ agglomeration of PtRu black under conditions quite

174

E, V vs. RHE0.40 0.45 0.50 0.55 0.60 0.65

i, m

A

0.0

0.2

0.4

0.6

0.8JM post 8 hr. CAJM initialTanaka post 8 hr. CATanaka initial

Figure 4.1 CO stripping data24 for the Johnson Matthey (red) and Tanaka (blue) catalysts before and after an 8-hour chronoamperometic test at 500 mV. The data have not been normalized for surface area.

175

similar to those in a fuel cell, with n (0, 20 and 40) potential-sweep cycles between 0.02

and 0.8 V. Two different commercial PtRu catalysts (Johnson-Matthey and Tanaka) were

utilized in 1M trifluoromethanesulfonic acid (TFMSA) with 0.3 M methanol (MeOH)

and their properties compared. Shown in Figure 4.1 are CO stripping curves for the two

catalysts before and after the 8-hour chronoamperometry (CA) session at 500 mV. The

data were collected at a scan rate of 10mV/s. In the case of the JM sample, there is a shift

in the onset potential for the CO stripping while there is no such shift for the Tanaka

samples. Further, a positive shift in the peak potential of ca.50 mV is observed in both

catalyst samples. It is clear that the two catalysts exhibit different aging characteristics.

To explain this difference in behavior, in situ XAS data were obtained on the samples and

their change with cycling observed using the Δµ-XANES and EXAFS analysis

techniques. It will be shown that the sensitivity of the Δµ-XANES technique indeed

allows changes in the particles to be observed in this potential range even after as few as

10-20 cycles.

4.2 Experimental Methods and Data Analysis

4.2.1 Electrode preparation and XAS cell assembly

Two different commercial PtRu (approximately 1:1 atomic ratio) based

electrocatalysts were used for preparation of the working electrodes. The first obtained

from Johnson-Matthey (HiSpec6000) and the second from Tanaka (TEC90110)

(hereafter referred to as JM and Tanaka). The working electrodes were prepared in-house

with a metal loading of 4.7 and 9.3 mg/cm2 of Pt and Ru respectively by hand painting

the inks (pre-weighed catalyst powder, deionized H2O and 5 wt% Nafion) onto carbon

cloth (PANEX 30, Zoltek Corperation) by a standard method described previously.35 The

176

total geometric area of the electrodes was 5 cm2. The metal loading was chosen based on

absorption cross sections for Pt and Ru to ensure an XAS step height of close to unity.

The in situ XAS cells used were similar to a previously reported design36 and assembled

by placing a Nafion 112 membrane between the PtRu working electrode and a Grafoil

(GrafTech International) counter electrode. In all cells the current collectors used were

0.5 mm Au wires (99.999%, Alfa Aesar) and mechanically pressed to the rear of each

electrode in a location that did not impinge the x-ray beam path in the photon window.

The cells were sealed using silicone gaskets (Auburn Chemical Co.). The electrolyte used

in these experiments was 1 M TFMSA (triply distilled, Strem Chemicals) with the

addition of 0.3 M methanol. An Autolab PGSTAT 30 potentiostat/galvanostat (Metrohm

USA, formerly Brinkmann Instruments) was employed for potential control in all

experiments.

4.2.2 In Situ XAS measurements

No catalyst pretreatment (i.e. potential cycling) was performed prior to XAS

measurements aside from soaking the working electrodes in electrolyte under vacuum for

1 hour prior to experiment for the purpose of wetting. Full range EXAFS scans were

collected (-200 eV below the edge to k = 18Ǻ-1 above the edge) at the Pt-LIII edge (11564

eV) as well as the Ru-K edge (22117 eV). A summary of the data collected is presented

in Figure 4.2. Briefly, data were collected at both edges at the indicated potentials after

0, 20 and 40 cycles. In each case, all the data at a given edge were first collected in 1M

TFMSA and then with the addition of 0.3M Methanol, wherein a fresh electrode from the

same batch of catalyst was used. In all, four electrodes were used for each catalyst in

order to perform the designated experiments (with and without methanol at the Pt and Ru

177

Ru K

0.24 0.70 0.840.54

20 40

1M TFMSA + 0.3M CH3OH

0Potential, V

1M TFMSAClean, reference

Electrolyte

0Cycles

Pt LIIIEdges

JM & TanakaCatalysts

Ru K

0.24 0.70 0.840.54

20 40

1M TFMSA + 0.3M CH3OH

0Potential, V

1M TFMSAClean, reference

Electrolyte

0Cycles

Pt LIIIEdges

JM & TanakaCatalysts

Figure 4.2 Summary of XAS data collected.

178

edges). Measurements were performed at beam line X23-A2 at the National Synchrotron

Light Source (Brookhaven National Laboratry, Upton, NY) which employs a piezo-

feedback stabilized Si(311) monochromator. XAS was collected in transmission mode

with the cell placed between two gas ionization chambers (I0, incident beam and I1,

transmitted beam) and a Pt foil/Ru powder (325 mesh, Alfa Aesar) between I1 and I2

(beam intensity for reference sample) . The Pt and Ru reference foil scans were used to

correct for any drift in the beam energy during data collection.

4.2.3 Electrochemical Measurements

All electrochemical measurements were carried out using 1M TFMSA (Strem

Chemicals), which was triply distilled by a method described elsewhere.37 For surface

area determination, aqueous solutions of 2 mM CuSO4 (Alfa Aesar) were prepared in 1M

TFMSA and employed by the method outlined by Green et al.38 Catalyst suspensions

were produced by mixing a pre-weighed quantity of Pt, PtRu or Ru/C, deionized H2O

(18.2 MΩ, Millipore MilliQ) and 5 wt% Nafion solution (suspended in low molecular

weight alcohols, LQ-1105, Ion Power Inc). The suspensions were sonicated for 15

minutes, stirred for two hours then deposited onto a 9 mm polycrystalline Au slug. The

total electrocatalyst target loading was 100 μg·cm-2, although some of the loosely bound

particles may have washed off the surface prior to testing. The Au slug was connected to

a threaded 316 stainless steel current collecting rod which never contacted the electrolyte.

The cells were comprised of a jacketed 50 cm-3 glass beaker with machined PTFE lid. In

all cases a sealed glass reference hydrogen electrode (RHE, 1M TFMSA internal) was

used and connected to the cell via a glass capillary which terminated at a fine porous frit

to minimize Cu2+, MeOH and Run+ crossing into the RHE. The counter electrode was a

179

Pt wire (Alfa Aesar, 99.999%) with a surface area of 1.7 cm2 as determined by

integrating the Hupd. Ru surface area was determined by immersing the slug into a

solution of Ar purged (UHP, Middlesex Gasses) 1M TFMSA + 2 mM CuSO4 and

polarizing to 0.3 V for 60 seconds. Subsequently, the potential was scanned anodically to

0.8 V. The Pt surface area measurement was carried out the same way except a starting

potential of 0.4V (also for 60 seconds) was used as Cu2+ will only deposit on Pt when

potentiostatically controlled in that region. The electrode was then removed from the

Cu2+ containing electrolyte, rinsed with deionized H2O and placed into an identical cell

containing clean 1M TFMSA. A total of 500 cyclic voltammograms (CVs) were

collected at a sweep rate of 50 mV s-1 between 0.02 – 0.8 V vs. RHE. Following the 500

cycles, surface area determinations were repeated by the procedure outlined above to

illustrate any loss of electrocatalyst as a result of cycling. CO stripping voltammograms

were collected by saturating the 1M TFMSA with gaseous CO (5.0 Grade, Middlesex

Gasses) and scanning the potential anodically at a sweep rate of 10 mV s-1.

4.2.4 XANES and EXAFS analysis

Analysis of the XANES region of the XAS data was carried out using the Δμ

technique34, 39-42 previously applied to adsorption of H, O/OH, and Cl on Pt43-45 and Pt-M

(M = Cr, Fe, Co, and Ni) cathodes46 and Pt-Ru anodes47 in an electrochemical cell, and

even to Pt and PtRu anodes in an operating direct methanol fuel cell.27 A brief summary

is given here for clarity and to highlight slight differences from the previous methods.

All XAS data were processed using the ATHENA code developed by Ravel and

Newville.48 The pre-edge background is removed using the AUTOBK algorithm,

described completely elsewhere,49 followed by normalization over the 50 to 150 eV

180

(relative to E0) range for XANES analysis. This procedure was carried out for both the

sample data in transmission mode (ln I0/I) and the reference foil data (ln I/Iref). The foil

data were then calibrated and aligned to the theoretical edge energy, and the resultant

energy differences were transferred to the sample data; i.e., the ΔE shifts determined for

the foil at any given potential is added to the energy of the sample data at the same

potential. This energy calibration corrects for shifts due to photon beam drift. This

energy calibration is crucial for the success of the Δμ technique to ensure full

cancellation of the atomic contribution in the XANES, which dominates the spectrum;

the resulting Δμ-XANES signal intensity is typically only about 1-5% of the total μ

signal.

The difference Δμ = μ(V) - μ(Vref) is generally determined by subtracting the μ-

XANES at an appropriate reference potential Vref, from μ−XANES at other potentials to

isolate the effect of adsorbates. The reference potential Vref is usually taken to be the

potential at which the electrode is relatively free of adsorbates. However, the optimal

choice of reference can change based on the nature of the inquiry, the sample, the

adsorption edge, and the operating conditions. In this work, the electrode in 1 M TFMSA

without methanol was used as the reference, at 0.54 V at the Pt edge (when the Pt is

relatively free of both H and O), and at 0.02 V at the Ru k-edge (when the Ru is mostly

free of Ox) with no cycling. Therefore, the Δμ signals (the notation used is the potential

followed by the number of cycles) were obtained from the following differences:

Δµ(V_cycles) = µ(V_cycles, MeOH) - µ (0.54_0, no MeOH). For Pt (Eq. 4.1)

Δµ(V_cycles) = µ(V_cycles, MeOH) - µ (0.02_0, no MeOH). For Ru (Eq. 4.2)

181

As will be shown in Figure 4.7, CO and O(H) were visible as different features by

this method, allowing for simultaneous indications of their coverages. The final Δμ

curves were background corrected (80 eV smooth) and smoothed to remove random

noise (5 eV) using a standard Savitzky–Golay smoothing routine with the indicated

energy range given in parentheses.

The EXAFS fitting analyses were performed using the ARTEMIS code,48 by

employing only 2 first-shell metal atom scattering paths for each sample (either Pt-Pt and

Pt-Ru or Ru-Ru and Ru-Pt) included at each edge. All fits were carried out on k2

weighted χ(k) data using a Kaiser-Bessel window over a k-range of 1.574 < k < 13.769

Å-1, and an R-window of 1.448 < R < 3.201 Å. Details of the method for extracting

coordination numbers (CN) been described elsewhere.47 Several fits were first carried out

on the data obtained, allowing all four parameters per path (N, R, σ, and Eo) to vary. It

was found that Debye-Waller factor (σ2) derived from the fits were in the range of 0.004-

0.006 Å2. Therefore σ was eventually fixed at the value of 0.005 Å2 for all fits (thereby

allowing only 6 parameters total to vary) to reduce scatter in the CNs, and FEFF 8.0 was

used to calculate all of the other necessary parameters including the many-body

amplitude reduction term S 20 (0.916 for Ru and 0.934 for Pt).

4.2.5 FEFF 8.0 calculations

The FEFF 8.0 code was used to model the adsorbate Δμ signatures. The Δµ(Ads)

was determined by subtracting the µ-XANES of a clean “Janin”26 type Pt4M2 cluster from

the µ-XANES of a cluster containing an adsorbate molecule in the atop, bridged or n-fold

position; i.e. Δμ(O) = μ(O/Pt4M2) - μ(Pt4M2). The Janin cluster, used in much of our

previous work,34, 40, 47, 50-55 was chosen here because it is the smallest cluster that contains

182

atop, bridged, fcc, and hcp sites. The bond distances used in the clusters were the same

as those in our previous FEFF 8.0 calculations.47, 56 In general, Pt-Pt & Pt-Ru distances

used were 2.77 Å as is known from crystallographic determinations. Oxygen in the atop

position was treated as OH (since the scattering from H has been shown to be

indiscernible)40, 54 while oxygen in an n-fold position was treated as O. This is consistent

with density functional theory (DFT) calculations,26, 57, 58 which show that OH prefers to

be singly coordinated, and O doubly or triply bonded to the Pt surface.

4.3 Results


Surface Area and the Effects of Potential Cycling. As described above, the

electrochemically active surface area for Pt and PtRu catalysts can be determined by

anodically stripping the underpotential deposited (upd) cupric ions that form a monolayer

on the catalyst surface. This process is illustrated in Figure 4.3, which shows CVs of

PtRu in the absence and presence of cupric ions. The CVs in the presence of cupric ions

reveal anodic peaks which correspond to the ‘stripping’ currents of Cu upd on the

indicated surface atoms. As evident from the figure, there is a potential dependence that

governs which surface atom will accept the Cu. For instance, if the potential is fixed at

0.4V, Cu will only deposit on Pt. However, if the potential is fixed at 0.3V it will deposit

on both the Ru and the Pt surface atoms. If the potential is cycled below 0.3V, a bulk Cu

layer is deposited and subsequently removed at 0.28V. Cyclic voltammograms for the

two PtRu materials being investigated are presented in Figure 4.4. The insets contain the

Cu stripping curves used to calculate the electrochemically active surface area for current

183

E, V vs. RHE0.2 0.4 0.6 0.8

i, μA

cm

-2

-5

0

5

10

15Bulk Cu

Ru

Pt

Figure 4.3 Cupric ion stripping voltammograms recorded in 1 M TFMSA + 2 mM CuSO4 taken at a sweep rate of 10 mV s-1. Cyclic voltammograms of a typical PtRu black catalyst in the presence (dot dashed) and absence of Cu ions (dotted) showing the various underpotentially deposited regions on both Pt and Ru sites as well as the bulk Cu deposition region.

184

density normalization and surface area analysis. As evident from the figure, both

materials undergo significant changes in surface area between cycles 5 and 50; the

integrated results presented in Table 4.1. Both materials show a decrease in the

magnitude of the Cu stripping off of Ru sites (E fixed at 0.3V), indicating loss of surface

Ru either through Ru dissolution or formation of RuOxHy islands as Cu ions will not

underpotentially deposit on RuOxHy but only on available, free metallic Ru sites. It is

worth stressing this latter point as it not only explains the Cu upd data but provides us

with a mechanistic insight into the aging process, all of which will be shown to be in

good agreement with results from the XAS analysis and electrochemical data. It should

be noted that the Tanaka PtRu surface area loss is more than double that of the JM (on

both the Pt and Ru). Interestingly, the Pt surface area of the Tanaka material decreased

considerably whereas the JM Pt surface area actually increased slightly. The increase in

Pt surface area for JM is likely the result of Ru dissolution from the surface of the particle

unveiling newly uncovered Pt surface atoms. The loss of Pt surface area in the Tanaka

sample can be attributed to either a) Pt dissolution or b) the formation of larger Ru

islands which mask more of the Pt surface sites. Both of the above possibilities are

discussed in greater detail in the EXAFS section below.

There are also significant changes to the CV profiles illustrating particle aging as a

result of potential cycling. For instance, in Figure 4.4 which overlays the three CVs for

the JM and Tanaka PtRu samples at 5, 50 and 500 cycles, there is a very noticeable

change in the Hupd region. At cycle 5 for the JM sample there are no discernable peaks

typical of H desorption, however, as the material is cycled it develops features which are

not unlike the Hupd region of pure Pt; for the Tanaka samples however, it appears to stay

185

Table 4.1 Summary of Cu stripping results for surface area analysis

Catalyst Cycle # Surface Metal Surface Area (cm2)

% Change

0 Ru 12.05 500 Ru 10.61 - 12.0 %

0 Pt 5.69 JM

500 Pt 5.85 + 2.74 %

0 Ru 6.91 500 Ru 5.20 - 24.7 %

0 Pt 3.40 Tanaka

500 Pt 2.98 - 12.4 %

Surface area calculations were done by integrating the Cu stripping peaks for the catalysts as outlined by Green et al.38

186

E, V vs. RHE0.0 0.2 0.4 0.6 0.8

i, μA

.cm

-2

-80

-60

-40

-20

0

20

40

Cycle 5 Cycle 50 Cycle 500

-100

-50

0

50

Cycle 5 Cycle 50 Cycle 500

E, V vs. RHE0.3 0.4 0.5 0.6 0.7 0.8

I, A

0

5e-5

1e-4

2e-4

2e-4

3e-4

3e-4

0 Cycles500 Cycles

E, V vs. RHE0.3 0.4 0.5 0.6 0.7 0.8

I, A

0.0

2.0e-5

4.0e-5

6.0e-5

8.0e-5

1.0e-4

1.2e-4

1.4e-4

1.6e-4

0 Cycles500 Cycles

a

b

Figure 4.4 Cyclic voltammograms for (a) Johnson Matthey, and (b) Tanaka catalysts after 5, 50 and 500 cycles showing differences in aging properties. The actual data is the same as in reference 24 except normalized to initial Cu stripping surface area for cycles 5 and 50, and normalized to Cu stripping area post cycle 500 for the 500th scans. The inset shows the detail of Cu upd data for the two catalysts.

187

more alloy-like (i.e. no discernable features). These observations may provide some

information about the size of the clear Pt regions; i.e. larger in the JM case because of the

smaller Ru islalnds. Larger regions of free Pt will more clearly resolve the H adsorption

at faces and corner/edges that cause these separate features. These results are consistent

with the Cu upd data above, which suggest the JM material undergoes Ru dissolution to

uncover more Pt-like surface. The Tanaka low potential region also reveals a larger

decrease in Hupd current than the JM suggesting more growth in Ru(OxHy) coverage,

consistent with the lack of Pt-like CV features. This is also consistent with the Cu

stripping data and is suggestive of case b mentioned above.

4.3.2 EXAFS

A representative FT-EXAFS least-squares fit is shown in Figure 4.5 for the Ru K-

edge data in R space and the agreement between experiment and theory of a two-path Ru-

Ru and Ru-Pt fit. Since the σ2 values were held constant at 0.005 Å2, this fit was actually

obtained using 6 parameters (N, R, and Eo for each path). Figure 4.6 illustrates the Ru-

Ru and Ru-Pt coordination numbers (CNs) obtained from fits similar to that in Figure 4.5

for 3 different potentials, 0.0, 0.24 and 0.54 V (relatively small changes with potential

were observed). The results in Figure 4.6 are the average of those obtained at 3 potentials

after 0, 20, and 40 potential cycles between 0.02 and 0.8 V (2.6 min per cycle). Note that

the ratio Ru-Ru/Ru-Pt is much larger for the Tanaka sample compared to the JM sample.

Also, it is noticeable that the CNs increase with cycling and that this increase is much

larger for the JM sample. The increase in Ru-Pt CN with cycling in the JM samples is

consistent with Ru dissolution assuming the Ru with low CN at the surface preferably

leaches. The increase in Ru-Ru CN suggests that the remaining Ru islands are larger,

188

k, Å-1

0 2 4 6 8 10 12 14 16 18

χ(k)

*k2 , Å

-2

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

Tanaka, 0.54 V, 20 cyclesBest Fit

FT [ χ2

(k) ]

R, Ǻ

0 1 2 3 40.0

0.2

0.4

0.6

0.8

1.0

1.2Tanaka, 0.54V, 20 cyclesFit

FT [ χ2

(k) ]

R, Ǻ

0 1 2 3 40.0

0.2

0.4

0.6

0.8

1.0

1.2Tanaka, 0.54V, 20 cyclesFit

Figure 4.5 Representative k-space (top) and Fourier Transformed (bottom) EXAFS data and fit for Tanaka PtRu sample at the Ru-K edge taken at 0.54 V after 20 cycles. The 2 path (Ru-Ru and Ru-Pt) fit was performed in R-space (1.574 < k < 13.769 Å-1, k2 weighted.

189

0 10 20 30 40Cycles

3.0

3.2

3.4

3.6

3.8

4.0

4.2

RuPtRuRu

JMTanC

oord

inat

ion

Num

ber

Figure 4.6 Changes in average Ru-Ru and Ru-Pt CNs with cycling for both the JM and Tanaka catalysts. Error bars of ±0.1 are representative of the relative error, but the absolute error is probably larger.

190

either due to Ostwald ripening, or just the lower coordinated Ru leaving the surface. The

CN ratio Ru-Ru/Ru-Pt reflects the Ru island size on the particle surface, since

presumably the Ru in the interior of the cluster is more alloyed with Pt. Thus the Ru

islands in the Tanaka sample are significantly bigger and therefore little change is noticed

with cycling. In contrast the small islands on the JM catalyst undergo Ru dissolution as

well as particle ripening (enlargement) with cycling.

The data in Figure 4.6 strongly suggest that much more Ru exist on the particle

surfaces of the Tanaka samples and thus the Ru islands are larger. This is entirely

consistent with the CNs obtained at the Pt edge as summarized in Table 4.2. Since they

did not change much with cycling we report just the average with potential and cycling

(i.e. average of 9 results, 3 potentials of 3 cycling levels each) in Table 4.3. Note that the

sum of the CNs, Pt-Pt + Pt-Ru, are nearly the same indicating nearly identical particle

sizes (around 1.0-1.5 nm based on spherical particles)59 in the two samples. However, the

Tanaka samples have larger Pt-Pt and smaller Pt-Ru CNs which is consistent with more

Ru existing at regions of lower coordination, i.e. at the surface.

4.3.3 Δµ-XANES Analysis

In order to more easily understand the Δµ-XANES data, Figure 4.7 shows

representative (qualitative) coverages for CO and O(H) on Pt and Ru as obtained from

our previous studies of 3 different PtRu catalysts in methanol.60 It reveals that at

potentials below 0.3 V, the CO coverage on Pt and even on the Ru islands (the latter true

only if Ru island clusters are relatively large) is nearly complete in methanol. At

potentials above 0.6 V, the coverage of OH and O are nearly complete, and the CO

191

Table 4.2 Summary of coordination numbers obtained from Pt-L III edge data.*

Sample Pt-Pt Pt-Ru Pt-Pt + Pt-Ru η = 1 – NPt-Ru / NRu-Pt Tanaka 4.8 2.0 6.8 ~ 0.4

JM 3.9 2.8 6.7 ~ 0.3 *The relative uncertainty in CN is ca. 0.1, but the absolute uncertainty is larger.

192

0 .1 .2 .3 .4 .5 .6 .7 .8

CO/Ru

OH/PtnRu

OH/PtOH/Ru

Ru:BF PtnRu:Dsl Pt

Pot (V RHE)

Cov.

CO/Pt

Figure 4.7 Representative CO and O(H) coverages for a PtRu anode in methanol as reported previously using the Δµ-XANES technique.60

193

coverage has dropped to unobservable levels. This occurs because of water activation

(Eq. 4.3) and the widely accepted mechanism for CO oxidation (Eq. 4.4):61-64

H2O + M → OH/M + H+ + e- (Eq. 4.3)

CO/Pt +OH/M → CO2 + Pt + M (Eq. 4.4)

Here, M can be either surface Pt or Ru atoms, and as illustrated in Figure 4.7, water

activation occurs at the lowest potential on Ru, followed by Pt near the Ru islands (PtnRu)

and finally on the Pt atoms. Figure 4.7 also indicates generally where the different CO

oxidation mechanisms (differentiated by the source of the OH) dominate. The

bifunctional (BF) mechanism dominates below 0.3 V when the facilitating OH comes

from the Ru, the direct surface ligand (DsL) mechanism dominates in the range 0.3 -0.5

V when the OH comes from the Pt atoms near the Ru (PtnRu), and the direct mechanism

above 0.5 V when the OH comes generally from the Pt atoms. Further, we found that

large Ru islands generally are more oxidized and hence exert a larger ligand effect on the

nearby Pt atoms, while smaller Ru islands experienced a “reverse” ligand effect from the

Pt, and were less reactive and were not oxidized at lower potentials. Therefore, the small

Ru atoms were available to activate water below 0.3 V and hence carry-out the BF

mechanism, while the larger Ru islands were heavily oxidized making the BF mechanism

inactive. The BF mechanism is seemingly more effective when small Ru islands exist

and the DsL mechanism when the larger Ru(OxHy) islands exist. Further, an educated

guess is made regarding the valence state of the Ru on the surface: what we term ‘heavily

oxidized’ most likely contain Ru in an oxidation state of 1.5-2.0 (as RuO) while those

194

Table 4.3 Summary of results from electrochemical, Cu upd and x-ray absorption data

Tanaka Johnson Matthey

Large Ru coverage and existing as larger RuOxHy islands on surface.

Smaller Ru coverage and existing in highly dispersed metallic Ru islands.

Islands relatively stable to Ru dissolution and growth, but RuOxHy – Pt interface regions grows as more Ru moves to surface.

Significant Ru dissolution and island growth and/or agglomeration with islands becoming more oxidized and leaving larger Pt open regions.

CO oxidation chiefly occurs via ligand-effect mechanism.

Enables CO oxidation chiefly via bifunctional mechanism initially but converts to a ligand-effect mechanism on aging.

195

termed ‘slightly oxidized’ really indicate primarily metallic Ru islands and probably have

a much lower oxidation state of between 0.30-0.60.

Figures 4.8 and 4.9 show a sampling of the Δµ-XANES results taken at both the Ru K

and Pt LIII edges. With 3 different potentials and 3 cycling levels at both the Ru K and Pt

LIII edges on 2 samples, a total of 36 different Δμ curves were constructed; only 11 are

shown in Figures 8 and 9 for clarity and to reveal the most important trends in the data.

Fig. 8 includes Ru K edge data using a “clean” PtRu electrode in only TFMSA at 0.02 V

as the reference, i.e., obtained using Equation 4.2, along with FEFF 8.0 calculated Δμ

signatures65 for CO/Ru and OH/Ru obtained as described above. The results at 0.02 V

and no cycling for the Tanaka and JM samples are compared in Figure 4.9b. Note the

division by 8 to provide comparable magnitudes for the Tanaka and JM samples. This is

consistent with much more Ru at the surface in the Tanaka sample since the Δμ intensity

primarily reflects changes at the surface, and as suggested by Figure 4.8 at 0.02 V the Ru

islands are covered with CO. The Δμ data for the Tanaka sample shows a CO/Ru

signature, consistent with that found previously in methanol on large Ru islands. At 0.54

V the Ru is expected to be nearly fully oxidized, and indeed the data in Fig. 8b are

consistent with O(H)/Ru.

Results are also shown at the Pt LIII edge (Figure 4.9), obtained using Equation 4.1

and compared with theoretical Δμ signatures obtained previously27, 60 for O(H)/Pt and

CO/Pt using the procedures described above. The 3 peaks in the O(H)/Pt signature

correspond to O(H)/Pt near Ru, O(H)/Pt, and O/Pt respectively enabling these species to

be separately observed, and hence the representative results in Figure 4.7. Note that the

Δμ curves for the Tanaka sample are much smaller, consistent with much of the Pt being

196

-0 010

-0.006

-0.002

0.002

0.006

0.010

.02_0

.02_40

.02_0(Tan)/8.

-2 0 -1 0 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0-0 .0 3 0

-0 .0 1 8

-0 .0 0 6

0 .0 0 6

0 .0 1 8

0 .0 3 0

E(eV rel. edge)

Δμ

Δμ

Ru K JM

Ru K JM

.54_40.54_40

CO/Ru

OH/Ru

Figure 4.8 Comparison of Δμ(V_cycles) at the Ru K edge, using Equation 4.3b. Also

shown are theoretical Δμ signatures denoted OH/Ru and CO/Ru. Note that the Δμ for the Tanaka sample has been scaled by a factor of 8 to place it on the same scale.

197

-0 03

-0.02

-0.01

0.00

0.01

0.02

-10 0 10 20 30E (eV rel. edge)

-0.04

-0.02

0.00

0.02

0.04

Δμ

CO/Pt

O(H)/Pt Pt L3 Tan

.02_0.02_40

Δμ

Pt L3 JM.24_0

.54_0.54_40.24_40

Figure 4.9 Comparison of Δμ(V_cycles) lineshapes at the Pt LIII edge using Eq. 4.3a. Also indicated are theoretical signatures for O(H)/Pt,27, 60 and CO/Pt. The three features in the OH/Pt signature correspond to OH/Pt near a Ru site, OH/Pt away from the Ru islands, and O/Pt.

198

covered by the large Ru islands. Thus the EXAFS and Δμ data consistently reveal that

the Tanaka sample has a much larger component of Ru and hence larger islands on the

surface when compared with the JM samples.

4.4 Discussion

The results above straightforwardly show that much more Ru initially exists on the

surface of the Tanaka samples compared with the JM samples. But careful comparison of

the changes in Δμ reveal much more interesting changes with potential cycling and the

nature of the Ru islands themselves. Such observations shall be discussed in greater

detail in the sections that follow.

4.4.1 Oxidation state of Ru islands

The ratio of the Pt-Ru/Ru-Pt CN’s reflect the relative oxidation level of the Ru islands

in the two samples. To understand this, consider a simple ensemble of 3 Pt atoms

coordinated to 1 Ru. There will be 3 Pt-Ru “interactions” so the average Pt-Ru CN = 1

and Ru-Pt = 3 in this case, i.e. each Pt sees one Ru atom and each Ru atom 3 Pt atoms.

Therefore the ratio Pt-Ru/Ru-Pt (i.e. 1/3) reflects the ratio of Ru/Pt atoms. However, the

catalysts contain an equal number of Pt and Ru atoms, but the oxidized Ru atoms are

essentially taken out of the metal-metal scattering if they are surrounded by O atoms.

Therefore this ratio reflects the fraction of unoxidized Ru atoms, and η = 1-(Pt-Ru/Ru-Pt)

the fraction oxidized. These oxidized fractions are listed in Table 4.2. Of course

considerable Ru may exist in the interior of the particles, so this is not the total fraction of

Ru at the surface that is more oxidized (those will be much higher), but these fractions

are consistent with a greater fraction of Ru at the surface in the Tanaka samples. It is also

worth mentioning that the larger islands are more oxidized than the smaller, more

199

dispersed ones, which tend to be more metallic in nature, as found previously. 60 Another

recent study which employed XAS (ex situ) and TEM on PtRu catalysts also provide

support for the existence of oxidized Ru islands in PtRu catalysts.66 Is it possible to make

any comments on the size of these Ru islands on the catalysts? While the absolute size of

the islands likely cannot be determined by any of the methods used in this study, we

would like to point out that it is the disparity in island size between the highly dispersed,

smaller and largely metallic Ru islands, and the larger and more oxidized Ru islands that

are observed and account for some of the differences in the aging processes occurring on

the two catalysts.

The Δμ magnitude and signatures for the Ru islands on the Tanaka sample do not

show much change with cycling, as expected for larger islands. Further, all signatures

reflect CO/Ru at all potentials ≤ 0.54 V, consistent with Figure 4.7. But these larger

RuOxHy islands exert a larger electronic or ligand effect on the nearby Pt atoms27, 60

increasing the oxophilicity of those Pt atoms. This trend is clearly seen in Figure 4.9b

showing some O(H)/Pt near the Ru islands already in the Tanaka samples at 0.02 V,

compared with the JM sample requiring 0.54 V for this to be evident. The following

sections discuss the catalyst aging process as a metal dissolution-agglomeration process

as a function of number of cycles and a visual summary of the two different aging

processes are depicted in Figure 4.10

4.4.2 Ru dissolution and agglomeration

At 0.54 V, any Ru at the surfaces is expected to be covered with O (see Figure 4.7),

and this is consistent with the Ru Δμ signatures in Figure 4.9b for the JM samples.

However, it shows that the magnitude of this signature decreases with cycling, indicating

200

Figure 4.10 Schematic representation of the primary PtRu nanoparticle aging processes occurring in the (a) Johnson Matthey and (b) Tanaka catalyst.

a.

b.

Pt

Ru

Slightly oxidized

Heavily oxidized RuAging

Run+

Run+ ions

201

that Ru is leaving the surface; i.e. dissolution of Ru. At 0.02 V however, the Δμ

signature obtained at the Ru edge reflects CO/Ru at zero cycles, but more of an O(H)

signature with a component of CO/Ru after 40 cycles. This clearly suggests that the Ru

islands, as they get larger, become more oxidized as indicated above. The growth in Ru-

Ru CN number seen in the EXAFS can come from both particle agglomeration and Ru

dissolution, since the smallest Ru islands presumably undergo dissolution the fastest. The

Δμ data clearly show that both are occurring in the JM sample. Similar phenomena have

been observed using High-Resolution Transmission Electron Microscopy (HR-TEM) and

Secondary-Ion Mass Spectrometry (SIMS) in a recent study on the decomposition of

PtRu anode catalysts.67

4.4.3 Pt dissolution and agglomeration

Although not reported in Table 4.2, the Pt-Pt CNs in the Tanaka sample do show

some variation with cycling (5.1 (0 cycles) down to 4.8 (20 cycles) and then back to 5.2

(40 cycles)). This suggests that the smaller PtRu particles initially underwent dissolution,

followed by agglomeration of some of the remaining particles. The Δμ data are consistent

with this change. The Δμ signatures at 0.02 V (Figure 4.9), when the Pt surface should be

nearly covered by CO, reveal a decrease in magnitude with cycling and indeed, even the

presence of a small amount of OH/Pt at 0.54 V after 40 cycles. This suggests an increase

in the efficiency of CO oxidation with cycling. This OH/Pt near the Ru islands is directly

evident in the Δμ(0.02_40) signature, showing the strong ligand effect of the large Ru

islands in the Tanaka sample. The increase in CO oxidation (i.e. reduction in CO on

surface) with cycling may either be due to elimination of the smaller PtRu particles,

202

where the CO oxidation might be less efficient, or at least indicates a smoothing of the Pt

surface perhaps from Pt dissolution of the corners or edges. This would enable a greater

fraction of more mobile CO, or more likely simply a reduction in Pt surface area

consistent with the Cu upd data in Table 4.1. Komanicky et al. found that the nanofaceted

surface dissolves faster indicating the edges and corners are the main sources of

dissolution. 7 In any event, the extent of CO on Pt appears to decrease in magnitude with

cycling for the Tanaka catalysts (Figure 4.9b). This is in contrast to that occurring on the

JM catalysts, where Figure 4.9a clearly shows the opposite trend, consistent with the Cu

upd data in Table 4.1.

4.4.4 Interpretation of CO stripping curve changes

We can now understand the changes in the CO stripping curves for the catalysts

before and after an 8-hour CA test as shown earlier in Figure 4.1. The Ru particles on the

Tanaka samples are heavily oxidized to begin with and therefore show no significant

change in RuOxHy content on aging. There is consequently hardly any change in the

onset potential for the CO oxidation as there is no change in the nature of the Ru islands.

The increased CO oxidation current and peak potential is likely due to increased

availability of the interface Ru(OxHy)-Pt sites where the ligand mechanism is active,

consistent with the schematic in Figure 4.10 showing more interface regions and less

clear Pt regions. The JM sample on the other hand, has much smaller, metallic Ru islands

on the surface to begin with, and on aging, undergoes dissolution and Ru island growth

via oxidation and agglomeration. The growth in island size changes the islands from

mostly Ru to Ru(OxHy), and hence the dominant CO oxidation mechanism changes from

the bifunctional to the direct surface ligand effect, which moves CO oxidation to slightly

203

higher potential as illustrated in Figure 4.7. It has been shown previously that available

RuOxHy species are critical for the CO oxidation properties of a PtRu alloy catalyst. 68, 69

Further, it is interesting to note that the CO stripping curves for the aged JM catalyst

begins to look more like the CO stripping curves for the Tanaka catalyst. This is

consistent with the fact that more of the Ru islands are now bigger and oxidized, just as

the Tanaka catalysts were in their initial state prior to the aging process.

4.5 Conclusions

The electrochemical, EXAFS and Δµ-XANES analyses consistently show that the

Tanaka sample has much more Ru segregated to the surface, exist in larger islands and

are present in more oxidized, stable Ru(OxHy) forms. The smaller Ru islands in the JM

sample were found to undergo faster dissolution of Ru as well as agglomeration with

cycling or chronoamperometric aging in methanol. The findings from this work are

summarized in Table 4.3. These results suggest that the smaller Ru islands, which

facilitate CO oxidation more favorably via the bi-functional mechanism at lower

potential, are relatively unstable in methanol at the surface of unsupported PtRu particles.

Therefore in a DMFC, larger Ru islands, which are less susceptible to dissolution and

induce a larger ligand effect (albeit at somewhat higher potential compared to the BF

mechanism) will be much more stable and effective. They also corroborate previous

findings that available ruthenium oxide and hydroxide phases rather than metallic Ru

along with Pt are essential for stable CO oxidation properties of a PtRu catalyst.

It could certainly be argued that with the proposed aging mechanisms evident in this

work, the JM particles should eventually become more like the Tanaka catalysts,

consistent with the CO oxidation curves in Figure 4.1. However, the CVs in Figure 4.4

204

suggest that there are still significant differences after cycling. It is thus apparent that in

the case of the JM catalysts, while some of the smaller Ru particles are lost to dissolution

and others grow in size on the surface due to Ostwald ripening and agglomeration, the

islands are still a bit smaller and hence less oxidized compared with those found on the

Tanaka catalysts. The XAS and CO stripping results above show that the Tanaka sample

in comparison with the JM catalyst, actually showed some signs of improvement after 40

cycles, consistent with some RuOxHy island growth even in the Tanaka catalysts.

4.6 Acknowledgments

Financial support for this project was provided by the Army Research Office via both

single investigator and Multi University Research Initiatives (P.I. Case Western Reserve

University). The authors are also grateful for the use of beamline X23-A2 at the National

Synchrotron Light Source, Brookhaven National Laboratory, which is supported by the

U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under

Contract No. DE-AC02-98CH10886.

205

4.7 References

1. Fuel Cells Durability and Performance: Stationary, Automotive, Portable (William Andrew Applied Science Publishers Norwich, NJ, 2006).

2. Borup, R., Davey, J., Wood, D., Garzon, F. & Inbody, M. PEM fuel cell durability. DOE hydrogen program report, VII.I.3, 1039 (2005).

3. Dam, V. A. & de Bruijn, F. J. Electrochem. Soc. 154, B495 (2007).

4. Ferreira, P. J. et al. Journal of the Electrochemical Society 152, A2256 (2005).

5. Johnson, D. C., Napp, D. T. & Bruckenstein, S. Electrochim. Acta 15, 1493 (1970).

6. Kinoshita, K., Lundquist, J. T. & Stonehart, P. J. Electroanal. Chem. 48, 157 (1973).

7. Komanicky, V. et al. Journal of the Electrochemical Society 53, B446 (2006).

8. Ota, K., Nishigori, S. & Kamiya, N. J. Electroanal. Chem. 247, 205 (1988).

9. Ralph, T. R. & Hogarth, M. P. Platinum Met. Rev. 46, 117 (2002).

10. Rand, D. A. J. & Woods, R. J. Electroanal. Chem. 36, 57 (1972).

11. Wang, X., Kumar, R. & Myers, D. J. Electrochem. and Solid-State Lett. 9, A225 (2006).

12. Wilson, M. S., Garzon, F., Sickafus, K. E. & Gottesfeld, S. J. Electrochem. Soc. 140, 2872 (1993).

13. Xie, J., Wood, D. L., More, K. L., Atanassov, P. & Borup, R. L. Journal of the Electrochemical Society 152, A104-A113 (2005).

14. Yasuda, K., Taniguchi, A., Akita, T., Ioroi, T. & Siroma, Z. Phys. Chem. Chem. Phys. 8, 746 (2006).

15. Antolini, E., Salgado, J. R. C. & Gonzalez, E. R. J. Power Sources 160, 957 (2006).

16. Chen, W. et al. J. Power Sources 160, 933 (2006).

17. Holstein, W. L. & Rosenfeld, H. D. J. Phys. Chem. B 109, 2176 (2005).

18. Hwang, B. J. et al. J. Am. Chem. Soc. 127, 11140 (2005).

19. Jeon, M. K. et al. J. Power Sources 158, 1344 (2006).

206

20. Liu, J. et al. Phys. Chem. Chem. Phys. 6, 134 (2004).

21. Lukaszewski, M. & Czerwinski, A. J. Electroanal. Chem. 589, 38 (2006).

22. Sarma, L. S. et al. J. Power Sources 167, 358 (2007).

23. Christian, E., Piotr, P., John, D. & Piotr, Z. Recoverable Cathode Performance Loss in Direct Methanol Fuel Cells. J. Electrochem. Soc. 153, A171-A178 (2006).

24. Lajos, G., Nazih, H., Brian, H. & Sanjeev, M. Dissolution of Ru from PtRu Electrocatalysts and its Consequences in DMFCs. ECS Transactions 3, 607-618 (2006).

25. Piela, P., Eickes, C., Brosha, E., Garzon, F. & Zelenay, P. J. Elecrochem. Soc. 151, A2053 (2004).

26. Janin, E., von Schenck, H., Gothelid, M., Karlsson, U. O. & Svensson, M. Phys. Rev. B 61 13144 (2000).

27. Scott, F. J., Roth, C. & Ramaker, D. E. J. Phys. Chem. C 111, 11403-11413 (2007).

28. Murthi, V. et al. Meet. Abstr. - Electrochem. Soc. 702, 403 (2007).

29. Darling, R. M. & Meyers, J. P. J. Elecrochem. Soc. 150, A1523 (2003).

30. Anderson, A. B. & Shiller, P. J. Phys. Chem. B 102, 2696 (1998).

31. Wang, Z.-B. et al. Catalyst failure analysis of a direct methanol fuel cell membrane electrode assembly. J. Power Sources 177, 386-392 (2008).

32. Wang, Z.-B. et al. Investigation of the performance decay of anodic PtRu catalyst with working time of direct methanol fuel cells. J. Power Sources 181, 93-100 (2008).

33. Zhang, J., Sasaki, K., Sutter, E. & Adzic, R. R. Science 315, 220 (2007).

34. Ramaker, D. E. & Koningsberger, D. C. Comment on "Effect of Hydrogen Adsorption on the X-Ray Absorption Spectra of Small Pt Clusters". Phys. Rev. Lett. 89, 139701 (2002).

35. McBreen, J. & Mukerjee, S. J. Elecrochem. Soc. 142, 3399 (1995).

36. McBreen, J., O'Grady, W. E., Pandya, K. I., Hoffman, R. W. & Sayers, D. E. Langmuir 3, 428 (1987).

37. Enayetullah, M. A., DeVilbiss, T. D. & Bockris, J. O. J. Electrochem. Soc. 136, 3369 (1989).

207

38. Green, C. L. & Kucernak, A. Determination of the Platinum and Ruthenium Surface Areas in Platinum−Ruthenium Alloy Electrocatalysts by Underpotential Deposition of Copper. I. Unsupported Catalysts. The Journal of Physical Chemistry B 106, 1036-1047 (2002).

39. Koningsberger, D. C., de Graaf, J., Mojet, B. L., Ramaker, D. E. & Miller, J. T. The metal-support interaction in Pt/Y zeolite: evidence for a shift in energy of metal d-valence orbitals by Pt-H shape resonance and atomic XAFS spectroscopy. Applied Catalysis A: General 191, 205-220 (2000).

40. Teliska, M., O'Grady, W. E. & Ramaker, D. E. J. Phys. Chem. B 108, 2333 (2005).

41. Ramaker, D. E., Mojet, B.L., Garriga Oostenbrink, M.T., Miller, J.T. and Koningsberger, D.C. . Physical Chemistry Chemical Physics 1, 2293 (1999).

42. Koningsberger, D. C., de Graaf, J., Mojet, B. L., Ramaker, D. E. & Miller, J. T. The metal-support interaction in Pt/Y zeolite: evidence for a shift in energy of metal d-valence orbitals by Pt-H shape resonance and atomic XAFS spectroscopy. Appl. Catal. A 191, 205-220 (2000).

43. Arruda, T. M., Shyam, B., Ziegelbauer, J. M., Mukerjee, S. & Ramaker, D. E. Investigation into the Competitive and Site-Specific Nature of Anion Adsorption on Pt Using In Situ X-ray Absorption Spectroscopy. J. Phys. Chem. C (2008).



46. Teliska, M., Murthi, V. S., Mukerjee, S. & Ramaker, D. E. J. Elecrochem. Soc. 152, A2159 (2005).

47. Scott, F. J., Mukerjee, S. & Ramaker, D. E. CO coverage/oxidation correlated with PtRu electrocatalyst particle morphology in 0.3 M methanol by in situ XAS. J. Electrochem. Soc. 154, A396-A406 (2007).

48. Ravel, B. & Newville, M. J. Synchrotron. Rad. 12, 537 (2005).

49. Newville, M., Livins, P., Yacoby, Y., Rehr, J. J. & Stern, E. A. Phys. Rev. B 47, 14126 (1993).

50. Koningsberger, D. C., de Graaf, J., Mojet, B. L., Ramaker, D. E. & Miller, J. T. J. Catal. A 191, 205 (2000).

208

51. Koningsberger, D. C., Oudenhuijzen, M. K., Bitter, J. H. & Ramaker, D. E. Top. Catal. 10, 167 (2000).

52. Mojet, B. L., Miller, J. T., Ramaker, D. E. & Koningsberger, D. C. J. Catal. 186, 373 (1999).

53. Mojet, B. L., Ramaker, D. E., Miller, J. T. & Koningsberger, D. C. Catal. Lett. 62, 15 (1999).

54. Ramaker, D. E., Mojet, B. L., Oostenbrink, G., Miller, J. T. & Koningsberger, D. C. Phys. Chem. Chem. Phys. 1, 2293 (1999).


56. Scott, F. J., Roth, C. & Ramaker, D. E. J. Phys. Chem. C (In Press).

57. Balbuena, P. B. et al. J. Phys. Chem. A 108, 6378 (2004).

58. Xu, Y., Ruban, A. V. & Mavrikakis, M. J. Am. Chem. Soc. 126, 4717 (2004).

59. Hardeveld R.V., M. A. V. Surface Science 4, 396 (1969).

60. Scott, F. J., Mukerjee, S. & Ramaker, D. E. CO coverage/oxidation correlated with PtRu electrocatalyst particle morphology in 0.3 M methanol by in situ XAS. Journal of the Electrochemical Society 154, A396-A406 (2007).

61. M. Watanabe, S. M. Electrocatalysis by ad-atoms Part II. Enhancement of Oxidation of Methanol on Platinum by Ruthenium Ad-atoms. J. Electroanal. Chem. 60, 275-273 (1975).

62. Ticanelli, E., Beery, J. G., Paffett, M. T. & Gottesfeld, S. An electrochemical, ellipsometric, and surface science investigation of the PtRu bulk alloy surface. Journal of Electroanalytical Chemistry 258, 61-77 (1989).

63. Kim, H., Rabelo de Moraes, I., Tremiliosi-Filho, G., Haasch, R. & Wieckowski, A. Chemical state of ruthenium submonolayers on a Pt(111) electrode. Surf. Sci. 474, L203-L212 (2001).

64. Richard J.K. Wiltshire, C. R. K., Abigail Rose, Peter P. Wells, Hazel Davies, Martin P. Hogarth, David Thompsett, Brian Theobald, Fredrick W. Mosselmans, Mark Roberts, Andrea Russell. Effects of composition on structure and activity of PtRu/C catalysts. Physical Chemistry Chemical Physics 11, 2305-2313 (2009).

65. Zabinsky, S. I., Rehr, J. J., Ankudinov, A., Albers, R. C. & Eller, M. J. Multiple-scattering calculations of x-ray-absorption spectra. Physical Review B 52, 2995 (1995).

209

66. Nitani, H. et al. Methanol oxidation catalysis and substructure of PtRu bimetallic nanoparticles. Appl. Catal., A 326, 194-201 (2007).

67. Kim, J.-W. & Park, S.-M. In Situ XANES Studies of Electrodeposited Nickel Oxide Films with Metal Additives for the Electro-Oxidation of Ethanol. Journal of The Electrochemical Society 150, E560-E566 (2003).

68. Long, J. W., Stroud, R. M., Swider-Lyons, K. E. & Rolison, D. R. How To Make Electrocatalysts More Active for Direct Methanol Oxidation Avoid PtRu Bimetallic Alloys! J. Phys. Chem. B 104, 9772-9776 (2000).

69. Vijayaraghavan, G., Gao, L. & Korzeniewski, C. Methanol Electrochemistry at Carbon-Supported Pt and PtRu Fuel Cell Catalysts: Voltammetric and in Situ Infrared Spectroscopic Measurements at 23 and 60 °C. Langmuir 19, 2333-2337 (2003).

210

Chapter 5

Fundamental Aspects of Spontaneous Cathodic Deposition of Ru onto

Pt/C Electrocatalysts and Membranes under Direct Methanol Fuel Cell

Operating Conditions: An In situ X-ray Absorption Spectroscopy and

Electron Spin Resonance Study∗

5.1 Introduction

Direct Methanol Fuel Cells (DMFCs) offer the promise of high energy density power

for both portable and stationary applications. Electronic devices are often sited as being

the greatest benefactors of DMFCs, which could offer a 10-fold increase in power density

in comparison to lithium-ion batteries.1 Despite the above mentioned qualities, DMFCs

have faced significant technological hurdles, hence impeding large scale

∗ Published in the Journal of Physical Chemistry C, 2010, 114 (2), 1028–1040 Authors: Thomas M. Arruda, Badri Shyam, Jamie S. Lawton, Nagappan Ramaswamy, David E. Budil, David E. Ramaker, and Sanjeev Mukerjee Sample preparation and electrochemistry data were collected by authors affiliated with Northeastern University. XAS experiments and data analysis were carried out by authors from The George Washington University and Northeastern University.

211

commercialization. Much of these issues rest on materials challenges such as activity and

stability of anode electrocatalysts and concomitant role of membranes.

As indicated in prior reviews by Stuve et al.2 and Wieckowski et al., 3,4

electrocatalysis of the six-electron methanol oxidation reaction (MOR) can be considered

as a two step process. The first is the initial dehydrogenation step involving the

abstraction of the first hydrogen by breaking the C-H bond in methanol (the next two

dehydrogenation steps being more facile). The second is the oxidation of the CO and

CHO moieties formed on the surface following the dehydrogenation steps. The current

state of the art electrocatalysts rely on the ‘bifunctional approach,’ in which a second

element, such as Ru, initiates oxidation of the CO or CHO species by activating water

(hence forming surface oxygenated species such as OH) at lower potentials. However, as

reported previously,5 these dual electrocatalytic requirements cause a simple bifunctional

catalyst with good nucleation of oxygenated species at lower overpotential to fail as a

good electrocatalyst for methanol oxidation, despite excellent CO oxidation

characteristics. This has been shown previously for PtSn electrocatalyst 5, 6 and more

recently for PtMo.7 In the latter case, while PtMo/C exhibited more than three fold

enhancement for CO oxidation as compared to the current state of the art PtRu/C, there

was no concomitant increase in activity for methanol oxidation. The activity for MOR

was closer to that for pure Pt despite its enhanced ability to oxidized CO. At the moment,

PtRu remains as the electrocatalyst of choice, however, in contrast to CO electro-

oxidation, supported electrocatalysts have shown limited ability to sustain electrocatalytic

activity beyond 0.3 A/cm2. This has necessitated the use of either unsupported

212

electrocatalysts or those with high electrocatalyst loading, which are an order of

magnitude higher than the current state of the art low Pt loading electrodes.

From an electrocatalytic perspective, stability of PtRu in its various forms (supported,

unsupported and in some cases as decorated nano particles (i.e., Ru decorated on Pt and

vice versa) is of paramount interest not only from the perspective of its dissolution and

other changes in anode electrocatalyst morphology but also from the perspective of any

dissolved adducts migrating to the cathode electrode and associated effect on the

membrane. In 2004, Piela and co-workers showed that PtRu anodes are susceptible to Ru

dissolution in an actual working fuel cell stack.8, 9 Although the concept of Ru dissolution

had been previously known,10-16 the direct consequences of spontaneous Ru deposition

onto the cathode catalyst from the anode had not been previously illustrated. To further

complicate matters, they also observed Ru ions in the polymer electrolyte membrane

(PEM).8

As reported by us earlier, the deposition of Ru on Pt under cathode operating

conditions results can translate to ~ 40 – 200 mV overpotential.17 To fully investigate the

extent of Ru poisoning, a fundamental investigation through rotating disk electrode

(RDE) studies was also carried out. It was noted that only μM quantities of dissolved Ru

has dramatic negative effects on the oxygen reduction reaction (ORR) electrode kinetics.

An additional overpotential of 160 mV was observed in comparison to pristine Pt, and the

Ru remained stable on the surface in the entire ORR potential window (0 – 1.2 V vs.

RHE). In addition, the cyclic voltammograms (CV) of the Ru contaminated Pt reveal

increasing double layer capacitance known to be caused by RuOx species, 18, 19 and

decreases in Pt-O formation/reduction peaks and Hupd charge.

213

Most of the earlier studies on Ru deposition on Pt either by electrochemical

deposition20-25 or spontaneous deposition,4, 11, 12, 14, 15, 24-33 have attempted to exploit

deposited Ru for the enhanced oxidation of methanol,11, 16, 25, 28, 31, 34, 35 ethanol,36 formic

acid37 and recently dimethyl ether.38 The latter method being particularly favored for

producing surfaces which are quite stable upon voltammetric cycling. However, we are

unaware of any comprehensive study investigating the effects of Ru adatoms on Pt

surfaces during typical ORR operating conditions. Although the details may vary

slightly, the overall theme centers on enhancement of methanol oxidation kinetics for Ru

decorated Pt in comparison to Pt alone, or in some cases even PtRu alloys. For example,

Waszczuk et al.31 showed that spontaneously deposited Ru on unsupported Pt

nanoparticles produces an electrocatalyst that is twice as active as commercially available

PtRu alloys for methanol oxidation. Their observations suggested that the electrocatalytic

enhancement may be a direct result of Ru edge atoms being under-coordinated by Ru or

surface Pt atoms, which could result in enhanced H2O activation and hence, an improved

bifunctional mechanism.

In light of the importance of the spontaneous deposition of Ru - as it pertains to the

above applications in electrocatalysis - many investigations have been conducted to

elucidate the surface structure of the deposited Ru.29, 30, 39, 40 Ex situ techniques such as

auger electron spectroscopy (AES),28 x-ray photoelectron spectroscopy (XPS)41 and low

energy electron diffraction (LEED)42, 43 have contributed greatly to the understanding of

such surfaces and their properties. A series of scanning tunneling microscopy (STM) and

electrochemical investigations by Crown et al.29, 30 indicated that Ru deposits on low

index Pt(hkl) surfaces without site preference. Surface coverage obtained for Pt(111),

214

Pt(100) and Pt(110) after a single deposition process were 0.20 ML, 0.22 ML and 0.10

ML respectively and found to be mostly in the form of monolayer-thick islands. The rate

of island formation (though described as slow in comparison to Os/Pt(111) deposition)

was shown to form 0.07 ML after 20 seconds with a maximum coverage obtained after

120 seconds. Iwasita et al.44 have shown that the coverage on such single crystals can be

increased by a process of repeated spontaneous deposition. In another study, Ku et al.39

found that Ru formed a ( 33 × )R30º RuO+ adlayer on a Pt(111) surface which is quite

stable even after voltammetric cycling. A comprehensive investigation by Strbac et al.40

also employed in situ STM to study Ru and Os spontaneous deposition on Pt(111) and

Au (111) surfaces. Interestingly, they found that the Ru island growth process is different

on the two surfaces. On Au (111), Ru prefers to deposit on steps and terraces relatively

quickly with a multi-layer thickness and hexagonal surface structure. On Pt(111) the

deposition time was also relatively fast however, only monolayer thickness (0.18 ML

saturation) could be observed after a single deposition period of 3 minutes. When a

second 3-minute deposition was applied, the coverage did not increase significantly (0.22

ML), however the height of the Ru adlayer increased as evidenced in the STM cross

section. Further, the Ru island size and shape (typically 2 – 5 nm in width) was shown to

be dependent on Ru oxidation state and could be manipulated by varying the potential.

Other in situ techniques such as electron quartz crystal microbalance (EQCM)

measurements have also been employed to study Ru deposition. Such a study was first

carried out by Frelink et al.45 to measure the Ru surface content of electrodeposited Ru

onto a Pt film electrode. They found a strong correlation between the potential of the

surface oxide reduction peak and the Ru content, and showed that it is possible to

215

accurately monitor the surface Ru content using this technique. Another study by Vigier

et al.16 - correlating measured growth rates with Fickian diffusion models - revealed a

likely 2-dimensional deposition in nature. Furthermore, on the assumption that every Ru

atom occupies one surface Pt atom, (i.e. Ru occupies an atop site), they found that their

estimates of surface coverage of Ru on Pt were in good agreement with other values in

the literature. We will show later in the discussion that we also find atop adsorption

albeit, chiefly at lower potentials, while at open circuit, we find that the Ru atoms at

higher coverage prefer to be more highly coordinated and occupy 3-fold sites on the

oxygen covered surface.

In the past, x-ray absorption spectroscopy (XAS) has been employed to study

fundamental electrode processes in electrochemistry. XAS is an ideally suited method for

examining nano-scale materials because it is typically performed in situ in modern

synchrotron facilities.46, 47 Although XAS is traditionally a bulk-averaging method, nano-

scale materials afford us the luxury of having ~ 50 % or more of their atoms on the

surface (depending on size and geometry) where electrochemical processes occur. As

such, small changes in coordination number (N) or bond distance (R) can be detected

during an electrochemical reaction by analyzing the extended x-ray absorption fine

structure (EXAFS). In addition, the newly developed Δμ (sometimes referred to as

Δ−XANES) method of x-ray absorption near edge structure (XANES) analysis has been

successfully employed to provided fundamental accounts of adsorbate binding sites on Pt

and Pt alloys.48-51 For example, the Δμ technique recently revealed that chloride anions

specifically adsorbed on Pt in a 3-fold configuration with virtually no Pt-O formation at

high chloride concentrations, while at lower concentrations a mixture of 3-fold Cl and 3-

216

fold Pt-O was observed.48 Prior to this, ultra-high vacuum (UHV) techniques and single

crystal studies have shown only that chloride adsorption occurs in disordered in-plane

structures with Pt-Cl separation of ~ 2.4 Å.52 Other materials have also been probed by

the Δμ method with success including porphyrins53 and metal-chalcogenides.54, 55

In other studies of Ru crossover, Ru cations have been observed in the solid

electrolyte membrane by x-ray fluorescence spectroscopy (XRF).8 Investigation of Ru3+

exchanged into a Nafion membrane offers the possibility for fundamental studies of

morphological changes caused by multivalent Ru ions leaching into the membrane from

catalyst layers. PtRu composites in the membrane have been shown to decrease the

proton conductivity of the membrane.56 To date however, we are unaware of any

comprehensive studies on the behavior of Ru ions inside the micropores of Nafion and

the effects of such species. Previously, electron spin resonance (ESR) was used to

measure the micro-viscosity of the fluid phase of the membrane.57 ESR has also been

used to observe the effects of mono and multivalent ions on the membrane.58 These

investigations showed that different cations exchanged in the membrane alter the water

uptake characteristics of the membrane as well as the micro-viscosity of the fluid phase,

ultimately by changing the free volume available to the solvent. Since water

management issues are important in fuel cells, and the presence of ions in the membrane

can affect hydration, it is important to understand the effects of Ru in the membrane as

well as on the cathode.

In this work, it is our intention to further the understanding of Ru poisoning by

traditional electrochemical methods (CV and RDE analysis) as well as in situ XAS and

ESR. For the first time, we have observed specifically adsorbed metal cations (Run+) in

217

the ORR potential window using the Δμ method. The presence of specifically adsorbed

Run+ cations result in a lower diffusion limiting current (Levich) and a small increase in

ORR overpotential in uncontaminated electrolyte. When ORR was performed in

electrolyte contaminated with 2.0 mM Run+, the overpotential was increased dramatically

and no diffusion limiting current was obtained. Further, the deposited Ru appears to be

stable on the surface and resists removal upon potential cycling. Interestingly, the

spontaneous deposition of Ru occurs to a great extent when the electrodes are allowed to

go to open circuit while much less deposition occurs when potential control is maintained

at high ORR overpotentials (i.e., closer to anode electrode operating conditions),

accentuating the effects of an uncontrolled fuel cell shut down.

5.2 Experimental Section


Cyclic voltammetry and rotating disk electrode studies were carried out using a Pine

Instruments MSR model dual contact RDE setup. All RDE measurements were

conducted by a procedure which has been discussed in great detail previously.59 Briefly,

catalyst suspensions were comprised of 10.5 mg of 30 wt. % Pt/C (BASF Fuel Cells, Inc.,

Somerset, NJ), 10 mL of 2-propanol (GFS Chemicals, 99.5 % min.) and 40 μL of 5 wt. %

Nafion in lower alcohols (Ion Power Inc.). Prior to the addition of 2-propanol, the

catalyst powder was passivated with a few drops of deionized water to prevent

spontaneous combustion of the support. The suspensions were magnetically stirred for 1

hour and sonicated for 10 minutes prior to use. Thin films of catalyst were cast onto a

polished glassy carbon (GC) RDE tip of 5.61 mm diameter (Pine Instruments). A total of

10 μL of suspension was used via two 5 μL applications, resulting in a final loading of 14

218

μg cm-2 Pt. All measurements were made at room temperature in a jacketed glass beaker

type cell fit with a PTFE machined lid. Electrolyte used was 1 M HClO4 (GFS

Chemicals, doubly distilled) for ‘clean electrolyte’ experiments. Ru contamination

experiments were conducted using the same 1 M HClO4 after it was subject to a

sacrificial Ru electrode procedure, which will be outlined below. A typical 3-electrode

setup was used including the GC disk working (WE) electrode, Pt wire/mesh counter

electrode (CE) with an area of 19.3 cm2 (by integration of the Hupd) and a sealed glass

reference hydrogen electrode (RHE) containing clean 1M HClO4. For the experiments

where Ru contaminated electrolyte was used, a separate Pt wire (1.42 cm2) CE and RHE

salt bridge (sealed with vycor frit, BAS Inc.) was used to avoid contamination of the

clean cell. The electrolyte was purged with Ar gas for CV measurements and O2 for

ORR (both UHP 5.0 grade, Middlesex Gasses). An Autolab PGSTAT 30 potentiostat

(Metrohm USA, formerly Brinkman Instruments) equipped with a SCANGEN module

was used for all electrochemical measurements.

All electrochemistry experiments were carried out by activating the catalyst via

potential cycling approx. 50 times at 50 mV s-1 or until a steady state CV was obtained.

Also, CVs were recorded at 20 and 10 mV s-1 for Hupd integration to determine the

electrochemically active surface area (ECSA). Once activated, the clean 1M HClO4 was

purged with O2 and ORR curves were measured via cyclic voltammetry prior to Ru

contamination. Ru deposition was achieved by placing the RDE tip into a separate cell

containing 1M HClO4 + 2.0 mM Run+ and kept at open circuit potential, which was

monitored by running zero-current chronopotentiometry for 1.5 hours. Following Ru

deposition the electrode tip was rinsed off with copious amounts of deionized H2O and

219

placed back into the clean 1M HClO4 cell for post-contamination CVs and ORR

measurements. All measurements were performed at room temperature (20 ºC). It should

be mentioned that we plot the CVs and ORR polarization curves using raw current

obtained from the experiments to illustrate changes that occur only as a result of Ru

deposition. Although a minimum of three experiments were performed for all cases, we

only compare the electrochemical results obtained from a single catalyst deposition as

current density normalization was not employed.

A sacrificial Ru electrode was used to produce 1 M HClO4 contaminated with Run+

ions by the following procedure. A catalyst suspension was produced by mixing 90.1 mg

of 60 wt% Ru/C electrocatalyst (BASF Fuel Cell Inc.) with 1 mL deionized water and 1

mL of 2-propanol. The mixture was then sonicated for 10 minutes and applied to a piece

of carbon weave (Panex 30, Zoltek Corp.) over an area of 10 cm2 for a final Ru loading

of 5.4 mg cm-2 and capped off with 4 drops of 5 wt. % Nafion solution. The electrode

was placed into a beaker cell with clean 1M HClO4 and cycled between 0.05 – 1.4 V vs.

RHE using a Pt wire CE and Ag/AgCl reference electrode (BAS Inc.). After approx. 50

CVs, the potential was fixed at 1.2 V vs. Ag/AgCl for 1 hour. The resulting solution was

filtered (Whatman 52 filter paper) three times to remove loose catalyst debris. Ru ion

concentrations were determined by ion-coupled plasma mass spectrometry (ICP-MS).

Prior experiments in a working DMFC at different anode overpotentials as a function of

time was used as a measure of expected Ru ion concentrations at the cathode electrode.

These ranged between 0.1 and 3 mM.

5.2.2 Flow-through Cell Design.

220

All in situ XAS experiments were performed in new flow-through type spectro-

electrochemical cell designed to accommodate for the introduction of contaminants

without the need to disassemble the cell amid experiment. The cell was machined out of

a Teflon block as depicted in Figure 5.1 with similar dimensions as the previously

designed model.60 The cell is designed as a multi-function spectro-electrochemical cell

which can be used to collect data in either transmission or fluorescence mode. The thin

wall side of the cell has a shallow internal electrolyte reservoir (3.81 cm × 4.064 cm ×

0.254 cm) in which the electrode is affixed in place by a rectangular PTFE bracket.

Directly across from the four corners of the holding bracket, on the thick wall side, are

electrolyte flow holes which terminate with standard PTFE compression fittings. All four

corners are bored out to allow the cell to be ambidextrous in that fluorescence

measurements may be performed at beam lines that have a detector on either the left or

right side of the cell. Two of the ports are used for the electrolyte inlet and return to the

exterior reservoir, while a third port supplies a salt bridge to the reference electrode,

which also resides in the exterior reservoir. The x-ray window is bored all the way

through the cell with a 45 degree chamfer on the rear of the thick side cell block. The

windows are sealed off to prevent electrolyte leakage using PTFE tape (3M). A variable

rate peristaltic pump (Cole Parmer Masterflex L/P) is used to circulate electrolyte. The

cell is sealed with a single pre-cut silicone gasket (Auburn Chemical Co.) which is placed

between the two cell halves. This cell configuration is particularly advantageous as it

221

Figure 5.1 Schematic illustration of the specially designed flow-through style, spectro-electrochemical XAS cell

222

allows the user to oxygenate/deoxygenate the electrolyte or introduce poisons during

the experiment which was not possible with other cell models.

5.2.3 In situ XAS Data Collection

The Pt WE were prepared by loading 4.97 mg cm-2 (metal loading) of 30 % Pt/C

(same lot as above) onto carbon weave and cut into 3.5 cm × 0.5 cm size pieces. The

electrode preparation method has been described in detail elsewhere.48 Briefly, the

catalyst suspension was made by mixing pre-weighed catalyst powder with 1:1 deionized

H2O/2-propanol and 5 wt. % Nafion with a catalyst weight to Nafion weight ratio of 95:5.

The Pt loading was chosen to yield an absorption cross section of ~ 1. Prior to cell

assembly, the electrodes were wetted via vacuum impregnation in 1M HClO4. The cell

also consisted of a pre-washed (0.5 M H2SO4, 80oC) Grafoil (GrafTech International Inc.)

CE, which was situated directly across from the WE. Grafoil was chosen as a CE

because it is inert and does not significantly attenuate the x-ray beam. In all cases, 0.1

mm thick Au foil (99.999%, Alfa Aesar) was used as current collectors for both

electrodes. The electrolyte inlet and outlet tubes (Viton, Cole Parmer) connected to the

cell terminated in a 150 mL beaker containing the 1M HClO4 or 1M HClO4 + 2.0 mM

Run+ electrolyte(s). An RHE was seated into a salt bridge sealed off with a vycor frit

(BAS) to minimize Run+ ions migrating into the RHE. The entire RHE assembly was

placed into the external electrolyte beaker where it was in ionic contact with the inside of

the cell via the inlet/outlet tubes and salt bridge tube.

Experiments were performed at beam lines X-18B, X-11A and X-11B at the National

Synchrotron Light Source, Brookhaven National Laboratory, Upton NY. Prior to the

start of each experiment, the working electrodes were activated by potential cycling (0.05

223

to 1.2 V vs. RHE at 20 mV s-1) in clean 1 M HClO4 until a steady state CV was obtained.

Full range Pt L3 EXAFS were collected (-200 eV to 18 k) with the working electrodes

fixed at various static potentials as described by the following scheme: (i) xV, Ar sat. 1M

HClO4; (ii) xV, O2 sat. 1M HClO4; (iii) xV, O2 sat. 1M HClO4 + 2.0 × 10-3 M Run+ (1

hr.); (iv) xV, Ar sat. 1M HClO4 (return scan). The scans recorded at open circuit

observed an OCP between 0.9 and 0.95 V. Throughout the progression of the scans

potential control was maintained at all times. This was achieved as the electrolyte could

be easily exchanged by draining the external reservoir of its contents down close to

empty and replacing with the next electrolyte without having the RHE become

disconnected. Prior to step (iv) above, the cell would be filled and drained with clean 1M

HClO4 3 times (~ 400 mL) to remove trace Ru contamination.

Data were collected in transmission mode using the typical three gas ionization

detector setup (I0, It and Iref) with a nominal N2/Ar mixture to allow for 10% photon

absorption in I0, 50 – 70% in It and Iref. The sample was placed between I0 and It, while

Pt reference foil (4 μm, Alfa Aesar) EXAFS were collected between It and Iref. At each

beam line a Si(111) monochromator was employed and detuned by 40% to remove

higher harmonics.

5.2.4 EXAFS Analysis

All EXAFS analyses were performed using the IFEFFIT suite, version 1.2.9

(Copyright 2006, Matthew Newville, University of Chicago,

http://cars9uchicago.edu/ifeffit/).61 All scans are carefully aligned and calibrated using

the reference foil to account for any changes in beam energy throughout the course of the

experiment. Background subtraction and normalization was performed using the

224

AUTOBK62 algorithm in Athena (Bruce Ravel, © 2006), a subroutine of IFEFFIT. The

normalized EXAFS data was then imported into the ARTEMIS program where EXAFS

fits were carried out using a k-range window of 2.0 – 15 Å-1 (Kaiser-Bessel) and an R-

window of 1.5 – 3.5 Å.

5.2.5 Δμ Analysis

The Δμ procedure has been described in great detail elsewhere.63-66 Briefly, all XAS

scans are carefully aligned and normalized over a much more narrow energy range (~ 25

to 130 eV) to only consider the XANES region. Difference spectra are calculated based

on the normalized XANES using the relationship

Δμ = μ(Pt-xelect., V) – μ(Pt-Ar, clean, V) Eq. 5.1)

where μ(Pt-xelect., V) is the XANES at a particular electrode potential in either a clean or

Ru contaminated electrolyte, and μ(Pt-Ar, clean) is the reference scan in clean electrolyte

at the same potential. This scan is chosen as the reference so as to remove any other

electrode processes occurring simultaneously (i.e. H2O activation) and emphasize only

the effect of Run+. This of course assumes that the O(H) adsorption levels are about the

same with and without Ru, which is not necessarily true, but we will see below that this a

reasonable assumption below 0.8 V.

Experimental Δμ spectra are then compared to theoretical Δμ signatures calculated

using the FEFF 8.0 code.67 This is achieved by calculating theoretical XANES curves

using small Pt clusters, in this case the “Janin” Pt6 cluster,68 with and without the

adsorbate placed in various geometries. The resulting XANES can then be subtracted

using the relationship:

Δμt = μ(Pt6X) – μ(Pt6) (Eq. 5.2)

225

where μ(Pt6X) is the theoretical XANES for the Pt6 cluster with adsorbate X in a

particular binding geometry and μ(Pt6) is that of the blank Pt6 cluster. It should be noted

that care needs to be taken to ensure that scans are properly aligned prior to subtraction.

Also, for optimal comparison to the experimental data, theoretical signatures are

sometimes shifted by 1 – 5 eV and/or scaled by a multiplication factor.

5.2.6 Electron Spin Resonance

Nafion™117 membranes were first purified by heating to 75°C for 1 hour in 3 %

hydrogen peroxide followed by 1 hour in deionized water, 1 hour in 0.5M sulfuric acid,

and again 1 hour in deionized water. The membranes were ion exchanged to varying

extents by soaking in Ru Nitrosylnitrate (Alfa Aesar) solutions ranging in concentration

from 0.5 mM to 100 mM for 2,5, and 7 days respectively. Upon removal from the Ru

solution, the membranes were rinsed to remove any surface Ru and to terminate the

exchange. The swollen membranes were weighed to gravimetrically determine the total

water uptake. Before and after the exchange process the membranes were thoroughly

dried and weighed and the extent of exchanges was determined gravimetrically using a

Cahn C-33 microbalance.

ESR measurements were accomplished by soaking the exchanged membranes in 0.1

mM 2,2,6,6-tetramethyl-4-piperidone N-oxide (TEMPONE) spin probe in water. The

ESR spectra were collected on a Bruker EMX X-band spectrometer. For each spectrum,

three scans of 2048 points were averaged using magnetic field modulation of 0.02 mT at

100 kHz. The fitting method utilized a MATLAB (MathWorks) based version of EPRLL,

the slow-motional line shape program of Freed and co-workers69, 70 and was used to

determine the correlation time (τc) of the spin probe by monitoring the rotational rate of

226

E, V vs. RHE0.0 0.2 0.4 0.6 0.8 1.0 1.2

I, A

-8e-5

-6e-5

-4e-5

-2e-5

0

2e-5

4e-5

6e-5

Clean Pt/CPost Cleaning Step

-8e-5

-6e-5

-4e-5

-2e-5

0

2e-5

4e-5

6e-5

Clean PtPost Ru Contamination

a

b

Figure 5.2 Cyclic voltammograms of 30 wt. % Pt/C taken in Ar purged 1 M HClO4. The Pt/C was loaded onto a 5.56 mm diameter glassy carbon RDE tip with a rotation of 0 RPM, collected at a scan rate of 50 mV s-1 at 20 oC. (a) CV prior to contamination in 2.0 mM Run+ contaminated HClO4 (solid line) and after spontaneous Ru adsorption (OCP, 30 minutes), rinsing (DI H2O), and return to clean 1 M HClO4 (dashed). (b) clean catalyst CV (solid) overlaid with the CV after Ru cleaning step (dashed). The cleaning step involved performing 200 potential cycles between 0.05 – 1.2 V, followed by 50 cycles between 0.05 – 1.4 V clean 1 M HClO4 with a scan rate of 50 mV s-1.

227

the probe (R). The relation between τc and R is τc =1/6R where τc is related to the local

viscosity of the solution around the spin probe through the Stokes-Einstein relationship:

Tkr

B

ec 3

4 3ηπτ = (Eq. 5.3)

where η is the effective local viscosity, re is the hydrodynamic radius of the rotating

probe, and kB is the Boltzmann constant.

5.3 Results and Discussion


The cyclic voltammograms (CVs) shown in Figure 5.2a reveal significant changes to

the Pt surface following contact with Ru contaminated electrolyte. This is clearly visible

in all three regions of the CV; (a) the Hupd region, (b) the Pt-O formation/reduction

region, and (c) the double-layer charge region. In region (a) it is evident that the total

charge of the Hupd has decreased due to a loss of electrochemically active surface area

(ECSA) as it has become blocked by Ru. Likewise in region (b) the Pt-O formation and

subsequent reduction is muted by the presence of adsorbed Ru. The increase in double

layer capacitance in region (c) suggests the adsorbed Ru likely exists in the form of some

RuOx species, which are known to exhibit higher capacitance than Pt.18, 19

To examine the reversibility of Ru deposition, a series of experiments were

performed on the contaminated surface. This is of interest since some research has

suggested that Ru can be at least partly removed.8 Figure 5.2b shows the CV of Pt/C in

clean electrolyte in comparison to that of the Pt/C after a “cleaning step.” The cleaning

step was performed by running 200 CVs between 0.05 – 1.2 V, followed by 50 CVs

228

Table 5.1 Electrochemically active surface area determination results

Average ECSA (cm2) % Change (cm2)

Clean Pt 1.59 ± 0.16 na After Contamination 1.20 ± 0.19 - 24.7 After Cleaning Step 1.26 ± 0.11 + 5.11

Uncertainties reflect the standard deviation in ECSA determined as a result of performing at least three independent experiments.

229

between 0.05 – 1.4V on the deposited catalyst in clean electrolyte. It is evident in the Pt-

O formation/reduction region (b) that some Ru has been removed as the Pt-O peaks have

become better defined. However, the double layer capacitance was still widened and the

Hupd did not fully recover either, indicating that some Ru remains on the surface. The

electrochemically active surface area (ECSA) for each of the above mentioned situations

have been calculated and displayed in Table 5.1. The ECSA was determined by a

common practice of integrating the Hupd charge (after subtracting the double-layer

capacitance) and dividing by the value of 210 μC cm-2. A total of 24.7 % loss of ECSA

was observed as a result of Ru blocking Pt surface sites. Following the cleaning

treatment, an increase in ECSA was observed (~ 5 %), however, clearly all of the Ru had

not been entirely removed, even as the electrode had been cycled up to 1.4 V. This result

is consistent with the observations of Piela et al.,8 who has observed partial Ru

dissolution from contaminated cathodes when cycled to anodic potentials.

The effects of deposited Ru on ORR are easily discernable by inspection of the ORR

polarization curves in Figure 5.3. The clean Pt catalyst exhibits a commendable ORR

activity with an onset potential ~ 1.0 V and a well defined diffusion limiting current as

described by the Levich equation:

ilim = 0.62neFD2/3ω1/2ν-1/6Co (Eq. 5.4)

where ilim is the diffusion limiting current density, ne is the number of transferred

electrons, F is Faraday’s constant, D is the diffusion coefficient of O2 in the electrolyte, ω

is the rotation rate of the RDE, ν is the kinematic viscosity and Co is the concentration of

O2. Although the ORR onset overpotential only increased by ~ 15 – 20 mV going from

the clean Pt to Ru deposited Pt, the decrease in ilim is consistent with a residual Ru

230

E, V vs. RHE0.4 0.6 0.8 1.0 1.2

I, m

A

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

0.2

Clean PtPost Ru Cont.In Ru Cont.

Figure 5.3 ORR polarization curves (anodic sweep) for 30 wt. % Pt/C on a 5.56 mm diameter glassy carbon disk in O2 saturated 1 M HClO4 with a 20 mV s-1 sweep rate at 900 RPM. The solid line represents the clean Pt/C prior to contamination, the dashed line has been exposed to 2.0 × 10-3 M Run+ and subsequently “cleaned” via the cycling procedure and the dash-dot line was collected in 1 M HClO4 + 2.0 × 10-3 M Run+.

231

presence blocking the Pt surface, in agreement with the discussion of the CVs above.

The ORR curve also shifts negative in the mixed kinetics mass-transport region as a

result of some loss in activity. When ORR was conducted in Ru contaminated electrolyte

(dash-dot line), the overpotential increased by more than 200 mV and no discernable ilim

was obtained. Apparently, Run+ ions in solution participate in an electrochemical

deposition process which cause the above mentioned characteristic changes to the ORR

polarization. This is consistent with the report of “current assisted” deposition by Piela et

al.,8 in which Ru contamination on the cathode increased significantly as the anode

potential was increased.

The Tafel plots presented in Figure 5.4 were transformed from the ORR curves in

Figure 5.3 after being treated by the mass transport correction equation:

ik = ilim × i / (ilim – i) (Eq. 5.5)

where ik is the kinetic current, ilim is the diffusion limiting current described by Eq. 4 and

i is the measured current during the ORR polarization (anodic sweep). Overall, the shapes

of the Tafel curves remain relatively unchanged, indicating that there is no major change

in the ORR mechanism, such as an increase in the H2O2 pathway.71 Although the Tafel

slope fitting is not shown in the plot, values obtained were all close to the typical values

of –60 mV decade-1 (high E region) and –120 mV decade-1 (low E region) for ORR on Pt.

The decrease in the ORR activity observed in the Tafel curves reflects the increase in

ORR overpotential as a result of adsorbed Ru, which is consistent with a site-blocking

process.71

5.3.2 EXAFS Analysis

232

log(Ik)-5.0 -4.0 -3.0 -2.0 -1.0

E, V

vs.

RH

E

0.5

0.6

0.7

0.8

0.9

1.0

Clean PtPost Ru Cont.ORR in Ru

Figure 5.4 Mass transfer corrected Tafel plots shown at 900 RPM for the ORR polarization curves presented in Figure 5.3. Due to the changing active surface area, we utilize only geometric surface area for current density normalization.

233

R, Å0 1 2 3 4

|χ(R

)|, Å

-3

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

Pt/C at 0.80 VFit

k, Å-10 2 4 6 8 10 12 14 16

k2 *χ(k

), Å

-2

-1.0

-0.5

0.0

0.5

1.0

Pt/C at 0.80 VFit

b

a

Figure 5.5 (a) Pt-L3 edge EXAFS spectrum (Kaiser-Bessel window 2.0 < k < 15 Å-1, k2 weighted) and corresponding least-squares fit for 30 wt. % Pt/C in 1 M HClO4 + 2.0 × 10-3 M Run+ fixed at 0.80 V. (b) Fourier transformed EXAFS, fitting was performed in R space using a single shell Pt-Pt scattering path and a Kaiser-Bessel window (1.0 < R < 3.5 Å, k2).

234

It is standard practice to fully analyze the extended x-ray absorption fine structure

(EXAFS) prior to any Δμ XANES analysis. The reason for this is to ensure that no major

changes in Pt-Pt bond length occur under the given experimental conditions. The Δμ-

XANES analysis relies on crystallographic modeling using consistent bond lengths in

order to generate realistic Δμ simulations. The EXAFS data processing involves a

normalization/background removal process using a background spline function

(AUTOBK)62 in the ATHENA code.72 Once normalized, the EXAFS is imported into

ARTEMIS, where the physical parameters are elucidated via least-squares fitting. A

representative fit is shown in Figure 5.5a and 5.5b for Pt/C at open circuit potential in 1M

HClO4 + 2.0 mM Run+. Although evidence suggests that deposited Ru exists on the

surface (as will be illustrated by the Δμ analysis below), it is not directly visible to the

FT-EXAFS because it is too low in concentration and would likely be located in the

region of the main Pt-Pt scattering (~ 2.5 Å); nevertheless the effects of the Ru deposition

are evident in the NPt-Pt values.

Table 5.2 offers a summary of EXAFS parameters obtained by the methods described

above. In order to ensure a valid comparison of coordination numbers (NPt-Pt), the best

value of σ2 (mean-square radial disorder) was fixed (5.05 × 10-3 Å2) and used for all fits.

Although no significant changes to the Pt-Pt distance were observed, small changes in

NPt-Pt were observed as the particles tend to distort when adsorbates are present as the Pt-

Pt scattering near the surface is altered by adsorbates.48, 51 These changes are illustrated

more clearly in Figure 5.6 (left axis). The value of NPt-Pt of 7.6 appears to represent the

clean cluster. The scan taken at 0.3 V shows a small increase in NPt-Pt, typical of that

235

Table 5.2 Summary of EXAFS parameters derived from first-shell fits

NPt-Pt

ΔN = ± 0.28a RPt-Pt, Å

(ΔR) E0, eV (ΔE0)

σ2, Å2 × 10-3

OCP Clean Pt 7.16 2.73 (-0.045) 6.35 (0.77) 5.05

1 hr. Ru exposure 6.24 2.73 (-0.038) 7.20 (1.20) 5.05 0.80 V

Clean Pt 7.57 2.73 (-0.039) 7.29 (0.75) 5.05 1 hr. Ru exposure 7.71 2.73 (-0.040) 7.19 (0.66) 5.05

0.70 V Clean Pt 7.41 2.73 (-0.041) 7.21 (0.66) 5.05

1 hr. Ru exposure 7.49 2.73 (-0.041) 7.29 (0.71) 5.05 0.50 V


0.40 V Clean Pt 7.64 2.73 (-0.040) 7.59 (0.60) 5.05

1 hr. Ru exposure 7.71 2.73 (-0.039) 7.39 (1.05) 5.05 0.30 V


aValue represents the largest statistical error of all the least-squares fits determined by ARTEMIS. NPt-Pt was calculated using the FEFF8 value of 2

0S (0.934) for Pt L3 edge.

236

Figure 5.6 Plot of NPt-Pt (solid lines, left axis) for Pt/C in 1 M HClO4 plotted as a function of potential. Also shown are the Ru Δμ magnitudes (Equation 5.1) for Ru deposition on Pt (dashed line, right axis). The dominant Ru adsorption site (n-fold or atop) as indicated by the Δμ spectral line-shape is also given.

E, V v.s RHE

0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

NPt

-Pt

6.0

6.2

6.4

6.6

6.8

7.0

7.2

7.4

7.6

7.8

8.0

Δμ Magnitude

0.00

0.02

0.04

0.06

NPt-Pt CleanNPt-Pt, 60 min. Ru

Δμ Magnitude (E)

OH OH

OCP

n-foldatop

n-fold

n-fold

E, V v.s RHE

0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

NPt

-Pt

6.0

6.2

6.4

6.6

6.8

7.0

7.2

7.4

7.6

7.8

8.0

Δμ Magnitude

0.00

0.02

0.04

0.06

NPt-Pt CleanNPt-Pt, 60 min. Ru

Δμ Magnitude (E)

OH OH

OCP

n-foldatop

n-fold

n-fold

237

seen after H adsorption, and again a small increase at 0.80 V, which is typical of that

seen when atop O(H) adsorption occurs. We have shown many times previously49, 51 that

atop anion (e.g. O(H)- and Cl-) adsorption generally increases NPt-Pt, and 3-fold

adsorption decreases it. The increase occurs as a result of the overall morphology of the

nanoparticle

becoming more spherical in the presence of atop adsorbates and 3-fold adsorption

generally directly decreases the Pt-Pt scattering. Note that the NPt-Pt values after Ru

deposition are larger than for the clean at 0.3 and 0.4 V, smaller at 0.5 V, and then larger

again at 0.7 V; i.e. exactly opposite that expected for an anion. We have previously

noted48, 50 that cation or neutral species adsorption (H+ and S) even in n-fold sites (n = 2

or 3) can increase NPt-Pt and apparently in 3-fold increase it. Thus the changes in NPt-Pt are

consistent with that expected for cation/neutral adsorbates in 3-fold sites at low coverage,

except at 0.5 V when adsorption occurs more in atop sites as indicated by the Δμ line-

shapes discussed below. Interestingly, both values of NPt-Pt decrease at OCP because of

some O adsorption in 3-fold sites as expected, but now the largest change in NPt-Pt

between the clean and Ru deposited results also exist due to significant Ru deposition. It

is a bit surprising, however, that now the Ru mostly adsorbed in 3-fold sites additionally

decreases NPt-Pt relative to the clean. This may provide information about the charge on

the Ru species in the presence of co-adsorbed O atoms (i.e. less positively charged and

behaving more as an additional ‘anion’) at these potentials. The preferred Ru deposition

site will be discussed in greater detail in the following section.

5.3.3 Experimental Δμ Analysis

238

E, eV (rel. to Pt L3 edge)-10 0 10 20

Δμ In

tens

ity a

.u.

-0.04

-0.02

0.00

0.02

0.04

0.06

0.0820 min. Ru40 min. Ru60 min. Rureturn, clean HClO4

Figure 5.7 Pt L3 edge Δμ = μ(2.0 × 10-3 M Run+, OCP) – μ(clean, 0.50 V) spectra for 30 wt. % Pt/C using the μ obtained in 2.0 × 10-3 M Run+ in 1 M HClO4 at open circuit, and 0.50 V in clean HClO4.

239

The Δμ curves presented in Figure 5.7 illustrating spontaneous Ru deposition were

constructed according to Equation 5.1, where the electrode was maintained at open circuit

and the electrolyte was 2.0 mM Run+ in 1M HClO4 unless otherwise noted. After 20

minutes of exposure, a positive peak developed approx. 5 eV past the Pt L3 edge with a

magnitude of ~ 4 % of the total XANES signal. The subsequent scans at 40 and 60

minutes reveal a negative dip that precedes the larger positive peak in the same region as

the former. This can be explained by two separate, simultaneous processes; changes in

the Δμ magnitude typically indicate an increase of adsorbate surface coverage (or

decrease depending on the direction of the change) and the modification of the line shape

suggests there is an adsorbate binding site transformation. Interestingly, the ‘return’ scan

reveals little change in the Δμ spectrum despite the cell being drained of all Ru

contaminated electrolyte and rinsed with clean HClO4, although no cyclic potential

scanning was performed. This suggests that adsorbed Ru is stable on the Pt, at least in the

context of these experimental conditions. In order to determine the Ru binding site(s), the

overall line shapes require modeling with FEFF8.0 and shall be discussed in full detail in

section 5.3.4.

The Δμ procedure was also used to investigate Ru deposition on Pt electrodes where

potential control was maintained throughout the duration of the XAS measurements. The

objective was to mimic the situation that fuel cell cathodes are subject to when operated

under a constant load in the presence of Ru contamination. The results are presented in

Figure 5.8. The Δμ curves (calculated by Equation 5.1) for the Pt/C electrodes were

subject to Ru contamination for 60 minutes at the indicated static potentials and again

flushed with clean electrolyte. For the sake of clarity, each curve in the figure was offset

240

E, eV (rel. to Pt L3 edge)-20 -10 0 10 20 30

Δμ a

.u.

-0.02

0.00

0.02

0.04

0.06

0.08

0.10

0.3 V

0.4 V

0.5 V

0.7 V

0.8 V

Figure 5.8 Pt L3 edge Δμ = μ(2.0 × 10-3 M Run+, xV) – μ(no Run+, 0.5 V) spectra for 30 wt. % Pt/C taken after 60 minutes exposure to Run+ contaminated 1 M HClO4.

241

on the Δμ axis by an increment of 0.02. All Δμ magnitudes here are ≤ 0.02, indicating

relatively low Ru adsorbate coverage. The spectral line-shapes reveal some

inconsistencies that will be discussed in the theoretical section below. For further

analysis, Figure 5.6 plots the amplitude of the experimental Δµ (right axis) obtained from

Figure 5.8, which signifies the change in relative Ru coverage with potential, along with

the NPt-Pt as already discussed. While under potential control, the Δμ amplitude reaches a

maximum at 0.5 V, decreases with potential until a sharp increase is observed at OCP.

There are two factors that seem to affect Ru deposition as suggested in the Figure, the

coverage of other adsorbates and Coulombic forces between Run+ in the electrolyte and

already adsorbed species. The large deposition at OCP seems reasonable as one would

expect that Run+ ions are not particularly attracted to a positively charged Pt surface, but

as the oxide forms above 0.8 V, the Run+ ion are attracted to the negatively charged O

atoms in the oxide layer and eventually co-deposit on the surface. To show that O(H)

adsorption is still occurring on Pt with Ru present, and to determine the effect of this Ru

on O(H) adsorption, we calculated the Δμ in Figure 5.9a using the relationship:

Δμ = μ(xV, Ru) – μ(0.5 V, Ru) (Eq. 5.6)

in order to isolate the Pt-O interactions by consequently subtracting out any Ru

contributions. The value of 0.5 V was used as it resides in the double-layer region where

typically no adsorbates are present. Recall that the 0.8 V line in Figure 5.8 did not reveal

any Pt-O signature because it had been subtracted out. That particular Δμ was calculated

using the clean reference at the same potential (Eq. 5.1), which as mentioned above

cancels out any process unassociated with Ru adsorption. The obtained line shapes in

Figure 5.9a do indeed indicate the presence of adsorbed O, however, no Ru because that

242

0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

0.00

0.01

0.02

0.03

0.04

0.05

0.06

E, V vs. RHE

Mag

nitu

de Δ

µ

Atop O(H)

N-fold O

Some Oxide

ArO2

O2 & Ru


Δμ, a

.u.

-0.005

0.000

0.005

0.010

0.015

0.020

0.025

0.70 V0.80 V

OH(far)

Oads

OH(near)

a

b

0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

0.00

0.01

0.02

0.03

0.04

0.05

0.06

E, V vs. RHE

Mag

nitu

de Δ

µ

Atop O(H)

N-fold O

Some Oxide

ArO2

O2 & Ru


Δμ, a

.u.

-0.005

0.000

0.005

0.010

0.015

0.020

0.025

0.70 V0.80 V

OH(far)

Oads

OH(near)

a

b

Figure 5.9 (a) Pt L3 edge O-adsorption Δμ = μ(2.0 × 10-3 M Run+, xV) – μ(2.0 × 10-3 M Run+, 0.5 V) spectra for 30 wt. % Pt/C taken after 60 minutes exposure to Run+ contaminated 1 M HClO4. (b) Maximum magnitude of similar O Δµ vs. potential under 3 different indicated conditions; i.e. when the 1 M HClO4 electrolyte de-oxygenated with Ar, when saturated with O2, and when saturated with O2 after 60 minutes of Run+ exposure. The shaded arrows indicate the dominant adsorbate as reflected in the Δµ spectral line-shape and discussed in the text.

243

was removed by the difference. The Δμ lines in Figure 5.9a indicate normal H2O

activation (atop Pt-O(H) at 0.7, 3-fold 0.8 V) and are in agreement with previous

observations.49 Note that the Δμ taken at 0.7 V reveals an additional shoulder (noted as

OH(near)) about 2 eV to lower energy. This has been observed many times previously

when Ru islands exists on the Pt surface, and is attributed to OH bonded to Pt at sites

next to the RuOx islands.73-76 It arises because of an electronic effect exerted by the RuOx

on the nearby Pt atoms shifting the core-level binding energy and hence shifting the

energy of the Δμ feature. Similar O(H) Δμ spectra (not shown) using Equation 5.6 but

taken before Ru deposition do not show this additional feature.

The effects of Ru deposition on the oxide coverage are also observed (see Figure

5.9b). The absolute magnitudes of the oxide Δμ line-shapes similar to those in Figure

5.9a (using Eq. 5.6) are plotted as a function of electrode potential. In the case of de-

oxygenated 1 M HClO4, H2O activation proceeds as previously observed in many of our

studies.48, 49, 76 The O(H) coverage steadily increases with potential; the Δμ line-shape

below 0.7 V reflects atop OH, then 3-fold O, and finally above 0.8 V that of an oxide

(with subsurface O). The data in 1 M HClO4, saturated with O2 shows a similar trend,

only the atop coverage increases much faster due to atop O adsorbed at cluster corner and

edge sites. Such lower coordinated Pt sites are known to be more reactive with O2, and

the adsorbed O in such sites will exhibit a lower coordination with Pt (i.e. atop-like).49

These Pt sites probably do not participate in the ORR (because they are blocked by

strongly adsorbed O), but this adsorbed O on the corners/edges appears to decrease

strongly the amount of 3-fold O on the Pt(111) planes above 0.7 V, which may enhance

the ORR rate on those sites. Such differences between the in situ and operando O(H)

244

coverage has been considered previously77 and shall not be further discussed here. The

sub-surface O appears similarly above 0.8 V with and without the presence of O2. The

effect of Ru deposition is quite interesting: a) It appears to slow O from going sub-

surface to form the oxide, b) it definitely hinders atop adsorption below 0.7 V, apparently

because the Ru and O compete for these sites; and c) it appears to exert a ligand effect

around 0.7 V; although it is difficult to separate the competing effects of atop O and Ru

on the surface. However, it is clear from the Δμ data just above 0.7 V that deposited Ru

in the presence of O2 can increase the adsorption of O(H), which blocks sites for the

ORR.

5.3.4 FEFF Modeling

As seen above, the Δμ spectral line-shapes can provide valuable evidence of the binding

sites of various adsorbates such as H, OH, and O on a surface using previously modeled

Δµ spectral line-shapes for O adsorption. No direct line-shape assignments can be made

without first theoretically simulating the adsorption event, which has not been previously

performed for the Ru Δμ signature. Therefore, theoretical XANES modeling (and hence

Δμt) was performed for Ru/Pt using the spatial coordinates of the Janin Pt6/Pt6-Ruads

clusters,68 along with the appropriate input parameters (Hedin-Lundqvist potentials,

NOHOLE card etc.), and evaluated for full multiple scattering by FEFF8.0. The Ru-Δμ

signatures for the commonly used 1-fold, 2-fold and fcc 3-fold adsorption sites are

presented in Figure 5.10. The 1 and 2-fold Ru line shapes are too similar and therefore

will be treated as impossible to distinguish experimentally. The 3-fold signature on the

other hand contains a negative dip just preceding the edge position, followed by a large

positive peak that is in very close resemblance to the OCP scans in Figure 5.7 taken

245

Figure 5.10 FEFF 8.0 generated Δμ = μ(Pt6-Ru, site) – μ(Pt6) theoretical spectra for the indicated Ru adsorption sites. The Pt-Ru bond distances used were ~ 2.6 Å.

246

Figure 5.11 Comparison of Δμ spectra obtained after 60 minutes exposure in Run+ contaminated HClO4 with the theoretical 3-fold fcc adsorbed Pt6-Ru cluster.


Δμ, a

.u.

-0.02

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

3-fold Ru FEFF Signature

60 Minute Ru Exposure, Open-Circuit Pot.


Δμ, a

.u.

-0.02

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

3-fold Ru FEFF Signature

60 Minute Ru Exposure, Open-Circuit Pot.

247

Table 5.3 Summary of estimates of Ru adsorption coverage on various Pt catalysts

Type of

electrode Deposition

method Ru coverage

(ML) Technique Reference

Pt film electrode

Spontaneous 0.32, 0.55 EQCM Ref. 45

Pt/C (E-TEK), 4.2 nm

Spontaneous

Electrochemical

0.22

0.46

EQCM Ref. 25

Pt(111) Spontaneous 0.20 STM Ref. 30

Pt(100) Pt(110)

Spontaneous 0.21 0.10

STM Ref. 29

Pt(111)

Spontaneous

Electrochemical

0.12, 0.18a

0.31b

AES Ref 28

Pt(111) Pt(100) Pt(110)

Spontaneous 0.10 0.24 0.05

Electrochemical Ref. 12

Quartz supported Pt

electrode

Spontaneous

Electrochemical

0.10

max ca. 0.5c

EQCM Ref. 16

Pt black Spontaneous (multiple)

0.20, 0.25, 0.35, 0.40d

ICP, Electrochemical

Ref. 31

Pt(111) Spontaneous (multiple)

0.11, 0.39, 0.63, 0.13e

Electrochemical Ref. 44

Pc platinum wire

Spontaneous 0.22 – 0.25f Electrochemical Ref. 35

Pt/C (E-TEK), 3.5 nm

Spontaneous 0.10, 0.33g XAS This work

a50 µM and 500 µM solutions of RuCl3 + 0.1 M HClO4 (2 min.) b2 mM Ru(NO)(NO3)3 + 0.5 M H2SO4 after a single voltammetric scan c0.5 M Ru(NO)(NO3)3 + 0.1 M HClO4 at 0.05 V vs. RHE d1 mM RuCl3 + 0.1 M HClO4 (1 hr.) after 1, 2, 3 and 4 successive voltammetric scans e10 µM-1.0 M RuCl3 + 0.1 M HClO4 (10 s - 10 mins); θRu of 0.13 obtained with H2 redn. f0.5 mM RuCl3 + 0.2 M H2SO4 at OCP (2 min.) g2 mM Run+ ions prepared in 1 M HClO4 (see experimental section)

248

with a Ru exposure time > 20 minutes. To further exemplify this, an overlay plot of

the experimental OCP scan and the 3-fold theory curve are provided in Figure 5.11 for

visualization purposes. We believe this to be compelling evidence that spontaneous Ru

deposition occurs primarily in 3-fold geometries when Pt (along with adsorbed O) is

allowed to remain at OCP in the presence of Run+ ions.

Similar findings have also been established in the literature by means of voltammetry and

AES. For example, a recent report by Bonilla et al.35 revealed surface concentrations (θ)

of Ru using the decreased Hupd charge and assuming that each Ru3+ adsorbed onto three

Pt sites. Their θ values were comparable to those obtained using AES,44 which supports

the 3Pt:1Ru ratio that we have described above. In further consideration, the ratio of

Ru:Pt by this model is 1:3 or 0.33 ML. This value is consistent with many of the Ru

coverage values observed in the literature as indicated in Table 5.3. Interestingly, the Δμ

at OCP in Figure 5.7 taken at a Ru exposure time of 20 minutes resembles the line-shape

of the 1 or 2-fold Ru adsorption signatures. It is entirely reasonable to suggest that

initially Ru adsorbs in lower coordinated sites (1 or 2) likely when O coverage is low,

and subsequently fills in the 3-fold sites as more Ru adsorbs. However, in Figure 5.8 the

Δμ scan at 0.5 V - where the coverage is largest under potential control - reveals 1 or 2-

fold spectral line-shape (atop or bridged), while those at lower coverage reflect a 3-fold

fcc line-shape. Together, these data suggest that Ru prefers to deposit on atop/bridged

sites on clean Pt, but on 3-fold sites when co-adsorbates (e.g. H, O(H)) are present. This

is not surprising when one considers that the Hupd at low coverage (i.e. that at potentials

above 0.3 V)78 and the OH adsorbed on the atop sites leave only 3-fold sites available for

Ru deposition. When the H and OH coverage get larger (at lower and higher potential

249

Time, minutes0 10 20 30 40 50 60

Δμ M

agni

tude

, a.u

.

-0.01

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

OCP0.50 V

Figure 5.12 Relative coverage of Ru on Pt at 0.5V vs. OCP (ca. 0.9V) by comparison of experimental Δμ-magnitudes at the two potentials.

250

respectively, excluding OCP) Ru deposition apparently does not occur at all or only

occurs to an extent which is undetectable by Δμ. Likewise at OCP, the Ru initially

deposits on atop sites (probably along the corners/edges of the Pt clusters, see Figure 5.7)

on O covered Pt (the O takes the 3-fold sites) and then moves over to the 3-fold sites at

higher coverage. Thus the Run+ deposition on an O covered Pt surface behaves

remarkably similar to H adsorption on clean Pt. That is, adsorption occurs initially on

corner/edges in atop sites and later on 3-fold sites because of lateral interactions.79

5.3.5 Deposition time dependence and coverage

Finally, Figure 5.12 shows the time dependence of the Ru deposition at both OCP and

0.5 V. It is clear that the amount of Ru adsorbed at OCP is much larger than under

potential control at 0.5 V. The marked difference in this time-dependence may also be

reflecting a different deposition mechanism at the two potentials. At OCP, the Coulomb

enhanced deposition (i.e. attractive interaction between Oδ- ions on the surface and Run+

ions in solution) apparently occurs quite rapidly, reaching an asymptote already after

about 40 minutes and yielding a logarithmic type plot. While the deposition at 0.5 V

appears to increase in a linear fashion as illustrated by the regression line, and therefore

controlled by a different, possibly slower diffusion process near the surface. In any event,

this plot reveals the detrimental effect of bringing a cathode to OCP relative to

maintaining potential control. Not only is the total coverage enhanced at OCP, it reaches

this larger coverage all together, in a relatively shorter period of time.

Spontaneous deposition saturation times have been reported to be on the order of

seconds to minutes depending on Ru concentration in the bulk electrolyte, 25 although

much larger periods of 60 minutes or more have been observed in this and other studies.16

251

These discrepancies can likely be explained by factors such as presence of anions (i.e.

chlorides, nitrates etc.), effective surface area, size of solvated Run+ ions and catalytic

activity of the Pt surface. For example, any pre or co-adsorbed anions would impede

available sites for Ru deposition, hence requiring longer deposition times, lower coverage

or both. Many of the cited investigations have been performed on Pt (hkl) surfaces, which

would have much lower electro-active surface area than nanoparticles, possibly resulting

in longer requisite exposure times to achieve saturation. Also, it has been well established

that the various Pt(hkl) surfaces have different catalytic activity in terms of ORR and

adsorption and therefore would be expected to yield different results (see Table 5.3).52

Finally, as the size of the solvation sheath of the Run+ ion increases, it would likely

increase deposition time and/or decrease coverage due to steric hindrance and lower

charge density.

Figure 5.12 provides an estimate of the Ru coverage. The theoretical signature here

has a comparable magnitude with the amplitude of the experimental line. Note that the

Ru atom is in a 3-fold site on the surface for this calculation. Therefore, every Pt atom

should ‘see’ approximately 3 Ru atoms (the model only had 1 Ru), so at full coverage,

the theoretical line-shape should be approximately 3 times larger. Assuming the intensity

of scattering off of neighboring atoms varies directly with the number of such neighbors,

the estimated experimental coverage then becomes 1/3 or ca. 0.33 ML. This can be

compared with 0.1 ML when under potential control after 1 hour; although the coverage

appears to be linearly increasing still after 1 hour. Here, it is also worth drawing a

comparison to the experimental Hupd data shown in Table 5.1 which revealed that ~ 25 %

of the ECSA was lost during spontaneous deposition of Ru. This value is consistent with

252

the Δμ magnitudes analysis, suggesting it is a relatively reliable method of determining

adsorbate coverage.

These values compare quite well with other estimates in the literature (see summary

in Table 5.3) for saturation coverage in the case of spontaneous deposition of Ru on Pt.

An estimate of the coverage on nanoparticles with Δμ-XANES is only possible at very

high Pt dispersion (> 50%), thus giving rise to quite unambiguous ∆µ-XANES signatures

with sufficiently high data quality. The Ru coverage on Pt has been shown to occur in

both monolayer and multilayer fashion.14, 27, 31, 44 Using the Δμ-XANES method of

determining coverage on nanoparticles, it is not possible to determine the nature of the

adsorbed layer except for determining the binding site of the first layer of atoms on the

catalyst surface. To the best of our knowledge, this is the first estimate of Ru coverage on

a Pt catalyst using XAS. It is commonly accepted that x-ray methods are inherently bulk-

averaging techniques, and that deriving such information from either interfaces or

surfaces is rather difficult.34 However, we show in this study that with sufficiently good

data and appropriately designed experiments, the ∆µ-XANES analysis makes it possible

to obtain surface-sensitive information from XAS and that such an analysis may be of

value in cases where only in situ measurements are realistic.

5.3.6 ESR results

While much of this work has focused on the result of Ru deposition on the cathode

after dissolution at the anode and subsequent crossover through the membrane, in this

section, the effect of Ru crossover in the membrane is considered. Figure 5.13 shows that

as more Ru enters the membrane, a nearly linear decrease in water uptake occurs. This is

consistent with the findings of Lawton et al.58 and Ahmed et al.,80 where the presence of

253

Figure 5.13 Plot of gravimetrically measured water uptake versus extent of Ru exchange in Nafion membranes. Data are fit with a linear trend with a slope of -4.3 and y-intercept 11.5.

254

Figure 5.14 Plot of correlation time, τc, versus extent of Ru exchange in Nafion membranes calculated from the rotational diffusion of Tempone spin probe measured using X-Band ESR spectroscopy. Data are fit with a linear trend with slope 1.0711 × 10-9 and y-intercept 1.4037 × 10-9.

255

Al3+ and Fe3+ in the Nafion membrane caused much lower swelling than lower valence

cations. Their conclusion58 - considering the modeling works of Niemark and

Vishnyakov81, 82 wherein trivalent ions were shown to rigidify the membrane’s backbone

and side chain regions due to ionic crosslinking83 - was that the membrane interactions

with the trivalent ions hinder swelling and minimize solvation.

Lawton et al. have also studied the effects of hydration in the membrane using a free

volume model. They observed that lower hydration levels lead to a lower rate of probe

rotation. 58 Figure 5.14 depicts the effect of Ru exchange on the τc of the probe. Based on

the Stokes-Einstein relationship in Equation 5.3, this suggests that the micro-viscosity of

the fluid state increases linearly with the presence of Ru3+ in the membrane. This could

result in slower vehicular diffusion across the membrane as a result of lower hydration

levels and an altering of the membrane’s free volume by ionic interactions with the Ru.

Reports of Al3+ ionomer exchange81, 82 and an in depth study of Ca2+ contamination in

the Nafion membrane84 have indicated that the multivalent ions have a higher binding

affinity than protons to the anion groups in the membrane. This suggests that over the

lifetime of fuel cell operation, higher levels of Ru could build up inside the membrane.

Ion contaminants with higher valance have been shown to reduce the proton conductivity

as well as dehydrate the membrane,85 which is consistent with our findings. This is of

further concern as transition metals existing in the membrane have been found to catalyze

radical attack on the membrane that leads to degradation.86

Even when present at lower levels, the trivalent Ru3+ ion decreases the equilibrium

hydration level and increases the micro-viscosity of the fluid state, which could alter the

water management attributes of the membrane in the fuel cell. More studies need to be

256

accomplished to fully understand the long term effects of Ru cross over considering

membrane degradation, proton conductivity, and vehicular diffusion.

5.4 Summary and Conclusions

Cyclic voltammograms of Pt taken after exposure (at open circuit) to 2.0 mM Run+

contaminated HClO4 reveal a significant increase in double-layer charge capacitance,

decreased Pt-O[H] formation/reduction and lower Hupd charge when compared to the CVs

taken in clean HClO4. An attempt to remove the spontaneously adsorbed Ru revealed that

some of the Ru could be removed by potential cycling to 1.4 V vs. RHE, however, not all

the Ru could be removed. Integration of the Hupd charge area indicated that adsorbed Ru

decreased the ECSA by approximately 25 %, of which only 5 % could be reclaimed upon

the cleaning procedure. These results suggest that spontaneously deposited Ru leached

out from a DMFC anode could impart irreversible damage to a fuel cell cathode via

catalytic site blocking.

ORR polarization sweeps reveal an increased overpotential by approximately 20 mV

after being subject to 2.0 mM Run+ for 90 minutes at open circuit potential. Although a

diffusion limited current was obtained, the magnitude of the current was slightly

decreased, supporting a site blocking theory. When the ORR polarization was performed

in the Ru contaminated electrolyte (2.0 mM), the overpotential increased ~ 150 mV and

no diffusion limiting current was obtained. The Tafel curves revealed a normal Pt

response with slopes close to –60 and –120 mV/decade for clean Pt, spontaneously

deposited Ru/Pt and ORR in Run+ contaminated HClO4. The Tafel line shapes were

relatively unchanged, consistent with the theory that there is no overall change to the rate

257

determining step of the reaction. However, the decrease in E (as well as exchange

current density) illustrates the adverse effect that Ru contamination has on the

overpotential of the reaction.

The EXAFS analysis revealed small but significant changes to the coordination

number NPt-Pt as a function of potential and the presence of Ru. The largest change to

NPt-Pt occurred when the clean Pt electrode was subject to Ru contaminated electrolyte at

OCP. Smaller changes in NPt-Pt occurred when the electrode potential was maintained, but

even these small changes could be related to atop vs. 3-fold Ru deposition, supporting the

Δμ results. The large decrease in NPt-Pt at OCP (7.16 to 6.24) is reflective of O adsorption

in 3-fold sites, as well as Ru deposition primarily in the 3-fold fcc sites at higher

coverage, resulting from the Pt particles becoming more flat. The Δμ−XANES analysis

support the EXAFS results and suggests Ru adsorption may proceed by an initial

atop/bridge Ru adsorption followed by adsorption onto the 3-fold fcc sites of the faces,

when the electrode is maintained at OCP as suggested by the FEFF 8.0 line shape

assessment. The Δμ curves, for which potential control was maintained, reveal smaller

magnitudes (< 2 % of the XANES signal) over a potential range of 0.3 – 0.8 V, indicating

relatively low Ru coverage (hindered by H or OH adsorption) and then a significant

increase at OCP apparently assisted by Coulombic forces involving the adsorbed O

atoms. The surface coverage is < 1 ML as shown by Figure 5.6 and also evident in the

CVs and the ESCA results in Table 5.1. It is further established that Ru adsorption may

enhance H2O activation by a ligand effect above 0.7 - 0.8 V, but partially impedes it by

blocking Pt sites for O2 dissociation at lower potentials.

258

The ESR analysis of Ru3+ in the Nafion membrane indicated changes in the

characteristics of the membrane in the presence of Ru ions. Increasing numbers of Ru

ions per sulfonic acid group in the membrane were observed to decrease the water uptake

of the membrane and also increase the micro-viscosity of the fluid regions possibly as a

result of a change of the free volume of the membrane. Ru ions in the membrane could

slow vehicular diffusion, decrease proton conductivity, and catalyze degradation of the

membrane by radical attack.

5.5 Acknowledgements

Financial support for this project was provided by the Army Research Office via a

single investigator grant and a Multi-University Research Initiative (Case Western

Reserve University, PI). We are also grateful for the use of beam lines X-18B, X-11A

and X-11B at the National Synchrotron Light Source, Brookhaven National Laboratory,

Upton, NY, which is supported by the U.S. Department of Energy, Office of Science,

Office of Basic Energy Sciences under Contract No. DE-AC02-98CH10886. We also

acknowledge Geolabs, Inc., Braintree, MA for ICP-MS measurements.

259

5.5 References

1. Deluca, N. W. & Elabd, Y. A. J. Polym. Sci., Part B: Polym. Phys. 44, 2201-2225 (2006).

2. Jarvi, T. D. & Stuve, E. M. (eds.) Fundamental Aspects of Vacuum and Electrocatalytic Reactions of Methanol and Formic Acid on Platinum Surfaces (Wiley-VCH, 1998).

3. Waszczuk, P., Crown, A., Mitrovski, M. & Wieckowski, A. (eds.) Methanol and Formic Acid Oxidation on Ad-Metal Modified Electrodes (Wiley-VCH, 2003).

4. Park, Y. S., Wieckowski, A. & Weaver, M. J. J. Am. Chem. Soc. 125, 2282 (2003).

5. Mukerjee, S. & McBreen, J. Journal of the Electrochemical Society 146, 600 (1999).

6. Wang, K., Gasteiger, H. A., Markovic, N. M. & Ross, P. N. Electrochimica Acta 41, 2587 (1996).

7. Mukerjee, S. et al. Investigation of enhanced CO tolerance in proton exchange membrane fuel cells by carbon supported PtMo alloy catalyst. Electrochem. Solid-State Lett. 2, 12-15 (1999).

8. Piela, P., Eicks, C., Brosha, E., Garzon, F. & Zelenay, P. J. Electrochem. Soc. 151, A2053-A2059 (2004).

9. Eicks, C., Piela, P., Davey, J. & Zelenay, P. J. Electrochem. Soc. 153, A171-A178 (2006).

10. Cao, D. & Bergens, S. H. An organometallic deposition of ruthenium adatoms on platinum that self poisons at a specific surface composition. A direct methanol fuel cell using a platinum-ruthenium adatom anode catalyst. J. Electroanal. Chem 533, 91-100 (2002).

11. Chrzanowski, W., Kim, H. & Wieckowski, A. Enhancement in methanol oxidation by spontaneously deposited ruthenium on low-index platinum electrodes. Catal. Lett 50, 69-75 (1998).

12. Chrzanowski, W. & Wieckowski, A. Ultrathin Films of Ruthenium on Low Index Platinum Single Crystal Surfaces: An Electrochemical Study. Langmuir 13, 5974-5978 (1997).

13. Crown, A. Surface studies of spontaneously deposited ruthenium on Pt(hk) electrodes. A summary report to The Electrochemical Society for the 2000 Colin Garfield Fink Summer Research Fellowship. Electrochem. Soc. Interface 10, 49-50 (2001).

260

14. Crown, A., Johnston, C. & Wieckowski, A. Growth of ruthenium islands on Pt(hkl) electrodes obtained via repetitive spontaneous deposition. Surf. Sci. 506, L268-L274 (2002).

15. Kim, H., Rabelo de Moraes, I., Tremiliosi-Filho, G., Haasch, R. & Wieckowski, A. Chemical state of ruthenium submonolayers on a Pt(111) electrode. Surf. Sci. 474, L203-L212 (2001).

16. Vigier, F., Gloaguen, F., Leger, J. M. & Lamy, C. Electrochemical and spontaneous deposition of ruthenium at platinum electrodes for methanol oxidation: an electrochemical quartz crystal microbalance study. Electrochim. Acta 46, 4331-4337 (2001).

17. Gancs, L., Hult, B. N., Hakim, N. & Mukerjee, S. Electrochem. Solid-State Lett. 10, B150-B154 (2007).

18. Wen, J. & Zhou, Z. Pseudocapacitance characterization of hydrous ruthenium oxide prepared via cyclic voltammetric deposition. Mater. Chem. Phys. 98, 442-446 (2006).

19. Wu, G., Li, L. & Xu, B.-Q. Effect of electrochemical polarization of PtRu/C catalysts on methanol electrooxidation. Electrochim. Acta 50, 1-10 (2004).

20. Cramm, S. et al. Anal. Chem. 358, 189 (1997).

21. Friedrich, K. A., Geyzers, K. P., Dickinson, A. J. & Stimming, U. J. Electroanal. Chem 524, 261 (2002).

22. Friedrich, K. A., Geyzers, K. P., Linke, U., Stimming, U. & Stumper, J. J. Electroanal. Chem 402, 123 (1996).

23. Friedrich, K. A., Geyzers, K. P., Marmann, A., Stimming, U. & Vogel, R. Z. Phys. Chem. 208, 137 (1999).

24. Maillard, F., Gloaguen, F., Hahn, F. & Leger, J. M. Fuel Cells 2, 143 (2002).

25. Maillard, F., Gloaguen, F. & Leger, J. M. Preparation of methanol oxidation electrocatalysts: ruthenium deposition on carbon-supported platinum nanoparticles. J. Appl. Electrochem. 33, 1-8 (2003).

26. Christensen, P. A., Jin, J.-M., Lin, W.-F. & Hamnett, A. Identification of CO Adsorbed at Ru and Pt Sites on a Polycrystalline Pt/Ru Electrode and the Observation of Their Oxidation and Free Interchange under Open Circuit Conditions. J. Phys. Chem. B 108, 3391-3394 (2004).

27. Herrero, E., Feliu, J. & Wieckowski, A. Langmuir 15, 4944 (1999).

261

28. Tremiliosi-Filho, G. et al. Reactivity and activation parameters in methanol oxidation on platinum single crystal electrodes 'decorated' by ruthenium adlayers. J. Electroanal. Chem. 467, 143-156 (1999).

29. Crown, A. & Wieckowski, A. Scanning tunneling microscopy investigations of ruthenium- and osmium-modified Pt(100) and Pt(110) single crystal substrates. Phys. Chem. Chem. Phys. 3, 3290-3296 (2001).

30. Crown, A., Moraes, I. R. & Wieckowski, A. Examination of Pt(111)/Ru and Pt(111)/Os surfaces: STM imaging and methanol oxidation activity. J. Electroanal. Chem. 500, 333-343 (2001).

31. Waszczuk, P. et al. Methanol electrooxidation on platinum/ruthenium nanoparticle catalysts. J. Catal. 203, 1-6 (2001).

32. Babu, P. K., Tong, Y. Y., Kim, H. & Wieckowski, A. Nanostructured electrode surfaces studied by electrochemical NMR. J. Electroanal. Chem 524, 157 (2002).

33. Lu, G. Q., Waszczuk, P. & Wieckowski, A. Oxidation of CO adsorbed from CO saturated solutions on the Pt(111)/Ru electrode. J. Electroanal. Chem 532, 49 (2002).

34. Maillard, F., Lu, G. Q., Wieckowski, A. & Stimming, U. Ru-Decorated Pt Surfaces as Model Fuel Cell Electrocatalysts for CO Electrooxidation. J. Phys. Chem. B 109, 16230-16243 (2005).

35. Bonilla, S. H. et al. Catalytic effects of ruthenium and osmium spontaneous deposition on platinum surfaces toward methanol oxidation. J. Colloid Interface Sci. 288, 377-386 (2005).

36. Colle, V. D., Giz, M. J. & Tremiliosi-Filho, G. Spontaneous deposition of Ru on Pt(100): Morphological and electrochemical studies. Preliminary results of ethanol oxidation at Pt(100)/Ru. J. Braz. Chem. Soc. 14, 601-609 (2003).

37. Waszczuk, P., Barnard, T. M., Rice, C., Masel, R. I. & Wieckowski, A. A nanoparticle catalyst with superior activity for electrooxidation of formic acid. Electrochem. Commun. 4, 599-603 (2002).

38. Lu, L., Yin, G., Wang, Z., Cai, K. & Gao, Y. Electrocatalytic oxidation of dimethyl ether on ruthenium modified platinum single crystal electrodes. Catal. Commun. 10, 971-974 (2009).

39. Ku, B., Jung, C. & Rhee, C. K. Atomic arrangements inside Ru and Os nanoislands spontaneously deposited on Pt(111). J. Phys. Chem. B 110, 13425-13429 (2006).

262

40. Strbac, S., Johnston, C. M., Lu, G. Q., Crown, A. & Wieckowski, A. In situ STM study of nanosized Ru and Os islands spontaneously deposited on Pt(111) and Au(111) electrodes. Surf. Sci. 573, 80-99 (2004).

41. Vericat, C., Wakisaka, M., Haasch, R., Bagus, P. S. & Wieckowski, A. Binding energy of ruthenium submonolayers deposited on a Pt(111) electrode. J. Solid State Electrochem. 8, 794-803 (2004).

42. Lin, W. F., Zei, M.S., Eisworth, M., Ertl, G. Electrocatalytic Activity of Ru-Modified Pt(111) Electrodes toward CO Oxidation. J. Phys. Chem. B. 103, 6968 (1999).

43. Park, K.-W. & Sung, Y.-E. J. Phys. Chem. B 109, 13585 (2005).

44. Iwasita, T., Hoster, H., John-Anacker, A., Lin, W. F. & Vielstich, W. Methanol Oxidation on PtRu Electrodes. Influence of Surface Structure and PtRu Atom Distribution. Langmuir 16, 522-529 (2000).

45. Frelink, T., Visscher, W. & Van Veen, J. A. R. Langmuir 12, 3702 (1996).

46. Mukerjee, S., Ziegelbauer, J., Arruda, T. M., Ramaker, D. & Shyam, B. Understanding Electrocatalytic Pathways in Low and Medium Temperature Fuel Cells: Synchrotron-based In Situ X-Ray Absorption Spectroscopy. Electrochem. Soc. Interface 17, 46-52 (2008).

47. Russell, A. E. & Rose, A. X-ray Absorption Spectroscopy of Low Temperature Fuel Cell Catalysts. Chem. Rev. 104, 4613-4635 (2004).

48. Arruda, T. M., Shyam, B., Ziegelbauer, J. M., Mukerjee, S. & Ramaker, D. E. Investigation into the Competitive and Site-Specific Nature of Anion Adsorption on Pt Using In Situ X-ray Absorption Spectroscopy. J. Phys. Chem. C 112, 18087-18097 (2008).


50. Teliska, M., Murthi, V. S., Mukerjee, S. & Ramaker, D. E. Correlation of Water Activation, Surface Properties, and Oxygen Reduction Reactivity of Supported Pt–M/C Bimetallic Electrocatalysts Using XAS. J. Electrochem. Soc. 152, A1259 (2005).


52. Marković, N. M. & Ross, P. N. Surface science studies of model fuel cell electrocatalysts. Surf. Sci. Rep. 45, 117-229 (2002).

263

53. Ziegelbauer, J. M. et al. Direct Spectroscopic Observation of the Structural Origin of Peroxide Generation from Co-Based Pyrolyzed Porphyrins for ORR Applications. J. Phys. Chem. C 112, 8839-8849 (2008).

54. Ziegelbauer, J. M. et al. Fundamental Investigation of Oxygen Reduction Reaction on Rhodium Sulfide-based Chalcogenides. J. Phys. Chem. C (accepted).


56. Jung, E. H. et al. Methanol crossover through PtRu/Nafion composite membrane for a direct methanol fuel cell. Int. J. Hydrogen Energy, 32 (7), 903-907 (2007).

57. Lawton, J. S., Smotkin, E. & Budil, D. E. Electron Spin Resonance Investigation of Microscopic Viscosity, Ordering, and Polarity in Nafion Membranes Containing Methanol−Water Mixtures. J. Phys. Chem. B 112, 8549 (2008).

58. Lawton, J. S. & Budil, D. E. J. Phys. Chem. B In Press (2009).


60. McBreen, J., O'Grady, W. E., Pandya, K. I., Hoffman, R. W. & Sayers, D. E. EXAFS study of the nickel oxide electrode. Langmuir 3, 428 (1987).

61. Newville, M. J. IFEFFIT. Synchrotron Radiat. 8, 322 (2001).

62. Newville, M., Livins, P., Yacoby, Y., Stern, E. A. & Rehr, J. J. Near-edge x-ray-absorption fine structure of Pb: A comparison of theory and experiment. Phys. Rev. B 47, 14126 (1993).

63. van Dorssen, G. E., Koningsberger, D. C. & Ramaker, D. E. Separation of double-electron and atomic XAFS contributions in x-ray absorption spectra of Pt foil and Na2Pt(OH)6. J. Phys.: Condens. Matter 14, 13529-13541 (2002).

64. Ramaker, D. E., Mojet, B. L., Miller, J. T. & Koningsberger, D. C. An atomic X-ray absorption fine structure study of the influence of hydrogen chemisorption and support on the electronic structure of supported Pt particles. Top. Catal. 10, 157-165 (2000).

65. Koningsberger, D. C., Mojet, B. L., Miller, J. T. & Ramaker, D. E. XAFS spectroscopy in catalysis research: AXAFS and shape resonances. J. Synchrotron. Radiat. 6, 135-141 (1999).

264

66. Ramaker, D. E., Mojet, B. L., Koningsberger, D. C. & O'Grady, W. E. Understanding Atomic X-ray Absorption Fine Structure in X-ray Absorption Spectra. J. Phys.: Condens. Matter 10, 8753-8770 (1998).

67. Ankudinov, A. L., Ravel, B., Rehr, J. J. & Conradson, S. D. Real Space Multiple Scattering Calculation and Interpretation of X-ray Absorption Near Edge Structure. Phys. Rev. B 58, 7565-7576 (1998).

68. Janin, E., von Schenck, H., Göthelid, M. & Karlsson, U. O. Bridge-bonded atomic oxygen on Pt(110). Phys. Rev. B 61, 13144-13149 (2000).

69. Schneider, D. J. & Freed, J. H. Biol. Magn. Reson. 8, 1 (1989).

70. Schneider, D. J. & Freed, J. H. Adv. Chem. Phys. 73, 387 (1989).

71. Schmidt, T. J., Paulus, U. A., Gasteiger, H. A. & Behm, R. J. The oxygen reduction reaction on a Pt/carbon fuel cell catalyst in the presence of chloride anions. J. Electroanal. Chem. 508, 41-47 (2001).

72. Ravel, B. & Newville, M. Athena & Artemis. J. Synchrotron Radiat. 12, 537-541 (2005).

73. Roth, C., Benker, N., Mazurek, M., Scheiba, F. & Fuess, H. Pt-Ru fuel cell catalysts subjected to H2, CO, N2 and air atmosphere: An X-ray absorption study. Appl. Catal., A 319, 81-90 (2007).

74. Scott, F. J., Mukerjee, S. & Ramaker, D. E. CO coverage/oxidation correlated with PtRu electrocatalyst particle morphology in 0.3 M methanol by in situ XAS. J. Electrochem. Soc. 154, A396-A406 (2007).

75. Scott, F. J., Roth, C. & Ramaker, D. E. J. Phys. Chem. C 111, 11403-11413 (2007).


77. Roth, C. & Ramaker, D. E. (ed. White, R. E.) (Springer, 2009).

78. Ramaker, D. E., Teliska, M., Zhang, Y., Stakheev, A. Y. & Koningsberger, D. C. Understanding the influence of support alkalinity on the hydrogen and oxygen chemisorption properties of Pt particles. Comparison of X-ray absorption near edge data from gas phase and electrochemical systems. Phys. Chem. Chem. Phys. 5, 4492-4501 (2003).

79. Koningsberger, D. C., Oudenhuijzen, M. K., De Graaf, J., Van Bokhoven, J. A. & Ramaker, D. E. In situ X-ray absorption spectroscopy as a unique tool for obtaining information on hydrogen binding sites and electronic structure of

265

supported Pt catalysts: towards an understanding of the compensation relation in alkane hydrogenolysis. J. Catal. 216, 178-191 (2003).

80. Ahmed, M., Elmuntasir, I., Huang, K.-L. & Holsen, T. M. Nafion-117 Behavior during Cation Separation from Spent Chromium Plating Solutions Ind. Eng. Chem. Res. 48, 6805-6810 (2009).

81. Vishnyakov, A. & Neimark, A. V. Specifics of solvation of sulfonated polyelectrolytes in water, dimethylmethylphosphonate, and their mixture: A molecular simulation study. J. Chem. Phys. 128, 164902 (2008).

82. Vishnyakov, A. & Neimark, A. V. Molecular Dynamics Simulation of Nanoscale Distribution and Mobility of Water and Dimethylmethylphosphonate in Sulfonated Polystyrene. J. Phys. Chem. B 112, 14905–14910 ( 2008).

83. Suleiman, D., Napadensky, E., Sloan, J. M. & Crawford, D. M. Thermogravimetric characterization of highly sulfonated poly(styrene-isobutylene-styrene) block copolymers: Effects of sulfonation and counter-ion substitution. Thermochim. Acta 460, 35-40 (2007).

84. Okada, T., Nakamura, N., Makoto, Y. & Sekine, I. Ion and water transport characteristics in membranes for polymer electrolyte fuel cells containing H+ and Ca2+ cations. J. Electrochem. Soc. 144, 2744-2750 (1997).

85. Kelly, M. J. et al. Conductivity of polymer electrolyte membranes by impedance spectroscopy with microelectrodes Solid State Ionics 176, 2111-2114 (2005).

86. Halalay, I., Merzougui, B. A. & Mance, A. M. Three mechanisms for protecting the PEM fuel cell membrane, plates and catalyst. ECS Trans. 16, 969-981 (2008).

266

Chapter 6

Probing the Influence of Polyvinyl Pyrrolidone (PVP) on Supported

Platinum Electrocatalysts in 0.1M HClO4 Using in situ X ray

Absorption Spectroscopy∗

6.1 Introduction

The number of papers reporting the synthesis and catalytic activity of metal

nanoparticles synthesized with the aid of polymers containing functional groups,

commonly known as capping agents, has increased significantly over the last decade.

Many of these polymers are known to stabilize colloidal suspensions and catalyst

nanoparticles by preventing coalescence and reducing agglomeration. The coalescence of

nanoparticles is responsible in part for the aging of catalysts resulting in an increase in

the average particle size that invariably leads to a reduced electrochemically active

∗ To be published in The Journal of Physical Chemistry C Authors: Badri Shyam, Ceren Susut, Yu Ye Tong and David E. Ramaker Sample preparation and electrochemistry data collected by authors affiliated with Georgetown University. XAS data collection and analysis carried out by authors from the The George Washington University.

267

surface area (ECSA). Given that nanoparticle growth by agglomeration is one of the main

reasons for the loss of catalytic activity1, 2 in a fuel cell, the use of capping agents to

stabilize the nanoparticles holds great promise.3 Some commonly used

stabilizing/capping agents include oleylamine (OA), poly(3-thiophene acetic acid),

polyvinyl alcohol (PVA), polystyrene-4-sulfonate (PSS), poly(3,4-

ethylenedioxythiophene) (PEDOT), poly-2-ethyloxazoliine (POX), polyethylene oxide

(PEO), polyethyleneimine (PIM), polyamyl alcohol (PAA) and poly-N-vinyl 2–

pyrrolidone (PVP), to name a few. Using various principles such as controlling the

strength of the reducing agent and varying the molar ratios of the metal precursor and

capping agent, one can vary the kinetics of formation of the nanoparticles, leading to

particle shape-controlled syntheses,4, 5 a field of research first explored by Ahmadi and

co-workers.6 Since then, several other reports 7-12 and review articles 13-16 discussing the

synthesis and properties of polymer-stabilized or specifically shaped-nanoparticles and

catalysts have been published.

PVP is the polymeric form of n-vinyl pyrrolidone, which is a 5-membered pyrrole

ring having a vinyl group and carbonyl group (see Figure 6.1). While this polymer is

hydrophilic and water-soluble, it also contains a hydrophobic hydrocarbon backbone

(essentially a surfactant) – a combination which provides functional polymers with very

interesting chemical properties. Polymeric PVP may have molecular weights ranging

from 2500-360,000 and has many applications due to its interesting chemical properties.

It is known to be non-toxic 17 and soluble in both organic and aqueous solvents, and

hence is widely used in medicinal applications 18 and personal care products. 19 PVP is

hygroscopic and also has a tendency to hydrogen bond to water. 20 As it effectively

268

stabilizes most metal colloids and nanoparticles, is has been used extensively to

synthesize stabilized colloids and nanoparticles. For more information on the physico-

chemical properties and spectroscopic characterization of PVP, the reader is referred to

early comprehensive studies on this capping agent. 17, 21-25

There have been a number of papers showcasing the various catalytic properties of

PVP-capped colloidal Pt nanoparticles, 26-30 alloy nanoparticles 31-33 and shaped-

nanoparticles. 34-40 Apart from the common shapes such as cubes, tetrahedra and

octahedral nanoparticles, more exotic shapes such as triangular plates, hollow-tetrahedra,

star-shaped NPs, and rod-like NPs have also been synthesized 30, 41-43. Most recently, PVP

has been used in the synthesis of core-shell nanoparticles. 44-47 Studies on PVP-M (where

M is a metal other than Pt) are also numerous. 46-56 It is interesting that PVP is used so

ubiquitously in colloidal catalysts that catalytic studies on PVP-M materials have been

used for baseline experiments in order to highlight the effect of further additives such as

salts of Fe, Co, Ni etc. 13

Given the large amount of literature involving the widespread use of PVP as a

stabilizing agent, it is almost surprising that so little is known about the actual influence

of PVP on the catalytic properties of platinum and other metals. This is an important

question as it is now generally accepted that residual PVP is almost always present in the

synthesized nanoparticles and indeed, even essential for preserving the shapes of these

nanoparticles. 57-59 But does the presence of the PVP influence their catalytic properties

beyond just stabilizing and maintaining the nanoparticle shape? The answer appears to be

yes, but it has also been acknowledged, not surprisingly, that the role and degree of

269

Figure 6.1 Chemical formula for PVP polymer (a), illustration of PVP carbonyl-Pt interaction (b) and (c); illustration of PVP polymer on Pt (d). Models after Borodko et al.66

a) PVP b) PVPN c) PVPCT/Pt d) (PVP)n/Pt

270

influence that the stabilizing agent has on the native properties of the substrate metal will

depend on the reaction being catalyzed. 60, 61 The precise reason for the enhancement of

the catalytic activity for some reactions on Pt and other metals by PVP is not yet known.

62 There exists a paucity of papers wherein the Pt-PVP interaction has been directly

observed and studied. Even in these few studies, the stabilized Pt colloids and

nanoparticles were mostly dried powders that were studied either in vacuum or in air

using spectroscopic techniques such as x-ray photoelectron spectroscopy (XPS), Fourier-

transform infrared (FTIR) spectroscopy and UV-Raman spectroscopy. 63-66 We are not

aware of any spectroscopic study of the Pt-PVP interaction in an aqueous environment

such as existing in a fuel cell, except for our previous work on unsupported PVP-Pt black

catalysts, when SEIRAS was utilized.59 Indeed, as was noted above, direct evidence for

the interaction between a capping agent and the metal catalyst is rare, and often carried

out under conditions that make cross-study comparisons hard. 67, 68

In this work, we attempt to understand the Pt-PVP interaction under electrocatalytic

reaction conditions for supported Pt/C catalysts using in situ XAS and electrochemical

measurements. In order to effectively exploit the various properties of stabilizing agents,

especially for applications in catalysis, a systematic effort towards understanding the

influence of capping agents on Pt and other commonly employed catalytic metals is

required. We report findings from an in situ X-ray Absorption Spectroscopy (XAS) study

of PVP-capped Pt/C catalysts utilizing the novel Δμ-XANES (X-ray absorption near edge

structure) technique as well as the more common EXAFS (Extended x-ray absorption

fine structure) analysis as illustrated in Figure 6.2. We compare results on these PVP-

271

Figure 6.2 Illustration of EXAFS and Δµ-XANES analysis procedure, with pre-edge background removal, normalization, and then isolation of the EXAFS signal and fit to model functional in EXAFS, and isolation of the adsorbate effect on the XANES by taking the difference, Δµ. After Roth et al.72

E(eV

Background Normalizatio

E(eV

Background Normalizatio Chi(k) Fourier

E(eV k(Å-1) R(Å

EXAFS

XANES

EXAFS

μ(ad/Pt)

μ(Pt) Δμ

E(eV E(eV E(eV

Fit

272

Pt/C catalysts with our previous work on the PVP-Pt black catalysts, but the current work

is the first to utilize XAS. In situ XAS is a powerful technique to study electrocatalysis,

as EXAFS analysis furnishes information on coordination number and bond lengths of

the catalysts while changes in their oxidation state can be followed by analyzing the

XANES region. In addition to the conventional XAS analysis just described, in this study

we also used the ∆µ-XANES method 69-71 to understand the interaction of PVP with Pt as

well as to obtain site-specific adsorbate information on the adsorbates (H, OH and O)

existing on these catalysts. The method has been successfully used in the past to study

adsorption of various ions and species on Pt catalysts, 70-78 ligand effects of alloyed

metals in Pt alloy catalysts, 75, 76 and more recently, even the effects of ligand

(triphenylphosphine triphosphonate, or TPPTP) stabilized Pt nanoparticles on water

activation.79 These Δμ-XANES results are correlated with water activation, methanol

oxidation, CO stripping and chronoamperometric measurements in 0.5M H2SO4 and 0.1

M HClO4, revealing interesting details regarding the nature of the Pt-PVP interaction.

Various mechanisms of interaction will be presented and discussed along with others’

findings in order to better understand the influence of PVP on supported Pt nanoparticle

electrocatalysts.

6.2 Experimental

6.2.1 Catalyst synthesis

PVP-capped carbon-supported Pt samples were synthesized as follows. For each

synthesis, 2.5 ml of ethylene glycol (EG, J.T. Baker) was refluxed for 5 minutes. Portions

of the PVP (MW = 55,000, Sigma Aldrich) and the Pt/C catalysts (Johnson-Matthey, 40

273

wt.% Pt/C) were added to the boiling EG every 30 seconds over a 16 minute period. Two

different samples were prepared (Pt-PVP12 and Pt-PVP24) by using two PVP:Pt molar

ratios viz., 1:12 and 1:24. The resulting mixture was then refluxed for an additional 60

minutes. The resulting catalysts were centrifuged repeatedly at 5000 rpm (Sorvall, RC 5C

PLUS) to remove any unreacted PVP. It should be emphasized that this catalysts

preparation procedure is significantly different from those utilized to prepare particularly

shaped Pt particles, as the Pt particles in Pt/C in the current work have already been

formed, and we assume little or no re-shaping of the Pt particles occurs upon mixing with

the PVP. Thus, in this work we are only examining the affect of the Pt-PVP interaction

for supported Pt/C catalysts, and not a shape effect.

6.2.2 XAS - sample preparation and data collection

In situ XAS experiments were conducted at the National Synchrotron Light Source

(NSLS) located at Brookhaven National Laboratory in Upton, N.Y. The data were

collected at beamline X-11B at the Pt L3 edge (11564 eV) in transmission mode using a

standard 3-detector setup. The beam energy was selected using a Si(111) double-crystal

monochromator and energy calibration was accomplished by recording the XAS

spectrum of a 7 µm Platinum foil placed between the second and third ionization

detectors (It and Iref). The storage ring operated at an energy of 2.81 GeV with ring

currents varying between 60 and 200 mA. The beam spot size measured ca. 1 mm Χ 4

mm in area. Full Pt L3 edge EXAFS scans (-200 eV to 16k) were collected at difference

potentials by holding the cell potentiostatically during every measurement. The data in

274

the XANES region (-50 to 80 eV) were collected in steps of 0.5 eV while all data in the

EXAFS region (E > 80 eV) were collected in steps of 3.5 - 4.0 eV.

All XAS measurements were carried out in a specially designed in situ XAS

spectroelectrochemical cell (Courtesy of the Mukerjee group, Northeastern University),

which contains an electrochemical half-cell and an arrangement to accommodate a

reference electrode. Catalyst inks of the Pt/C and PVP-Pt/C materials were made in

deionized water and applied to a carbon cloth (Panex ® 30 woven carbon fiber fabric,

Zoltek) with a paintbrush. Through the application of multiple coats, the carbon cloth

electrodes were repeatedly dried and weighed till the requisite loading of ca. 3-4 mg/cm2

was obtained on the electrodes. The assembled in situ XAS cell was then placed between

the first two detectors (I0 and It) in order to obtain an x-ray absorption spectrum of the

catalysts. The electrode potential was controlled by a potentiostat (EcoChemie Autolab

PGSTAT-30) and all potentials were measured against a standard hydrogen electrode.

6.2.3 Electrochemical measurements

All electrochemical measurements were carried out in an Ar-blanketed conventional

3-electrode electrochemical cell using either an EG&G 273A potentiostat (Princeton

Applied Research) controlled by a PC with a CoreWare software package or a CHI

electrochemical work station (CHI 660C). Unless otherwise noted, all cyclic

voltammograms (CV) were recorded at a scan rate of 50 mV/s. A Pt wire and Ag/AgCl

(3M) standard electrode were used as counter and reference electrodes respectively. All

chronoamperometric (CA) measurements were recorded at a constant potential of 0.39 V

(vs. RHE).

275

The supporting electrolytes used were 0.1M HClO4 and 0.5M H2SO4 solutions made

up using mill-Q water (18.2 MΩ). Specifically, 0.1 M HClO4 solutions were prepared

with high-purity perchloric acid (GFS chemicals, sulfate content 0.0001%). Note that

while all electrochemical data were physically measured against the Ag/AgCl electrode,

the data are expressed with respect to the Reversible Hydrogen Electrode (RHE) in all

figures and the text. The conversion numbers are 0.29 V for 0.1M HClO4, 0.26 V for

0.5M H2SO4. These numbers were determined by measuring the open circuit potential

(OCP) between the Ag/AgCl (3M) reference electrode and the Pt wire counter electrode

under H2 bubbling. All methanol oxidation experiments were performed in 0.5M H2SO4

+ 0.5M CH3OH or 0.1M HClO4 + 0.5M CH3OH solutions.

For CO stripping measurements, the CO was adsorbed by saturating the

electrochemical cell with CO for 10 minutes while holding the electrode potential at 0.1V

vs. an Ag/AgCl electrode. The cell was then purged with Ar for 30 minutes before the

CO stripping CV was recorded. All currents reported were normalized to the Pt surface

area calculated by the total hydrogen desorption charge (220 µC/cm2 for commercially

available Pt black).

6.3 EXAFS analysis

EXAFS analysis on the data was accomplished using the programs available in the

IFEFFIT suite. 80 The ATHENA program was used for all the raw data processing such

as energy calibration of the reference foil spectra, alignment and normalization while

ARTEMIS was used for carrying out the EXAFS fits to the data. In order to the extract

coordination numbers (NPt-Pt) and bond distances (RPt-Pt) from the XAS data, standard

procedures were used. All XAS spectra were first normalized in the EXAFS region (120-

276

930 eV above the Pt L3 edge) and the oscillatory component χ(k) extract. Only single-

shell fits were carried out on the data using crystallographic parameters for a face-

centered cubic (fcc) Pt crystal. The set of rectangular coordinates for the Pt cluster were

generated using the ATOMS program that is also available as part of the IFEFFIT suite.

81 The k2-weighted χ(k) data were fit over a k-range of 2 Å-1 < k < 13.95 Å-1 and an R-

window of 1.699 Å < r < 3.229 Å. The data were Fourier-transformed using a Kaiser-

Bessel window. The experimental NPt-Pt values were calculated by using the relation:

NPt-Pt (exp.) = 1/S02(theo.) * [NPt-Pt (theo.) * amp (exp.)] (Eq. 6.1)

where S02 (theo.) is the theoretical many-body amplitude reduction factor and is

calculated using FEFF 8.0; a value of 0.934 was obtained for Pt as absorber atom. NPt-Pt

(theo.) is the theoretical coordination around Pt in a face-centered cubic lattice and is

equal to 12. The final variable in the equation, amp (exp.), is the amplitude value

obtained from each fit. As is generally performed in all of our in situ XAS analyses, the

fits are carried out twice over in order to arrive at a meaningful variation in the NPt-Pt as a

function of the electrode potential. It is well-known that coordination numbers, in this

case NPt-Pt, are positively correlated with the Debye-Waller factors (σ2) for EXAFS fits.

Therefore, a fixed σ2 value is used for all fits in order to arrive at a more accurate

variation in NPt-Pt as a function of electrode potential. An averaged σ2 is obtained by

fitting the entire data set once allowing all four fitting parameters to vary. The final set of

NPt-Pt values are obtained from a second round of fits using a value of σ2.fixed at the

averaged value obtained previously.

277

6.4 XANES analysis

As the ∆µ-XANES method is a surface-sensitive method, it is advantageous to have a

significant dispersion of the Pt catalyst in order to obtain data of sufficiently good

quality. Thus, even though features in CV’s may be clearer for data collected on

unsupported Pt catalysts (Pt black), 82 carbon-supported Pt nanoparticles were preferred

to Pt black for the XAS experiments.

All XAS spectra were first normalized in the 25-170 eV range. The Pt foil spectra

were then calibrated to the Pt L3 edge energy of 11,564 eV and were aligned to one

reference foil spectrum. Any energy shifts associated with this alignment are

automatically applied to the respective sample spectra in ATHENA. The experimental

∆µ-XANES spectra (also referred to as difference spectra in parts of the paper) were

calculated as ∆µ(Vi) = µ(Vi) - µ(0.54V)ref where µ(Vi) is the absorption spectrum at the

potential of interest and µ(0.54V)ref is the reference spectrum collected at 0.54 V (double-

layer region) when the surface can be expected to be free of any adsorbate(s). Note that

both the μ(Vi) and μ(0.54V) spectra are taken either without PVP or with PVP (see

Figure 6.2). Thus in both cases the Δμ difference should reflect the adsorption due to any

adsorbate(s) (e.g. H, OH, O) on the Pt surface, but also in the latter case possibly from

changes in the nature of the PVP-Pt interaction. These spectra were compared to

theoretical difference spectra, ∆µth = µ(Pt6-adsorbate) - µ(Pt6), obtained using the FEFF

8.0 code 83 where the adsorbate may be H or O in a particular binding site and Pt6 is a six-

atom platinum cluster. Further details on the modeling and calculations are given in the

following section.

278

6.4.1 FEFF 8.0 modeling and theoretical calculations

The theoretical near-edge absorption spectra (XANES) for both the clean Pt cluster as

well as the clusters with bonded PVP in various orientations were calculated using the

FEFF 8.0 code. Pt-Pt bonds in the cluster were maintained at 2.77 Å for all calculations,

the established interatomic bond distance in bulk platinum (fcc).

The theoretical spectra for the Pt/C catalyst was a 6-atom cluster (sometimes called

the Janin cluster) and is the smallest cluster that contains atop, bridged, fcc and hcp

adsorption sites. 84 In order to calculate the theoretical spectra for the PVP-Pt/C catalyst,

a modified PVP molecule bonded to the Pt cluster was used. Essentially, the monomer

consisting of only the pyrrolidone ring was ‘bonded’ to the Pt cluster via the carbonyl

group. More specifically, the ‘PVP’ molecule was placed over the Pt6 cluster such that

the carbonyl group occupied, in each case, either an atop, bridged or 3-fold-like position

on the Pt6 cluster. ArgusLab™, a computational modeling and visualization software was

used to run a molecular modeling (MM) simulation with a unified force field (UFF)

geometry optimization procedure. A convergence of at least 0.1 kcal/mol/Angstrom for

the gradient was arrived at by allowing a maximum of 500 iterations. Care was taken to

ensure that the Pt-Pt bond distance of 2.77 Å and the overall geometry of the Pt6 cluster

was unchanged by suitably constraining their parameters during the geometry

optimization calculation. The final coordinates of the Pt cluster along with that of the

adsorbed ‘PVP’ molecule were imported and used for the FEFF calculations. Theoretical

difference spectra were finally calculated by taking the difference between the spectra

obtained for the PVP-Pt/C calculation and the ‘clean’ Pt6 cluster used to represent the

Pt/C catalyst.

279

6.5 Results

Shown in Figure 6.3 are results of water activation in both 0.1M HClO4 and 0.5M

H2SO4. Both sets of data indicate that there is an increase in the capacitive current in the

0.40-0.60 V region reflecting the absorption of the partially insulating PVP on the Pt

surface. Taking this higher capacitive current into account, it is also clear that a decrease

in the electrochemically active surface area (ECSA) exist on the PVP-Pt/C catalysts when

compared to Pt/C catalysts, as evidenced by the smaller Hupd area and the corresponding

loss of features typically assigned to underpotential deposited (upd) H on specific sites in

this region (0.05 < V < 0.35 V vs. RHE). Further, while the upd H on the (100) sites and

corners and edges (attributed to the 0.3 V feature anodically) is significantly reduced in

the PVP-Pt/C catalysts when compared to that seen on Pt/C, the adsorption onto the (111)

planes (attributed to the 0.8 V feature anodically) is clearly shifted to lower potentials and

results in much larger currents.

Figure 6.4 shows CO stripping data for the two catalysts in both electrolytes. The

effect of PVP on the Pt/C electrocatalysts is reflected in marked changes in the onset of

CO oxidation and the peak potential. The onset of CO oxidation is shifted to slightly

higher potentials in both electrolytes, albeit to a much higher potential in sulfuric acid

when compared to data collected in perchloric acid. Methanol oxidation catalyzed by the

two catalysts in both electrolytes was also studied (Figure 6.5). A remarkable

enhancement in the methanol oxidation was observed in both media. However, there

appeared to be no change in the actual onset of the oxidation, which is interesting given

our earlier observation that the onset of CO oxidation on the PVP-Pt/C catalysts is

delayed when compared to that on Pt/C. Finally, chronoamperometric (CA)

280

Figure 6.3 Comparison of the CV curves for water activation on Pt/C and PVP-Pt/C taken in a) 0.5M sulfuric acid and b) 0.1M perchloric acid. The data were collected at a scan rate of 50 mV/s.

a.

(111) (100)/edges

b.

281

Figure 6.4 Comparison of CO stripping curves for Pt/C and PVP-Pt/C taken in a) 0.5M sulfuric acid and b) 0.1M perchloric acid

a.

b.

282

Figure 6.5 Comparison of the methanol oxidation data for Pt/C and PVP-Pt/C with 0.5M methanol in a) 0.5M sulfuric acid and b) 0.1M perchloric acid.

a.

b.

283

measurements were carried out at 0.39V (vs.RHE) on both catalysts, in both

electrolytes, the results of which are shown in Figure 6.6. In both sulfuric acid and

perchloric acid, a better transient response was obtained for the PVP-capped Pt/C

catalysts, suggesting a better tolerance to CO poisoning when compared to commercial

Pt/C, similar to that found previously for the PVP-Pt black.

The ∆µ-XANES curves for data collected as a function of electrode potential on the

plain Pt/C catalyst sample are shown in Figure 6.7. All difference spectra shown here

were obtained as described earlier (see experimental section). For sake of clarity, the

entire set of spectra between 0 and 1400 mV have been separated into two parts. Figure

6.7a contains the ∆µ-XANES spectra at lower potentials (< 400 mV) while Figure 6.7b

contains spectra at more positive potentials all the way up to 1400 mV. In Figure 6.7a,

the difference spectra are characteristic of 3-fold adsorbed upd H on a Pt surface and are

identical to that seen in all our previous in situ XAS studies for data collected at the Pt L3

edge.70, 71 Further, as expected, the amount of upd H decreases with increase in potentials.

As the potential is increased to the double-layer region and beyond, the Pt surface begins

to adsorb O(H) species from the electrolyte due to water activation (Figure 6.7b). This is

seen as a small positive feature centered around 0 eV (relative to the edge energy). This

feature increases in both intensity and energy i.e. as the surface adsorbs more O(H)

species with increasing potential, and the adsorbed species move from lower-coordinated

OH in atop sites to the more highly-coordinated O in bridged and 3-fold sites with

resultant shift upward in energy.71 These results for Pt/C have been reproduced many

times in our previous work.69-72, 74 It is interesting to note that a broad feature between -30

eV and 0 eV is observed, and decreases with an increase in potential. While

284

Figure 6.6 Chronoamperometry data for Pt/C and PVP-Pt/C in 0.5M methanol and a) 0.5M sulfuric acid and b) 0.1M perchloric acid.

a.

b.

285

Figure 6.7a Experimental ∆µ = µ (Vi) - µ (0.54V) curves for Pt/C at potentials below 0.40 V (vs. RHE) showing adsorbed upd hydrogen.

-30 -20 -10 0 10 20 30 40 50 60 70-0.20

-0.15

-0.10

-0.05

0.00

0.05

0.10

0.06 V0.40 V

Δμ

Erel, eV

a.

286

Figure 6.7b Comparison of experimental ∆µ = µ (i) – µ (0.54) at the indicated potentials, and theoretical ∆µ = µ (ad/Pt6) – µ (Pt6) in the indicated adsorption sites. Note the gradual shift in energy of the experimental peak around 0.5 V indicating the shift from atop OH to n-fold O.

b.

Δμ

Erel, eV

Atop OH 2-fold O 3-fold O

Theoretical ∆µ-XANES

3- fold Atop Bridged

-30 -20 -10 0 10 20 30 40 50 60 70-0.04

-0.01

0.02

0.05

0.08

0.11

0.140.60 V0.70 V0.80 V0.90 V1.00 V1.20 V

287

Figure 6.8 Experimental delmu XANES curves for a) Hupd region (below 0.40 V) and b) oxidation region (above 0.60 V) for PVP/Pt/C. Note the absence of a shift in peak energy for data below 1.1 V.

-30 -20 -10 0 10 20 30 40 50 60 70-0.06

-0.04

-0.02

0.00

0.02

0.04

0.066 V0.20 V0.30 V0.40 V

-30 -20 -10 0 10 20 30 40 50 60 70-0.05

0.00

0.05

0.10

0.15

0.200.60 V0.70 V0.80 V0.90 V1.00 V1.10 V1.30 V1.40 V

Δμ

Δμ

Erel, eV

a.

b.

288

the origin of this feature is not clear, it has been seen in some other recent studies

within our research group.85 We can only speculate that it may be due to scattering from a

loosely-bound water layer above the Pt surface.

The ∆µ-XANES spectra for the PVP-capped Pt particles are shown in Figure 6.8. The

most striking feature of this set of curves is the absence of a shift in the energy of the

‘OH’ like peak all the way up to 1.00 V in contrast to that for Pt/C. However, after 1.00

V, a clear shift in the location of the peaks to higher energies is observed, reflecting the

typical n-fold O/Pt lineshape.

Finally, Pt L3 EXAFS results are shown in Figure 6.9 at 0.70 V to illustrate the

quality of a model single path Pt-Pt fit to the EXAFS data on the PVP-Pt/C catalyst.

Table 6.1 gives a summary of all of the fit parameters using a single Pt-Pt path. Here the

four parameters for the Pt-Pt single scattering path are characterized by the Pt-Pt

coordination number (NPt-Pt), the Pt-Pt bond length (RPt-Pt), the Debye Waller factor (σ2),

which takes into account both the dynamic and static disorder, and the inner potential

(Eo), which accounts for small shifts in the edge due to adsorbates, charging and other

factors. In these fits, the Debye-Waller factor was fixed at 0.0055 Å-2, since the strong

correlation between NPt-Pt and σ2 makes the values of NPt-Pt highly uncertain. The fixed

value was determined by taking the average of the values obtained when this parameter

was allowed to vary. (See EXAFS analysis section for more details)

Along with results of FEFF 8.0 modeling to interpret the experimental ∆µ-XANES

difference spectra, the electrochemical data will be discussed in greater detail in the

following section. Our objective is to understand the PVP interaction and changes in this

interaction with applied potential, and thereby understand the role of PVP in significantly

289

Figure 6.9 A model single-shell EXAFS fit to the PVP-Pt/C data collected at 0.70 V vs.

RHE. The data was collected at the Pt L3 edge.

0 1 2 3 4 5 6-0.10.10.30.50.70.91.11.31.5

FT (χ

2 (k))

PVP-Pt/C, 0.70 Vfit

R, Å

290

Table 6.1 Summary of EXAFS results for the Pt/C and PVP-Pt/C catalyst samples

Pt/Ca PVP-Pt/Ca

V,volts NPt-Ptb RPt-Pt

c(Ǻ)

Eo

d(eV) NPt-Ptb RPt-Pt

c(Ǻ) Eod(eV)

0.54 8.04 2.75 7.96 9.60 2.77 9.47 0.60 8.40 2.76 7.38 9.87 2.77 9.73 0.70 8.67 2.76 7.43 9.20 2.77 8.67 0.80 8.96 2.76 7.61 9.11 2.77 7.90 0.90 9.02 2.76 7.20 9.11 2.77 7.69 1.00 8.39 2.76 6.76 8.74 2.77 7.47 1.10 7.71 2.75 5.79 8.42 2.76 6.98

aDebye-Waller factor (σ2) fixed at 0.0055 Å-2 for all fits bAbsolute estimated uncertainty about 20% but change in values more accurate cEstimated uncertainty about ±0.02Ǻ dEstmated uncertainty about ±1 eV.

291

enhancing the methanol oxidation activity of Pt nanoparticles and/or their tolerance to

CO poisoning.

6.6 Discussion

6.6.1 How does the PVP increase the Δμ-XANES signal above 0.5 V?

In order to compare the differences in the adsorbate coverage on the plain Pt/C and

PVP-Pt/C, the magnitudes of the positive features in the Δμ-XANES spectra are plotted

and shown in Figure 6.10. All assignments of the adsorbates in various potential regions

(H3f, OH, O etc.) are determined by comparing previous theoretical difference spectra 70,

71 calculated using the FEFF 8.0 code to the experimental Δμ spectra (as illustrated in

Figure 6.7). The most noticeable difference is the increased intensity of the O(H)/Pt-like

signature in the region 0.50-0.90 V on the PVP-Pt/C catalysts when compared with the

Pt/C.

The reasons for the apparent increased OH coverage on PVP-Pt/C are likely due to

one or more of the following mechanisms –

a. Change in PVP binding: It is currently believed that PVP bonds via a >C = O-

Pt interaction between the free carbonyl of the pyrrolidone ring and the Pt surface.

Borodko et. al. 65, 66 showed using Raman/FTIR that on reduction of the catalyst in H2,

the interaction is weak, but after oxidation, becomes stronger with charge transfer (see

Figure 6.1). In our data, the Δµ-XANES curves could be reflecting this changed >C = O-

Pt interaction directly, since the O is effectively in an atop site in the stronger bonded

case, and hence should appear to be ‘OH-like’ in the Δμ spectra (we discuss this further

in the next section). Perhaps at lower potentials the interaction is weaker in nature or

292

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

0.000.020.040.060.080.100.120.140.16 PtBlk

Ptpvp

Pt/PVP-C

Pt/C

Hat H3f OH O Ox

PVPblocking sites

Atop O(H) shape yet

Δµm

agni

tude

Potential (V vs. RHE)

Figure 6.10 ∆µ-XANES magnitudes for positive features of lineshapes seen in Figures 6.7 and 6.8. Note the marked increase in an atop OH-like feature between 0.50 and 1.00 V for PVP/Pt/C.

293

more disordered so that the PVP-Pt interaction is much less visible in the Δμ-XANES.

Then the Δμ difference spectrum probably reflects the nature of the more strongly bonded

PVP. The larger Δμ magnitude is observed only above 0.50V when the platinum surface

is charged suggesting that the changed PVP-Pt interaction might involve charge transfer.

There is precedence for adsorbates that are invisible to the Δμ. This would be similar

to the bisulfate anion on the surface, which although present on the Pt surface in sulfuric

acid electrolyte, is not visible in the Δμ-XANES because it is not adsorbed in registry or

in specific sites on the surface. After addition of OH above 0.60 V, the bisulfate is forced

to bind in specific sites in between the OH groups so it becomes visible in the Δμ as O-

like groups in the 3-fold sites albeit with longer Pt-O bondlength.78 Even H at low

coverage, because it is so mobile on the surface, is invisible in the Δμ, until it is forced to

localize at higher coverage. 70 Thus, it is interesting that at lower potentials (<0.40V), the

observed Δμ spectra appear to reflect changing amounts of upd H only (see Figure 6.8a).

The Δμ lineshapes are typical of a 3-fold upd H/Pt 70 and decrease in intensity as the

potential is increased from 66 mV up to around 400 mV, and only a blocking effect is

apparent.

Thus the Δμ data suggest a potential-dependent interaction of PVP with the Pt

surface: at lower potentials, a more weakly-bonded form of PVP exists, which we will

term PVP neutral (PVPN), whereas at higher potentials, a more strongly-bonded form of

PVP exists, termed PVP charge transfer (PVPCT) with a stronger interaction between the

charged Pt surface and the lone pairs of the carbonyl groups of the pyrrolidone rings.

These are consistent with those indicated previously by Borodko et al. 65, 66 and illustrated

294

in Figure 6.1. We will provide much further evidence for this mechanism below both

from the literature and our own electrochemical results.

b. A ligand effect: It is also possible that the PVP simply exerts a ligand (or

electronic) effect on the Pt, so that the increased Δμ-XANES indeed reflects increased

adsorption of OH between 0.6-1.0 V. This is consistent with methanol oxidation data

where increased reactivity is observed for Pt-PVP/C when compared with just Pt/C.

Adsorbed CO species are oxidized to CO2 usually through reaction with adsorbed OH

species in the normal bifunctional mechanism. As such, an increased amount of OH on

the surface would clearly be advantageous and would enhance the methanol oxidation

reaction. However, it is unlikely that this is happening for two reasons: A ligand effect

should alter the onset for all 3 species, namely OH, O, and H adsorption. The results here

show no change in the onset for H and O, in fact not even a change in the onset for OH

exists, and only a much larger magnitude for the assigned OH signature is seen.

Secondly, it is well-established that PVP adsorbs onto surface sites of the Pt

nanoparticles. If this is true, then it can only reduce the number of catalytically active

surface sites, leaving us with no reason to assume that the amount of adsorbed OH

species can increase (relative to that seen on Pt/C) in the presence of bonded PVP. An

increased number of catalytically active sites despite significant site-blocking by PVP can

only occur if the PVP-Pt/C catalysts were much more dispersed and possessed a much

smaller particle size than the Pt/C catalyst. However, a cursory look at the NPt-Pt values

for the two catalysts obtained through EXAFS fits (See Table 6.1) seem to indicate that

both the Pt/C and the PVP-Pt/C catalysts have nearly the same particle size and rules out

this possibility entirely.

295

c. A bifunctional-like mechanism: In this mechanism, the carbonyl group on the

PVP could directly assist in breaking the H2O bond because the H+ (proton) can add to

one of the lone pairs on the nearby O atom in the carbonyl instead of going into solution.

In this mechanism the PVP would play the role of the Ru in the typical commercial PtRu

catalysts. Thus, more OH could possibly adsorb in the 0.60-1.0 V region. While the onset

potential for O and H would not be expected to change, it is surprising that the onset

potential for even OH adsorption does not appear to change, showing differences only in

the amount of OH on the surface. Thus we are inclined to rule out this mechanism as

well.

d. A Geometric effect: Kweskin et al. studied CO adsorption and oxidation on

PVP-stabilized, cubic Pt nanoparticles (ca. 9.4 nm) catalysts using IR-Visible SFG and

estimate that close to 65 % of the surface may be covered with PVP, water and other

organics.108 As such, if a significant number of PVP chains are bonded to the Pt surface,

the bonded carbonyls may be sufficiently large in number to actually force the incoming

adsorbate molecules into otherwise “less desirable” unoccupied sites, or onto sites that

are occupied normally only under more full coverage. We term such an effect a

geometric effect, as it actually modifies the adsorption geometry of the CO and OH and

then could lead to altered reaction rates on the Pt surface. How would such an adsorption

behavior affect the CO stretching frequency observed by either SEIRAS or SFG? We

believe that it would not be unlike that of CO adsorbed onto a surface containing

significant amounts of oxide on the surface, wherein the CO stretching frequency would

be red-shifted due to a weakening of the COads/Pt due to strong lateral interactions

between the CO and adsorbed >C=O groups from the PVP.

296

There is some evidence in the literature that lends further support to a geometric effect

due to PVP. It has been reported that PVP-stabilized Pt and Rh catalysts alter the rates of

reactions of small adsorbed molecules such as CO, OH and O. 27, 62, 86 Somorjai and co

workers found that CO adsorbs in bridge sites on a PVP-altered Rh surface, which may

actually reduce the activation barrier for the CO oxidation reaction by providing a

geometrically advantageous orientation for the adsorbed CO and OH species. According

to the authors, some evidence for the possibility of such an enhancement is also

supported by theoretical calculations. 87 In a study on the effect of PVP on the rates of

ethylene hydrogenation and CO oxidation on monodisperse Pt nanoparticles, it was

demonstrated that PVP is capable of altering the reaction rates through adsorption to

surface sites. However, it was unable to change the intrinsic activity of active Pt sites, a

conclusion which was supported by surface-normalized reaction rate data. 88 Our Δμ-

XANES results indicate an increased atop OH contribution that remains until higher

potentials. This could mean that the PVP-Pt interaction forces the OH to remain in the

atop sites to higher potentials, providing adsorbed OH species to react with CO even at

much higher potentials. Such a mechanism would also lead to increased methanol

oxidation activity as it has been established that adsorbed OH is much more reactive for

CO oxidation than O in an n-fold site. However, we are inclined to believe that this

mechanism is not the primary reason for the enhanced Δμ-XANES magnitudes seen for

the PVP-Pt/C catalyst in Figure 6.10.

e) PVP decomposition product: There are reports in the literature indicating that if

PVP degrades or decomposes on a Pt surface, both the oxygen from the free carbonyl

group as well as the nitrogen from the pyrrolidone ring may be strongly bonded to the Pt

297

surface. 66 The enhanced Δμ signature might be reflecting the additional N/Pt bonding.

However, FEFF8 calculations below along with CV results will help to eliminate this

possibility.

6.6.2 The PVP-Pt binding site via FEFF calculations

Theoretical calculations using the FEFF 8.0 code were carried out to interpret the

experimental Δμ spectra and determine the possible binding site of the PVP to the Pt

surface, assuming the Δμ spectra indeed reflect the PVPCT interaction. In order to

calculate the theoretical spectra for PVPCT/Pt, a truncated PVP molecule bonded to the Pt

cluster was used i.e., the monomeric unit of the polymer consisting of only the

pyrrolidone ring bonded to the Pt cluster via the free carbonyl group and with no

hydrocarbon backbone. Shown in Figure 6.11 are the theoretical ∆µ-XANES spectra for

PVP bonded to the Pt surface in an atop site (singly coordinated) and a 3-fold site (triply

coordinated). Although the positive features of the difference spectra differ by as little as

2-3 eV, the negative feature seen in the 3-fold signature easily distinguishes one from the

other. Comparison of the theoretical signatures for atop PVPCT (Figure 6.11) and OH

(Figure 6.7b) on Pt are indeed similar, so this makes it somewhat difficult to distinguish

them. Nevertheless, we believe the experimental ∆µ-XANES spectra shown in Fig. 8b is

dominated by the PVP and therefore indicates that the PVPCT is bonded predominantly in

the atop sites between 0.5 and 0.9 V. At higher potentials, the appearance of a negative

feature and shifts of the positive feature in Δμ to higher energies suggests that the Δμ is

now dominated by the O/Pt also present at these potentials, and that this is O in a more

highly-coordinated 3-fold site

298

Figure 6.11 Theoretical FEFF 8.0 calculations for atop (1-fold) and 3-fold bonded PVP-Pt/C.

3-fold

11560 11580 11600 11620 Photon energy (eV)

Atop

0

0.2

0.1

299

To help eliminate the PVP decomposition product interpretation mentioned above, a

FEFF calculation for PVP bonded to the Pt surface through both the O and the N atoms

was carried out to see if there were any similarities to the experimental ∆µ-XANES

curves. Interestingly, the lineshape is very similar to that of PVPCT bonded to the Pt

surface just via the oxygen with the positive features shifted by only 1-2 eV. However, if

the PVP indeed degrades on the surface, specific redox features should have been

observed in the CVs (Figures 6.3 and 6.5) and since these are absent, we reject this

possibility.

6.6.3 Further evidence for a change of PVP interaction from EXAFS results.

Figure 6.12 shows a plot of the bulk Pt-Pt coordination number (NPt-Pt) values from

Table 6.1 as obtained from EXAFS analysis along with a plot of the Δμ magnitudes. We

consider the results for Pt/C first. When atop OH comes on, this raises the NPt-Pt slightly

because atop OH likely makes the clusters a bit more spherical. This has the effect of

increasing the Pt-Pt coordination. Evidence for such a structural change upon adsorption

in atop sites has been seen several times in our previous work. 71, 74 As the atop OH

moves over to become O in more highly-coordinated fcc sites, a change which is

reflected in the Δμ signature, a drop in the coordination number occurs because the O in

these sites “interferes” a bit with the Pt-Pt scattering. The continued decrease in

coordination past 1.1 V can also be attributed to a place-exchange mechanism that occurs

above this potential wherein some fraction of adsorbed oxygen atoms go subsurface 89

and actually goes directly between Pt atoms. This no doubt reduces the NPt-Pt even

further.

300

In the case of the PVP-capped Pt/C nanoparticles, the coordination number starts out

larger than that for Pt/C, apparently because the PVPN interaction again causes the Pt

particles to be more spherical. The PVPN bind onto atop Pt sites preferably on edges and

100 sites as inferred from the CV to be discussed further below, and this should increase

the NPt-Pt, just as H does in this potential range. The NPt-Pt then begins to decrease slightly

exactly where the Δμ magnitude begins to increase. This perhaps is a bit surprising

because the Δμ signatures reflecting the PVPCT do not indicate that the PVPCT moves

over into the fcc sites, but stays in the atop sites at the edge. So why does the change

from PVPN to PVPCT cause this reduction in NPt-Pt. Clearly the stronger PVPCT-Pt

interaction somehow does not have as great an effect on the morphology of the Pt cluster

as the weaker PVPN-Pt interaction because between 0.8-0.9 V the NPt-Pt for both catalysts

are similar. Thus we can conclude that OH (present on the Pt/C) has about the same

effect on the cluster shape as the PVPCT does in the PVP/Pt case. The PVPN has about the

same effect or even greater than H does at lower potentials and in general we have found

previously that H appears to have a larger effect than OH. 71, 74

Above 0.90 V, NPt-Pt again decreases with adsorption of OH(O), but note that the

decrease is less dramatic compared with Pt/C. This is completely consistent with our

hypothesis that a significant number of surface sites are blocked-off by the PVP, which

would reduce the amount of surface oxide that can be formed on the nanoparticles and

probably result in a lesser amount of oxide going subsurface. This aspect of stabilized-

nanoparticles, whereby the amount of surface oxide is reduced despite possessing better

catalytic activity for certain reactions appears to be a promising method of hindering the

aging process in nanoparticle catalysts.

301

6.6.4 Evidence for two forms of bonded PVP from other spectroscopic

techniques

Table 6.2 summarizes other spectroscopic studies that indicate the bonding of

PVP to metals can change with conditions. As briefly discussed earlier in the paper,

Somorjai and co-workers find that PVP interacts more strongly with Rh than Pt and also

that this interaction depends strongly on the oxidation state of the metal; i.e., the polymer

‘breathes’ around the nanoparticle as it is oxidized and reduced 65, 66. A particle size-

dependent PVP-Pt interaction has also been observed in an FTIR and XPS study by

Bonet et al. 63 and Qiu et al.64. Direct evidence for a particle size-effect on the PVP-Au

interaction on Au clusters as studied using FTIR, XPS and XAS has also been reported 90

although the sizes of the Au clusters studied were not dissimilar enough to be completely

convincing. It all these studies, it was shown that PVP binds weakly to larger, more

metallic clusters and more strongly to the surface of smaller clusters or partially oxidized

clusters. This effect is probably due to the decreased surface charge present in smaller

nanoparticles or partially oxidized ones, when compared with extended metal surfaces

present in larger nanoparticles, and thus charge transfer occurs between the electron rich

302

Table 6.2 Summary of literature showing two different types of PVP binding

PVPN/metals PVPCT/metals Technique(s) utilized

Large Au clusters (>2.5 nm)

Small Au clusters (<2 nm)

XPS, XANES and reactivity 90

Large Pt clusters (>24nm)

Small Pt clusters (<7nm) XPS study 64

Reduced Pt clusters (ca. 9 nm particles – metallic)

Oxidized Pt clusters (ca. 9 nm particles – metallic)

Deep UV, FTIR, Raman, SERS 66

3 nm Pt clusters Low Potential (<0.5 V vs. RHE) Non- site specific binding

3 nm Pt clusters High Potential (>0.7 V vs. RHE) Site-specific binding

This study

303

PVP and the metal surface.91 Thus the change in PVP-Pt interaction from PVPN to PVPCT

seen above 0.5 V RHE in our Δμ data is totally consistent with these previous

observations.

6.6.5 Does the PVP-Pt interaction affect the metal nanoparticle?

6.6.5.1 Previously reported results:

While some findings suggest that PVP in some way alters the metal substrate,

others have found no evidence of such an effect. In an early study on the possible

influence of PVP on the electronic properties of Pt, it was shown that the white-line

intensity at the Pt L3 edge (1s-5p) is reduced for PVP-stabilized colloids when compared

to pure Pt nanoparticles. This finding suggested that there is a donation of charge from

the PVP to the Pt (as we suggest in the PVPCT case), thus leading to a decrease in the

unoccupied density of states for the Pt. They also found a shorter Pt-Pt bond distance and

a larger Debye-Waller factor in the PVP stabilized colloids when compared with that of

bulk Pt metal. 92 Busser et al. studied changes in the 13C NMR spectra collected in an

experiment wherein the Rh salt concentration was steadily increased relative to the PVP

concentration. The peaks changed significantly with the increase in Rh salt concentration

strongly suggesting that PVP interacts strongly with Rh metal ions present in solution. 93

Contrary to the findings in these studies, other investigations on PVP-capped

nanoparticles suggested that the metal surfaces were unaffected by the polymer. The

results of an EC-NMR study of 195Pt in Pt/Pd nanoparticles by Tong et. al. suggested that

the polymer had little or no effect on the electronic properties of the nanoparticles. 94 De

Caro et al., using infrared spectroscopy, showed that PVP-Pt nanoparticles possessed CO

304

adsorption properties very similar to that of clean Pt surfaces in air, suggesting that while

PVP could block active catalytic sites, it likely did not affect the properties of the Pt

surface or of the CO adsorbed onto it. 95 In a SERS/Raman study of PdCu bimetallic

nanoparticles capped with PVP, Toshima et al. found no evidence for a detectable

interaction of PVP with the metal surface. 52 Dassenoy et. al. carried out a study to

determine the packing (f.c.c. vs. h.c.p) of PtRu alloy nanoparticles by systematically

varying their composition. 96 They too found no evidence of an interaction between PVP

and either Pt or Ru.

Care must be taken in comparing the influence of PVP across various metals as the

interaction may vary widely based on the nature of the metal. Thus, it is apparent that

there is still no consensus or model for a complete representation of the interaction

between PVP and the metal substrate. Nevertheless, many papers indicate that PVP-

stabilized metals and alloys have enhanced catalytic properties. PVP has been reported to

enhance the oxidation of CO to CO2, 97, 98 the catalysis of hydrogenation and

dehydrogenation reactions 27, 88 and even the methanol oxidation properties of Pt, as seen

in this study and work by some others.59, 99 However, a few other independent studies

indicate no such enhancement of the methanol oxidation properties on PVP-capped Pt

nanoparticles. 32, 82, 100

6.6.5.2 Evidence from current work:

Electrochemical data for the Pt/C and PVP-Pt/C catalysts were collected in

both, 0.1M HClO4 and 0.5M H2SO4. Note that the oxidation currents for water activation

as well as CO stripping, shown in Figures 6.3 and 6.4, are higher in HClO4 than in

H2SO4. The current densities for methanol oxidation in perchloric acid are also higher

305

than that in sulfuric acid (Figures 6.5 and 6.6). This is generally attributed to blocking of

Pt sites by the strongly adsorbing bisulfate ions, which adsorb in the potential range

between 0.4-0.90. The perchlorate anions adsorb only at very high potentials (V>0.90 V)

if at all. In contrast, enhanced methanol oxidation activity (PVP/Pt/C relative to Pt/C) is

seen in both electrolytes indicating unequivocally that PVP enhances the electrocatalytic

activity of Pt nanoparticles. Since in situ XAS data were only collected in perchloric acid

medium, our discussion below will largely pertain to the electrochemical data collected in

perchloric acid.

6.6.5.2.1 PVPN at lower potentials

All of the electrochemical data viz. water activation, CO stripping and methanol

oxidation, display an increased double-layer capacitive current in the 0.40-0.60 V region,

reflecting the presence of a less-conductive material on the catalyst surface. The more

weakly-bonded PVPN as expected, affects the Hupd region (0<V<0.35 V) as indicated by

the clear loss of Hupd features as well as the reduction in total area under the CV curve in

the upd H region. This of course can be attributed to the presence of weakly-adsorbing

PVPN on the (100) sites and corners and edges at lower potentials. The loss in Hupd

current is consistent with the experimental Δμ-XANES magnitudes for adsorbed

hydrogen (see Figure 6.10), which show a reduction of less than 40% and this only below

0.2 V.

In Figure 6.5, which shows the methanol oxidation curves in perchloric acid, the Hupd

region for PVP-Pt/C is again smaller compared to that seen on the clean Pt/C catalyst.

Quite surprisingly, the converse is observed in case of methanol oxidation in sulfuric

acid, in which case the area corresponding to the Hupd regions for PVP-Pt/C is

306

substantially larger than that seen for the plain Pt/C catalyst. We will revisit this

observation and address the possible reasons for this later on in the discussion.

In general, we note that the amount of upd H site blocking by the PVP is not

unexpected. If indeed the PVPN blocks primarily the corner and edge sites, along with

some of the (100) planes (Figure 6.3), these sites constitute only a small fraction of the

total surface sites in a typical Pt nanoparticle that is around 3-4 nm in size (about 25-30%

assuming cubooctahedral particles101). Further, the Δμ-XANES curves indicate that this

weak, neutral-like interaction exists up to around 0.60 V. The changing nature of the

PVP-Pt interaction at around 0.65 V, most clearly visible in the Δμ-XANES curves, is not

directly reflected in any of the CVs, rather the charge transfer may be occurring over the

entire potential region between 0.4 and 0.8 V, causing in part the capacitive current

visible in the PVP/Pt samples.

6.6.5.2.2 PVPCT at higher potentials

As methanol oxidation occurs primarily above 0.60 V, we will discuss this

potential region in greater detail. At potentials above 0.60 V, as the experimental Δμ-

XANES curves suggest, an enhanced PVP-Pt interaction is observed and we have

attributed it to a change in the nature of adsorbed PVP from a more neutral-like

interaction (PVPN) to a more strongly-bonded form of PVP that undergoes some form of

charge-transfer with the Pt surface (PVPCT). This should lead to more significant site

blocking on the Pt surface, reduce the amount of adsorbed OH and O and therefore

reduce the water activation currents. This site blocking by PVPCT is confirmed by the

reduced NPtPt coordination drop in Figure 6.12 even though this is not obvious from the

Δμ magnitudes in Figure 6.10.

307

Figure 6.12 EXAFS fit results showing changes in Pt-Pt coordination number,NPt-Pt, as a function of applied potential. Also shown (on right) for comparison are the delmu magnitudes originally shown in Figure 6.10.

0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2

Potential, V vs. SHE

6

7

8

9

10

NPt

-Pt

NPt-Pt Pt/CNPt-Pt PVP-Pt/C

Δµ Mag.

308

This hypothesis, based on the experimental XAS results, is completely consistent

with the water activation curves shown in Figure 6.3. While the amount of oxide on the

Pt surface is not clear on the anodic scan, some sense of the extent of oxidation is

obtained by closely looking at the oxygen reduction region on the cathodic scan (0.70 <

V < 0.90). Figure 6.3b clearly shows that the ORR current is significantly reduced in case

of the PVP-Pt/C catalyst when compared to the pure Pt/C catalyst. As the voltammetric

curves were collected under identical conditions, the reduced ORR currents can be

attributed to a reduced amount of surface oxide on the PVP-Pt/C catalyst. Previously

reported experiments lend further support to these observations as the suppressed ORR

features have also been seen when PVP is mixed-in with Pt black. 82

It is now fairly well-established that OH adsorption on Pt first occurs on atop sites at

the nanoparticle edges and perhaps the 100 planes at lower potentials and then becomes

more highly coordinated to form n-fold O on the (111) planes at higher potentials. 71

Some previously reported electrochemical evidence exists that supports our findings that

PVPCT primarily blocks edge/100 sites. 82 While water activation results on pure Pt black

in 0.1M HClO4 shows clear signs of oxidation on edge/100 sites through a marked

oxidation current already at around 0.65 V, there is no such feature during water

activation on the PVP-Pt catalyst samples. Note that this is exactly the region where we

notice a significant >C=O-Pt interaction in the experimental ∆µ-XANES curves (see

Figure 6.10) and suggests that a large number of these sites are blocked by PVPCT.

CO stripping results in both electrolytes are shown in Figure 6.4. The presence of

adsorbing anions from the electrolytes are clearly seen as the voltammetric curve for the

Pt/C catalyst in sulfuric acid is quite different from that in perchloric acid. Although the

309

onset potential is only marginally different in both electrolytes, the peak current for CO

oxidation is higher in perchloric acid and can be explained as being due to the difference

in adsorption strengths between HClO4- amd HSO4

- anions. Most important, we observe

that on PVP/Pt the onset of CO oxidation is delayed (shifted to higher potentials) in both

electrolytes, although less so in case of perchloric acid. This would suggest that CO

oxidation is thermodynamically less favorable in the presence of adsorbed PVP. If indeed

the PVPCT blocks a significant number of edge/100 sites, then we should have a smaller

amount of adsorbed OH, and further the reaction between OH and CO on the Pt surface

can be expected to be hindered due to severely attenuated surface diffusion processes.

Thus the PVP hinders CO oxidation as expected.

Since CO oxidation appears to be hindered, then we cannot explain the enhanced

methanol oxidation activity by simply attributing it to a more facile CO oxidation (and

hence less CO poisoning) due to increased OH on the surface. These results then lend

support to our hypothesis that the increased methanol oxidation on the PVP-Pt/C catalyst

over plain Pt/C is most likely due to a weakening of the Pt-CO bond on the (111) planes

caused either by increased lateral interactions arising from the presence of a large amount

of atop bonded PVPCT (i.e. a geometric effect) and/or perhaps a ligand effect (electronic

effect).

Further evidence for a weakening of other adsorbate bonds to the (111) planes

comes from previously reported experiments on Pt black. [83] First the upd H peak on

the (111) planes (similar to that in Figure 6.3) is significantly shifted to lower potential

indicating that H/Pt is weakened. This is clearer in the Pt black data than in Figure 6.3.

Second, the effects on the bisulfate anion adsorption reflect the same behavior for

310

bisulfate on the (111) planes. In this previously reported study, two different samples

with different Pt:PVP molar ratios (Pt-PVP12 and Pt-PVP24) were investigated in 0.5M

H2SO4 and 0.1M HClO4. In sulfuric acid, the Pt black samples showed a clear feature at

0.55V which is attributed to the well-established bisulfate adsorption on Pt(111) planes.

This feature is really a result of a mild oxidative current between 0.45 and 0.65 V due to

the lifting of the adsorbed bisulfate adlayer. In the case of the PVP-Pt black samples, this

feature at 0.55 V gradually disappears as the molar ratio of PVP:Pt was increased; i.e.

this feature is suppressed by the PVP. We attribute this to a weakening of the bisulfate/Pt

interaction. Thus the PVP, whether in the PVPN or PVPCT form appear to destabilize

adsorbate binding on the (111) planes.

We believe that there is enough evidence in the aforementioned studies, many of

them which have been carried out under various conditions, to show that PVP interacts

fairly strongly with the Pt surface through the free carbonyl groups on the pyrrolidone

rings of the polymer chains. The interaction is also strong enough to actually change the

adsorption site (or at least binding energy) of important species such as H, H2SO4-, and

CO on Pt sites not even blocked by PVP.

6.6.6 Selective or preferential binding of PVP to certain low-index Pt planes

(other reports)

As mentioned earlier, PVP has been used extensively to control the shape and growth

of crystal surfaces leading to the synthesis of nanoparticles of various shapes. While the

preferential binding of PVP to certain crystallographic planes of various metals has only

been inferred indirectly, 4, 97, 102-105 we believe that the CV results and inferences drawn

311

from the discussion presented in this paper provide strong evidence that the adsorption of

PVP is indeed stronger on the (100) sites and corners & edges than on the (111) planes.

The Hupd current in both electrolytes (see Figures 6.3 and 6.4) strongly suggests that the

PVP possesses a site-preference and therefore blocks upd H on the Pt(100) sites to a

larger extent when compared to the more closely-packed Pt(111) planes. A recent study

on the synthesis, characterization and growth mechanisms of silver nanostructures also

reveals that PVP binds stronger on the (100) faces when compared to that on the (111)

faces and hints that the binding preference of PVP to the (100) facets may in fact be a

general feature of PVP adsorption onto metal surfaces, 106 as it is fairly well-established

that the open (100) faces, corners and edges are much more reactive than the more

tightly-packed and stable (111) face.

In contrast to the above findings, at least two other authors claim that PVP binds

stronger onto the (111) planes when compared to the (100) planes of platinum

nanoparticles.97, 107 Therefore, more evidence will be required before a generalization can

be made.

Regardless of which planes the PVP prefers to bind onto, it appears that PVP bonds to

the Pt surface in two different forms which we term PVPN and PVPCT, the latter

significantly weakening the C=O-Pt bond, thus greatly enhancing methanol oxidation by

reducing the extent of CO poisoning. Further, this effect is more pronounced for

increasing methanol oxidation on the (111) planes, as the Pt-CO bonding on Pt (111) is

weaker than on Pt (100) to begin with, and hence the weakening on the (111) planes

allows for CO to be replaced by upd H, just as we have found previously for PtSn and

312

PtMo (via the ligand mechanism), as opposed to PtRu which enhances water activation

(i.e. via the bifunctional mechanism).75

6.7 Conclusions

The interaction between the stabilizer/capping agent and the metal catalyst is often

inferred through changes in the vibrational modes of the stabilizing molecule. In this

study, using the novel ∆µ-XANES method, we provide the first direct evidence for the

interaction of PVP with a Pt/C catalyst in 0.1M HClO4. It was found that interaction

occurs primarily through Pt atop sites, and increases in strength between potentials of

0.60 and 1.00 V (vs. RHE). We arrive at the following conclusions:

1. It appears that there is a weaker form of neutral bonding (PVPE) and a stronger

bonded form involving charge transfer (PVPCT) from the PVP to the Pt surface.

PVPN is still strong enough to block most of the H-adsorption onto Pt(100) planes

and apparently weakens the H binding on the (111) planes.

2. PVPCT primarily bonds onto atop sites via the carbonyl group of the pyrrolidone

ring.

3. The binding appears to be preferentially on the (100) planes and corners and

edges when compared to the (111) planes as suggested by CV data in the Hupd

region.

4. PVP in effect reduces the binding of adsorbates by a direct site-blocking

mechanism, and probably weakens all adsorbate bonding on even non-blocked

sites on the surface viz. that of H, OH and CO as it raises the lateral interaction

energy between the bonded PVP and adsorbates (a geometric effect) and/or ligand

effect, leading to a destabilization and thus a weaker Pt-adsorbate bond.

313

5. The PVP interaction greatly enhances methanol oxidation by apparently reducing

the extent of CO poisoning. This effect is expected to be more pronounced on the

(111) planes, as the Pt-CO bonding on Pt (111) is weaker than on Pt (100) to

begin with, and hence the weakening on the (111) planes allows for H

replacement, just as we have found previously for PtSn and PtMo (via the ligand

mechanism), as opposed to PtRu which enhances water activation and hence CO

oxidation (i.e. via the bifunctional mechanism).

We note that in general, the interaction between important stabilizing agents and metal

nanoparticles is poorly understood and has rarely been probed directly in solution phase,

where many important catalytic reactions are known to occur. A proper understanding of

the influence of various organic capping agents on metal catalysts could open up the

possibility of organic ligand-controlled reaction selectivity within the field of

heterogeneous catalysis.

314

6.8 References 1. McCarty, J. G., Malukhin, G., Poojary, D. M., Datye, A. K. & Xu, Q. Thermal

Coarsening of Supported Palladium Combustion Catalysts. The Journal of Physical Chemistry B 109, 2387-2391 (2004).

2. Cozzoli, P. D., Kornowski, A. & Weller, H. Low-Temperature Synthesis of Soluble and Processable Organic-Capped Anatase TiO2 Nanorods. Journal of the American Chemical Society 125, 14539-14548 (2003).

3. Yoojung, K., Min Gyu, K., Yoojin, K., Youngil, L. & Jaephil, C. Effect of Capping Agents in Tin Nanoparticles on Electrochemical Cycling. Electrochemical and Solid-State Letters 9, A34-A38 (2006).

4. Xiong, Y. et al. Poly(vinyl pyrrolidone): A Dual Functional Reductant and Stabilizer for the Facile Synthesis of Noble Metal Nanoplates in Aqueous Solutions. Langmuir 22, 8563-8570 (2006).

5. Chen, J., Herricks, T., Geissler, M. & Xia, Y. Single-Crystal Nanowires of Platinum Can Be Synthesized by Controlling the Reaction Rate of a Polyol Process. Journal of the American Chemical Society 126, 10854-10855 (2004).

6. Ahmadi, T. S., Wang, Z. L., Green, T. C., Henglein, A. & El-Sayed, M. A. Shape-Controlled Synthesis of Colloidal Platinum Nanoparticles. Science 272, 1924-1925 (1996).

7. Khomutov, G. B. et al. Organized planar nanostructures from ligand-stabilized nanoclusters: a route to molecular nanoelectronic devices. Applied Surface Science 226, 149-154 (2004).

8. Tian, N., Zhou, Z.-Y. & Sun, S.-G. Platinum Metal Catalysts of High-Index Surfaces: From Single-Crystal Planes to Electrochemically Shape-Controlled Nanoparticles. The Journal of Physical Chemistry C 112, 19801-19817 (2008).

9. Sun, Y. & Xia, Y. Shape-Controlled Synthesis of Gold and Silver Nanoparticles. Science 298, 2176-2179 (2002).

10. Somorjai, G. A., Contreras, A. M., Montano, M. & Rioux, R. M. Clusters, surfaces, and catalysis. Proceedings of the National Academy of Sciences 103, 10577-10583 (2006).

11. Viswanath, B., Kundu, P., Halder, A. & Ravishankar, N. Mechanistic Aspects of Shape Selection and Symmetry Breaking during Nanostructure Growth by Wet Chemical Methodsâ€ The Journal of Physical Chemistry C 113, 16866-16883 (2009).

315

12. El-Sayed, M. A. Some Interesting Properties of Metals Confined in Time and Nanometer Space of Different Shapes. Accounts of Chemical Research 34, 257-264 (2001).

13. Yu, W. W. & Liu, H. Singular modification effects of metal cations and metal complex ions on the catalytic properties of metal colloidal nanocatalysts. Journal of Molecular Catalysis A: Chemical 243, 120-141 (2006).

14. Burda, C., Chen, X., Narayanan, R. & El-Sayed, M. A. Chemistry and Properties of Nanocrystals of Different Shapes. Chemical Reviews 105, 1025-1102 (2005).

15. Welch, C. & Compton, R. The use of nanoparticles in electroanalysis: a review. Analytical and Bioanalytical Chemistry 384, 601-619 (2006).

16. Ferrando, R., Jellinek, J. & Johnston, R. L. Nanoalloys: From Theory to Applications of Alloy Clusters and Nanoparticles. Chemical Reviews 108, 845-910 (2008).

17. Oster, G. & Immergut, E. H. Ultraviolet and Infrared Spectral Studies of Polyvinylpyrrolidone1. Journal of the American Chemical Society 76, 1393-1396 (1954).

18. Kushchevskaya, N. F. Surface Modification of Highly Dispersed Powders of Iron and Its Composites with Platinum and Silver by Polyvinylpyrrolidone. Powder Metallurgy and Metal Ceramics 40, 533-535 (2001).

19. http://www.medic8.com/medicines/Poly-vinyl-pyrrolidone.html.

20. Van Rheenen, P. R., McKelvy, M. J. & Glaunsinger, W. S. Synthesis and characterization of small platinum particles formed by the chemical reduction of chloroplatinic acid. Journal of Solid State Chemistry 67, 151-169 (1987).

21. Miller, L. E. & Hamm, F. A. Macromolecular Properties of Polyvinylpyrrolidone:Molecular Weight Distribution. The Journal of Physical Chemistry 57, 110-122 (1953).

22. Molyneux, P. & Frank, H. P. The Interaction of Polyvinylpyrrolidone with Aromatic Compounds in Aqueous Solution. Part I. Thermodynamics of the Binding Equilibria and Interaction Forces1. Journal of the American Chemical Society 83, 3169-3174 (1961).

23. Molyneux, P. & Frank, H. P. The Interaction of Polyvinylpyrrolidone with Aromatic Compounds in Aqueous Solution. Part II. The Effect of the Interaction on the Molecular Size of the Polymer. Journal of the American Chemical Society 83, 3175-3180 (1961).

24. Molyneux, P. & Frank, H. P. The Interaction of Polyvinylpyrrolidone with Aromatic Compounds in Aqueous Solution. III. A Model for the Molecular

316

Expansions Caused by Anionic Cosolutes. Journal of the American Chemical Society 86, 4753-4757 (1964).

25. Sidelkovskaya, F. P. Chemistry of N-polyvinylpyrrolidone and its Polymers [in Russian]. Nauka, Moscow (1970).

26. Zakarina, N. & Bekturov, E. Platinum Nanoparticles Stabilized by Polyvinylpyrrolidone for Hydrogenation. Chinese Journal of Catalysis 29, 1165-1168 (2008).

27. Lange, C. et al. Polymer-Induced Selectivity Enhancement in the Hydrogenation of 2-Hexyne Catalyzed by Poly(vinylpyrrolidone)-Stabilized Platinum Colloids in an Amorphous Mixed Metal Oxide Support. Langmuir 15, 5333-5338 (1999).

28. Narayanan, R. & El-Sayed, M. A. Catalysis with Transition Metal Nanoparticles in Colloidal Solution: Nanoparticle Shape Dependence and Stability. The Journal of Physical Chemistry B 109, 12663-12676 (2005).

29. Liu, M., Han, M. & Yu, W. W. Hydrogenation of Chlorobenzene to Cyclohexane over Colloidal Pt Nanocatalysts under Ambient Conditions. Environmental Science & Technology 43, 2519-2524 (2009).

30. Mahmoud, Tabor, C. E., El-Sayed, M. A., Ding, Y. & Wang, Z. L. A New Catalytically Active Colloidal Platinum Nanocatalyst: The Multiarmed Nanostar Single Crystal. Journal of the American Chemical Society 130, 4590-4591 (2008).

31. Park, J. Y., Zhang, Y., Grass, M., Zhang, T. & Somorjai, G. A. Tuning of Catalytic CO Oxidation by Changing Composition of Rhâˆ’Pt Bimetallic Nanoparticles. Nano Letters 8, 673-677 (2008).

32. Liang, H.-P., Jones, T. G. J., Lawrence, N. S., Jiang, L. & Barnard, J. S. Understanding the Role of Nanoparticle Synthesis on Their Underlying Electrocatalytic Activity. The Journal of Physical Chemistry C 112, 4327-4332 (2008).

33. Dan Zhao, B.-Q. X. Platinum covering of gold nanoparticles for utilization enhancement of Pt in electrocatalysts. Physical Chemistry Chemical Physics 8, 5106-5114 (2006).

34. Xiao, C. et al. Shape-Controlled Synthesis of Palladium Nanorods and Their Magnetic Properties. The Journal of Physical Chemistry C 113, 13466-13469 (2009).

35. Englisch, M., Jentys, A. & Lercher, J. A. Structure Sensitivity of the Hydrogenation of Crotonaldehyde over Pt/SiO2and Pt/TiO2. Journal of Catalysis 166, 25-35 (1997).

317

36. Narayanan, R. & El-Sayed, M. A. Shape-Dependent Catalytic Activity of Platinum Nanoparticles in Colloidal Solution. Nano Letters 4, 1343-1348 (2004).

37. Lee, I., Morales, R., Albiter, M. A. & Zaera, F. Synthesis of heterogeneous catalysts with well shaped platinum particles to control reaction selectivity. Proceedings of the National Academy of Sciences 105, 15241-15246 (2008).

38. Chen, H. M. et al. Hollow Platinum Spheres with Nano-Channels: Synthesis and Enhanced Catalysis for Oxygen Reduction. The Journal of Physical Chemistry C 112, 7522-7526 (2008).

39. Tsung, C.-K. et al. Sub-10 nm Platinum Nanocrystals with Size and Shape Control: Catalytic Study for Ethylene and Pyrrole Hydrogenation. Journal of the American Chemical Society 131, 5816-5822 (2009).

40. Dan, Z. & Bo-Qing, X. Enhancement of Pt Utilization in Electrocatalysts by Using Gold Nanoparticles13. Angewandte Chemie International Edition 45, 4955-4959 (2006).

41. Washio, I., Xiong, Y., Yin, Y. & Xia, Y. Reduction by the End Groups of Poly(vinyl pyrrolidone): A New and Versatile Route to the Kinetically Controlled Synthesis of Ag Triangular Nanoplates. Advanced Materials 18, 1745-1749 (2006).

42. Byungkwon, L., Yujie, X. & Younan, X. A Water-Based Synthesis of Octahedral, Decahedral, and Icosahedral Pd Nanocrystals13. Angewandte Chemie International Edition 46, 9279-9282 (2007).

43. Chen, A. et al. Formation Process of Silverâˆ’Polypyrrole Coaxial Nanocables Synthesized by Redox Reaction between AgNO3 and Pyrrole in the Presence of Poly(vinylpyrrolidone). The Journal of Physical Chemistry B 109, 18283-18288 (2005).

44. Alayoglu, S. & Eichhorn, B. Rhâˆ’Pt Bimetallic Catalysts: Synthesis, Characterization, and Catalysis of Coreâˆ’Shell, Alloy, and Monometallic Nanoparticles. Journal of the American Chemical Society 130, 17479-17486 (2008).

45. Alayoglu, S., Nilekar, A. U., Mavrikakis, M. & Eichhorn, B. Ru-Pt core-shell nanoparticles for preferential oxidation of carbon monoxide in hydrogen. Nat Mater 7, 333-338 (2008).

46. Xuan, S., Wang, Y.-X. J., Yu, J. C. & Leung, K. C.-F. Preparation, Characterization, and Catalytic Activity of Core/Shell Fe3O4@Polyaniline@Au Nanocomposites. Langmuir 25, 11835-11843 (2009).

318

47. Liu, Z. et al. Novel Hybrid Electrocatalyst with Enhanced Performance in Alkaline Media: Hollow Au/Pd Core/Shell Nanostructures with a Raspberry Surface. The Journal of Physical Chemistry C 113, 16766-16771 (2009).

48. Mori, K., Hara, T., Mizugaki, T., Ebitani, K. & Kaneda, K. Hydroxyapatite-Supported Palladium Nanoclusters: A Highly Active Heterogeneous Catalyst for Selective Oxidation of Alcohols by Use of Molecular Oxygen. Journal of the American Chemical Society 126, 10657-10666 (2004).

49. Grandjean, D., Benfield, R. E., Nayral, C., Maisonnat, A. & Chaudret, B. EXAFS and XANES Study of a Pure and Pd Doped Novel Sn/SnOx Nanomaterial. The Journal of Physical Chemistry B 108, 8876-8887 (2004).

50. Lu, P., Teranishi, T., Asakura, K., Miyake, M. & Toshima, N. Polymer-Protected Ni/Pd Bimetallic Nano-Clusters: Preparation, Characterization and Catalysis for Hydrogenation of Nitrobenzene. The Journal of Physical Chemistry B 103, 9673-9682 (1999).

51. Bradley, J. S., Via, G. H., Bonneviot, L. & Hill, E. W. Infrared and EXAFS Study of Compositional Effects in Nanoscale Colloidal Palladiumâˆ’Copper Alloys. Chemistry of Materials 8, 1895-1903 (1996).

52. Lu, P., Dong, J. & Toshima, N. Surface-Enhanced Raman Scattering of a Cu/Pd Alloy Colloid Protected by Poly(N-vinyl-2-pyrrolidone). Langmuir 15, 7980-7992 (1999).

53. Guo, L., Huang, Q.-J., Li, X.-Y. & Yang, S. PVP-Coated Iron Nanocrystals:â€‰ Anhydrous Synthesis, Characterization, and Electrocatalysis for Two Species. Langmuir 22, 7867-7872 (2006).

54. Wang, C., Shao, C., Zhang, X. & Liu, Y. SnO2 Nanostructures-TiO2 Nanofibers Heterostructures: Controlled Fabrication and High Photocatalytic Properties. Inorganic Chemistry 48, 7261-7268 (2009).

55. Lin, Y.-S. & Haynes, C. L. Synthesis and Characterization of Biocompatible and Size-Tunable Multifunctional Porous Silica Nanoparticles. Chemistry of Materials 21, 3979-3986 (2009).

56. Yan, X., Liu, H. & Liew, K. Y. Size control of polymer-stabilized ruthenium nanoparticles by polyol reduction. Journal of Materials Chemistry 11, 3387-3391 (2001).

57. Lee, E. P. et al. Electrocatalytic Properties of Pt Nanowires Supported on Pt and W Gauzes. ACS Nano 2, 2167-2173 (2008).

58. Younan, X., Yujie, X., Byungkwon, L. & Sara , E. S. Shape-Controlled Synthesis of Metal Nanocrystals: Simple Chemistry Meets Complex Physics? Angewandte Chemie International Edition 48, 60-103 (2009).

319

59. Susut C., C. G. B., Samjeské G., Osawa M., Tong Y.Y. . An unexpected enhancement in methanol electro-oxidation on an ensemble of Pt(111) nanofacets: a case of nanoscale single crystal ensemble electrocatalysis. Physical Chemistry Chemical Physics 10, 3712 (2008).

60. Narayanan, R. & El-Sayed, M. Some Aspects of Colloidal Nanoparticle Stability, Catalytic Activity, and Recycling Potential. Topics in Catalysis 47, 15-21 (2008).

61. Daniela, F. et al. Colloidally Prepared Pt Nanoparticles for Heterogeneous Gas-Phase Catalysis: Influence of Ligand Shell and Catalyst Loading on CO Oxidation Activity. ChemCatChem 2, 198-205.

62. Michael , E. G. et al. A Reactive Oxide Overlayer on Rhodium Nanoparticles during CO Oxidation and Its Size Dependence Studied by In Situ Ambient-Pressure X-ray Photoelectron Spectroscopy13. Angewandte Chemie 120, 9025-9028 (2008).

63. Bonet, F., Tekail-Elhsissen, K., Vijaya Sarathy, K. Study of interaction of ethylene glycol/PVP phase on noble metal powders prepared by polyol process. Bulletin of Material Science, Indian Academy of Sciences 23, 165-168 (2000).

64. Qiu, L., Liu, F., Zhao, L., Yang, W. & Yao, J. Evidence of a Unique Electron Donorâˆ’Acceptor Property for Platinum Nanoparticles as Studied by XPS. Langmuir 22, 4480-4482 (2006).

65. Borodko, Y. et al. Probing the Interaction of Poly(vinylpyrrolidone) with Platinum Nanocrystals by UV, Raman and FTIR. The Journal of Physical Chemistry B 110, 23052-23059 (2006).

66. Borodko, Y., Humphrey, S. M., Tilley, T. D., Frei, H. & Somorjai, G. A. Charge-Transfer Interaction of Poly(vinylpyrrolidone) with Platinum and Rhodium Nanoparticles. The Journal of Physical Chemistry C 111, 6288-6295 (2007).

67. Zhang, P. & Sham, T. K. Tuning the electronic behavior of Au nanoparticles with capping molecules. Applied Physics Letters 81, 736-738 (2002).

68. Bradley, J. S., Tesche, B., Busser, W., Maase, M. & Reetz, M. T. Surface Spectroscopic Study of the Stabilization Mechanism for Shape-Selectively Synthesized Nanostructured Transition Metal Colloids. Journal of the American Chemical Society 122, 4631-4636 (2000).

69. Ramaker, D. E. & Koningsberger, D. C. Comment on "Effect of Hydrogen Adsorption on the X-Ray Absorption Spectra of Small Pt Clusters". Physical Review Letters 89, 139701 (2002).

70. Teliska, M., O'Grady, W. E. & Ramaker, D. E. Determination of H Adsorption Sites on Pt/C Electrodes in HClO4 from Pt L23 X-ray Absorption Spectroscopy. The Journal of Physical Chemistry B 108, 2333-2344 (2004).

320

71. Teliska, M., O'Grady, W. E. & Ramaker, D. E. Determination of O and OH Adsorption Sites and Coverage in Situ on Pt Electrodes from Pt L23 X-ray Absorption Spectroscopy. The Journal of Physical Chemistry B 109, 8076-8084 (2005).

72. Roth, C. et al. Determination of O[H] and CO Coverage and Adsorption Sites on PtRu Electrodes in an Operating PEM Fuel Cell. Journal of the American Chemical Society 127, 14607-14615 (2005).

73. Scott, F. J., Roth, C. & Ramaker, D. E. Kinetics of CO Poisoning in Simulated Reformate and Effect of Ru Island Morphology on PtRu Fuel Cell Catalysts As Determined by Operando X-ray Absorption Near Edge Spectroscopy. The Journal of Physical Chemistry C 111, 11403-11413 (2007).

74. Arruda, T. M., Shyam, B., Ziegelbauer, J. M., Mukerjee, S. & Ramaker, D. E. Investigation into the Competitive and Site-Specific Nature of Anion Adsorption on Pt Using In Situ X-ray Absorption Spectroscopy. The Journal of Physical Chemistry C 112, 18087-18097 (2008).

75. Scott, F. J., Mukerjee, S. & Ramaker, D. E. Contrast in Metal Ligand Effects on PtnM Electrocatalysts with M Equal Ru vs Mo and Sn As Exhibited by in Situ XANES and EXAFS Measurements in Methanol. The Journal of Physical Chemistry C 114, 442-453 (2009).

76. Shyam, B., Arruda, T. M., Mukerjee, S. & Ramaker, D. E. Effect of RuOxHy Island Size on PtRu Particle Aging in Methanol. The Journal of Physical Chemistry C 113, 19713-19721 (2009).

77. Arruda, T. M. et al. Fundamental Aspects of Spontaneous Cathodic Deposition of Ru onto Pt/C Electrocatalysts and Membranes under Direct Methanol Fuel Cell Operating Conditions: An in Situ X-ray Absorption Spectroscopy and Electron Spin Resonance Study. The Journal of Physical Chemistry C 114, 1028-1040 (2009).

78. Teliska, M., Murthi, V. S., Mukerjee, S. & Ramaker, D. E. Site-Specific vs Specific Adsorption of Anions on Pt and Pt-Based Alloys. The Journal of Physical Chemistry C 111, 9267-9274 (2007).

79. Gatewood, D. S. et al. Characterization of Ligand Effects on Water Activation in Triarylphosphine-Stabilized Pt Nanoparticle Catalysts by X-ray Absorption Spectroscopy. The Journal of Physical Chemistry C 112, 4961-4970 (2008).

80. Ravel, B. & Newville, M. ATHENA, ARTEMIS, HEPHAESTUS: data analysis for X-ray absorption spectroscopy using IFEFFIT. Journal of Synchrotron Radiation 12, 537-541 (2005).

81. Ravel, B. ATOMS: crystallography for the X-ray absorption spectroscopist. Journal of Synchrotron Radiation 8, 314-316 (2001).

321

82. Susut, C. Tuning the Activity of Platinum Electrocatalysts via Size, Shape and Capping Polymer. Ph.D. Thesis, Georgetown University, Washington D.C. (2009).

83. Rehr, J. J. & Albers, R. C. Theoretical approaches to x-ray absorption fine structure. Reviews of Modern Physics 72, 621 (2000).

84. E. Janin, H. v. S., M. Gothelid, U.O. Karlsson, M. Svensson. Phys. Rev. B: Condensed Matter Mater. Phys. 61, 13144 (2000).

85. Anna Korovina, C. R., David E. Ramaker. To be published.

86. Grass, M. E., Joo, S. H., Zhang, Y. & Somorjai, G. A. Colloidally Synthesized Monodisperse Rh Nanoparticles Supported on SBA-15 for Size- and Pretreatment-Dependent Studies of CO Oxidation. The Journal of Physical Chemistry C 113, 8616-8623 (2009).

87. Gong, X.-Q., Liu, Z.-P., Raval, R. & Hu, P. A Systematic Study of CO Oxidation on Metals and Metal Oxides: Density Functional Theory Calculations. Journal of the American Chemical Society 126, 8-9 (2003).

88. Kuhn, J. N., Tsung, C.-K., Huang, W. & Somorjai, G. A. Effect of organic capping layers over monodisperse platinum nanoparticles upon activity for ethylene hydrogenation and carbon monoxide oxidation. Journal of Catalysis 265, 209-215 (2009).

89. Savinova, D. V., Molodkina, E. B., Danilov, A. I. & Polukarov, Y. M. Surface and Subsurface Oxygen on Platinum in a Perchloric Acid Solution. Russian Journal of Electrochemistry 40, 683-687 (2004).

90. Tsunoyama, H., Ichikuni, N., Sakurai, H. & Tsukuda, T. Effect of Electronic Structures of Au Clusters Stabilized by Poly(N-vinyl-2-pyrrolidone) on Aerobic Oxidation Catalysis. Journal of the American Chemical Society 131, 7086-7093 (2009).

91. Navin, J. K., Grass, M. E., Somorjai, G. A. & Marsh, A. L. Characterization of Colloidal Platinum Nanoparticles by MALDI-TOF Mass Spectrometry. Analytical Chemistry 81, 6295-6299 (2009).

92. Dominique Richard, J. W. C., John M. Thomas. Structural and electronic properties of finely-divided supported Pt-group metals and bimetals. Faraday Discussions 92, 109-119 (1991).

93. Busser, G. W., van Ommen, J. G. & Lercher, J. A. Preparation and Characterization of Polymer-Stabilized Rhodium Sols. I. Factors Affecting Particle Size. The Journal of Physical Chemistry B 103, 1651-1659 (1999).

322

94. Tong, Y. Y., Yonezawa, T., Toshima, N. & van der Klink, J. J. 195Pt NMR of Polymer-Protected Pt/Pd Bimetallic Catalysts. The Journal of Physical Chemistry 100, 730-733 (1996).

95. de Caro, D. & Bradley, J. S. Surface Chemistry of Colloidal Platinum: Vibrational Coupling in the Infrared Spectrum of CO Adsorbed on Colloidal Platinum in Liquid Dispersion. Langmuir 14, 245-247 (1998).

96. Dassenoy, F. et al. Size and composition effects in polymer-protected ultrafine bimetallic PtxRu1-x (<x<1) particles. Physical Review B 63, 235407 (2001).

97. Tang, Z., Geng, D. & Lu, G. Size-controlled synthesis of colloidal platinum nanoparticles and their activity for the electrocatalytic oxidation of carbon monoxide. Journal of Colloid and Interface Science 287, 159-166 (2005).

98. Chun-Liu Fang, K. Q., Jianhua Zhu, Shangbin Wang, Xiaoxuan Ly, Shu-Hong Yu. Monodisperse alpha-Fe2O3@SiO2@Au core/shell nanocomposite spheres: synthesis, characterization and properties. Nanotechnology 19, 125601 (2008).

99. You-Jung Song, J.-K. O., Kyung-Won Park. Pt nanostructure electrodes pulse electrodeposited in PVP for electrochemical power sources. Nanotechnology 19, 355602 (2008).

100. Tian, Z. Q., Jiang, S. P., Liu, Z. & Li, L. Polyelectrolyte-stabilized Pt nanoparticles as new electrocatalysts for low temperature fuel cells. Electrochemistry Communications 9, 1613-1618 (2007).

101. Benfield, R. E. J. Chem. Soc. Faraday Trans. 88, 1107 (1992).

102. Sun, Y., Gates, B., Mayers, B. & Xia, Y. Crystalline Silver Nanowires by Soft Solution Processing. Nano Letters 2, 165-168 (2002).

103. Benjamin, W., Yugang, S., Brian, M. & Younan, X. Shape-Controlled Synthesis of Metal Nanostructures: The Case of Silver. Chemistry - A European Journal 11, 454-463 (2005).

104. Jingyi, C., Yujie, X., Yadong, Y. & Younan, X. Pt Nanoparticles Surfactant-Directed Assembled into Colloidal Spheres and used as Substrates in Forming Pt Nanorods and Nanowires13. Small 2, 1340-1343 (2006).

105. Seo, D. et al. One-Dimensional Gold Nanostructures through Directed Anisotropic Overgrowth from Gold Decahedrons. The Journal of Physical Chemistry C 113, 3449-3454 (2009).

106. Cai-Xia Kan, J.-J. Z., Xiao-Guang Zhu. Silver nanostructures with well-controlled shapes: synthesis, characterization and growth mechanisms. J. Phys. D: Appl. Phys. 41, 155304 (2008).

323

107. Yu, Y. & Xu, B. Selective formation of tetrahedral Pt nanocrystals from K2PtCl6/PVP. Chinese Science Bulletin 48, 2589-2593 (2003).

108. Kweskin, S.J., Rioux, R.M., Habas, S.E., Komvopoulos, K., Yang, P. and Somorjai, G.A. Carbon Monoxide Adsorption and Oxidation on Monolayer Films of Cubic Platinum Nanoparticles Investigated by Infrared-Visible Sum Frequency Generation Vibrational Spectroscopy. The Journal of Physical Chemistry B 110, 15920-15925 (2006).

324

Chapter 7

Conclusions

In this thesis, in situ XAS was employed as a probe to study the poisoning and

degradation of platinum (and Pt-based) electrocatalysts under operating conditions. The

spectroscopic data was complemented with electrochemical data from techniques such as

cyclic voltammetry, chronoamperometry, rotating disk electrode and copper

underpotential deposition experiments. While many of the problems studied were of

fundamental interest, they also bear a direct relevance to the field of applied fuel cell

research. Investigating the performance of catalysts by following the behavior of small

molecular adsorbates such as H, OH, CO, O etc. on various classes of catalysts has been

the focus of many previous studies within our group. However, the aim of this thesis was

directed towards obtaining a better understanding of the mechanisms of poisoning and

aging of low-temperature fuel cell electrocatalysts, a major bottleneck that exists in

attempts to commercialize such fuel cells for widespread application.

It must be mentioned that studies on catalyst aging are by no means new. While

325

an extensive literature exists for such studies in gas-phase catalysis, little attention was

given to electrocatalyst degradation until low-temperature fuel cells such as PEMFCs and

DMFCs were poised to enter the market in the early years of the last decade. Since then,

numerous studies have been carried out on electrocatalyst aging, an active field of

research in many fuel cell research groups around the world. Even so, many of these

studies were (and still are) carried out ex situ, commonly employing microscopic and

spectroscopic techniques that require ultra-high vacuum such as IRAS, TEM, XPS, and

LEED, etc. It is now generally accepted that the catalytically active state of a catalyst

may be very different from what is observed in an ex situ characterization of the catalyst.

Further, unless great pains are taken to characterize and study catalysts in situ, they are

almost always oxidized due to the presence of atmospheric oxygen. While this fact has

been appreciated by researchers for decades, it is only over the past 5-10 years that the

development of third-generation synchrotron sources and related advances in technology

have made it possible to study catalysts without having to remove them from their natural

operating conditions, and thus these in situ studies can provide valuable insights into the

nature of catalytic activity. All our research, the work in this thesis being no exception,

has focused on trying to study electrocatalysis either in situ, employing conditions very

closely mimicking that of the actual operating conditions (e.g. a half-cell set up) or in

operando, where the experiments are carried out in a real operating system in its entirety

(e.g. a complete working PEM fuel cell).1 The XAS experiments were all carried out at

the National Synchrotron Light Source (NSLS), one of the many large-scale research

facilities located at Brookhaven National Lab, Long Island, NY. and is supported by the

U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences

326

Nanoparticles, as opposed to bulk materials of the same element, are a very

interesting class of materials as they display a rich diversity in their physical, chemical,

optical, electronic, magnetic and catalytic properties. While we are still quite far from

completely exploring and understanding the basis of this diverse behavior, it is their

catalytic properties that will prove particularly valuable in helping us transform the way

we produce the energy required to fuel a sustainable future. As a field of basic and

applied research, few other areas have the potential to immediately impact our standard

of living like research in hetereogeneous catalysis. Such catalytic processes are employed

to produce virtually all products used in daily life and are of immense importance.

Heterogeneous catalysis is a surface phenomenon i.e., occurs primarily at interfaces. As

such, the chief value of nanoparticle catalysts becomes immediately apparent:

nanoparticles possess extremely high specific surface areas and as one approaches the 1-3

nm regime, nearly all their constituent atoms are at the particle surface. Furthermore,

because of the relatively small number of atoms in a given cluster, adsorption/desorption

events can significantly affect the overall morphology of the nanoparticle during

catalysis, a phenomenon that has only been observed very recently. 2 This is a classic

example of a finding that could only have been discovered through the use of in situ

techniques. In the case of alloy catalysts, these changes become significant as it can lead

to phase segregation, wherein the more reactive elements preferentially segregate to the

surface of the nanoparticles, leading to a completely different surface morphology, and

eventually resulting in an altered catalytic activity from that originally exhibited by the

catalysts. 3 Besides the direct relevance to applied fuel cell research, an intention of the

327

work described in this thesis was also to discover some fascinating new aspects of

electrocatalysis.

Chapter 2 of the thesis described in great detail the various reasons that in situ XAS is

ideally suited to study electrocatalysts, as it is one of the few spectroscopic techniques

that is able to provide both structural details as well as information on the oxidation state

of elements in the material under investigation. Equally important is the fact that the

technique is element specific, making possible detailed structural and morphological

studies of alloyed materials. Further, many nanomaterials used in catalysis have x-ray

absorption edge energies in the 5-30 keV energy range. This is a tremendous advantage

because there is little attenuation of the beam light by the surrounding electrolyte, making

possible in situ spectroscopic studies in an electrochemical environment. The ∆µ-XANES

method nicely adds a valuable component to XAS – that of surface-sensitivity. While

there are other vibrational techniques to study adsorbed CO and other organic species on

surfaces such as Sum-Frequency Generation (SFG), Subtractively Normalized Infra-Red

Absorption Spectroscopy (SNIFTRS) and Surface-Enhanced Infra-Red Absorption

Spectroscopy (SEIRAS), they are all severely attenuated by the presence of electrolyte

around the electrode, giving rise to very noisy data in many instances. Further, they may

also require specially-prepared catalyst surfaces in order to obtain high-quality data. XAS

appears to have none of these constraints and therefore becomes a very versatile and

robust technique to study the solid-liquid interface. The problems studied as part of this

thesis utilized the ∆µ-XANES method along with conventional EXAFS analysis and

helped provide a very detailed molecular-level picture of the poisoning and aging

328

processes on Pt-based catalysts, which are widely used in low-temperature fuel cells.

What follows is a summary of the major findings from the studies described in this thesis.

7.1 Poisoning of Pt/C nanoparticles

Low-temperature fuel cells contain electrocatalysts that are chiefly Pt-based: the

anode most often consist of alloyed nanoparticles of platinum and ruthenium, while the

cathode is still mostly pure platinum nanoparticles because it provides the best

performance for the oxygen reduction reaction (ORR). However, polarization of the

cathode is more serious than that of the anode in H2, and this is attributed to the sluggish

ORR kinetics. In other words, a large fraction of the overall polarization loss leading to

diminished operating cell potentials is due to the slow reduction reaction on the cathode.

This problem is exacerbated in the presence of strongly adsorbing anions such as

bisulfate, chloride and sulfide ions, as they inhibit the already slow reaction by blocking

catalytically active surface sites. These anions are either present as electrolytic species or

find their way into the fuel as a by-product of catalyst synthesis or as impurities in the

fuel feed stream, and can result in a potential loss of a few hundred millivolts, drastically

affecting the power output of the fuel cell.

7.1.1 Chloride poisoning

The study carried out and reported in chapter 3 describes an investigation into the

competitive and site-specific nature of chloride poisoning on carbon-supported Pt

catalysts (Pt/C) in HClO4 using in situ XAS and electrochemical techniques such as

cyclic voltammetry (CV) and rotating disk electrode (RDE) experiments. Through the use

329

of the Δμ-XANES method, we were able to provide the first conclusive spectroscopic

evidence for site-specific adsorption of Cl- ions on 3-fold sites on the (111) planes of Pt

nanoparticles. It was also found that atop chloride exists in a small potential window (0. 4

– 0.7 V vs. SHE) when compressed adlayers are formed as a result of increasing chloride

coverage on the surface. On examining the effect of chloride as a poison for cathode

catalysts, RDE experiments establish that adsorbed chloride drastically affects the

catalysis by blocking active surface sites and increases the overpotential for the oxygen

reduction reaction by approximately 85 mV for every 10-fold increase in chloride

concentration (in electrolyte). EXAFS analysis on data collected at several potentials,

interestingly enough, reveals that the Pt-Pt coordination decreases with an increase in

choride ion concentration. Chloride ions adsorbed in 3-fold sites were found to be largely

responsible for this decrease in Pt-Pt coordination, while those adsorbed in atop sites

increased the Pt-Pt coordination number. Such a dependence of substrate metal

coordination on adsorbed species (previously seen in case of OH and O species, also on

Pt/C catalysts) is attributed to minor shape-changes that occur as a result of adsorption of

these species: ions adsorbed in atop sites tend to make the clusters more spherical,

effectively increasing the Pt-Pt coordination, whereas ions adsorbed in 3-fold sites more

directly decrease the Pt-Pt scattering and hence decrease the apparent Pt-Pt coordination

number. In light of these findings, a reinterpretation of existing electron quartz crystal

nanobalance (EQCN) studies on halide adsorption in H2SO4 suggests that while bisulfate

ions, once adsorbed on the Pt surface at lower potentials cannot be replaced by chloride

ions, this layer is disturbed at higher potentials by oxygen adsorption, after which

chloride adsorbs very strongly, and these cannot be displaced by the bisulfate ions present

330

in the electrolyte. Bromide ions, on the other hand, were able to displace any bisulfate on

the Pt surface at both lower and higher potentials, confirming previous studies in the

literature on the relative adsorption strengths of the two halides on platinum surfaces.

These relative anion adsorption preferences revealed in this study help explain the

dependence of the important ORR on anion adsorption, and also suggests that the effect

of chloride poisoning may be quite dependent on the particle size.

7.1.2 Ru dissolution and poisoning

While PtRu black catalysts are still the state-of-the-art material for many low-

temperature fuel cell anodes, Piela and co-workers at Los Alamos National Lab were the

first to show that Ru dissolution from the anode could lead to the transport of Run+ ions

through the membrane and detrimental deposition on the cathode, a process termed

ruthenium crossover, and thus highlighted a major limitation of such alloy catalysts for

use in PEM fuel cell anodes. 4 Our study on the spontaneous deposition of Ru onto Pt

cathodes under DMFC operating conditions (see Chapter 4) not only corroborate their

findings, but also builds on it by exploring the ramifications of such a crossover of

dissolved Ru ions. We showed that even millimolar amounts of Run+ present in the

electrolyte around the cathode are sufficient to result in a spontaneous deposition onto a

Pt/C cathode at potentials commonly encountered during the oxygen reduction reaction.

The deposited ruthenium species were shown to severely affect the ORR activity of the

cathode. Through the use of in situ XAS, and specifically the ∆µ-XANES method, it was

found that ruthenium ions deposit onto a Pt surface much more readily at OCP (ca. 0.95

V vs. SHE) than under potential control. The site-specificity of the Δμ-XANES method

331

also allowed us to determine the binding site of Ru on the Pt surface, wherein Ru was

found to bind chiefly in 3-fold sites, and onto atop sites only within a very small potential

window. To the best of our knowledge, this was the first XAS study of the spontaneous

deposition of Ru on Pt/C nanoparticle catalysts. Finally, conclusive evidence through

ESR studies for Run+ deposited in the polymer electrolyte membrane and their effects on

membrane hydration levels and pore micro-viscosity were also reported and discussed.

Taken together, these findings indicate the severe implications of Ru poisoning on the

transport properties of PEM fuel cell membranes and cathodes. The study also highlights

the critical nature of keeping the fuel cell under a constant load in order to minimize the

effects of Ru poisoning on fuel cell cathodes, and suggests a method by which prolonged

fuel cell operating lifetimes may be obtained.

7.2 PtRu electrocatalyst degradation through morphological changes

The two primary forms of catalyst degradation (essentially loss in catalytically active

surface area) are: i) poisoning of nanoparticle surface sites and ii) nanopartaicle

morphological changes such as metal atom dissolution, preferential segregation of

alloyed metal atoms, particle growth via coalescence or aggregation, Ostwald ripening

through metal atom dissolution and redeposition etc. The former degradation mechanism

(namely poisoning) was investigated using certain ions (Cl- and Run+), which are

commonly encountered during fuel cell operation, and were described in Chapters 3 and

5. In Chapter 4, the second degradation mechanism, namely the aging and degradation of

PtRu anode electrocatalysts due to metal atom dissolution, particle growth and other

morphological changes, were studied using in situ XAS and appropriate electrochemical

332

methods. Specifically, the aging properties of two commercial PtRu black catalysts,

Johnson Matthey HiSpec6000 and Tanaka TEC90110, were studied in TFMSA with

methanol as fuel to adequately represent the electrochemical environment around such

PtRu catalysts employed in a typical DMFC. The samples were aged both, through

voltammetric potential cycling between 0.02 and 0.80 V, as well as an 8-hour

chronoamperometry test at 0.50 V. A detailed electrochemical investigation coupled with

an analysis of the in situ XAS data through EXAFS analysis as well as the ∆µ-XANES

method revealed why the two catalysts age differently – the two catalysts have

considerable morphological differences. The Tanaka sample had much more Ru

segregated to the surface and was present as heavily oxidized islands (RuOxHy) while the

Johnson Matthey catalysts contained smaller, more metallic Ru islands on their surface.

The ∆µ-XANES curves show that the smaller, metallic Ru islands, which facilitate CO

oxidation on the Pt largely via the bifunctional mechanism, were found to be more

susceptible to dissolution than the larger RuOxHy islands present in the Tanaka catalyst,

and which are known to exert a stronger ligand effect on the surface Pt atoms. It was also

found that the smaller Ru particles present in the JM catalyst sample grew faster than

their more oxidized counterparts present on the Tanaka catalyst. Thus, the changes in

catalytic activity, as revealed by the electrochemistry, were correlated with observable

changes in catalyst morphology, as revealed by the EXAFS and ∆µ-XANES data, over

the potential cycling window number (40-60). Other than a study by Holstein and

Rosenfeld 5, we are unaware of any other detailed investigation into the aging process on

fuel cell catalysts as studied using in situ XAS, especially as the data were collected at

both Pt L3 and Ru K edges as a function of potential and number of cycles.

333

Efforts to mitigate the degradation of such alloy nanoparticles through the addition of

more noble metals such as Au have already been undertaken. 6, 7 The idea is that Au,

being nobler, is less susceptible to oxidation at lower potentials and may actually

passivate the Ru (or Pt) against oxidation via a ligand effect. In a similar study, albeit a

more detailed one, Au-stabilized PtRu nanoparticles have been prepared using the

microemulsion technique by Dr.Mukerjee’s group at Northeastern University. The initial

findings suggest that not only do these catalysts exhibit an increased stability in the

surface morphology (as demonstrated through Cu upd measurements), they also seem to

have a marginally improved catalytic activity. In situ XAS data has also been collected

on these catalysts recently and we are in the process of understanding the exact cause of

this enhanced stability and activity. While this approach may appear promising, it is

hoped that alternative catalysts with less expensive elements that are innately more stable

and active will be synthesized, in order to avoid these expensive and rare noble metal

catalysts altogether.

7.3 Probing the interaction between a stabilizing agent (PVP) and Pt/C

electrocatalysts

The size and morphology of nanomaterials can be controlled through the use of

specific synthetic methods. Organic, polymeric stabilizing agents commonly called

capping agents have been used to control the size distribution of nanoparticles and

synthesize specifically shaped-nanoparticles. Polyvinyl Pyrrolidone, or PVP, is one such

widely used capping agent. These organic polymers are also used to prevent coalescence

and aggregation in nanoparticles, a practice first used to stabilize colloidal particles,

334

effectively slowing down the aging process that commonly occurs during catalysis.

However, in order to do so, they have to bind to the surface of the materials, blocking

active catalytic sites in the process. This raises concerns for their use in preventing aging

as it would result in a reduced catalytic activity of the nanoparticles. We are then left with

trading one problem for another. Research carried out in collaboration with the Tong

research group at Georgetown over the last couple of years has showed that PVP-capped

Pt nanoparticles display a remarkable enhancement in their methanol oxidation activity

over plain Pt/C nanoparticles. This finding presents us with an instance of capping agents

that can not only stabilize nanoparticles against aging, but also play a role in enhancing

their catalytic activity for certain reactions.

An effort towards understanding the precise reason for the enhancement of the

methanol oxidation activity was undertaken in the study described in chapter 6 of this

thesis. The effects of PVP on the electrocatalytic activity of supported platinum

nanoparticles was studied using in situ x-ray absorption spectroscopy (XAS) at the Pt LIII

edge using both EXAFS and the Δμ-XANES analysis techniques. Water activation in

0.1M HClO4 was carried out on 40 wt. % Pt/C with and without PVP using the XAS.

The experimental ∆μ-XANES analysis revealed a marked increase in an atop O(H)-like

lineshape between 0.60-1.0 V (vs. RHE) for the PVP-capped Pt nanoparticles, when

compared to that seen on the plain Pt/C nanoparticles. We attribute this increase to a

change from a more neutral PVP-Pt interaction (PVPN) to a stronger interaction involving

charge transfer (PVPCT) from the PVP to Pt, consistent with that suggested previously in

the literature. Theoretical FEFF 8.0 calculations modeling the PVPCT/Pt suggest that the

PVP bonds to platinum in atop sites, while the PVPN are apparently not in registry or

335

ordered on the surface since they are much less visible in the Δμ. However, the EXAFS

data show that PVPN strongly affects the morphology of the Pt clusters, much like H at

low potentials, and the PVPCT less so at higher potentials. Cyclic voltammetry data

suggest that the PVPN preferentially blocks sites on the (100) faces for H adsorption and

the PVPCT strongly enhances methanol oxidation on the PVP-capped Pt/C nanoparticles

in 0.5M H2SO4 and 0.1M HClO4, confirming earlier findings for such an enhancement on

Pt black. The PVP strongly enhances methanol oxidation apparently by reducing the Pt-

CO bond strength and therefore allowing H replacement at least on the (111) planes, even

though some site blocking by the PVP occurs on the (100) sites.

While the interaction between PVP and Pt has been studied earlier, we are unaware of

any such study in an aqueous environment and hope that the work described in this

chapter of the thesis leads to more such in situ studies, in order that we may eventually

obtain a comprehensive understanding of the effect of organic capping agents on metal

nanoparticles.

7.4 Some limitations of XAS (experiment and theory)

One drawback of having to use synchrotron radiation (as a general user of a

synchrotron facility) for research is that not all planned experiments can be executed due

to various constraints such as ease of access to the facility, availability of beamtime and

the actual allocated time for your experiments, etc. Further, the status of operation of the

synchrotron beam may fluctuate due to technical issues while one is on-site and running

experiments. Often, one is left with having to run only the most critically important

336

experiments to answer the questions at hand, making thorough, detailed investigations

very hard and rare. Some limitations of XAS as a spectroscopic technique to study

catalysts include the necessity for complex EXAFS data analysis routines, and

considerable expertise with their use before one can comfortably fit and analyze EXAFS

data. Coordination numbers obtained from EXAFS analysis are generally accurate to

around +/- 10-15 %, and may be insufficient for observing very small perturbations in

morphology. However, it must be mentioned that carefully prepared samples and high

quality data may increase the accuracy to around 5 %. Finally, it must be pointed out that

changes in coordination number as a function of say, electrode potential, as was carried

out in many of our in situ XAS studies, are much more accurate (on a relative number

basis) and hence have revealed many interesting details about the dynamic changes

occurring in the particle morphology during electrocatalysis.

Just as any other analytical method or technique, the Δμ-XANES method of studying

adsorbates on nanoparticle surfaces also has its own limitations. A prerequisite is that

high surface areas (preferably over 50 % dispersion) are required for obtaining data with

sufficiently high signal/noise ratio. This would preclude studying for example, single

crystal surfaces using this method, although many interesting insights may have been

obtained by such studies. Since any XAS spectrum is essentially an averaged signal over

all atoms sampled by the x-ray beam, the Δμ-XANES signatures reflect binding site

information only from the dominant binding site. Similar to the common prodecure of

fitting XANES spectra of multivalent samples with standards of known valence states, it

might be possible to merge more than one theoretical Δμ-XANES spectra to obtain the

best fit to experimental data. One would encounter such a situation when larger

337

adsorbates (e.g. organic molecules such as benzene, pyridine etc.) are studied. Because of

their size and geometry, and given that they can possibly adsorb in different energetically

favorable positions, their different adsorption geometries would lead to the adsorbing

atoms of one molecule occupying various surface sites. In fact, this approach has already

been tried in a previous study on the catalytic activity of the Laccase enzyme, a multi-

copper oxidase isolated from the fungi trametes versicolor and appears to be promising. 8

However, this approach has not yet been used in any of our electrocatalytic studies on

fuel cell catalysts. Perhaps the greatest strength of the Δμ-XANES method is its ability to

provide coverage and site binding information on small molecular adsorbates such as H,

OH, O and CO. Owing to their relevance and simplicity (from a molecular standpoint),

they still are among the most widely studied species, and even if the Δμ-XANES method

were limited to primarily observing these species in situ on the surface of catalysts, it

would still remain a very valuable method to probe these molecules in the context of

catalytic reactions.

There are some current limitations in the FEFF 8.0 version of the code used to

generate theoretical x-ray absorption spectra. Some knowledge of the active site is

required as a model structure has to be input into the program in order to generate a

theoretical absorption spectrum. Further, only 7 unique potentials may be assigned to any

cluster used for the calculations. This implies that many of the atoms far removed from

the central absorbed atom may have to be multiply assigned to the same potential. While

this may not be a severe limitation in many cases, it would be beneficial to be able to

assign different, more appropriate potentials to calculate the theoretical absorption

spectrum for larger clusters. In some biological catalysts, it is believed that there may be

338

multiple sites for the absorption of certain reactant molecules. In the FEFF 8.0 code, only

one atom can be assigned as the ‘absorber’ atom. Thus, in order to accurately calculate a

theoretical absorption spectrum using FEFF in such a situation, many separate

simulations would have to be run by changing the location of the absorber atom. A

merged spectrum of these separate calculations is bound to represent the experimental

absorption spectra more accurately than those obtained with a single calculation, using a

single-absorber atom only.

However, many of the discussed limitations are offset by the numerous advantages of

the technique. Over the last decade, in situ XAS has grown from a largely novel and

exploratory technique to a mature one. We demonstrate that in situ XAS, with support

from various electrochemical techniques, is capable of probing the many aspects of

poisoning and aging seen in nanoparticle catalysts. Given the utility and versatility of

XAS, there is little doubt that the number of researchers using XAS to study the various

aspects of electrocatalytic reactions will only grow in the years to come.

7.5 Looking ahead

There are various questions concerning catalysis on nanoparticles that are yet to be

completely understood for many catalytic reactions such as:

1. What is the nanoparticles exact internal and surface structure? Does it change during

operation? If so, how does it change and what factors affect these changes?

339

2. What is the exact chemical state of the catalysts during operation? How much are

these catalysts perturbed during catalytic activity? Can we take advantage of this

understanding to design catalysts with greater efficiency and selectivity?

Many of the dynamic changes that occur during catalysis can be adequately probed

using time-resolved XAS, a technique that is still being developed in many labs around

the world. As was pointed out earlier, the absorption event itself is of the order of 10-15s

and essentially provides an instantaneous picture of the molecule(s) under study. Time-

resolved XAS could play the role of a stroboscope for reactions in aqueous media and

such studies have the potential to provide an unprecedented insight into the kinetics of

catalytic activity.

A Δμ-XANES analysis carried out for an in situ XAS study of electrocatalysts often

provides us with a qualitative picture of the coverage of various adsorbates as a function

of electrode potential. However, quantitative or semi-quantitative estimates of adsorbate

coverage may also be obtained through such an analysis. Previously published work from

our research group has shown that this is possible. Especially notable is a discussion on

the quantitative estimates obtained for the CO coverage on PtRu catalysts in methanol

using in situ XAS at various potentials. These estimates were made through the use of

certain assumptions that are completely reasonable under conditions of the experiment

carried out. 9 Such an estimate was also derived from the in situ XAS data collected for

the Ru deposition on Pt/C electrocatalysts (see chapter 5), wherein the first XAS-based

quantitative estimate of Ru on a Pt/C nanoparticle catalyst was made, resulting in a fairly

good agreement with other studies in the literature. However, in order for such analyses

340

to be put on a firmer basis, a detailed in situ XAS study would have to carried out on an

electrochemical system in which well-established, electrochemically determined

adsorbate coverages are correlated with those obtained from a Δμ-XANES analysis.

Again, some preliminary work in this direction has already been carried out through an in

situ XAS study of Cu upd layers on Pt/C.10 This system was chosen as underpotential

deposition of certain metals (like copper, lead, tin etc.) on noble metal electrodes (like

Ag, Au, Pt, Rh etc.) have been extensively studied and several estimates of the metal

adsorbate coverage at various potentials exist in the literature.

Over the course of the study described in chapter 6, it was found that interactions

between capping agents and nanoparticles are, in general, poorly understood. We believe

that this is a potentially exciting field of study and is currently under-explored. Further, it

is now possible to prepare nanoparticles of virtually any desired shape using various

organic capping agents. Shaped nanoparticles have the potential to bridge the gap

between single-crystal studies and those on nanoparticles. For instance, catalytic studies

for a given reaction on cubic nanoparticles, which expose primarily the (100) facets

prepared using capping agents, may be compared to the activity of a clean (100) metal

surface. A better understanding of catalytic behavior at the nanoscale should take us

closer towards being able to design catalysts with high activity and selectivity. In our

collaborative work with Dr.Tong’s group at Georgetown University, some steps in this

direction have already been taken, wherein the catalytic activity of cubic and

tetrahedral/octahedral Pt nanoparticles were studied using in situ XAS in order to

understand the catalytic activity of these specifically-shaped nanoparticles towards

methanol and formic acid oxidation. However, due to lack of time, the work had to be left

341

unfinished. It is hoped that such problems will be studied again in greater detail as some

insights may be obtained into the structure-property relationships that are at the basis of

catalytic activity.

7.5 References

1. Christina Roth, N. B., Thorsten Buhrmester, Marian Mazurek, Matthias Loster, Hartmut Fuess, Diederik C. Koningsberger, David E. Ramaker. Determination of O[H] and CO Coverage and Adsorption Sites on PtRu Electrodes in an Operating PEM Fuel Cell. Journal of the American Chemical Society 127, 14607-14615 (2005).

2. Mark a. Newton, C. B.-C., Arturo Martinez-Arias, Marcos Fernandez-Garcia. Dynamic in situ observation of rapid size and shape change of supported Pd nanoparticles during CO/NO cycling. Nature Materials 6, 528-532 (2007).

3. Mayrhofer, K. J. J., Hartl, K., Juhart, V. & Arenz, M. Degradation of Carbon-Supported Pt Bimetallic Nanoparticles by Surface Segregation. Journal of the American Chemical Society 131, 16348-16349 (2009).

4. Piela, P., Eickes, C., Brosha, E., Garzon, F. & Zelenay, P. Ruthenium Crossover in Direct Methanol Fuel Cell with Pt-Ru Black Anode. Journal of The Electrochemical Society 151, A2053-A2059 (2004).


6. Zhang, J., Sasaki, K., Sutter, E. & Adzic, R. R. Stabilization of Platinum Oxygen-Reduction Electrocatalysts Using Gold Clusters. Science 315, 220-222 (2007).

7. Liang, Z. X., Zhao, T. S. & Xu, J. B. Stabilization of the platinum-ruthenium electrocatalyst against the dissolution of ruthenium with the incorporation of gold. Journal of Power Sources 185, 166-170 (2008).

8. Arruda, T. M. X-ray absorption investigations into the stability and activity of fuel cell electrocatalysts. Ph.D. Thesis, Northeastern University, Boston, MA (2009).

9. Scott, F. J., Mukerjee, S. & Ramaker, D. E. CO Coverage/Oxidation Correlated with PtRu Electrocatalyst Particle Morphology in 0.3 M Methanol by In Situ XAS. Journal of The Electrochemical Society 154, A396-A406 (2007).

342

10. Nagappan Ramaswamy, B. S., David Ramaker, Sanjeev Mukerjee. Underpotential deposition of copper on Pt/C in 0.1 M HClO4: an in situ XAS study. manuscript in preparation.

343

Biography

Badri Shyam was born on November 22nd, 1982 in Bangalore, India. He attended

Baldwin Boys’ High School and then completed pre-university at M.E.S. College of Arts

& Sciences. He received his B.E. in Chemical Engineering from R.V. College of

Engineering, Bangalore, in 2005.

While in college, he was actively involved in many student organizations and greatly

enjoyed planning and organizing events. During his senior year, he carried out research

with Prof.Shukla at the Solid State and Structural Chemistry Unit, Indian Institute of

Science (IISc), Bangalore. The project involved the design and fabrication of a direct

borohydride fuel cell for low-power applications.

In fall 2005, deciding to pursue an academic research career in the physical sciences,

he joined Prof.Ramaker’s research group at GW to carry out research in the

electrocatalysis of fuel cells and specifically, to study aging processes occurring in

electrocatalysts using in situ X-ray absorption spectroscopy. As such, since 2006, he has

been a general user of the National Synchrotron Light Source (NSLS), a synchrotron

research facility located at Brookhaven National Lab, Upton N.Y. He also spent one

summer (2008) as a visiting researcher in Dr.Sanjeev Mukerjee’s research group which is

part of the Northeastern University Center for Renewable Energy Technology

(NUCRET). The research was also supported by a Sigma-Xi summer research

fellowship.

He was also part of a graduate student committee which planned and organized a

symposium at the 237th National Meeting of the American Chemical Society (2009) held

in Salt Lake City, Utah. The symposium carried out under the auspices of the Graduate

344

Student Symposium Planning Committee (an ACS program) and titled ‘Naturally Nano’,

featured renowned researchers in the field who are pushing the frontiers of the discipline.

The symposium also hosted a panel discussion on the responsible and sustainable

development of nanotechnology.

Badri was supported as a graduate teaching assistant (GTA) in the department

throughout graduate school and was a lab instructor for the General Chemistry labs and

Organic Chemistry labs. He is a recipient of the 2008 Benjamin Van Evera Memorial

Prize for the most effective graduate teaching assistant in the Chemistry Department and

the 2008-2009 Philip Amsterdam Graduate Teaching Award for Teaching Excellence

from the George Washington University.

345

Publications 1. Observation of PtRu Particle Aging in Methanol with X-ray Absorption Spectroscopy B. Shyam, T. Arruda, J.M. Ziegelbauer, S. Mukerjee and D.E. Ramaker ECS Transactions, 11 (1) 1359-1368 (2007) 2. An Investigation into the Competitive and Site-Specific Nature of Anion Adsorption on Pt Clusters Using In Situ X-ray Absorption Spectroscopy Thomas Arruda, Badri Shyam, Joseph Ziegelbauer, Sanjeev Mukerjee, and David E. Ramaker, Journal of Physical Chemistry C, 2008, 112 (46), 18087-18097 3. Understanding Electrocatalytic Pathways in Low and Medium Temperature Fuel Cells: Synchrotron-based In Situ X-ray Absorption Spectroscopy S. Mukerjee, J. Ziegelbauer, T. Arruda, D.E. Ramaker and B. Shyam Interface, 17(4), 46 (2008) 4. Promoting effect of CeO2 in the electrocatalytic activity of rhodium for ethanol electrooxidation Q.He, S. Mukerjee, B. Shyam, D. Ramaker, S. Parres-Esclapez, M.J. Illan-Gomez and A. Bueno-Lopez, Journal of Power Sources, Volume 193, Issue 2, 2009, 408-415 5. Effect of RuOxHy Island Size on PtRu Particle Aging in Methanol Badri Shyam, Thomas Arruda, Sanjeev Mukerjee and David E. Ramaker Journal of Physical Chemistry C, 2009, 113 (45), 19713-19721 6. Fundamental Aspects of Spontaneous Cathodic Deposition of Ru onto Pt/C Electrocatalysts and Membranes under Direct Methanol Fuel Cell Operating Conditions: An in situ X-Ray Absorption Spectroscopy and Electron Spin Resonance Study Thomas M. Arruda, Badri Shyam, Jamie S. Lawton, Nagappan Ramaswamy, David E. Budil, David E. Ramaker, and Sanjeev Mukerjee Journal of Physical Chemistry C, 2010, 114 (2), 1028–1040 7. Probing the Influence of PVP on Pt/C Nanoparticles in 0.1M HClO4 using in situ X-ray Absorption Spectroscopy Badri Shyam, Ceren Susut, Yu Ye Tong and David E. Ramaker Journal of Phyiscal Chemistry C (to be submitted)

346

Conference Presentations Probing the Influence of Polyvinyl Pyrrolidone on Platinum Electrocatalysts Using In Situ X-Ray Absorption Spectroscopy Badri Shyam, David E. Ramaker, Ceren Susut, and YuYe Tong 216th ECS Meeting, Oct. 4-9, Vienna, Austria 2009 Towards Mediation of Phosphate Anion Poisoning to Anodic Pt/C Catalyst by alloying Pt with Ni in Phosphoric Acid Fuel Cell Sanjeev Mukerjee, Qinggang He, Nagappan Ramaswamy, Badri Shyam, and David Ramaker 215th ECS Meeting, May 24-29, San Francisco 2009 In situ XAS Studies on PVP Stabilized and Specifically Shaped Pt Nanoparticles in 1M HClO4 Badri Shyam, Ceren Susut, Thomas Arruda, Sanjeev Mukerjee, YuYe Tong, and David Ramaker 214th ECS Meeting, Oct. 12-17, Honolulu, Hawaii 2008 The Spontaneous Deposition of Ru Onto Pt/C Electrocataylsts: An In-situ XAS Study Thomas M. Arruda, Badri Shyam, David E. Ramaker, Vivek Murthi, and Sanjeev Mukerjee 213rd ECS Meeting, May 18-22, Phoenix, Arizona 2008

Observation of PtRu Particle Aging in Methanol with X-ray Absorption Spectroscopy Badri Shyam, Thomas Arruda, Joesph Ziegelbauer, Sanjeev Mukerjee, and David Ramaker

In Situ XAS Investigation of Electrocatalysts Surface Poisoning by Halides Thomas Arruda, Joesph Ziegelbauer, Andrea Gulla, Badri Shyam, David Ramaker, and Sanjeev Mukerjee Understanding the Anodic Dissolution of Ru from Select PtRu Electrocatalysts During DMFC Operating Environment Vivek Murthi, Thomas Arruda, Lajos Gancs, Badri Shyam, David Ramaker, and Sanjeev Mukerjee 212nd ECS Meeting, Oct. 7-12, Washington D.C. 2007

Understanding the Poisoning, Aging and Degradation of Low ...

Documents